Notes
Article history
The research reported in this issue of the journal was funded by the HTA programme as project number 14/26/08. The contractual start date was in September 2015. The draft report began editorial review in August 2017 and was accepted for publication in June 2019. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.
Declared competing interests of authors
Rona Moss-Morris has published papers that met the criteria for inclusion in the review, and she was previously an advisor to the NHS Improving Access to Psychological Therapies programme. Peter White does consultancy work for a re-insurance company. He also is a member of the Independent Medical Experts Group, a non-departmental body, which advises the UK Ministry of Defence regarding the Armed Forces Compensation Fund. Peter White was also an unpaid chairperson of One Health between 2002 and 2010. One Health is a not-for-profit company that was set up to promote the British Psychological Society model within medicine. Andrew Booth is a member of the National Institute for Health Research (NIHR) Complex Reviews Advisory Board, the NIHR Health Services and Delivery Research Funding Board and the NIHR Systematic Review Advisory Group.
Permissions
Copyright statement
© Queen’s Printer and Controller of HMSO 2020. This work was produced by Leaviss et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.
2020 Queen’s Printer and Controller of HMSO
Chapter 1 Background
Definition of medically unexplained symptoms
The term ‘medically unexplained symptoms’ (MUS) is used to describe a wide range of persistent bodily complaints for which adequate examination does not reveal sufficient explanatory structural or other specified pathology (reproduced with permission from the Royal College of General Practitioners). 1 Henningsen et al. 2 describe three main types of MUS: pain in different locations, for example headache, back pain, non-cardiac chest pain; functional disturbance of organ systems; and complaints of fatigue or exhaustion. The term MUS may be applied to patients presenting with single symptoms, multiple symptoms or clusters of symptoms that are related to one another and are specific to a certain organ system or medical specialty; for example, chronic fatigue syndrome (CFS), irritable bowel syndrome (IBS) or fibromyalgia. CFS, IBS or fibromyalgia are often referred to as functional somatic syndromes (FSSs). 3 For patients reporting multiple symptoms, these may vary in range and type. MUS may also vary in terms of reported severity (i.e. number/duration of symptoms) and their effects on functional disability or quality of life.
The term MUS is controversial, and debate regarding its use is ongoing. To many patients with symptoms that are not readily explainable by disease, a diagnostic label is important, but the label ‘MUS’ can be regarded as offensive. 4 Creed et al. 5 suggest that the use of the term ‘MUS’ is a barrier to improved care and, presented a review of the challenges associated with terminology in this area. They suggested alternative terms, such as functional or persistent symptoms. ‘MUS’ is a portfolio term covering a wide range of presentations. The term ‘medically unexplained’ does not exclude physical pathology.
The debate surrounding an appropriate alternative to ‘MUS’ is ongoing and the current review does not seek to contribute to this, nor to address ‘causes’ of MUS.
Classification and diagnosis of medically unexplained symptoms
Diagnostic criteria for MUS are varied. Most of the FSSs are diagnosed according to published diagnostic criteria that include specified symptom criteria alongside the exclusion of medical and/or psychiatric conditions that may mimic similar symptoms [e.g. CFS may be diagnosed by the Fukuda Diagnostic Criteria,6 functional gastrointestinal disorders may be diagnosed by the Rome 111 Diagnostic Criteria,7 fibromyalgia may be diagnosed by the American College of Rheumatology (ACR) 2010 Diagnostic Criteria8]. Patients visiting their general practitioner (GP) frequently with unexplained symptoms are not necessarily offered a formal diagnosis. Where diagnosis of MUS is made, this may be either by use of a validated instrument, such as the Patient Health Questionnaire (PHQ) 15,9 Screening for Somatoform Symptoms (SOMS),10 the Brief Symptom Inventory (BSI),11 or by clinical judgement, usually by a GP. Hoedeman et al. 12 describe a continuum of severity for MUS, ranging from short term or incidental to persisting and recurrent. Fink et al. 13 argue that research into the FSSs and related disorders and their treatment is restricted by the lack of a valid and reliable diagnostic classification. There is overlap between the diagnostic categories of functional somatic disorders and, therefore, patients with similar symptoms and clinical presentations may receive different diagnostic labels. Fink et al. 13 have gone on to describe ‘bodily distress syndrome’ as an alternative to MUS.
The presence of MUS is also a key feature of a range of somatoform disorders. These include somatisation disorder, somatoform pain disorder, undifferentiated somatoform disorder and unspecified somatisation disorder. Diagnosis of any of the somatoform disorders is made by clinical structured interview, with patients meeting diagnostic criteria according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV,14 or V15 or the International Classification of Diseases (ICD)-916 or 10. 17 Although the DSM-IV specifically refers to symptoms being medically unexplained, the DSM-V classification no longer has a requirement that symptoms should lack an explanation. Somatic symptom disorder in the DSM-V is characterised by ‘the presence of one or more distressing and disabling somatic symptoms that disrupt daily functioning’. 18 Significant, moderate to severe somatic symptoms are required to be present, accompanied by excessive, illness-related thoughts, feelings or behaviours. Criticism of the DSM-V definition of somatic symptom disorder centres around the removal of the requirement for symptoms to be ‘unexplained’, and the focus on ‘excessive responses’. As the DSM-V classifies mental disorders, it is argued that this extended scope risks mislabelling many people with physical conditions, such as cancer, heart disease, IBS or fibromyalgia, as mentally ill. 19 Frances20 raises concerns that mislabelling a patient with somatic symptom disorder causes harms (e.g. missed diagnosis of underlying medical causes), subjecting patients to stigma, inappropriate drugs, psychotherapy and iatrogenic disease, and that it may also cause patients to be disadvantaged with regard to employment, education and health care. The 2016 Rome IV guidelines suggest removing the term ‘functional’ from gastrointestinal disorders such as IBS and replacing them with terminology relating to ‘brain–gut interaction’. 21,22 Smith and Dwamena23 propose a clinical spectrum of severity for MUS, from normal/mild, featuring few, minor transient symptoms and little accompanying depression/anxiety, to very severe, which includes the somatoform disorders. Other acknowledged somatoform disorders that have their own diagnostic criteria include bodily distress syndrome, bodily distress disorder and complex somatic symptom disorder.
The current review uses a broad definition of MUS, which encompasses all of the above definitions, so that the term MUS will be used to refer to any of the following definitions: (1) the occurrence of physical symptoms in the absence of clear physical pathology, (2) to FSSs, such as CFS, IBS or fibromyalgia, (3) the DSM-IV (and more recently V) somatoform disorders and (4) somatoform disorders that have their own diagnostic criteria (e.g. bodily distress syndrome). The rationale behind this broad definition is that there is clear overlap between these groups and as yet no consensus as to the validity of one syndrome (i.e. MUS) versus many (i.e. the various FSSs). Whether patients are diagnosed with MUS as opposed to a more specific diagnosis can be an artefact of clinician or researcher preference rather a defining feature of the included patients. 3,24 A final point about classification of MUS is that there is preliminary evidence that several single FSSs are in fact themselves composed of multiple different conditions, united only by common symptoms, which may complicate our understanding of whether or not interventions work for MUS.
Clinical guidelines for MUS (Joint Commissioning Panel for Mental Health 201725) encourage a philosophy of care where physical and mental health are integrated. A recognition that MUS are ‘mind–body problems’ is also encouraged. Some authors have suggested that the biopsychosocial model itself is responsible for dissatisfaction and harm in patients with CFS,26 arguing that its application is biased towards the psychological, framing patients as mentally ill. It is argued that this risks distraction from research into the biological aetiology of symptoms and syndromes, for example Ghoshal and Gwee,21 de Vega et al. 27 and Gur and Oktayoglu. 28 Imposing the biopsychosocial model on patients, it is argued, can lead to ‘disputes over diagnosis, rejection of psychiatric diagnosis, as well as doctors being dismissive, sceptical and lacking in knowledge about the condition’. 29 The alternative view is that the biopsychosocial model allows the inclusion and integration of biological, psychological and social factors in understanding and treatment particularly of chronic conditions. 30
Prevalence and costs of medically unexplained symptoms
A range of prevalence rates of MUS have been estimated. Edwards et al. 31 report worldwide prevalence rates of primary care patients presenting with MUS of 25–50%. In the UK, Taylor et al. 32 report a MUS prevalence rate of 18% of consecutive attenders to UK GP practices. It is estimated that this creates an annual cost to the UK NHS in excess of £3.1B. 33,34 Taking into account quality of life and sickness absence, wider costs to the economy were estimated at over £14B. 33 The inappropriate management of MUS may result in patients undergoing invasive and potentially harmful tests and treatments. Some patients with MUS have comorbid depression/anxiety. 35 Health-care utilisation varies between patients with MUS due to the wide variability in symptom experience. Collin et al. 36 used a case–control study of nearly 8000 matched pairs to show that GP consultation rates for patients with CFS were 50% higher in adult cases than in the controls 11–15 years before diagnosis, and 56% higher 6–10 years after diagnosis, with a peak difference of more than twofold higher in the year of diagnosis. Similarly, a study of health-care resource use for patients with fibromyalgia37 found that patients had considerably higher resource use at least 10 years prior to their diagnosis of fibromyalgia. At the time of diagnosis, patients recorded an average of 25 visits per year compared with 12 visits for the matched controls. For IBS, health-care visits are considerably lower, with one study estimating one extra visit to primary care per year compared with controls. 38
A systematic review of the course and prognosis of MUS and somatoform disorders39 suggested that the prognosis for patients with MUS is influenced by the severity of the condition at baseline and by the number of symptoms. Creed et al. 40 showed, in a large epidemiological study, that symptom count predicted later quality of life. It has been estimated that between 50% and 75% of patients with MUS will improve, whereas between 10% and 30% will see their condition deteriorate. 39
Interventions for medically unexplained symptoms
A wide range of interventions has been implemented in the treatment of MUS. Pharmacological interventions (e.g. antidepressants) are sometimes used. Reviews of pharmacological interventions have shown these to produce some improvement in responsive patients in terms of symptom severity and functioning,12,41,42 but significant heterogeneity of efficacy between different FSSs.
Psychological therapies
Several types of psychological therapies have been implicated. Cognitive–behavioural therapy (CBT) for treatment of MUS is based on the model of CBT proposed by Beck43 and is one of the most common interventions used for this group of patients. CBT for MUS focuses on the perpetuating cycle that maintains symptoms, distress and disability. This type of therapy targets the relationship between cognitive, behavioural and physiological responses that are proposed to maintain symptoms. 44 Reattribution for MUS, although no longer commonly delivered, was designed to be delivered by GPs and is based on providing a psychological explanation for somatised mental disorders. Patients are encouraged to reattribute their symptoms and relate them to psychosocial problems. The three stages of therapy are feeling understood, changing the agenda and making the link. 45 Behaviour therapy may be delivered to MUS patients. In these cases, therapy aims to modify behaviours such as increased vigilance in detecting physical symptoms, or reducing inappropriate coping behaviours such as reassurance-seeking or inactivity. 46 Relaxation therapies may be used as treatments for MUS – these include biofeedback,47,48 meditation-based stress reduction49 and qigong. 49 Third-wave CBTs include mindfulness and acceptance and commitment therapy (ACT), which focuses on psychological flexibility, self-regulation of attention and acceptance. 50 Other psychological therapies such psychodynamic therapy have also been adopted for the treatment of MUS. 51
Physical therapies
A further category of interventions for MUS are physical therapies. Such physical therapies include graded exercise therapy (GET), whereby exercise is started gradually and increased over time, and may incorporate the psychological component of graded exposure to exercise alongside a range of aerobic or non-aerobic exercise, such as walking, pool-based exercise or strength training. 52–56 More physiologically based physical activity interventions include aerobic exercises or non-aerobic exercise (e.g. yoga/qigong), which may also be offered to patients with MUS. 57–59 Our review distinguishes between graded and other physical activity interventions, with the former defined as exercise with a defined behavioural model and the latter defined as exercise with a physiological rationale. Physiotherapy-based exercise interventions were considered provided they included an element of active behavioural participation. Manual therapies were not considered for inclusion if they were predominantly passive.
Other therapies
Other therapies that have been adopted for the treatment of MUS include alternative therapies such as hypnotherapy60 or acupuncture. 61 These are usually passive therapies and were not included in the review.
Not all of these treatments are available on the NHS and, therefore, some patients with MUS may pay to access treatments that they perceive to improve their own symptoms, and where they feel they have more time to express their concerns without the pressure of a time-limited GP consultation.
Setting
Interventions for MUS may be delivered in primary care settings, or after referral to secondary care (e.g. to one or more specialists, such as general physicians, immunologists, neurologists, haematologists, or psychiatrists). 62 In primary care, GPs may deliver behavioural modification interventions to MUS patients as part of enhanced care (encompassing techniques including CBT, reattribution or reframing). Alternatively, patients with MUS may receive collaborative care, where for example a psychologist may deliver CBT within the primary care setting. Delivering interventions in a primary care setting may offer additional benefits, for example patients with MUS may refuse referral to services outside the primary care setting. 63
Evidence for the effectiveness of interventions for medically unexplained symptoms
To our knowledge, there are currently no published systematic reviews that specifically evaluate behavioural modification interventions for patients fulfilling the broad definition of MUS patients as outlined above, within a primary care setting. However, a number of reviews have been conducted for specific subgroups of interventions or patients. Reviews of evidence for the effectiveness of interventions for MUS in general are less common than reviews of individual FSS. Reviews of FSSs have shown that, in the case of CFS, CBT and GET can improve symptom severity and functioning following treatment and are acceptable to patients. 64–67 In the case of fibromyalgia, CBT has been shown to improve physical symptoms and functioning,68 as have exercise therapies69,70 and multicomponent therapy. 71 In the case of IBS, psychological therapies have been shown to reduce symptoms as effectively as pharmacological therapies,42 whereas Zijdenbos et al. 72 found psychological interventions to be slightly superior to usual care or waiting list controls. For other conditions, Aggarwal et al. 73 found only weak evidence of effectiveness of psychosocial interventions including CBT and biofeedback for patients with chronic orofacial pain. Champaneria et al. 74 found psychological interventions improved pain scores for patients with chronic pelvic pain compared with no psychological intervention. van Dessel et al. 75 conducted a review of all non-pharmacological interventions for somatoform disorders and MUS but identified only studies of psychological interventions. The authors found that compared with usual care, treatment resulted in less severe symptoms at the end of treatment. The evidence for CBT was similar to other psychotherapies.
Primary care reviews
The majority of studies included in these reviews were conducted within secondary care. Fewer reviews addressed the effects of interventions in a primary care setting. A review of psychological interventions for MUS76 found that short-term psychotherapy demonstrated small effects for the improvement of physical symptoms in patients with medically unexplained physical symptoms (MUPS), with type and mode of therapy and profession of the therapist moderating the results (e.g. inpatient therapy was more effective, as was therapy delivered by mental health professionals). However, GP-delivered interventions were found to be more effective at reducing health-care utilisation. Rosendal et al. 77 reviewed enhanced care delivered by generalists for patients with functional somatic symptoms and concluded that the current evidence does not answer the question of whether or not there is an effect for these types of interventions. Gerger et al. 78 reviewed psychological therapies for MUS and compared the effectiveness of such interventions when delivered by GPs versus psychologists. They found a small effect for psychological therapies at the end of treatment for physical symptoms, physical functioning and psychological functioning. This effect was moderated by the provider, with delivery by a psychologist found to be more effective, but only for physical symptoms. There was no robust evidence for any long-term effects. Metaregression also showed moderating effects for the number of sessions, with more sessions being more effective. There was no moderating effect of severity of symptoms, although exploratory analyses indicated that psychological intervention delivered by a GP was more effective for more patients with more severe symptoms. Garcia-Campayo et al. 79 reported that psychological interventions may be no less effective in primary care than when conducted in secondary care settings. Edwards et al. 31 provide a narrative review of the literature on the treatment of MUS in primary care, which outlines some of the issues related to the delivery of interventions in a primary care setting, for example the importance of the doctor–patient relationship, involving family members in interventions and the importance of cultural considerations. The authors concluded that no single approach would effectively treat all MUS patients in primary care, and that care must be taken to investigate which intervention is appropriate for individual patients. Our qualitative review and realist synthesis will add to these findings.
Definitions of behavioural modification interventions
As evidenced by the existing literature and described above, interventions for MUS are, in general, based around pharmacological, psychological or physical therapeutic models. Our review will focus specifically on interventions that aim to promote behavioural change. Although there are a number of theoretical models of behavioural change, attempting to assign interventions designed for patients with MUS to any of these theoretical frameworks presents difficulties. For example, for psychological therapies, there may be little behaviour modification theory or practice in ‘pure’ cognitive therapies but it has been shown empirically that in practice not many therapists will practice pure cognitive therapy – most will incorporate behavioural elements. 80 Similarly, for physical therapies, if an intervention is based around a model of physical fitness rather than behaviour re-engagement, then it could be argued that this no longer meets the criteria of a behavioural modification intervention. Many physical fitness methods involve predetermined goals based on a patient’s physiology, which are set by the physiologist or sport scientist and may not be considered as ‘therapy’, although they still constitute an intervention. We will therefore adopt a liberal definition of ‘behavioural modification interventions’ as ‘interventions aimed to achieve behavioural change’. Interventions will include ‘named’ behavioural interventions such as CBT, behavioural therapy and GET (GET incorporates principles of systematic desensitisation and behaviour modification with the aim of gradually increasing physical activity, see, for example, Bagnall et al. 62). However, we will also include any intervention that meets the criteria described above. Owing to wide variation in these interventions, we will categorise them by subtype rather than attempt to treat them as one homogeneous intervention type.
Modifying effects
Results of existing reviews suggest that the effectiveness of treatments for MUS may be modified by a number of factors and that treatment may depend on how MUS is defined. There is currently no consensus on whether or not to use a generic intervention protocol, where all patients with MUS receive the same treatment protocol regardless of key presenting symptoms and/or level of disability versus the use of a very specific protocol, developed for patients with a defined functional somatic or DSM syndrome. There is some suggestion from previous reviews that more specific protocols may have larger treatment effects but this has yet to be investigated systematically. 76
Furthermore, the type of control condition used in randomised controlled trials (RCTs) may influence an intervention’s effectiveness. Some studies have shown that patients with IBS respond well to placebo,72 whereas patients with CFS do not respond well. 81 This highlights the importance of recognising differences in the design and conduct of control conditions. Where the control condition is inactive (e.g. waiting list or treatment as usual), good effect sizes for the experimental intervention have been found, whereas trials with active control interventions have shown small effect sizes. 82 Our review will take account of these issues by extracting information from individual studies for a number of potential modifiers, including mode of delivery of the intervention, MUS population (e.g. diagnosed FSSs), multiple MUS, and chronic unexplained pain as described in Chapter 3, Description of the evidence. Potential modifying effects for intervention type will be explored by categorising by broad type of behavioural modification intervention (e.g. CBT, GET, behaviour therapy). Details of all types of controls will be synthesised for all included trials.
Acceptability of primary care interventions for medically unexplained symptoms
Several authors suggest that the relationship between service users and service providers is key to the success of primary care interventions. 83–85 Poor communication between the GP and the patient as well as lack of emotional and practical support are suggested as barriers to effective treatment of MUS. Creating a safe, therapeutic environment, and the importance of offering effective reassurance, are highlighted as important enabling factors for effective treatment of MUS. 84 Therefore, the current review aims to add greater depth to the clinical effectiveness data by retrieving qualitative data relating to potential barriers to and facilitators of effectiveness and conducting realist synthesis of these data. This is of particular importance as a good proportion of these patients hold strong views about the biological nature of their condition and view the suggestion of a more psychosocial approach to treatment as invalidating their symptoms. 86 Understanding ways in which to make behavioural approaches more acceptable may increase their uptake.
Chapter 2 Definition of the decision problem
Decision problem
The assessment addressed the question: what is the clinical effectiveness, cost-effectiveness and acceptability of behavioural modification interventions for MUS in primary care or community-based settings?
Intervention
Interventions that aimed to modify behaviour were sought. These included explicit behavioural interventions such as CBT, behaviour therapy and GET. Where the intervention was not explicitly named as a behavioural modification intervention (i.e. one of the above), a broad definition of behavioural change interventions was adopted. Interventions therefore included but were not exclusive to a range of psychotherapies, for example CBT, behavioural therapy, psychodynamic therapy, mindfulness and reattribution. Interventions also included other physical therapies, such as aerobic exercise and strengthening or stretching exercises. Interventions with multiple components were included where one of the components was considered a behavioural modification technique as defined by the above criteria. Individual and group interventions were noted as separate interventions; however, because of the limited number of studies per intervention type, both group and individual interventions of the same type were considered together for the purposes of the network meta-analyses, and sensitivity analyses conducted where possible. Interventions were also considered where primary care practitioners were trained to communicate a ‘behavioural’ message to patients during their consultations. In these cases, interventions required a stated explicit aim to train GPs to adopt a behavioural or biopsychosocial approach towards consultations with patients with MUS.
Population and relevant subgroups
Studies of populations meeting the criteria for MUS, MUPS, and somatoform disorders were included. Populations with defined FSSs were also included. Diagnostic/inclusion criteria used are discussed in more detail in Chapter 5, Scope the primary literature.
Relevant comparators
Any comparator was considered. Comparators are described in greater detail in Chapter 5, Scope the primary literature.
Overall aims and objectives of assessment
Research aim
This project evaluated the clinical effectiveness, cost-effectiveness and acceptability of behavioural modification interventions for MUS in primary care or community-based settings. The purpose of the project was to provide a comprehensive systematic review of both quantitative and qualitative studies, using rigorous methods for reviewing, evidence synthesis and cost-effectiveness modelling to evaluate the clinical effectiveness and cost-effectiveness of these interventions.
Research objectives
-
To determine the clinical effectiveness of behavioural modification interventions for MUS in primary care and community-based settings, by undertaking a full systematic review of quantitative literature.
-
To evaluate the barriers to and facilitators of effectiveness and acceptability of behavioural modification interventions for MUS from the perspective of both patients and service providers, by undertaking a realist synthesis following a systematic review of the available qualitative research literature.
-
To undertake a meta-analysis of the available evidence on clinical effectiveness, including a network meta-analysis (NMA) to allow simultaneous comparison of all identified intervention types where appropriate.
-
To identify and synthesise evidence on health economic outcomes such as health-care resource use (e.g. GP appointments), and health-related quality-of-life (HRQoL) data from the studies included in the clinical effectiveness review.
-
To provide new evidence on the cost-effectiveness of behavioural modification interventions for MUS conducted in a primary care or community setting, by conducting a systematic review of existing economic analyses and undertaking a de novo model-based evaluation where there is an absence of high-quality published analyses which are directly applicable to our research question.
-
To explain which circumstances influence the effects of behavioural interventions for MUS patients (via realist synthesis).
Patient and public involvement
Patients were involved throughout the review process. Two members of the public with a history of MUS contributed to the writing of the review protocol. They, along with a person with experience of fibromyalgia, went on to be active members of our Expert Advisory Group. The Expert Advisory Group was made up of subject experts, clinicians and our patient and public involvement (PPI) representatives (experts by experience). There were two whole-group meetings: the project team and one of the Expert Advisory Group, held at the School of Health and Related Research (ScHARR) but with an independent chairperson. These were at the beginning of the review to discuss plans and potential issues before the review got started, and at the end of the review to report the results of the review. Between these meetings, the Expert Advisory Group were e-mailed at key stages in the project to receive updates on progress and to be invited to contribute any feedback.
In addition to the two whole-group meetings, a meeting with JL, AS and MB was held in London solely for the PPI representatives. The purpose of this meeting was to allow a more informal discussion of the project, in particular the qualitative and quantitative reviews, with a focus on a patient point of view. The PPI representatives were also given a booklet containing plain English information on the systematic review process.
All of the PPI representatives made substantial and valuable contributions to the project. Providing a patient perspective at each stage of the review enabled the project team to gain a deeper understanding of the issues arising from the literature, and kept the importance of patient perceptions of their symptoms and health-care provision in focus.
One patient with MUS wrote:
Being involved with this review has opened up my understanding of how important it is when one has ‘unexplained symptoms’ to take part in one’s own recovery and health and how difficult it must be for doctors to have patients who look to them as saviours, not to be able to diagnose and then treat. I thoroughly enjoyed having an insight into both doctors’ and patients’ point of view into the frustrating world of MUS! It also gave me hope seeing the differing and varied interventions available. It was encouraging to see that nearly all symptoms under the various headings seemed to respond to CBT. It has been a great pleasure to be involved with this study and review. The team were brilliant in making a very complex subject accessible and interesting to a lay person.
The other patient with MUS wrote:
It has been a fascinating experience being a small part of a very carefully thought out and thorough review. Credit should go to the team that managed to filter through all of the studies and create a model that allowed for some conclusions to be made. While the review didn’t perhaps reach the clear conclusions it aimed for, there were a lot of interesting observations: From a patient-perspective, the fact that there were few significant effects for any GP intervention, i.e. reattribution, GP led CBT, or GP MUS management, is quite worrying. It is useful knowledge that multimodal and CBT interventions have an impact on the majority of MUS. These findings are something that should be embraced and addressed by the NHS (although it’s interesting to see that there is little evidence supporting the impact on long-term health). It seems that patient caution and stigma can still be attached to CBT and similar therapies so I would be interested to find how this problem could be tackled in the future. It’s also a pity that little could be found that would benefit patients with ‘somatisation and generic physical symptoms’. I’ve really enjoyed working on this project and would be happy to contribute to any further studies.
The person with experience of fibromyalgia wrote:
My experience of being part of the stakeholder group: It was an enormous piece of research that was undertaken and I observed that it was done with great care. I always felt that my opinions, written and oral, were taken seriously and followed-up. I am not sure how valuable my contributions were. I do know that I tried my best to look at all the information and paid attention to the details as much as I tried to look at the bigger picture. I do think that having patient representatives helps to keep the research grounded in the real world. I was astounded by some of the outcomes, as they seemed counter to generally held beliefs. This only shows how important it was to collate this evidence. I do hope that the report will help many people with pain and other unexplained symptoms to reach relief that they have not yet achieved.
Chapter 3 Assessment of clinical effectiveness
Methods for reviewing effectiveness
A systematic review of the literature and (network) meta-analysis (where appropriate) was undertaken to evaluate the clinical effectiveness of behavioural modification interventions for MUS, in a primary care or community setting. The review of the clinical evidence was undertaken in accordance with the general principles recommended in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.
Identification of studies
Searches were undertaken to identify relevant studies regarding clinical effectiveness, cost-effectiveness and acceptability (qualitative studies). The search strategies are reported separately for each below. Methods of searching for studies included in the realist synthesis are described in Chapter 7.
Screening and eligibility
A two-stage sifting process for inclusion of studies (title/abstract then full-paper sift) was undertaken. Titles and abstracts were scrutinised by one systematic reviewer (JL) according to the inclusion and exclusion criteria. There was no exclusion on the basis of quality. All studies identified for inclusion using the abstract alone, plus any study in which a decision on inclusion was not possible only from the abstract, were retrieved for more detailed appraisal. Agreement on inclusion at title/abstract sift was checked by a second systematic reviewer (CGC) for 20% of the total electronic search results. Further sifting processes were developed as it became apparent that there were many studies where inclusion/exclusion was unclear. A sifting meeting was held with all subject experts present to discuss general sifting issues and specific individual cases. A common issue was whether or not interventions met the primary care/community-based criteria. To address this issue, operational sifting criteria were developed in order to aid decision-making. Data were extracted regarding where diagnosis, recruitment and referral (to the study) took place and where the intervention took place, and outcomes were assessed. Inclusion was decided based on a combination of these factors and, where there was doubt, judgements were made via discussion among the review team. Inclusion of studies on setting was kept broad. Appendix 2, Table 40, shows the sifting criteria considered for setting for each included study. Another common issue was the nature of the symptoms meeting the ‘unexplained’ criteria. Studies were included if symptoms were explicitly described as ‘medically unexplained’, as were studies that explicitly stated that they included patients with ‘MUS’. Studies of populations with FSSs were included without a need for further reference to medical explanation. Inclusion issues became apparent in studies of patients with chronic pain but no description of whether or not the pain had a known organic cause. To address this issue, a sample of study authors of these papers were contacted to request further information about their populations. None of the studies of those who responded had lack of ‘medical explanation’ as a criterion for inclusion. Therefore, it would be impossible to determine whether the populations contained a mix of patients with chronic pain with known cause, for example arthritis, and patients with pain without known cause. Specifically, an explanation of the cause of pain was not deemed necessary in these studies, nor was it necessary for pain to be a target for the interventions. It was therefore decided to keep this inclusion criterion narrow, and to include only studies of patients with chronic pain that deliberately targeted pain of unknown or ‘unexplained’ origin.
Clinical effectiveness searches
A systematic search strategy was developed in consultation with the review team, to identify systematic reviews and RCTs relating to the defined population. The focus was on identifying studies in primary care or community-based settings; therefore, population terms were combined with terms to define the setting. A combination of free-text terms and thesaurus searching was used. Published methodological search filters to limit the study type (systematic review or RCT) were used where available. No other search limits were applied. Reference sections of included studies were scrutinised for additional potential studies to include, as were reference lists from relevant reviews.
Searches were conducted in the following sources:
-
MEDLINE via OvidSP (1946–20 November 2015)
-
MEDLINE In-Process & Other Non-Indexed Citations & Epub Ahead of Print & MEDLINE® without Revisions via OvidSP (2013–20 November 2015)
-
Cumulative Index to Nursing and Allied Health Literature (CINAHL) via EBSCOhost (1981–3 December 2015)
-
PsycINFO via OvidSP (1967–3 December 2015)
-
EMBASE via Ovid SP (1974–4 December 2015)
-
Cochrane Database of Systematic Reviews (CDSR) via the Cochrane Library (2005–4 December 2015)
-
Database of Abstracts of Reviews of Effects (DARE) via the Cochrane Library (1994–April 2015 – no longer updated, archive only searched 4 December 2015)
-
Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Library (1898–4 December 2015)
-
Health Technology Assessment (HTA) Database via the Cochrane Library (1989–4 December 2015)
-
Science Citation Index via Web of Science (1900–7 December 2015)
-
Social Sciences Citation Index via Web of Science (1956–7 December 2015).
Searches for systematic reviews and RCTs were conducted between 20 November and 7 December 2015. Detailed search strategies are provided in Appendix 1.
Qualitative searches
A systematic search strategy was developed in consultation with the review team to identify qualitative research relating to the defined population. The focus was on identifying studies in primary care or community-based settings; therefore, population terms were combined with terms to define the setting. A combination of free-text terms and thesaurus searching was used. Published methodological search filters to limit to study type (qualitative) were used where available. The qualitative research filter was combined with a geographic filter to identify UK studies only. No other search limits were applied.
Searches were conducted in the following sources:
-
MEDLINE via OvidSP (1946–4 July 2016)
-
MEDLINE In-Process & Other Non-Indexed Citations & Epub Ahead of Print & MEDLINE without Revisions via OvidSP (2013–4 July 2016)
-
EMBASE via Ovid SP (1974–4 July 2016)
-
CINAHL via EBSCOhost (1981–4 July 2016)
-
PsycINFO via OvidSP (1967–4 July 2016)
-
Science Citation Index via Web of Science (1900–4 July 2016)
-
Social Sciences Citation Index via Web of Science (1956–4 July 2016).
Searches for qualitative research were conducted on 4 July 2016. Detailed search strategies are provided in Appendix 1.
Economic searches
A systematic search strategy was developed in consultation with the review team to identify economic evaluations relating to the defined population. The focus was on identifying studies in primary care or community-based settings; therefore, population terms were combined with terms to define the setting. A combination of free-text terms and thesaurus searching was used. Published methodological search filters to limit to study type (economic evaluation) were used where available. No other search limits were applied.
Searches were conducted in the following sources:
-
MEDLINE via OvidSP (1946–15 August 2016)
-
MEDLINE In-Process & Other Non-Indexed Citations & Epub Ahead of Print & MEDLINE without Revisions via OvidSP (2013–15 August 2016)
-
EMBASE via Ovid SP (1974–25 August 2016)
-
CINAHL via EBSCOhost (1981–25 August 2016)
-
PsycINFO via OvidSP (1967–25 August 2016)
-
NHS Economic Evaluation Database (NHS EED) via the Cochrane Library (1968–April 2015 – no longer updated, archive only searched 25 August 2015)
-
Science Citation Index via Web of Science (1900–25 August 2015)
-
Social Sciences Citation Index via Web of Science (1956–25 August 2015).
Searches for economic evaluations were conducted between 15 and 25 August 2016. The search results were imported into EndNote [Clarivate Analytics (formerly Thomson Reuters), Philadelphia, PA, USA] and subsequently filtered to identify UK studies, using terms from line 29 of the EU economies search filter to search the EndNote library for potentially relevant references. Detailed search strategies are provided in Appendix 1.
Clinical effectiveness review
Details of the qualitative and cost-effectiveness review methods are detailed in Chapters 4–6.
Inclusion and exclusion criteria
Study design
Only RCTs were included as these represent the optimal study design for assessing intervention effectiveness. Scoping of the review indicated the availability of a substantial number of published RCTs. No minimum duration of follow-up was applied.
Intervention
A diverse range of interventions that met with our definition of behavioural interventions were identified. Interventions were subsequently ‘grouped’ by type. Definitions of intervention groups were created following review team discussions. These are presented in Table 1. Two reviewers initially grouped each included intervention into these groups. Where there were difficulties or disagreements, subject experts were consulted.
| Intervention group | Description |
|---|---|
| CBT – high intensity | CBT, delivered by a trained clinical specialist, ≥ 6 hours’ contact |
| CBT – low intensity | CBT, either delivered by a trained clinical specialist but < 6 hours’ contact time, or delivered by a non-specialist (may be > 6 contact hours) |
| Other psychotherapy | Any other psychotherapy (e.g. expressive, psychoanalytic) |
| Graded activity or exercise therapy | Exercise with a defined behavioural model |
| Strength, endurance, sport | Exercise with a physiological model (e.g. aerobic, strengthening) |
| Relaxation, stretching, social support/emotional support | Interventions designed to encourage relaxation or stress relief, general MUS-focused support, stretching |
| Guided self-help | Educational support, including information or self-management materials; visual presentations |
| Multimodal | An intervention that incorporates components from more than one category or was conceptualised as ‘multimodal’ by the study authors |
| GP interventions | |
| GP – reattribution | GP trained in reattribution according to Goldberg principles45 or modified reattribution |
| GP – CBT | GP trained in and delivered CBT |
| GP – other psychotherapy | GP trained in and delivered any other type of psychotherapy as described in general ‘other psychotherapy’ category |
| GP – MUS management | GP trained in the management of MUS. Must be focused on management using behavioural/biopsychosocial principles |
| GP – other | GP intervention consisting of multiple components, does not fit with any other category |
| Non-behavioural comparator interventions | |
| Medication | Any medication prescribed specifically as a comparator intervention (i.e. above patients’ usual regimen) |
| Usual care | Care as usual, also incorporates waiting list or treatment as usual |
| Usual care plus | Enhanced usual care or usual care with minor addition (e.g. a leaflet) |
Appendix 2, Table 28, describes the interventions at a study level, with a brief description, with their designated intervention groupings.
Population
Adults aged ≥ 18 years with MUS, MUPS or somatoform disorders were included. Diagnosis of MUS or MUPS could be either by validated instrument (e.g. PHQ-15, SOMS, BSI) or by clinician judgement. Diagnosis was not restricted by duration (except in the case of chronic pain the duration of which should be > 3 months) or severity (e.g. number of symptoms). Patients with single symptoms were included. Populations with FSSs were included (e.g. IBS, CFS, fibromyalgia). For somatoform disorders, diagnosis should have been made by formal clinical interview and should meet DSM-IV or DSM-V, or ICD-9 or ICD-10 criteria. Somatoform disorders included somatisation disorder, somatoform disorders, somatoform pain disorders, persistent physical symptoms, bodily distress syndrome, bodily distress disorder, FSS, medically unexplained syndrome.
Appendix 2, Tables 29–36, shows diagnostic/inclusion criteria used by condition for individual studies.
Comparator
Studies where ‘usual care’ was the comparator were included. Owing to variation in terminology, studies where the comparator is ‘treatment as usual’ or ‘waiting list’ ’were also included as usual care’. A ‘medication’ control group was included for studies where a comparator arm consisted of a specific medication regimen. Trials with a ‘placebo’ control (e.g. which control for time and attention) were also included. As a number of high-quality head-to-head trials of two or more experimental interventions were identified during scoping searches, head-to-head trials were also included where at least one intervention arm met the definitions outlined above.
Outcomes
Appendix 2, Table 37, presents information on primary and secondary outcomes measured in each study, with an indication of the scale used.
Primary outcomes
Patient level: improvement in symptoms, functioning and/or health-related quality of life (HRQoL). Measures of symptom improvement could be through assessment of severity or frequency and must have been assessed using a generic or symptom-specific validated instrument, for example EuroQol-5 Dimensions (EQ-5D)/Short Form questionnaire-36 items (SF-36) for HRQoL, symptom checklist for symptom severity and PHQ-15.
Health-care level: use of health-care resources (e.g. frequency of GP visits, diagnostic outpatient procedures, hospital admissions, emergency department attendances). Costs are reviewed in detail in the cost-effectiveness review Chapter 6.
Secondary outcomes
Emotional distress, including depression and anxiety as diagnosed by a validated instrument [e.g. Beck Depression Inventory (BDI) or Beck Anxiety Inventory (BAI)] or a composite measure, such as the mental health component from the SF-36; satisfaction with care; attrition (persistence and adherence).
Studies were diverse and, because of the differences in populations, types of symptoms measured were varied. Scales used to measure similar constructs (e.g. depression, quality of life) differed between studies. Appendix 2, Table 37, lists all the primary and secondary outcomes measured, with scales used, for all included studies. Outcome data for individual studies were extracted and commonalities were sought. Ten key outcomes were considered to have sufficiently similar data to be included in meta-analyses. These were individual physical symptoms: pain, fatigue and bowel symptoms; somatisation; composite measures of emotional distress; physical functioning; depression; anxiety; impact of symptoms on daily activities (including disability); and generic physical symptoms (e.g. severity of ‘main’ symptom, where no particular symptom is specified). In addition, data were extracted regarding satisfaction/adherence, adverse events and health-care utilisation. These outcomes were considered too heterogeneous to consider meta-analysis and are reported as a narrative synthesis only.
Outcome measurement time points
Studies measured outcomes at a range of time points. As well as variation in follow-up times (e.g. 3 months, 6 months, 1 year), there was variation in the definitions of these time points. As an example, Figure 1 shows three different ways of defining ‘6 months’ follow-up’.
FIGURE 1.
Variation in definitions of follow-up.
Although all of these variations may be described in individual studies as ‘6 months’ follow-up’, 6 months may refer to the time since one-off treatment, to the time since the end of treatment, to the time since the GP received training, or to the time since baseline (pre-treatment). As previous studies have shown that intervention effects can diminish once treatment has ended, time since end of treatment was considered important. Time points used in the meta-analyses were extracted as baseline, end of treatment (i.e. corresponding to duration of treatment), short-term follow-up (time since end of treatment < 6 months) or long-term follow-up (time since end of treatment ≥ 6 months). The longest follow-up time point within these categories was chosen where possible. Where studies did not explicitly report end-of-treatment time, this was calculated by subtracting duration from follow-up since baseline. Weeks were converted to months using a conversion of 1 week = 0.230137 months. Assessment time points as reported in individual studies are reported in the table of basic study design characteristics in Appendix 2, Table 42. Converted or calculated time points are reported for individual studies in Appendix 2, Table 39.
Settings
Studies in primary care or community-based settings were included. To be considered a primary care setting, interventions must have been conducted within a primary care or community-based setting, but they could have been delivered by any health-care discipline within that setting. Interventions could be face to face or delivered at a distance (e.g. via the internet or telephone), and may include computer-assisted interventions. However, to be considered primary care, a degree of involvement with primary health professionals (HPs) was necessary. Therefore studies of e-health, telephone interventions or self-help that were conducted by university research teams with no primary care practitioner involvement were excluded. For interventions delivered by a therapist (e.g. a psychologist or physiotherapist – not by the GP or primary care practice staff), these could have been delivered by the therapist while the patient was still regarded as a ‘primary care patient’, but not once the patient had been referred to secondary or tertiary care. Improving Access to Psychological Therapies (IAPT) interventions were included if delivered in a primary care or community-based setting.
Data extraction strategy
Data extraction was performed by one reviewer into a standardised data extraction form and independently checked for accuracy by a second. The extraction form was designed using the Template for Intervention Description and Replication (TIDieR) Checklist as a guide. Intervention information regarding setting, duration, provider (e.g. qualifications and training) and number of sessions, etc., was extracted. Basic demographic information for participants was also extracted. Discrepancies were resolved by discussion between the two reviewers and, if agreement could not be reached, then a third reviewer was consulted.
Critical appraisal strategy
The quality assessment of included RCTs was performed by one reviewer (JL) using Higgins’ risk-of-bias tool87 for RCTs, and 20% of completed checklists were independently checked for accuracy by a second reviewer (AS). Discrepancies were resolved by discussion between the two reviewers and, if agreement could not be reached, then a third reviewer was consulted.
Methods of data synthesis
Comparative effectiveness was evaluated using a NMA to allow a comprehensive synthesis of all evidence on all relevant interventions. NMA is an extension of pairwise meta-analysis and it can be used to combine direct and indirect evidence about treatment effects across studies to provide an internally consistent set of intervention effects while respecting the randomisation used in individual studies. 88 The NMAs were conducted using a Bayesian Markov chain Monte Carlo approach88 on the following outcomes: pain, fatigue, bowel symptoms, somatisation, generic physical symptoms, physical functioning, impact of illness on daily activities, anxiety, depression and emotional distress. This assumed a random-effects model to allow for heterogeneity in treatment effects across studies. Separate NMAs were performed for the three time points: immediately post treatment, short term (up to 6 months post treatment) and long term (> 6 months post treatment).
Definition of treatment effect (continuous outcome measures)
For each outcome of interest, the individual studies may have used one of several different (continuous) measurement scales (see Appendix 2, Table 38). To allow studies using different measurement scales to be included in a single NMA, standardised mean differences (SMDs) were computed for each study. The use of SMDs stems from the concept that the different reported measures are essentially quantifying the same effect and can be placed on a common scale by dividing the mean difference between the intervention and control arms in each study by the standard deviation (SD). Raw reported data, in the form of mean/median, SD/standard error (SE)/confidence interval (CI)/interquartile range (IQR), were used to calculate the SMD for each study using Hedges’ (corrected) g. 87 The SMD was computed based on absolute values at the end of follow-up rather than mean difference from baseline, as the within-study correlation would be needed for the latter and was not reported. All of the scales were transferred to be consistent across the scales used in the included studies so that a positive SMD indicates beneficial effect of a treatment in the intervention group when compared with the treatment in the control group.
Synthesis population
The synthesis population was defined following the inclusion criteria as all MUS. Condition groupings within MUS have been defined as chronic fatigue, chronic pain single site, chronic pain multiple sites, IBS or MUS/somatoform disorders. All condition groupings were synthesised in a single integrated analysis. Ideally, differential responses within each condition grouping would be explored through metaregression; however, the networks were too sparse to allow this.
Statistical model for the network meta-analysis
Let yik denote the observed SMD of arm k of trial i (i = 1 . . . ns, k = 1 . . . na), with variance Vik. We assume that the treatment effects are normally distributed such that:
The parameters of interest, θik, are modelled using the identity link function:
A random-effects model was assumed, so that the trial-specific treatment effects, δi,1k, are assumed to arise from a common population distribution with mean treatment effect relative to the reference treatment such that:
where dti1tik represents the mean effect of the treatment in arm k of study i (tik) compared with the treatment in arm 1 of study i (ti1) and τ2 represents the between-study variance in treatment effects (heterogeneity), which is assumed to be the same for all treatments.
Parameters were estimated in a Bayesian framework. Where there was sufficient sample data, conventional reference prior distributions were used:
-
between-study SD of treatment effects, τ ∼ U(0,1.1)
-
mean of treatment effects dti1tik∼N(0,1002).
In the case of there being relatively few studies, an informative prior distribution was assumed for the between-study SD. Rhodes et al. 89 proposed a t-distribution for log of the heterogeneity parameter for the SMD scale. The prior proposed by Rhodes et al. 89 still has probabilities of extremely high heterogeneity, which is implausible in the context that we are working on. For example, this prior represents the belief that the heterogeneity will be low, moderate, high or extremely high with probabilities of 22%, 41%, 16% or 20%, respectively. It has about 20% of the odds ratio in one study would be > 50 times greater than in another. Hence, the prior prosed by Ren et al. 90 is used, which is a truncated Turner et al. 91 prior [a log-normal (–2.56, 1.742)]. The truncation is based on the judgement that the odds ratio in one study would not be ≥ 50 times greater than in another. The resulting prior represents the belief that the heterogeneity will be low, moderate, high or extremely high with probabilities of 15%, 66%, 19% or 0%, respectively.
Inconsistency checking was performed by comparing the standard NMA consistency model with an inconsistency model. 92 In the inconsistency model, no consistency is assumed; that is, each of the pairwise comparisons represents an unrelated parameter to be estimated. The deviance information criteria (DIC) for both models are compared, as are the contributions to the deviance for both models, to determine if there is evidence to suggest inconsistency in the network.
All analyses were conducted in the freely available software packages WinBUGS93 (MRC Biostatistics Unit, Cambridge, UK) and R (The R Foundation for Statistical Computing, Vienna, Austria) using the R2Winbugs interface package. 94 Convergence to the target posterior distributions was assessed using the Gelman–Rubin statistic. 95 The chains converged within 18,000 iterations so a burn-in of 18,000 iterations was used. We retained a further 20,000 iterations of the Markov chain to estimate parameters using one chain. The absolute goodness of fit was checked by comparing the total residual deviance with the total number of data points included in an analysis.
Results are presented using the posterior median treatment effects, 95% credible intervals (CrIs) and 95% prediction interval (PrI). The 95% PrI indicates the extent of between-study heterogeneity by illustrating the range of SMDs that might be expected in a future study. The PrI is calculated based on the predictive distribution of the mean treatment effect. In a Bayesian Markov chain Monte Carlo setting, the predictive distribution is obtained by sampling from the distribution of effects N(d,τ2). Probabilities of treatment rankings were computed by counting the proportion of iterations of the Markov chain in which each intervention had each rank. Median treatment rankings and the probabilities of being the best treatment are presented.
Results
Quantity of research available
Characteristics of included studies
The searches identified 8925 citations for RCTs and 2929 citations for reviews. After deduplication, there were 5909 unique citations for RCTs and 2464 for reviews. For the RCTs, 281 full papers were retrieved as being potentially relevant. A total of 220 of these papers were excluded for at least one of the following reasons: pain was acute or subacute; symptoms did not meet the pre-defined review criteria for ‘unexplained’ as described above; pain was mixed explained/unexplained but populations could not be distinguished from one another in the results; setting was not primary care or insufficient primary care involvement; outcomes were not relevant; conference abstracts or dissertations; or not RCTs. Figure 2 shows the PRISMA flow chart. Studies excluded at full-paper stage are presented with reasons in Appendix 4.
FIGURE 2.
The PRISMA flow diagram.
Sixty-two papers provided data from 59 trials. There were a total of 9077 participants across all trials that randomised numbers in each arm. The number of participants in a single trial ranged from 1096 to 524. 97 Owing to the nature of some of the interventions (i.e. where GPs received training to deliver treatment of MUS), some studies were cluster randomised, whereas the rest were randomised at patient level. Basic study characteristics are presented in Appendix 2, Table 42.
Description of the evidence
Study characteristics
Population
Appendix 2, Tables 29–36, shows the population inclusion criteria for individual studies by condition grouping. Condition groupings were MUS/somatoform disorder (including single MUS or mixed MUS), chronic fatigue (including but not exclusive to CFS), chronic pain (single site), chronic pain (multisite, including fibromyalgia) and IBS.
Of 59 studies that met the inclusion criteria, 29 studied either ‘MUS’ or ‘somatoform disorder’. Approximately half of these studies required participants to meet the diagnostic criteria for either somatoform disorder (DSM-III-R, DSM-IV, ICD-9, or ICD-10) or abridged somatisation disorder. 98 The remaining studies included populations of patients with ‘MUS’. Criteria for inclusion were varied, from number of unexplained symptoms within a set time (e.g. two or more within the past 6 months,99 lifetime history of 6–12 unexplained symptoms,100 five or more symptoms meeting the definition of unexplained during past 6 months101) to more general criteria (e.g. ‘multiple unexplained symptoms’,102 ‘symptoms rated by the GP as psychosomatic in origin’,103 ‘GP confirmed medically unexplained nature of symptoms’104 or ‘primary care providers had recognised that emotional status may have been related to their patient’s symptoms’105). The remaining studies set duration of unexplained symptoms as their inclusion criteria (e.g. duration of unexplained complaints of at least 12 months,106 no documented organic disease to explain symptoms of at least 6 months’ duration107 and ≥ 3 months’ physical symptoms not explained by physical pathology108).
One of the 59 studies109 had a population of mixed diagnoses that included any functional disorders, and one further study included participants with a single MUS: medically unexplained vaginal discharge. 110
Twelve of the 59 studies were of participants with chronic fatigue, and 7 of these 12 included populations meeting diagnostic criteria for CFS. Most of these used the US Centers for Disease Control and Prevention (CDC) criteria for CFS,6 but one study used the Oxford criteria. 111 Two of the 12 studies included participants who either met US CDC criteria for CFS or scored ≥ 4 on the Chalder Fatigue Scale. 112,113 The remaining three studies did not include participants with CFS, with two requiring a score of ≥ 4 on the Chalder Fatigue Scale,114,115 and one requiring a score of ≥ 35 on the fatigue subscale of the Dutch Checklist of Individual Strength. 116,117
Six of the 59 studies were of chronic pain at a single site on the body. Four of these were of back pain,118–121 one was of headache122 and one was of neck pain. 123 All required the duration of pain to be ≥ 3 months, apart from Loew et al. ,122 in which the requirement was for ≥ 12 months’ duration.
Seven of the 59 studies were of chronic pain at multiple sites of the body. Four of these studies were of participants with fibromyalgia, and these used the 1990 ACR diagnostic criteria as their inclusion criteria. 124 The remaining three studies were of chronic widespread pain125,126 or mixed chronic multisite pain, for example chronic generalised or regional pain where organic explanation had been ruled out. 127,128
The remaining three studies were of IBS. Inclusion criteria were that patients met the Rome I diagnostic criteria129 or Rome I and II diagnostic criteria. 130 The third IBS study required a diagnosis of functional gastrointestinal symptoms diagnosed as IBS, but participants did not necessarily have to meet Rome criteria. 131
Setting
Fifty-six of the included studies were defined as ‘primary care’, with the remaining three studies defined as ‘community based’. Appendix 2, Table 40, shows that there was considerable heterogeneity in the details of the setting of the studies. Studies varied in the primary care involvement, although all were designed for primary care patients rather than patients already in tertiary care or who self-referred to a university-based study without co-ordination with a primary care practitioner. Variation in setting detail included study designs where:
-
Patients were recruited and treated by their own GP at their own GP practice.
-
Patients were recruited by their GP, but treated by another health-care professional at their own GP practice.
-
Patients were recruited and assessed by their GP, but treated by another health-care professional at an outside facility; for example a gymnasium or park.
-
Patients were recruited and co-ordinated by their GP, but treatment was self-directed (e.g. home-based exercises).
-
Non-UK studies where the organisation of primary care may differ from the UK health-care system (e.g. ‘primary care physiotherapy clinic’). These clinics are described as working in close co-operation with ordinary primary health systems.
Studies where the intervention itself was not delivered within the primary care practice tended to be sport- or exercise-based interventions, or use of self-help materials. Community-based interventions were included only if the study was explicit in its description and aim of the intervention as being community based. Appendix 2, Table 40, shows setting details for individual studies.
Interventions
Intervention arms were coded into one of the 13 pre-defined intervention groupings. Appendix 2, Table 28, reports the detail of the intervention arms for each study, as described by the authors, and the intervention grouping that the arm has been coded into. Control arms that were active rather than passive were coded into one of the intervention groupings, therefore the numbers reported below for each intervention group add up to greater than the number of studies. Passive control arms were coded either as medication or as usual care/usual care plus. There were a total of 127 intervention arms. Of these, 80 were active intervention arms (or were categorised as such by the review team; for example, where an education booklet/presentation was called usual care by the authors, this was categorised as guided self-help and, therefore, an active intervention) and 47 were passive control arms. There was considerable heterogeneity both between and within groupings. Numbers for types of intervention groups are listed below. Appendix 2, Table 41, presents a summary of intervention groupings for each study arm.
Active intervention arms:
-
GP reattribution (including modified), n = 5
-
GP MUS management, n = 6
-
GP-CBT, n = 1
-
GP other psychotherapy, n = 1
-
GP other, n = 1
-
CBT high intensity, n = 8
-
CBT low intensity, n = 7
-
other psychotherapy, n = 11
-
graded activity (GA), n = 7
-
strength/endurance/sport (SES), n = 7
-
relaxation, stretching, social support, emotional support (RSSE), n = 8
-
guided self-help, n = 6
-
multimodal, n = 12.
Passive control arms:
-
medication, n = 3
-
usual care, n = 39
-
usual care plus, n = 5.
The most common active intervention was multimodal therapy. There was wide variation in the nature of the multimodal interventions, with various combinations of the individual interventions represented. These may be specifically defined in the paper as ‘multimodal’ (e.g. Smith et al. 107) or may consist of components from different groups (e.g. van der Roer et al. 120 sport/exercise + education + behavioural programme; McBeth et al. 125 CBT + sport/exercise). Not only did the RSSE group encompass a wide range of different types of intervention, but these types of intervention were also commonly used as active controls (or were classed as active controls by the review team), as were guided self-help interventions, which were often a self-management information/education booklet. There was an element of overlap between some of the guided self-help interventions and low-intensity CBT (CBTLI), with the latter providing more structure and support than the guided self-help interventions. Other psychotherapy was the next most common intervention, with high-intensity CBT (CBTHI) and CBTLI being the third and fourth most common. Graded activity and sport/exercise interventions were equally represented, with seven arms each. Some interventions were conducted in groups and some were conducted individually. Some were conducted face to face whereas others were conducted at a distance, either by telephone or by e-mail. Appendix 2, Table 41, presents these details by arm for each study.
For most interventions, treatment was directed at patients, in a set number of sessions, for a recommended duration. However, some interventions were directed primarily at the GP. GPs would receive training in methods of treating patients with MUS. This training aimed to enable the GP to communicate a behavioural/biopsychosocial approach to unexplained symptoms to their patients. In these cases, a set study period was usually specified, during which time the GP would conduct consultations with patients in their usual distribution of surgeries rather than at a set number of GP/patient sessions. This study period could last up to 2 years. 97
There was also heterogeneity within intervention groups. Appendix 2, Tables 43 and 44, presents details of planned duration of treatment sessions for each arm, and Table 44 presents details of planned duration of the treatment period. Greater detail by intervention arm for actual mean sessions/duration for each arm is reported in the cost analysis in Chapter 6. There was no typical treatment duration, either between or within intervention groups. Treatment sessions ranged from 1 × 10- to 15-minute session with a HP followed by self-management114 to 10 × 90-minute sessions of treatment plus booster sessions. 132 The shortest treatment periods were around 6–8 weeks (e.g. McCleod et al. ,105 LeFort128 – 6 weeks; Macedo et al. ,119 Moss-Morris et al. 130 – 8 weeks), with a mid-range of 12 weeks (e.g. Marques et al. 133 and Ridsdale et al. 112). Longer-term treatment periods ranged up to 1 year (e.g. Kocken et al. 103 and Smith et al. 107).
Differences in intervention provider and the contact time spent with patients are shown in Appendix 2, Table 45. Interventions were delivered by a range of health-care professionals (e.g. GPs, psychologists, practice nurses, physiotherapists, psychiatrists, psychiatric nurses and health educators). There was little consistency in contact time between provider and patient across the studies.
Most usual-care arms were not specific in their descriptions of care received – this probably varies between individual patients in response to their particular needs, but may consist of giving information leaflets/medication. Some controls defined as ‘usual care’ by the study authors were categorised into active intervention groups. For example, an active study intervention ‘self-help’ consisting of a self-management booklet and minimal contact with a HP114 is categorised as guided self-help. The same booklet is used as a part of a ‘usual-care’ control arm in another study115 but, for the purposes of the review, has been categorised as guided self-help.
Outcomes
All primary and secondary outcomes assessed in each study, together with the scales used for assessment, are reported in Appendix 2, Table 37. Owing to differences in populations, outcomes varied between studies. Key outcomes across studies were identified and are presented in Appendix 2, Table 38. Condition-specific symptoms were measured, for example bowel symptoms for IBS studies, fatigue for chronic fatigue studies, and pain for chronic pain studies. Some of these condition-specific symptoms were also recorded for studies of ‘MUS’ patients, most commonly for pain, but never for bowel symptoms. Pain was most frequently measured using a visual analogue scale (VAS) or numerical rating scale (NRS) in studies of chronic pain populations, whereas, for studies of MUS patients, the SF-36 bodily pain subscale was more frequently used. Fatigue was almost always assessed using the Chalder Fatigue Scale. Psychological symptoms, most commonly anxiety and depression, were measured across all condition groups. For depression, the Hospital Anxiety and Depression Scale – Depression (HADS-D), BDI, Hamilton Rating Scale for Depression (HAM-D) and Symptom Checklist-90 – Depression (SCL-90-D) were the most frequently used scales. For anxiety, the Hospital Anxiety and Depression Scale – Anxiety (HADS-A), Hamilton Anxiety Rating Scale (HAM-A) and Symptom Checklist-90 – Anxiety (SCL-90-A) were the most frequently used scales. Composite measures of emotional distress were also reported, most commonly using the SF-36 mental health subscale, but also HADS total scores, or the General Health Questionnaire (GHQ-30). For ‘MUS/somatoform’ studies, ‘somatisation’ was commonly measured to assess symptom load, using a number of scales but most commonly the SOMS or PHQ-15 or Brief Symptom Inventory (BSI). Severity of main symptoms or number of symptoms was also measured (generic physical symptoms) in a minority of studies, with severity VAS or number of symptoms used as methods of assessment. Physical function was measured in studies of all conditions, almost always using the Short Form questionnaire-12 items (SF-12) or SF-36 physical functioning subscale. Finally, illness impact on daily activities was measured across conditions, using disability scales such as the Roland–Morris Disability Questionnaire (RMDQ), the Neck Disability Index or Oswestry Disability Index (ODI), or, for fibromyalgia, the Fibromyalgia Impact Questionnaire (FIQ) total.
Quality of the evidence
Individual risk-of-bias extractions and summary tables can be found in Appendix 3. The quality of individual studies ranged from low to high. Most commonly, studies were found to be at high risk of performance bias, with patients and intervention providers frequently not blinded because of the nature of the interventions and comparators. Over 25% of studies were at high risk of attrition bias, but over 50% of studies were found to be at low risk of attrition bias using the 20% cut-off point. 134 Reporting bias was assessed by reference to study protocols. Study protocols were sought using a number of methods and these are described, with success rates, in a separate paper. 135 Figure 3 shows the summary table for risk-of-bias assessments.
FIGURE 3.
Summary table of risk-of-bias assessments.
Study results
Raw data extracted from individual studies (means, SD/SE by outcome) can be found in Appendix 4. For the following reasons, studies do not contribute data to all outcomes in the NMA:
-
as is seen in Appendix 2, Table 38, key outcomes vary by study
-
as is seen in Appendix 2, Table 39, follow-up time points are not consistent between studies
-
studies could not be included in the NMA where both intervention arms were grouped into the same intervention category; for example, Aiarzeguena et al. 136 where both arms were grouped as GP reattribution (one arm being modified reattribution) or where both arms were GA (one arm being symptom contingent and one arm being time contingent)
-
data could not be included where no variance was given
-
data could not be included when only provided in graphical format
-
studies could not be included where no raw data were provided for non-significant outcomes.
For these reasons, results of the NMA should be interpreted with caution, and considered within the context of the narrative summaries of results from individual studies. Appendix 2, Tables 46–76, presents narrative summaries of results for key outcomes for individual studies.
Narrative overviews for key outcomes are presented below, complementing the summaries in Appendix 2, Tables 46–80. Studies varied in the manner in which they reported their results. Some presented information on significance of effects both within and between groups, whereas others reported only between-group differences. Some studies report significance controlling for baseline variables, or before and after controlling for multiple comparisons. Owing to these differences in reporting, it was considered inappropriate to present ‘counts’ or ‘percentages’ of significant effects for each outcome by time point. However, a summary of main effects by outcomes as reported in individual studies is presented in Appendix 2, Tables 46–80, along with key information regarding the patient population studied and the interventions being compared. Therefore, this indicates whether the trials had active or ‘do nothing’ (usual care/treatment as usual/waiting list) controls.
Physical symptoms
Pain
Twenty-three studies reported pain as an outcome. Thirteen of these were studies of the chronic pain population: six of single-site chronic pain118–123 and seven of chronic pain at multiple sites on the body. 125,127,128,137–140 The remaining 10 studies were in the MUS/somatoform disorder population (see Appendix 2, Table 46). 103,106,136,141–148 A number of intervention types were reported to show an improvement in pain intensity. These include CBTHI, SES, RSSE and multimodal (exercise + education). No other psychotherapy or any GP intervention showed a positive effect. There were no CBTLI studies. Narrative summaries for each individual study for pain results are presented in Appendix 2, Table 46.
Fatigue
Fourteen studies reported fatigue as an outcome. Most studies measured fatigue using the Chalder Fatigue Scale, although the Checklist of Individual Strength (CIS) fatigue severity subscale and the FIQ fatigue subscale were also used in a minority of studies. Twelve studies were undertaken in participants with chronic fatigue111–115,133,149–155 and two studies in a chronic multisite pain population. 125,139 Intervention types that were reported to show a positive improvement in fatigue included CBTLI, GA and RSSE. No GP interventions were found to be effective. There were no studies of CBTHI. Narrative summaries for each individual study for fatigue results are presented in Appendix 2, Table 47.
Bowel symptoms
Three studies reported bowel symptoms as an outcome and these were all in populations with IBS. Two studies used the IBS Symptom Severity Scale129,130 and one used the Clinical Global Impression Scale (CGIS) – Severity of Symptoms. 131 Two studies reported significant improvements in bowel symptoms after behavioural interventions. These were both CBTLI, with one also including medication, but CBT was reported to have a beneficial summary effect over medication alone, although the effect was lost by 12 months. The one study131 that found no significant effect for behavioural interventions compared an RSSE intervention with a multimodal intervention and a third arm of usual care. Narrative summaries for bowel symptoms results are presented in Appendix 2, Table 48.
Somatisation
Twenty-one studies reported somatisation as an outcome. Of these, 19 were of populations of patients in the MUS/somatoform disorder condition group,97,99,102,104–107,116,117,141–144,146–148,156–159 one study was of unexplained vaginal discharge110 and the remaining study was of patients with chronic fatigue. 133 Studies measured somatisation using the Symptom Checklist-90 – Somatisation (SCL-90-S), the BSI-somatic complaints, the PHQ-15 somatic complaints, the SOMS-7 and the Scale for Assessment of Somatic Symptoms (SASS)-somatisation. Intervention types that were reported to show positive improvement in somatic complaints after behavioural modification interventions included both CBTLI and CBTHI, other psychotherapy, multimodal therapy and GP reattribution, although this significant effect was lost after controlling for confounding factors. Narrative summaries for somatisation results are presented in Appendix 2, Table 49.
Generic physical symptoms
Five studies reported this outcome, which represents ‘symptom load’ or severity of unspecified symptoms where not measured by a validated scale (see ‘somatisation’ outcome for these). 96,104,109,156,160 This outcome includes measures such as ‘N unexplained symptoms’ or severity of ‘main’ symptom’ (where the symptom is not specified and not specific to a particular condition, e.g. IBS or fibromyalgia). All studies were of the MUS/somatoform disorder condition group. Intervention types that were reported to show positive improvement in symptom load included CBTLI, CBTHI and multimodal therapy.
Physical functioning and impact
Physical function
Thirty-two studies reported physical function as an outcome (see Table 38 for condition groups and scales used in individual studies). 97,100,106,107,111,114,118,119,125,127,128,131,133,136,139–153,155–158,161 Most of these studies used either the SF-12 or SF-36, either the physical functioning subscale or less commonly the physical component summary only. One study used the FIQ physical functioning subscale. Of the 32 studies, eight were in the chronic fatigue condition group, 16 were in the MUS/somatoform condition group, one was in the IBS group, and seven were in chronic pain populations (two single site and five multisite). A number of different intervention types showed significant positive improvement in physical function. These included RSSE, SES, CBTHI, GP reattribution and multimodal interventions. Narrative summaries for physical function results are presented in Appendix 2, Table 53.
Impact of illness on daily activities
Twenty-two studies reported measures of impact as an outcome (see Table 38 for conditions and scales used in individual studies). 103,112,113,115,118–121,123,127–130,137–141,146,154,159,160 These mostly used disability scales: the RMDQ, ODI, London Handicap Scale, NDI, or the SCL-90 – Global Wellness (SCL-90-G), the FIQ-total, or the Work and Social Adjustment Scale (WASA)162 impact subscale. Four studies were of populations in the chronic fatigue condition group, 11 in the chronic pain condition group (five single site and six multisite), two in IBS populations and five in the MUS/somatoform disorder condition group. A number of different intervention types showed significant improvements in ‘impact’ after behavioural interventions. These were both CBTLI and CBTHI, RSSE, SES and multimodal interventions.
Emotional distress
Anxiety
Thirty studies reported anxiety as an outcome. 102,104–108,110–113,115,127–130,133,137–139,141,142,144,145,148,149,151,154,156–159 A range of measures was used, most commonly the HADS-A and the HAM-A, but also the BAI, SCL-90-A, the BSI-anxiety and the FIQ anxiety subscale. Fifteen studies were in populations in the MUS/somatoform condition group, two were in populations of patients with IBS, eight studies were in populations in the chronic fatigue condition group, and four were in populations with chronic multisite pain. The remaining study was of unexplained vaginal discharge (see Table 38 for conditions and measures used for individual studies). A number of different intervention groups showed positive effects: CBTLI and CBTHI, RSSE, GP MUS management and GP reattribution, although the GP reattribution effect was lost when controlling for confounders. Narrative summaries for anxiety results are presented in Appendix 2, Table 51.
Depression
Depression, along with physical functioning, was the most commonly reported outcome, with 32 studies representing populations in four condition groups: seven for chronic fatigue, six for pain – multisite, one for IBS and 17 for the MUS/somatoform condition group, and one for unexplained vaginal discharge (see Table 38 for conditions and measures used for individual studies). 99,102,105–108,111–113,115,127–130,133,137–145,147,148,151,156–159 A range of measures were used, the most common being HADS-D, HAM-D and BDI, with other studies using CES-D, SCL-90-D and the PHQ-9. Studies reporting positive improvement in depression were found across a number of intervention types, including CBTHI, GA, SES and GP reattribution, although this effect was lost after controlling for confounders. Narrative summaries for depression are presented in Appendix 2, Table 52.
Emotional distress
Thirty studies reported composite measures of emotional distress. 97,100,103,106,107,114,116–119,125,128,129,131,133,136,140–148,150,153,155,157,158,160,161 All condition groups were represented, with five chronic pain studies (two single site, three multisite), four chronic fatigue studies, two IBS studies and 19 MUS/somatoform studies) (see Table 38 for conditions and measures used for individual studies). The majority of these studies used the SF-12 or SF-36 mental health subscale or mental component summary. HADS total, GHQ-30 psychological morbidity and sickness impact profile (SIP) psychological subscales were also used. Positive effects were found across a number of different intervention types: CBTLI and CBTHI, other psychotherapy, RSSE, GA, SES, GP reattribution and multimodal. Narrative summaries for emotional distress results are presented in Appendix 2, Table 50.
Satisfaction, acceptability and adherence
Satisfaction, acceptability and adherence were not measured or reported in a consistent way across studies and, therefore, no attempt was made to conduct a meta-analysis of the data. For GP interventions, ‘attendance’ (by patients to their treatment sessions) or by GPs (to their training) gave an indication of acceptability where no formal measure was taken. GPs were also asked for their confidence in dealing with patients with MUS, or their satisfaction with their training and its relevance to their practice. Patients were also asked about satisfaction with their care. For other interventions, satisfaction was rarely formally measured. Uptake, attendance and attrition data were presented for many studies, giving an indication of acceptability of the treatment offered. Appendix 2, Tables 54–62, presents a narrative synthesis of data relating to these outcomes for individual studies by intervention type. Narrative summaries of results for satisfaction, acceptability or adherence by intervention type are also presented in Appendix 2, Tables 54–62.
Health-care utilisation
The review of health economic literature in Chapter 6 presents data relating to costs associated with health-care utilisation. Owing to sparse reporting of health-care utilisation and heterogeneity of reported data, this is not included as an outcome in the NMA. Appendix 2, Tables 63–71, presents a narrative synthesis of data relating to this outcome, by intervention type. As most studies did not provide these data, only those that did are reported.
Adverse events
Very few studies reported adverse events in detail. It is unclear whether this is because of a lack of events or a lack of recording of events. Where data were reported, the most frequent adverse events were increase in pain with exercise interventions, increase in nausea and dry mouth for pharmaceutical interventions, and an increase in rumination for CBT in one study. One patient was found to have an incorrect diagnosis of CFS. Although other studies did report numbers of reported adverse events, the specific nature of these events was not reported. Appendix 2, Tables 72–76, presents a narrative summary of data relating to this outcome. As the majority of studies did not record/report adverse events, only those that did so are presented in the tables, by intervention type, where available.
Results for network meta-analysis
For all the results presented in this section, a positive SMD indicates a beneficial effect when compared with usual care. Cohen’s categories were used to describe the magnitude of the effect size, with SMD ≥ 0.2 to 0.5 being a small effect size, SMD ≥ 0.5 to 0.8 being a medium effect size, and SMD ≥ 0.8 being a large effect size. 163 SMD < 0.2 was labelled ‘not substantial’ effect size.
Physical symptom outcome measures immediately post treatment
Pain (post treatment)
Data were available from 10 studies presenting the pain score immediately post treatment. A NMA was used to compare the effects of usual care plus, medication, CBTHI, other psychotherapy, RSSE, guided self-help, SES and multimodal relative to usual care on pain score. Figure 4 presents the network of evidence.
FIGURE 4.
Pain score immediately post treatment: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 5 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI (a medium effect size, SMD 0.54, with 95% CrI 0.28 to 0.84) and multimodal (a small effect size, SMD 0.52, with 95% CrI 0.19 to 0.89) were statistically significant at a conventional 5% level. CBTHI was also statistically significant compared with usual care, based on the 95% prediction intervals (which illustrate the range of SMDs that might be expected in a future study) (see Figure 5). The interventions with the highest probabilities of being the best were CBTHI and multimodal (probability 0.31 and 0.30, respectively). The between-study SD was estimated to be 0.19 (95% CrI 0.04 to 0.44), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 5.
Pain score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Fatigue (post treatment)
Data were available from nine studies presenting the fatigue score immediately post treatment. A NMA was used to compare the effects of GP-CBT, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help, GA, SES and multimodal relative to usual care on fatigue score. Figure 6 presents the network of evidence.
FIGURE 6.
Fatigue score immediately post treatment: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 7 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of RSSE (a large effect size, SMD 0.87, with 95% CrI 0.20 to 1.55), CBTLI (a medium effect size, SMD 0.72, with 95% CrI 0.27 to 1.21), multimodal (a medium effect size, SMD 0.52, with 95% CrI 0.14 to 0.92) and GA (a medium effect size, SMD 0.51, with 95% CrI 0.14 to 0.93) were statistically significant at a conventional 5% level. Only the effects of RSSE and CBTLI were statistically significant compared with usual care based on the 95% prediction intervals (which illustrate the range of effect size that might be expected in a future study) (see Figure 7). The intervention with the highest probability of being the best was RSSE (probability 0.60). The between-study SD was estimated to be 0.18 (95% CrI 0.04 to 0.48), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 7.
Fatigue score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Bowel symptoms (post treatment)
Two studies (Kennedy et al. 129 compared multimodal with medication and Moss-Morris et al. 130 compared usual care vs. CBTLI) were available on the bowel symptoms outcome immediately post treatment. No NMA was performed because of the disconnect network. The estimated SMD of multimodal compared with medication was a small effect size, 0.45 (95% CI 0.13 to 0.77), from Kennedy et al. 129 The estimated SMD of CBTLI compared with usual care was a small effect size, 0.44 (95% CI –0.03 to 0.95), from Moss-Morris et al. 130
Somatisation (post treatment)
Data were available from 11 studies presenting the somatisation score immediately post treatment. A NMA was used to compare the effects of usual care plus, GP reattribution, CBTLI, CBTHI, other psychotherapy, RSSE and SES relative to usual care on somatisation score. Figure 8 presents the network of evidence.
FIGURE 8.
Somatisation score immediately post treatment: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 9 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of usual care plus, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, none of the results was statistically significant at a conventional 5% level and the largest beneficial effect was associated with CBTHI with a small effect size (SMD 0.32, with 95% CrI –0.12 to 0.75). None of the results was statistically significant based on the prediction interval (PrI) (see Figure 9). The intervention with the highest probability of being the best was CBTHI (probability 0.39). The between-study SD was estimated to be 0.16 (95% CrI 0.03 to 0.43), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 9.
Somatisation score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Generic physical symptoms (post treatment)
Data were available from two studies presenting the generic physical symptoms score immediately post treatment. A NMA was used to compare the effects of other psychotherapy relative to usual care on generic physical symptoms score. Figure 10 presents the network of evidence.
FIGURE 10.
Generic physical symptoms score immediately post treatment: network of evidence. OP, other psychotherapy; UC, usual care.
Figure 11 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. The estimated SMD of other psychotherapy was less effective than usual care (a small effect size, SMD –0.25, with 95% CrI –0.77 to 0.30). This effect was not statistically significant at a conventional 5% level, and it was also not statistically significant based on the PrI (see Figure 11). The intervention with the highest probability of being the best was usual care (probability 0.83). The between-study SD was estimated to be 0.13 (95% CrI 0.03 to 0.45), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 11.
Generic physical symptoms score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. OP, other psychotherapy; UC, usual care.
Physical functioning and impact outcome measures immediately post treatment
Physical functioning (post treatment)
Data were available from 14 studies presenting the physical functioning score immediately post treatment. A NMA was used to compare the effects of usual care plus, GP reattribution, GP-CBT, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help and GA, SES and multimodal relative to usual care on physical functioning score. Figure 12 presents the network of evidence.
FIGURE 12.
Physical functioning score immediately post treatment: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 13 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of usual care plus, GP-CBT, other psychotherapy and guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of multimodal (a small effect size, SMD 0.33, with 95% CrI 0.09 to 0.59) was statistically significant at a conventional 5% level, but none of the beneficial effects was statistically significant based on the PrI (see Figure 13). Guided self-help was significantly worse than usual care (a medium effect size, SMD –0.73, with 95% CrI –1.18 to –0.29), and this effect was also statistically significant on the PrI (see Figure 13). The intervention with the highest probability of being the best was multimodal (probability 0.35). The between-study SD was estimated to be 0.10 (95% CrI 0.03 to 0.29), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 13.
Physical functioning score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Impact of illness on daily activities (post treatment)
Data were available from nine studies presenting the impact score immediately post treatment. A NMA was used to compare the effects of medication, CBTLI, CBTHI, other psychotherapy, guided self-help, SES and multimodal relative to usual care on impact score. Figure 14 presents the network of evidence.
FIGURE 14.
Impact score immediately post treatment: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Figure 15 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help, SES and multimodal, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI (a large effect size, SMD 1.30, with 95% CrI 0.59 to 2.00) was statistically significant at a conventional 5% level, and it was also statistically significant based on the PrI (see Figure 15). The intervention with the highest probability of being the best was CBTHI (probability 0.72). The between-study SD was estimated to be 0.48 (95% CrI 0.31 to 0.55), which implies high heterogeneity of intervention effects between studies.
FIGURE 15.
Impact score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy.
Emotional distress outcome measures immediately post treatment
Anxiety (post treatment)
Data were available from 14 studies presenting the anxiety score immediately post treatment. NMA was used to compare the effects of usual care plus, medication, GP reattribution, CBTLI, CBTHI, other psychotherapy, guided self-help, GA and multimodal relative to usual care on anxiety score. Figure 16 presents the network of evidence.
FIGURE 16.
Anxiety score immediately post treatment: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 17 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, guided self-help and GA, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI with a medium effect size (SMD 0.52, with 95% CrI 0.06 to 0.96) was statistically significant at a conventional 5% level, but this effect was not statistically significant based on the PrI (see Figure 17). The intervention with the highest probability of being the best was CBTHI (probability 0.48). The between-study SD was estimated to be 0.22 (95% CrI 0.05 to 0.45), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 17.
Anxiety score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Depression (post treatment)
Data were available from 13 studies presenting the depression score immediately post treatment. A NMA was used to compare the effects of usual care plus, medication, GP reattribution, CBTLI, CBTHI, other psychotherapy, GA and multimodal relative to usual care on depression score. Figure 18 presents the network of evidence.
FIGURE 18.
Depression score immediately post treatment: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 19 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI, with a large effect size (SMD 0.80, with 95% CrI 0.26 to 1.38), was statistically significant at a conventional 5% level, but the result was not statistically significant based on the PrI (see Figure 19). The intervention with the highest probability of being the best was CBTHI (probability 0.66). The between-study SD was estimated to be 0.40 (95% CrI 0.24 to 0.54), which implies high heterogeneity of intervention effects between studies.
FIGURE 19.
Depression score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Emotional distress (post treatment)
Data were available from 14 studies presenting the emotional distress score immediately post treatment. A NMA was used to compare the effects of usual care plus, medication, GP reattribution, GP-CBT, GP MUS management, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help, GA, SES and multimodal relative to usual care on emotional distress score. Figure 20 presents the network of evidence.
FIGURE 20.
Emotional distress score immediately post treatment: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 21 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, guided self-help and GA, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of other psychotherapy (a medium effect size, SMD 0.58, with 95% CrI 0.05 to 1.13), RSSE (a medium effect size, SMD 0.66, with 95% CrI 0.18 to 1.28) and SES (a small effect size, SMD 0.49, with 95% CrI 0.03 to 1.01) were statistically significant at a conventional 5% level, but none of the effects was statistically significant based on the PrI (see Figure 21). Guided self-help was significantly worse than usual care (a large effect size, SMD –1.03, with 95% CrI –1.95 to –0.10), but the result was not statistically significant based on the PrI. The interventions with the highest probabilities of being the best were RSSE and CBTLI (probability 0.32 and 0.32, respectively), but the effect of CBTLI was inconclusive. The between-study SD was estimated to be 0.20 (95% CrI 0.04 to 0.49), which implies high heterogeneity of intervention effects between studies.
FIGURE 21.
Emotional distress score immediately post treatment: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Physical symptom outcome measures at short term
Pain (short term)
Data were available from six studies presenting the pain score at short term. A NMA was used to compare the effects of medication, GP MUS management, CBTHI, guided self-help, SES and multimodal relative to usual care on pain score. Figure 22 presents the network of evidence.
FIGURE 22.
Pain score at short term: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 23 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI with a medium effect size (SMD 0.73, with 95% CrI 0.10 to 1.39) was statistically significant at a conventional 5% level, but this effect was not statistically significant based on the PrI (see Figure 23). The intervention with the highest probability of being the best was CBTHI (probability 0.41). The between-study SD was estimated to be 0.45 (95% CrI 0.27 to 0.55), which implies high heterogeneity of intervention effects between studies.
FIGURE 23.
Pain score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; UC, usual care.
Fatigue (short term)
Data were available from seven studies presenting the fatigue score at short term. A NMA was used to compare the effects of usual care plus, CBTLI, CBTHI, other psychotherapy, RSSE, GA, SES and multimodal relative to usual care on fatigue score. Figure 24 presents the network of evidence.
FIGURE 24.
Fatigue score at short term: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 25 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTLI with a medium effect size (SMD 0.62 with 95% CrI 0.11 to 1.14) and RSSE with a medium effect size (SMD 0.51 with 95% CrI 0.06 to 1.00) were statistically significant at a conventional 5% level, but none of the results was statistically significant based on the PrI (see Figure 25). The interventions with the highest probability of being the best were CBTLI (probability 0.34). The between-study SD was estimated to be 0.18 (95% CrI 0.03 to 0.49), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 25.
Fatigue score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; MM, multimodal; OP, other psychotherapy; UC, usual care.
Bowel symptoms (short term)
Two studies (Kennedy et al. 129 compared multimodal vs. medication and Moss-Morris et al. 130 compared usual care vs. CBTLI) were available on the bowel symptoms outcome immediately post treatment. No NMA was performed because of the disconnect network. The estimated SMD of multimodal compared with medication was a small effect size, 0.42 (95% CI 0.10 to 0.74), from Kennedy et al. 129 The estimated SMD of CBTLI compared with usual care was a medium effect size, 0.66 (95% CI 0.16 to 1.16), from Moss-Morris et al. 130
Somatisation (short term)
Data were available from six studies presenting the somatisation score at short term. A NMA was used to compare the effects of GP MUS management, GP reattribution and multimodal relative to usual care on somatisation score. Figure 26 presents the network of evidence.
FIGURE 26.
Somatisation score at short term: network of evidence. GPRE, general practitioner-delivered reattribution; MM, multimodal; UC, usual care.
Figure 27 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. Only the estimated SMD of multimodal was greater than zero, suggesting a beneficial effect compared with usual care. However, this is a not substantial effect size (SMD 0.18, with 95% CrI –0.28 to 0.61) and it was not statistically significant at a conventional 5% level. None of the results was statistically significant based on the PrI (see Figure 27). The intervention with the highest probability of being the best was multimodal (probability 0.74), but the result of treatment effect was inconclusive. The between-study SD was estimated to be 0.15 (95% CrI 0.03 to 0.42), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 27.
Somatisation score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; MM, multimodal; UC, usual care.
Generic physical symptoms (short term)
Only one study,160 which compared multimodal versus GP reattribution, was available on the generic physical symptoms at short term. No NMA was performed because of the disconnect network. The estimated SMD of multimodal compared with GP reattribution was a large effect size, 1.13 (95% CI 0.62 to 1.64), from van der Feltz-Cornelis et al. 160
Physical functioning and impact outcome measures at short term
Physical functioning (short term)
Data were available from 10 studies presenting the physical functioning score at short term. A NMA was used to compare the effects of GP reattribution, GP MUS management, CBTLI, CBTHI, RSSE, guided self-help, SES and multimodal relative to usual care on physical functioning score. Figure 28 presents the network of evidence.
FIGURE 28.
Physical functioning score at short term: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Figure 29 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of multimodal, with a medium effect size (SMD 0.78, with 95% CrI 0.23 to 1.40), was statistically significant at a conventional 5% level, but none of the results was statistically significant based on the PrI (see Figure 29). The intervention with the highest probability of being the best was multimodal (probability 0.50). The between-study SD was estimated to be 0.40 (95% CrI 0.21 to 0.54), which implies high heterogeneity of intervention effects between studies.
FIGURE 29.
Physical functioning score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; MM, multimodal; UC, usual care.
Impact (short term)
Data were available from seven studies presenting the impact score at short term. A NMA was used to compare the effects of medication, CBTLI, CBTHI, other psychotherapy, guided self-help, GA and multimodal relative to usual care on impact score. Figure 30 presents the network of evidence.
FIGURE 30.
Impact score at short term: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy.
Figure 31 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI, with a large effect size (SMD 2.25, with 95% CrI 1.34 to 3.16), and the effect of medication with a medium effect size (SMD 0.79 with 95% CrI 0.02 to 1.56) were statistically significant at a conventional 5% level. The effect of CBTHI was also statistically significant based on the PrI (see Figure 31). The intervention with the highest probability of being the best was CBTHI (probability 0.98). The between-study SD was estimated to be 0.41 (95% CrI 0.20 to 0.54), which implies high heterogeneity of intervention effects between studies.
FIGURE 31.
Impact score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Emotional distress outcome measures at short term
Anxiety (short term)
Data were available from nine studies presenting the anxiety score at short term. A NMA was used to compare the effects of medication, GP reattribution, GP MUS management, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help and GA relative to usual care on anxiety score. Figure 32 presents the network of evidence.
FIGURE 32.
Anxiety score at short term: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Figure 33 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, GP MUS management and RSSE, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI with a medium effect size (SMD 0.74 with 95% CrI 0.14 to 1.37) was statistically significant at a conventional 5% level, but this effect was not statistically significant based on the PrI (see Figure 33). The intervention with the highest probability of being the best was CBTHI (probability 0.59). The between-study SD was estimated to be 0.25 (95% CrI 0.06 to 0.51), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 33.
Anxiety score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Depression (short term)
Data were available from 12 studies presenting the depression score at short term. A NMA was used to compare the effects of medication, GP reattribution, GP MUS management, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help, GA and multimodal relative to usual care on depression score. Figure 34 presents the network of evidence.
FIGURE 34.
Depression score at short term: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Figure 35 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions, with the exception of GP reattribution and GP MUS management, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI, with a large effect size (SMD 0.93, with 95% CrI 0.37 to 1.52), was statistically significant at a conventional 5% level, and this effect was also statistically significant based on the PrI (see Figure 35). The intervention with the highest probability of being the best was CBTHI (probability 0.64). The between-study SD was estimated to be 0.24 (95% CrI 0.07 to 0.49), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 35.
Depression score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Emotional distress (short term)
Data were available from 12 studies presenting the emotional distress score at short term. A NMA was used to compare the effects of medication, GP reattribution, GP MUS management, CBTHI, RSSE, guided self-help, SES relative and multimodal relative to usual care on emotional distress score. Figure 36 presents the network of evidence.
FIGURE 36.
Emotional distress score at short term: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; UC, usual care.
Figure 37 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, medication and GP MUS management, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of RSSE (a large effect size, SMD 0.82, with 95% CrI 0.02, 1.65) and multimodal (a small effect size, SMD 0.43, with 95% CrI 0.04 to 0.91) were statistically significant at a conventional 5% level, but none of the results was statistically significant based on the PrI (see Figure 37). The intervention with the highest probability of being the best was RSSE (probability 0.68). The between-study SD was estimated to be 0.32 (95% CrI 0.11 to 0.52), which implies high heterogeneity of intervention effects between studies.
FIGURE 37.
Emotional distress score at short term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Physical symptom outcome measures at long term
Pain (long term)
Data were available from seven studies presenting the pain score at long term. A NMA was used to compare the effects of usual care plus, medication, GP MUS management, GP other, CBTHI, other psychotherapy, guided self-help and multimodal relative to usual care on pain score. Figure 38 presents the network of evidence.
FIGURE 38.
Pain score at long term: network of evidence. GP-O, GP – other; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 39 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help, GP MUS management, other psychotherapy, multimodal and usual care plus, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. The largest beneficial effects were associated with medication (a small effect size, SMD 0.41, with 95% CrI –0.16 to 0.98) and CBTHI (a small effect size, SMD 0.36, with 95% CrI –0.21 to 0.94). However, none of the beneficial effects was statistically significant at a conventional 5% level, and none of the beneficial effects was statistically significant based on the PrI (see Figure 39). Guided self-help was significantly worse than usual care (a large effect size, SMD –2.27, with 95% CrI –3.30 to –1.23). The intervention with the highest probabilities of being the best was medication and CBTHI (a probability of 0.49 and 0.28, respectively), but the results of treatment effects were inconclusive. The between-study SD was estimated to be 0.14 (95% CrI 0.03 to 0.48), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 39.
Pain score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GP-O, GP – other; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Fatigue (long term)
Data were available from five studies presenting the fatigue score at long term. A NMA was used to compare the effects of GP-CBT, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help, GA, SES and multimodal relative to usual care on fatigue score. Figure 40 presents the network of evidence.
FIGURE 40.
Fatigue score at long term: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Figure 41 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions, with the exception of GP-CBT, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTLI, with a medium effect size (SMD 0.64, with 95% CrI 0.05 to 1.20), was statistically significant at a conventional 5% level, but none of the results was statistically significant based on the PrI (see Figure 41). The intervention with the highest probability of being the best was CBTLI (probability 0.73). The between-study SD was estimated to be 0.11 (95% CrI 0.03 to 0.42), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 41.
Fatigue score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care.
Bowel symptoms (long term)
Data were available from two studies presenting bowel symptoms at long term. A NMA was used to compare the effects of usual care plus, CBTLI and guided self-help relative to usual care on bowel symptoms. Figure 42 presents the network of evidence.
FIGURE 42.
Bowel symptoms at long term: network of evidence. GSH, guided self-help; UC, usual care.
Figure 43 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. The estimated SMDs of both CBTLI and guided self-help were greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTLI, with a large effect size (SMD 0.84, with 95% CrI 0.17 to 1.52), was statistically significant at a conventional 5% level, and also statistically significant based on the PrI (see Figure 43). The intervention with the highest probability of being the best was CBTLI (probability 0.95). The between-study SD was estimated to be 0.15 (95% CrI 0.03 to 0.49), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 43.
Bowel symptoms at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; UC, usual care.
Somatisation (long term)
Data were available from 11 studies presenting the somatisation score at long term. A NMA was used to compare the effects of usual care plus, GP MUS management, GP reattribution, GP other, CBTLI, CBTHI, other psychotherapy and multimodal relative to usual care on somatisation score. Figure 44 presents the network of evidence.
FIGURE 44.
Somatisation score at long term: network of evidence. GP-O, GP – other; GPRE, general practitioner-delivered reattribution; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 45 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP MUS management, GP reattribution, CBTLI and GP other, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. The largest beneficial effect was associated with CBTHI (a small effect size, SMD 0.47 with 95% CrI –0.30 to 1.29), but this effect was not statistically significant at a conventional 5% level, and it was also not statistically significant based on the PrI (see Figure 45). The intervention with the highest probability of being the best was CBTHI (probability 0.71), but the result of treatment effect was inconclusive. The between-study SD was estimated to be 0.11 (95% CrI 0.02 to 0.35), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 45.
Somatisation score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GP-O, GP – other; GPRE, general practitioner-delivered reattribution; GSH, guided self-help; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Generic physical symptoms (long term)
Only one study,104 which compared other psychotherapy and usual care, was available on the generic physical symptoms at short term. No NMA was performed because of the disconnect network. The estimated SMD of other psychotherapy compared with usual care was a small effect size, –0.30 (95% CI –0.76 to 0.16), from Kolk et al. 104
Physical functioning and impact outcome measures at long term
Physical functioning (long term)
Data were available from 13 studies presenting the physical functioning score at long term. A NMA was used to compare the effects of usual care plus, GP reattribution, GP MUS management, GP-CBT, CBTLI, CBTHI, other psychotherapy, RSSE, guided self-help, GA, SES and multimodal relative to usual care on physical functioning score. Figure 46 presents the network of evidence.
FIGURE 46.
Physical functioning score at long term: network of evidence. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 47 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP-CBT, other psychotherapy and guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. The largest beneficial effect was associated with CBTHI (a small effect size, SMD 0.47, with 95% CrI 0.19 to 1.44), but this effect was not statistically significant at a conventional 5% level, and none of the results was statistically significant based on the PrI (see Figure 47). Guided self-help was significantly worse than usual care (a large effect size, SMD –2.98, with 95% CrI –4.00 to –1.96), and the result was statistically significant based on the PrI. The intervention with the highest probability of being the best was CBTHI (probability 0.38). The between-study SD was estimated to be 0.21 (95% CrI 0.05 to 0.49), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 47.
Physical functioning score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Impact (long term)
Data were available from four studies presenting the impact score at long term. A NMA was used to compare the effects of medication, CBTLI, CBTHI, guided self-help and multimodal relative to usual care on impact score. Figure 48 presents the network of evidence.
FIGURE 48.
Impact score at long term: network of evidence. GSH, guided self-help; ME, medication; MM, multimodal; UC, usual care.
Figure 49 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of guided self-help and multimodal, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. The largest beneficial effect was associated with CBTLI (a large effect size, SMD 0.89, with 95% CrI 0.22 to 1.55), and it was also statistically significant at a conventional 5% level. Guided self-help was significantly worse than usual care (and a large effect size, SMD –1.10, with 95% CrI –2.08 to –0.07 for guided self-help). However, this result was statistically significant based on the PrI (see Figure 49). The intervention with the highest probability of being the best was CBTLI (probability 0.92), but the result of treatment effect was inconclusive. The between-study SD was estimated to be 0.15 (95% CrI 0.03 to 0.48), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 49.
Impact score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GSH, guided self-help; ME, medication; MM, multimodal; UC, usual care.
Emotional distress outcome measures at long term
Anxiety (long term)
Data were available from 11 studies presenting the anxiety score at long term. A NMA was used to compare the effects of usual care plus, medication, GP reattribution, GP MUS management, GP other, GP other psychotherapy, CBTLI, CBTHI, other psychotherapy, RSSE and GA relative to usual care on anxiety score. Figure 50 presents the network of evidence.
FIGURE 50.
Anxiety score at long term: network of evidence. GP-O, GP – other; GPRE, general practitioner-delivered reattribution; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 51 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of medication, GP other psychotherapy and CBTHI, the estimated SMD was smaller than zero, suggesting a negative effect compared with usual care. None of the effects was statistically significant at a conventional 5% level and based on the PrI (see Figure 51). The largest beneficial effect was associated with GP other psychotherapy (a not substantial effect size, SMD 0.18, with 95% CrI –0.40 to 0.76). The intervention with the highest probability of being the best was GP other psychotherapy (probability 0.32). The between-study SD was estimated to be 0.19 (95% CrI 0.04 to 0.46), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 51.
Anxiety score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GP-O, GP – other; GPRE, general practitioner-delivered reattribution; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Depression (long term)
Data were available from 14 studies presenting the depression score at long term. A NMA was used to compare the effects of usual care plus, medication, GP reattribution, GP MUS management, GP other, GP other psychotherapy, CBTLI, CBTHI, other psychotherapy, RSSE, GA and multimodal relative to usual care on depression score. Figure 52 presents the network of evidence.
FIGURE 52.
Depression score at long term: network of evidence. GP-O, GP – other; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 53 presents the SMD of each intervention relative to usual care, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of usual care plus, GP reattribution, GP other, GP other psychotherapy, CBTLI, other psychotherapy, RSSE and GA, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. The largest beneficial effect was associated with multimodal (a small effect size, SMD 0.51, with 95% CrI –0.02 to 1.13), but this effect was not statistically significant at a conventional 5% level. None of the results was statistically significant based on the PrI (see Figure 53). The intervention with the highest probability of being the best was multimodal (probability 0.66). The between-study SD was estimated to be 0.20 (95% CrI 0.05 to 0.46), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 53.
Depression score at long term: SMD for all interventions relative to usual care. a, Probability of being the best treatment. GP-O, GP – other; GPRE, general practitioner-delivered reattribution; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Emotional distress (long term)
Data were available from nine studies presenting the emotional distress score at long term. There were two separate networks for the emotional distress score at long term due to disconnect networks. A NMA was used to compare the effects of medication, GP reattribution, GP MUS management, GP-CBT, guided self-help and multimodal to usual care on emotional distress score. A separate NMA was used to compare the effect of CBTHI, other psychotherapy relative to usual case plus. Figure 54 presents the network of evidence.
FIGURE 54.
Emotional distress score at long term: network of evidence – (a) network 1; and (b) network 2. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; OP, other psychotherapy; UC, usual care; UC+, usual care plus.
Figure 55 presents the SMD from network 1, the median of treatment rankings and the probability of being the best treatment. For all interventions with the exception of GP reattribution, guided self-help, the estimated SMD was greater than zero, suggesting a beneficial effect compared with usual care. However, none of the beneficial effects was statistically significant at a conventional 5% level and based on the PrIs. The largest beneficial effect was associated with multimodel (a small effect size, SMD 0.56, with 95% CrI –0.31 to 1.45). Guided self-help was significantly worse than usual care (a large effect size, SMD –1.44, with 95% CrI –2.60 to –0.30), and this effect was also statistically significant based on the PrI (see Figure 55). The intervention with the highest probability of being the best was multimodal (probability 0.73), but the result of treatment effect was inconclusive. The between-study SD was estimated to be 0.15 (95% CrI 0.03 to 0.47), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 55.
Emotional distress score at long term: SMD for all interventions relative to usual care (network 1). a, Probability of being the best treatment. GPRE, general practitioner-delivered reattribution; GSH, guided self-help; ME, medication; MM, multimodal; UC, usual care.
Figure 56 presents the SMD from network 2, the median of treatment rankings and the probability of being the best treatment. The estimated SMDs of CBTHI and other psychotherapy were both greater than zero, suggesting a beneficial effect compared with usual care. However, only the effect of CBTHI with a median effect size (SMD 0.61, with 95% CrI 0.05 to 1.18) was statistically significant at a conventional 5% level, but none of the results was statistically significant based on the PrI (see Figure 56). The intervention with the highest probability of being the best was CBTHI (probability 0.94). The between-study SD was estimated to be 0.19 (95% CrI 0.03 to 0.49), which implies moderate heterogeneity of intervention effects between studies.
FIGURE 56.
Emotional distress score at long term: SMD for all interventions relative to usual care plus (network 2). a, Probability of being the best treatment. OP, other psychotherapy; UC+, usual care plus.
Model checking
The goodness of model fit is presented in Table 2. For all the analyses, the models fitted the data well, with the total residual deviance close to the total number of data points. Inconsistency checking was performed using the inconsistency model approach. 92 The DICs for both the consistency model and the inconsistency model are presented in Appendix 13, Figures 75–84. Plots of the posterior mean deviance of the individual data points in both the consistency model and the inconsistency model for each of the outcomes are presented in Appendix 11.
For all the outcomes, the consistency model provides a similar DIC as the inconsistency model, with the exception of physical functioning at post treatment and short term, and emotional distress at short term. The contributions to the deviance are very similar in the two models for all the outcomes, with the exception of pain short term (the comparison between guided self-help and multimodal, the comparison between usual care and medication, and the comparison between usual care and CBTHI), fatigue at short term (the comparison between usual care and RSSE), physical functioning at short term (the comparison between guided self-help and multimodal, and the comparison between usual care and GP MUS management), emotional distress at post treatment (the comparison between usual care and RSSE) and short term (the comparison between guided self-help and multimodal), and depression at short term (the comparison between usual care and medication, and the comparison between usual care and CBTHI). However, there are overlaps in the 95% CrIs for the estimates between the two models (see Appendix 11). Hence, overall there is no evidence to suggest inconsistency in the networks for all the outcomes.
| Outcome | Time point | Total number of data points | Total residual deviance | Inconsistency check | |
|---|---|---|---|---|---|
| Consistency model DIC | Inconsistency model DIC | ||||
| Pain | Post treatment | 14 | 16.21 | 4.27 | 6.19 |
| Short term | 9 | 10.38 | 0.96 | 1.25 | |
| Long term | 8 | 8.05 | 1.00 | 0.97 | |
| Fatigue | Post treatment | 12 | 13.38 | 2.95 | 3.79 |
| Short term | 11 | 13.54 | 3.96 | 4.60 | |
| Long term | 10 | 9.51 | –0.75 | 0.38 | |
| Bowel symptoms | Long term | 2 | 2.03 | 0.91 | 0.98 |
| Somatisation | Post treatment | 11 | 12.04 | 2.96 | 3.75 |
| Short term | 6 | 6.80 | –0.16 | –0.20 | |
| Long term | 11 | 10.99 | –0.90 | –0.74 | |
| Generic physical symptoms | Post treatment | 2 | 1.32 | 2.30 | 2.19 |
| Physical functioning | Post treatment | 17 | 16.32 | –1.44 | 2.74 |
| Short term | 13 | 15.16 | 4.57 | 1.39 | |
| Long term | 15 | 16.24 | 4.40 | 4.49 | |
| Impact | Post treatment | 11 | 11.51 | 9.34 | 10.15 |
| Short term | 9 | 10.43 | 9.34 | 10.15 | |
| Long term | 5 | 4.95 | 3.42 | 3.56 | |
| Emotional distress | Post treatment | 15 | 16.24 | 3.42 | 3.57 |
| Short term | 14 | 17.03 | 4.30 | 0.21 | |
| Long term for NMA1 | 7 | 7.23 | 3.21 | 3.15 | |
| Long term for NMA2 | 2 | 2.03 | 0.92 | No indirect evidence | |
| Anxiety | Post treatment | 17 | 19.05 | 7.78 | 8.70 |
| Short term | 12 | 13.35 | 6.95 | 5.45 | |
| Long term | 14 | 15.09 | 4.66 | 4.80 | |
| Depression | Post treatment | 16 | 17.39 | 6.20 | 7.33 |
| Short term | 15 | 16.11 | 5.69 | 2.04 | |
| Long term | 16 | 17.29 | 5.30 | 6.11 | |
All NMAs had moderate to high heterogeneity in treatment effects between studies. For most of the outcomes, there were only one or two studies to inform each of the pairwise comparisons/contrasts, with the exception for pain and anxiety both at post treatment, where one contrast was informed by four studies, somatisation at short term and depression at both post treatment and long term, where one contrast was informed by three studies. Hence, there were not enough data to update the prior for the heterogeneity parameter in most analyses. However, the use of plausible informative prior distribution for the heterogeneity ensures that the posterior distribution of the heterogeneity and treatment effects were in also in the plausible range.
It was not possible to perform a metaregression to explain the heterogeneity because there was insufficient replication of each treatment effect across studies. Hence, we compared the point estimate and CIs across condition groupings.
The outcome measures of pain, fatigue, physical functioning, anxiety, depression and emotional distress have been assessed in different condition groupings. For all of these outcomes, the point estimates of the pooled SMDs differed over condition groups, when they were estimated separately in each of these groups. The exceptions to this were for physical functioning score (guided self-help vs. usual care) at short term between the chronic fatigue group and the pain multiple sites group, and for depression score (CBTLI vs. usual care) at short term between the IBS group and the chronic fatigue group. In both of these cases, the SMDs for the individual condition groups were similar.
In all the cases, the CIs of the point estimates generally overlapped indicating that the difference in SMD between condition groups was not statistically significant. Overall, the evidence suggested that the relative treatment effects for the interventions listed in Table 3 were not significantly different across condition groupings.
| Time point | Outcome measure | Treatment (vs. usual care) | Condition grouping (first author and year of publication) | SMD (95% CI) | Comments |
|---|---|---|---|---|---|
| Immediately post treatment | Pain | CBTHI | MUS/somatoform (Zonneveld, 2012141) | 0.51 (0.20 to 0.82) | Different point estimates, but overlapping CIs |
| Pain multiple sites (Alda, 2011137) | 0.23 (–0.14 to 0.59) | ||||
| Pain multiple sites (Luciano, 2014138) | 1.20 (0.78 to 1.61) | ||||
| Pain multiple sites (McBeth, 2012125/Beasley 2015126) | 0.35 (0.08 to 0.61) | ||||
| Physical functioning | CBTHI | MUS/somatoform (Zonneveld, 2012141) | 0.39 (0.08 to 0.70) | Different point estimates, but overlapping CIs | |
| Pain multiple sites (McBeth, 2012125) | 0.15 (–0.11 to 0.41) | ||||
| Anxiety | CBTLI | Chronic fatigue (Tummers, 2012)149 | 0.14 (–0.21 to 0.50) | Different point estimates, but overlapping CIs | |
| IBS (Moss-Morris, 2010130) | –0.46 (–0.95 to 0.04) | ||||
| OP | MUS/somatoform (Kolk, 2004104) | –0.01 (–0.47 to 0.45) | Different point estimates, but overlapping CIs | ||
| MUS/somatoform (McLeod, 1997105) | 0.32 (–0.08 to 0.72) | ||||
| MUS/somatoform (Posse, 200496) | 0.46 (–0.67 to 1.59) | ||||
| Chronic fatigue (Wearden, 2010111) | –0.18 (–0.45 to 0.10) | ||||
| Depression | CBTLI | IBS (Moss-Morris, 2010130) | –0.03 (–0.51 to 0.45) | Different point estimates, but overlapping CIs | |
| MUS/somatoform (Martin, 200799) | –0.14 (–0.47 to 0.19) | ||||
| OP | MUS/somatoform (Kolk, 2004104) | –0.10 (–0.55 to 0.36) | Different point estimates, but overlapping CIs | ||
| MUS/somatoform (McLeod, 1997105) | 0.57 (0.16 to 0.97) | ||||
| Chronic fatigue (Wearden, 2010111) | –0.09 (–0.36 to 0.19) | ||||
| Short term | Physical functioning | GSH | Pain multiple sites (LeFort, 1998128) | 0.27 (–0.10 to 0.64) | Similar point estimates, and overlapping CIs |
| Chronic fatigue (Chalder, 1997114) | 0.26 (–0.06 to 0.58) | ||||
| Anxiety | CBTLI | IBS (Moss-Morris, 2010130) | –0.08 (–0.56 to 0.41) | Different point estimates, but overlapping CIs | |
| Chronic fatigue (Friedberg, 2013151) | 0.35 (–0.10 to 0.80) | ||||
| Depression | CBTLI | IBS (Moss-Morris, 2010130) | 0.29 (–0.20 to 0.78) | Similar point estimates, and overlapping CIs | |
| Chronic fatigue (Friedberg, 2013151) | 0.25 (–0.20 to 0.70) | ||||
| RSSE | MUVD (Kobeissi, 2012110) | 0.10 (–0.14 to 0.34) | Different point estimates, but overlapping CIs | ||
| Chronic fatigue (Friedberg, 2013151) | –0.02 (–0.48 to 0.42) | ||||
| Emotional distress | GSH | Pain multiple sites (LeFort, 1998128) | 0.38 (0 to 0.75) | Different point estimates, but overlapping CIs | |
| Chronic fatigue (Chalder, 1997114) | 0.28 (–0.37 to 0.60) | ||||
| Long term | Anxiety | OP | MUS/somatoform (Kolk, 2004104) | –0.34 (–0.80 to 0.13) | Different point estimates, but overlapping CIs |
| Chronic fatigue (Wearden, 2010111) | –0.15 (–0.42 to 0.13) | ||||
| CBTLI | IBS (Moss-Morris, 2010130) | –0.27 (–0.76 to 0.21) | Different point estimates, but overlapping CIs | ||
| Chronic fatigue (Friedberg, 2013151) | 0.18 (–0.26 to 0.63) | ||||
| Depression | OP | MUS/somatoform (Kolk, 2004104) | –0.31 (–0.77 to 0.15) | Different point estimates, but overlapping CIs | |
| Chronic fatigue (Wearden, 2010111) | –0.13 (–0.41 to 0.14) | ||||
| CBTLI | IBS (Moss-Morris, 2010130) | 0.35 (–0.14 to 0.83) | Different point estimates, but overlapping CIs | ||
| MUS/somatoform (Martin, 200799) | –0.26 (–0.60 to 0.06) | ||||
| Chronic fatigue (Friedberg, 2013151) | –0.07 (–0.52 to 0.37) |
Discussion of clinical effectiveness review
The review identified 59 studies that met the inclusion criteria. A broad approach to inclusion was taken, and this is reflected in the considerable heterogeneity found. Owing to the considerable differences in types of interventions, it was not appropriate to treat all interventions as one behavioural intervention group and, therefore, study arms were categorised into 13 types. Even within intervention types, there was variation in treatment duration and intensity, mode of delivery and intervention provider, and differences in the specifics of the primary care setting – from interventions delivered by the patient’s own GP within their own GP practice to those involving collaborative care with health-care professionals from outside the practice or primary care patients travelling to a gym or fitness facility. Studies included a range of populations, and this resulted in further heterogeneity, with, for example, several studies of CBTHI but these being in different populations. Ideally, these differences would have been explored through metaregression, but there was insufficient replication of each intervention type within the networks to allow this. These issues have resulted in limitations when drawing conclusions regarding clinical effectiveness. Further to this, studies did not measure the same outcomes at the same time points (i.e. end of treatment, short- and long-term follow-up) and, therefore, different studies inform the network at these different time points; these analyses are not repeated measures and do not account for correlations between time points.
The evaluation of clinical effectiveness necessitated making judgements regarding intervention groups, and inclusion judgements such as participants meeting ‘medically unexplained’ criteria (where populations did not meet the criteria for FSSs or other pre-defined diagnostic classifications) and primary care setting criteria. Although efforts were made to ensure that these judgements were made using objective criteria, and with consultation between project team members when borderline decisions were required, it is acknowledged that judgements could arguably have been made differently. A narrower scope may have reduced this possibility but using broad inclusion criteria has allowed a broader evaluation of the topic. With regard to setting, the review did not include studies of home-based or self-help behavioural interventions, such as e-health, where these had insufficient or no involvement of primary care practitioners (i.e. were conducted solely by university research teams). These interventions may well be feasible and effective in a primary care setting and when co-ordinated by primary care teams. However, as was evidenced by the wide variation in recruitment by GPs seen in the current review, it was considered inappropriate to include these interventions until they have been conducted in the ‘real world’ of primary care.
Appendix 12, Table 97, summarises the significant results identified in the NMAs, showing the SMDs for the significant intervention groups for each outcome and time point. Although the NMAs do not identify individual significant trials, SMDs for all individual trials are presented by outcomes in Appendix 12, Tables 98–107. For those intervention groups that were shown to have significant beneficial effects, a brief description of the key characteristics of the trials in those groups is presented below by outcome.
Pain
At the end of treatment, CBTHI and multimodal interventions were shown to be significantly more effective than usual care at reducing pain. CBTHI was the only intervention to be significantly more effective than usual care at short-term follow-up. No interventions were more effective than usual care at long-term follow-up.
High-intensity cognitive–behavioural therapy
There were four studies that informed the pain networks for CBTHI interventions. These were Zonneveld et al. ,141 Alda et al. ,137 Luciano et al. 138 and McBeth et al. 125 One of these studies141 was of a population of patients with unexplained physical symptoms, referred to the study by the GPs if their symptoms were considered to be unexplained by a physical condition. Patients were then included if they fulfilled the DSM-IV criteria for undifferentiated somatoform disorder or a chronic pain disorder. The CBT intervention was based on the consequences model, tailored for MUS in primary care, and delivered over 13 weeks by an outreaching mental health service. Patients in the control group were put on a waiting list. At the end of treatment, small to moderate SMDs were found in favour of the intervention for pain, physical functioning, emotional distress and somatisation.
The remaining three studies that informed this network were of populations of patients with chronic widespread pain. McBeth et al. 125 report results from the MUSICIAN trial [Managing Unexplained Symptoms (chronic widespread pain) In primary Care: Involving traditional and Accessible New approaches]. 125 Patients meeting the criteria were recruited from primary care practices in the UK and were allocated to one of four arms: (1) telephone CBT, consisting of seven weekly sessions of 45–60 minutes plus two further follow-up sessions. Patients were given a self-management CBT manual and offered some choice in the type of CBT they preferred, including behavioural activation, cognitive restructuring and lifestyle changes; (2) exercise, consisting of a leisure facility- and gym-based exercise programme, with recommended exercise duration of 20 to 60 minutes, at least twice per week; (3) combined, consisting of both the telephone CBT and exercise interventions; and (4) treatment as usual, consisting of usual care by their family physician. Small SMDs were found in favour of all three active intervention groups, with the SMDs for CBTHI and combined groups marginally larger.
Luciano et al. 138 studied a population of patients fulfilling the ACR 1990 criteria for fibromyalgia, recruited by GPs from primary health-care centres in Spain. The intervention arm received group ACT, adapted to fibromyalgia patients, consisting of sessions on mindfulness, cognitive defusion, committed action and observation of the self. Sessions were delivered by a therapist over eight 2.5-hour sessions, with daily homework exercises. A second intervention group received recommended pharmacological treatment, prescribed by the GP after a 2-hour training session. A third, control, group were put on a waiting list. At the end of treatment, large SMDs were found for patients in the CBTHI group, and moderate effects for those in the medication group, when compared with those on the waiting list.
Alda et al. 137 also included patients with fibromyalgia recruited by primary care doctors from primary health-care centres in Spain. The intervention arm received CBT based on Thorn’s model of pain catastrophising, adapted to people with fibromyalgia. The main components of the CBT were cognitive restructuring and coping. Ten 90-minute group CBT sessions were delivered over 10–12 weeks by trained therapists. A second intervention group received recommended pharmacological treatment, administered by a psychiatrist. A third, control, group received treatment as usual. At the end of treatment, SMDs showed no substantial effects on pain for CBT compared with usual care, although a small effect was found at long-term follow-up. There was a small effect for the medication group both at the end of treatment and in the long term.
Only the McBeth et al. 125 and Luciano et al. 138 studies informed the network for CBTHI at short-term follow-up, with the CBTHI group in Luciano et al. 138 continuing to show a large effect size, and McBeth et al. 125 showing a small effect size for CBTHI compared with usual care. By long-term follow-up, the NMA found no significant effects for CBTHI compared with usual care, with data from only one study remaining to inform the network (Alda et al. 137).
Multimodal interventions
Five studies informed the pain networks for multimodal (MM) interventions. These were Cuesta-Vargas et al. ,118 Schaefert et al. ,132 Walti et al. ,121 Luciano et al. 139 and McBeth et al. 125 McBeth et al. 125 had multiple arms and has been previously described under CBTHI. Cuesta-Vargas et al. 118 is described under guided self-help.
Schaefert et al. 132 reports results from the speciAL trial (specific collaborative group intervention for MUS patients in general practice). This was a cluster randomised trial including patients meeting the criteria of > 6 months’ bodily complaints without sufficient explanatory peripheral organ pathology, and with MUS as the main treatment issue. The multimodal group consisted of GP MUS management training, plus a syndrome-oriented psychosomatic group intervention for MUS patients conducted by GPs together with a psychosomatic specialist, in 10 weekly sessions of 90 minutes plus two booster sessions. The GP training consisted of a guideline-based curriculum in the diagnosis and management of patients with MUS. At follow-up, SMDs showed that there were no substantial differences between the two groups for pain scores.
Walti et al. 121 conducted a pilot RCT with a population of patients with non-specific lower back pain recruited and treated in a primary care physiotherapy centre in Switzerland. Patients had moderate to severe disability. The multimodal intervention group received patient education, sensory retraining and motor retraining. Sessions were delivered in 16 sessions over 8–12 weeks. The control group received usual physiotherapy care, which consisted primarily of active treatment, such as muscle-strengthening exercises and mobilisation or stretching exercises.
Luciano et al. 139 conducted a trial of 216 patients with fibromyalgia. The intervention was set in three general practices in Spain. GPs previously referred patients suspected of fibromyalgia for diagnosis at a rheumatology unit. A database of these patients is kept at the GP practice for monitoring. These patients were included in the trial, which was conducted by a multidisciplinary team within the three general practices. The multimodal intervention consisted of five 2-hour group sessions of education and four 2-hour sessions of autogenic training. The education element included information on symptoms, course of the condition, comorbidities and causes. The programme outlined the influence of psychosocial factors on pain, the benefits of exercise and behavioural change. The autogenic training focused on physical and mental relaxation, pain relief and stress reduction. The group sessions encouraged emotive exchange with other patients. The control group received usual care. The multimodal intervention group reported a greater reduction in pain than the control group. SMDs showed this to be a moderate effect.
Guided self-help
Guided self-help was shown to be less effective than usual care at long-term follow-up. Only one study of guided self-help informed the network. Cuesta-Vargas et al. 118 studied a population of 58 primary care patients with > 3 months of non-specific chronic lower back pain, who had been referred to the study by their GP. The intervention arm was deep-water running three times per week for 4 months plus an education intervention. This consisted of a 25-page educational booklet and verbal presentation on basic anatomy and physiology of the spine, principles of ergonomics for low back pain patients, and instructions for coping strategies. The booklet encouraged the patients to treat themselves instead of undergoing passive treatments. This was presented alongside usual care from their GP. Patients in the control arm received the education booklet and presentation alongside usual care from their GP. Both groups showed improvement over time, but the deep-water running/education group showed more improvement with large SMDs for pain, physical functioning, impact and emotional distress at all time points. The effect sizes for this study were particularly large, and inconsistency checking of the NMA showed that the direct evidence of this effect was not consistent with the indirect evidence.
Fatigue
At the end of treatment, RSSE, CBTLI, multimodal interventions and GA were shown to be significantly more effective than usual care at reducing fatigue. RSSE and CBTLI were significantly more effective than usual care at short-term follow-up. CBTLI was the only intervention more effective than usual care at long-term follow-up.
Relaxation/stretching/social support/emotional support interventions
Two RSSE interventions informed the fatigue network at the end of treatment. The first of these was Ho et al.,150 a Hong Kong study of people with CFS who were recruited from the community. Participants were recruited following screening via a web-based questionnaire and were required to meet the US CDC criteria for CFS; however, diagnosis was not confirmed by medical examination. The intervention group received 2-hour group qigong training for 5 weeks. The group sessions included basic theories of Chinese medicine and physiology of mind–body connections, mindful meditation for relaxation and gentle movement or body stretching in standing postures. A final session in qigong exercise training was delivered by a Daoist qigong master. Participants were also expected to continue to practise qigong exercise at home for the remaining 12 weeks of the study period. Participants in the control group received qigong training at the end of the study. SMDs showed a large beneficial effect on reduction in self-perceived fatigue at the end of the 5-week group training for the qigong group. At the 4-month follow-up, this beneficial effect remained, although it should be noted that it was expected that participants continued to practise qigong at home during this follow-up period.
A further RSSE intervention arm informed the network at short- and long-term follow-up (Friedberg et al. 151), which also included a CBTLI arm and is described below.
Low-intensity cognitive–behavioural therapy interventions
Two CBTLI interventions informed the fatigue network at the end of treatment,113,149 with a further study informing the network at short- and long-term follow-up. 151
Ridsdale et al. 113 studied a UK population of 123 patients presenting to their GPs with unexplained fatigue of > 3 months. The study compared two active interventions: CBT and graded exercise. Both treatments were delivered on the GP premises by CBT therapists and physiotherapists, in six 45-minute sessions over 12 weeks. The CBT treatment was manualised, based on a model of precipitating and perpetuating factors. It involved activity planning, establishing a sleep routine and other cognitive interventions, and addressed negative beliefs, self-expectations and self-esteem. The GA intervention (i.e. GET) was based on each individual patient’s physical capacity, and aimed to achieve a gradual but progressive increase in aerobic activity. A third, non-randomised cohort of patients received an educational booklet. Both groups saw a significant reduction in fatigue from baseline. SMDs showed a small beneficial effect for the CBT group compared with the GET group at the end of treatment, but showed no substantial difference between groups at follow-up. A total of 3% of patients in the CBT group showed an increase in fatigue after treatment, compared with 12% of patients in the GET group.
Tummers et al. 149 studied a population of 123 patients in Holland diagnosed with CFS (US CDC criteria) by either their GP or a consultant. The active intervention was a minimal intervention based on CBT, previously shown by the authors to be effective in a tertiary treatment setting. The intervention was applied in a community-based mental health centre, with no previous experience of treating patients with CFS. Psychiatric nurses were trained to deliver the intervention, which was based on a protocol for CBT for CFS, and was in the form of a guided self-instruction booklet delivered over 20 weeks. The programme focused on precipitating and perpetuating factors, challenging fatigue-related cognitions, reducing the focus of fatigue, establishing a sleep routine, assessing activity patterns and gradually increasing physical activity. The control group was put on a waiting list. A total of 12 out of the 123 randomised patients were found to have a misdiagnosis of CFS during the course of the trial but were not excluded from the analyses. Patients in the intervention group showed a significantly greater reduction in fatigue severity than those in the waiting list group at the end of treatment. SMDs showed this effect to be moderate.
Friedberg et al. 151 studied a population of 111 primary care patients with chronic fatigue, recruited from a family medicine/primary care practice in the USA. A total of 39% of the included sample met the US CDC criteria for CFS, and the remaining 61% did not meet the full criteria but had at least 6 months’ persistent fatigue not clearly attributable to identifiable medical conditions, with associated impairment in functioning. The study had three arms. The CBTLI intervention was ‘fatigue self-management’. This consisted of two nurse-led sessions based on a CBT treatment programme for CFS, with a self-help booklet for a period of home-based self-management. The nurse-led sessions covered diagnosis and possible causes, stress and sleep disturbance, balance between mental and physical exertion, unhelpful behaviours or illness beliefs, and development of more useful cognitive and behavioural strategies. The second arm was an attention control RSSE intervention. It included emotional support and self-monitoring of symptoms, affect and stress. The intervention also consisted of two sessions and a period of home-based activities. The third arm was usual medical care/no treatment. The results showed significantly greater reduction in fatigue in the CBTLI group than in both the RSSE attention control group and the usual-care group. SMDs showed moderate effects for the CBTLI group compared with usual care, and showed no substantial effects for the RSSE intervention compared with usual care.
Multimodal interventions
Two multimodal interventions informed the fatigue network, McBeth et al. 125 and Luciano et al. 139 For both of these studies, the populations were patients with chronic widespread pain/fibromyalgia. The studies are outlined above under Pain. For fatigue, SMDs showed a small effect for the intervention on reduction in fatigue at the end of treatment in Luciano et al. 139 compared with usual care, and a large effect at the end of treatment in McBeth et al. ,125 reducing to a small effect at long-term follow-up compared with usual care.
Graded activity
Four studies with GA arms informed the fatigue network: Ridsdale et al. 113 (described in Low-intensity cognitive–behavioural therapy interventions), Marques et al. ,133 Wearden et al. 111 and Ridsdale et al. 115
Marques et al. 133 included a study population of 99 patients meeting the CDC criteria for idiopathic chronic fatigue, recruited to the study by their medical doctor. The trial took place in four public primary care centres and one private practice, and in a patient association. The active intervention arm was the 4-STEPS programme, which was delivered by a health psychologist over a 12-week intervention period. The intervention was based on the use of self-regulation theory to promote physical activity. Features of the intervention were the use of motivational interviewing to increase motivation and confidence, and the formation of action plans with specific personal physical activity goals. The control condition (guided self-help) comprised an educational booklet containing information about the benefits of physical activity and physical activity guidelines for adults, and participants in this group were set a personal activity goal. At the end of the intervention, a significant beneficial effect for the GA intervention was found compared with the control of guided self-help. SMDs showed this to be a moderate effect.
Ridsdale et al. 115 included a study population of 222 primary care patients consulting with their GPs for fatigue of > 3 months’ duration as the main problem, but with no known physical condition that could explain the fatigue. The study had three arms. The graded exercise intervention was delivered by physiotherapists over eight 30-minute sessions at 2-weekly intervals in the patient’s own primary care practice. Patients were guided through supervised exercise (walking), tailored to their individual current physical capacity, and gradually building in intensity. A second arm was a psychotherapy intervention. This consisted of Rogerian client-centred non-directive counselling. The counselling was delivered by trained therapists in eight 50-minute sessions at 2-weekly intervals. The third arm was usual care with the addition of a booklet of self-help techniques based on CBT principles, and including information on the causes of fatigue (guided self-help). No significant differences in reduction in fatigue were found between the three intervention groups at either the 6- or the 12-month follow-up. SMDs showed no substantial effect for either guided self-help or GA when compared with other psychotherapy at either time point. The authors reported a high level of dissatisfaction with care, but less dissatisfaction in the GA group.
Wearden et al. 111 included a study population of 296 patients who met the Oxford criteria for CFS/myalgic encephalomyelitis, and who scored < 70% on the SF-36 physical functioning scale, and > 4 on the Chalder Fatigue Scale. Patients meeting these criteria were referred by their GP. The study had three intervention arms. The GA intervention (‘pragmatic rehabilitation’) consisted of a programme of graded return to activity, and a focus on regularising sleep patterns, addressing somatic symptoms of anxiety and addressing concentration and memory problems. The counselling intervention (‘other psychotherapy’) was a non-directive listening therapy, allowing the patient to discuss their concerns and problems. Both the GA and the counselling interventions were delivered over a 10-week period by general nurses with experience of working in primary care, but not of CFS/myalgic encephalomyelitis, in the patients’ own homes, with additional telephone sessions. Home visits were of 1 hour’s duration, and telephone sessions were of 30 minutes’ duration. The third arm was GP usual care. At the end of treatment, patients in the GA group had significantly reduced fatigue compared with patients in the usual-care group. This effect was no longer significant at long-term follow-up; however, at 70 weeks, GA was significantly better than usual care when the Chalder Scale was used with Likert scoring. There was no significant beneficial effect on fatigue of other psychotherapy compared with usual care at the end of treatment or follow-up. SMDs show small effects at the end of treatment and long-term follow-up for GA compared with usual care, but show no substantial effects for other psychotherapy compared with usual care at either time point.
Bowel symptoms
There were insufficient studies to form networks for bowel symptoms for end of treatment and short-term follow-up. Only two studies informed the network for long-term follow-up. Of these, only CBTLI was shown to have a significantly beneficial effect compared with usual care.
Low-intensity cognitive–behavioural therapy intervention
Moss-Morris et al. 130 reported results from a pilot study of a population of 64 patients meeting the Rome II criteria for IBS. Patients were recruited either from a database of known patients with IBS who had taken part in a previous primary care study or from patients presenting to the study GP with IBS symptoms. The study GP screened interested participants for eligibility. The intervention arm was a 7-week home-based manualised self-management programme. The programme included one 1-hour face-to-face session and two 1-hour telephone sessions with a health psychologist. The programme focused on assessment of symptoms and self-monitoring behaviours, behavioural management of symptoms and goal-setting, managing unhelpful thoughts, personal expectations and activity patterns, and relaxation and stress management. The control arm received treatment as usual, which included a sheet explaining that a range of tests had been conducted and had ruled out structural causes for their IBS. Results showed a significant reduction in bowel symptoms severity at the end of treatment and at long-term follow-up. SMDs showed this to be a moderate to large effect.
Somatisation
Data from 11 studies were available to inform the network for somatisation. Despite this, no interventions were found to have any significant beneficial effects on somatisation compared with usual care. Trials of a range of intervention types were identified. SMDs for individual trials show that, although a small number of trials found small beneficial effects at the end of treatment, these effects are mostly lost by long-term follow-up. One trial of CBTHI144,148 produced moderate beneficial effects both at end of treatment and at long-term follow-up. This was a study of patients meeting the criteria for Escobar’s Abridged Somatisation Disorder, recruited by GPs from 21 primary care centres in Spain. The CBT intervention was based on a protocol by Escobar et al. ,156 delivered over 10 weekly sessions. The programme included muscle relaxation, emotional mindfulness, cognitive restructuring and social skills.
Generic physical symptoms
Data from two studies were available to inform the network for generic physical symptoms at the end of treatment; however, no interventions showed significant beneficial effects. There were insufficient data to form networks for short- or long-term follow-ups.
Physical functioning
At the end of treatment and short-term follow-up, multimodal interventions were the only interventions that showed significant beneficial effects. No interventions showed significant beneficial effects at long-term follow-up. Studies that contributed data to the network for multimodal studies have been described previously. 107,118,125,132,139,142 SMDs for these studies showed that, in Cuesta Vargas et al.,118 large effects were found for multimodal compared with guided self-help at the end of both short- and long-term treatment; in Schaefert et al.,132 a small effect was found at short-term follow-up for multimodal compared with GP MUS management, but no substantial effect was found at long-term follow-up; in Luciano et al.,139 a small effect was found for multimodal compared with usual care at the end of treatment; and in McBeth et al.,125 a small effect was found in favour of multimodal compared with usual care at the end of treatment and at short-term follow-up.
Smith et al. 107 report results from a study of 206 high-utilising MUS patients recruited from a health maintenance organisation in the USA. MUS patients were identified through patient records, with eligibility assessed in patients who had more than eight visits per year. MUS was identified if symptoms of > 6 months’ duration were present that had no documented disease. Nurse practitioners with no prior experience in mental health received a 10-week training programme to deliver the intervention. The intervention consisted of patient-centred management to improve communication and establish positive patient–provider relationships, antidepressants, exercise, relaxation training, physical therapy and comorbid organic disease management. It was delivered in 12 20-minute visits with patients over a 1-year period. The control group received usual care. Patients in the intervention group were more likely to improve than those in the control group.
Smith et al. 142 report results from a pilot study of 30 high-utilising patients with MUS. Patients were identified from the Henry Ford Health System using ICD-9 codes to identify MUS. The intervention group received a multimodal intervention delivered by primary care physicians (PCPs) who had received training for the study. The intervention was similar to that of Smith et al.,107 with the exception that the prior study used nurse practitioners to deliver the intervention. It consisted of antidepressant medication (where PHQ-9 identified depression), structured CBT and patient-centred management aimed at maximising communication and the patient–provider relationship. The intervention was delivered over 1 year, with seven visits with the PCP, and three visits with a case manager. The intervention produced a significant beneficial effect on somatisation compared with the control group, which was usual care. SMDs show a moderate effect for the multimodal intervention compared with usual care at long-term follow-up.
Impact
At the end of treatment and at short-term follow-up, only the CBTHI interventions were shown to have a significant beneficial effect on impact of the illness on daily life. Two studies of CBTHI informed the network at the end of treatment. 137,138 Only data from Luciano et al. 138 informed the network at short-term follow-up. Both of these studies have been previously described.
Anxiety
At the end of treatment and at short-term follow-up, only the CBTHI interventions were shown to have a significant beneficial effect on anxiety compared with usual care. Four studies of CBTHI informed the network at the end of treatment. These were Escobar et al. ,156 Gili/Moreno,144,148 Alda et al. 137 and Luciano et al. 138 At short-term follow-up, only data from Luciano et al. 138 informed the network. Gili/Moreno,144,148 Alda et al. ,137 and Luciano et al. 138 have been previously described. No interventions showed significant beneficial effects on anxiety at long-term follow-up.
Escobar et al. 156 studied a population of patients meeting the criteria for Escobar’s abridged somatisation, recruited from primary care clinics in the USA. PCPs and nurses referred consecutive adult patients who had repeatedly sought care for MUS and for whom the symptoms were a source of distress. The intervention group received a CBT-type intervention delivered by therapists trained for the study. The intervention was delivered in 10 sessions of 45–60 minutes over 10–20 weeks. The control group received usual clinical care from their PCP, which included a consultation letter. SMDs showed a small beneficial effect on anxiety for the intervention group versus the control at the end of treatment. In Gili/Moreno,144,148 a moderate beneficial effect was found for the CBTHI group compared with usual care. For Alda et al.,137 no substantial effect was found compared with usual care. In Luciano et al.,138 a large beneficial effect was found for the CBTHI group compared with usual care.
Depression
At the end of treatment and at short-term follow-up, only the CBTHI interventions were shown to have significant beneficial effects. Four studies on CBTHI informed the network at the end of treatment. These were the same studies as for anxiety, and all have been described previously. 137,138,144,148,156 No interventions had significant beneficial effects at long-term follow-up. SMDs showed small beneficial effects in Escobar 2007156 and Gili/Moreno. 144,148 No substantial effects on depression were found in Alda et al.,137 but a large beneficial effect on depression was found in Luciano et al. 138 Only Luciano et al. 138 showed a beneficial effect at short-term follow-up.
Chapter 4 The acceptability of primary care or community-based behaviour modification interventions for medically unexplained symptoms: qualitative systematic review
This chapter aims to provide an overview of the evidence for patients’ and HPs’ perspectives on the acceptability, relative benefits and potential harms of primary care or community-based behaviour modification interventions for MUS.
Review methods
Screening and eligibility
A two-stage sifting process for inclusion of studies (title/abstract then full-paper sift) was undertaken. Titles and abstracts were scrutinised by one systematic reviewer (AS) according to the inclusion and exclusion criteria. There was no exclusion on the basis of quality. All studies identified for inclusion using the abstract, together with any in which a decision on inclusion was not possible from these brief details, was obtained for more detailed appraisal. Agreement on inclusion at title/abstract sift was checked by a second systematic reviewer (AB) for 20% of the total electronic search results; therefore, 345 records were sifted by both AS and AB. Agreement was calculated using the kappa statistic. Based on these data, the kappa statistic was 0.765. As the kappa statistic was above acceptable levels (i.e. 0.7), double-sifting was not deemed necessary. Where the reviewers disagreed on inclusion/exclusion at title/abstract sift, these records were retrieved to check at full text. In the event of disagreement regarding the inclusion of a study, the opinion of the MUS experts in the project team was sought.
The inclusion and exclusion criteria for the qualitative review are reported in Table 4. The inclusion and exclusion criteria for the project as a whole are presented in detail in Chapter 3. Any inclusion and exclusion criteria that are specific to the qualitative review are documented in more detail here.
| Inclusion criteria | Included | Excluded |
|---|---|---|
| P – population | Patients meeting the criteria for MUS, MUPS and somatoform disorders. Populations with FSSs were included (e.g. IBS, CFS, fibromyalgia). Health-care providers who had delivered behavioural modification interventions designed for these patients were also included | Subacute patients. Patients with intermittent pain (where current episode was < 3 months – or this information was not available from the paper/or they cannot be disentangled from the rest of the sample) |
| I – intervention | Behavioural modification interventions | Studies of management of MUS where evidence relating to a treatment of interest was not separately identifiable |
| C – comparator | N/A | N/A |
| O – outcomes | Qualitative data | N/A |
| S – study design | Qualitative research, mixed-methods research, qualitative data embedded in trial reports or process evaluations | Quantitative reports without qualitative evidence |
Evidence from health-care providers who delivered the behavioural interventions for included patients was sought, in addition to patients’ views. Studies were included only if they contained data related specifically to an intervention that met the review’s inclusion criteria, as the review was specifically aimed at evaluating perceptions of behavioural interventions delivered in a primary care setting. Studies containing qualitative data relating to general management of MUS and not to a specific included intervention were therefore excluded (e.g. Ax et al. ,164 Bayliss et al. ,165 Wallace et al. ,166 Deale and Wessely,167 Raine et al. 168).
Study types: (1) studies reporting qualitative research or qualitative data elicited via a survey or a mixed-methods study to include qualitative data on the perspectives and attitudes of patients who had received behaviour modification interventions in a primary care or community-based setting; (2) qualitative data, embedded in trial reports or in accompanying process evaluations, which can inform an understanding of how issues of acceptability are likely to affect the clinical effectiveness of eligible interventions; (3) qualitative data, either from separately conceived research or embedded within quantitative study reports, reporting the acceptability of interventions to health-care practitioners.
Quality assessment strategy
Assessment of confidence in the review findings:
The Confidence in the Evidence from Reviews of Qualitative (CERQual)169 research approach was used to summarise our confidence in the findings across the studies included in the review. CERQual is currently under development and draws on the principles used to develop the GRADE approach. An earlier version of this approach has been used in two Cochrane reviews169,170 and three non-Cochrane reviews. 171–173 CERQual assesses confidence in the evidence based on four key components:
-
The methodological limitations of included studies. Methodological quality of individual studies was appraised using an abbreviated version of the Critical Appraisal Skills Programme (CASP) quality assessment tool for qualitative studies. 174 Two reviewers (AS and AB) independently applied the set of quality criteria to each included study. In the event of a disagreement, a third reviewer (JL) was consulted. Studies were included in the review regardless of study quality.
-
The relevance of the included studies to the review question, which is the extent to which the review finding is applicable to the context (perspective or population, setting) specified in the review question.
-
The coherence of the review finding, which is the extent to which the pattern across the data that constitutes a review finding, is based on evidence that is consistent across multiple individual studies and/or incorporates convincing explanations for the patterns of evidence in the underlying studies, including explanations for variations across individual studies. The coherence of each review finding was assessed by looking at the extent to which a clear pattern across the data was identified and was contributed to by each individual study, and whether or not this was consistent across multiple contexts.
-
The adequacy of the data contributing to a review finding that refers to an overall determination of the degree of richness and quantity of data supporting a review finding.
Confidence in a finding will be weakened when:
-
the included studies have important methodological limitations
-
the contexts of the primary studies underlying a review finding are substantively different from the context of the review question (relevance)
-
variation is found across data from individual studies and there is no convincing explanation for this variation (coherence)
-
a review finding is supported by data from only one or few primary studies, settings or relevant groups, or the data supporting a finding are very thin (adequacy).
After assessing each of the four components, a judgement was made about the overall confidence in each review finding. Confidence was judged as high, moderate, low or very low. The starting point of ‘high confidence’ reflects a view that each review finding should be seen as a reasonable representation of the phenomenon of interest unless there are factors that would weaken this assumption.
Data extraction strategy
Data extraction from included qualitative studies was undertaken by one reviewer (AS) using a data extraction tool adapted and tailored for the purpose of this qualitative review. All data extractions were checked by a second reviewer (AB), with any discrepancies being discussed by both reviewers. Where data for included studies were missing, reviewers attempted to contact the authors at their last known e-mail addresses. For the purpose of data extraction, two principal approaches to decide what counts as qualitative evidence have been proposed. 175 In the first, only data from primary studies that are illustrated by a direct quotation from the respondent are extracted, whereas, in the second, all qualitative data identified in the primary studies and relevant to the review question are extracted. Given the anticipated paucity of relevant evidence (i.e. evidence relating specifically to behavioural modification interventions in primary care rather than perceptions and attitudes towards general management of MUS), the latter, more inclusive, approach to data type was adopted, together with a selective approach to extract data relevant to the specific research question. A framework for extraction was developed which focused specifically on data relating to the review question. This framework allowed the data to be extracted into broad themes relating to the research question, as illustrated in Table 5. Within these broad themes, more specific subthemes were then generated by coding the data. Although in a number of papers the authors had coded the data and had arranged it in themes, these themes were not always used in this review; in some cases, these themes were adapted and in other cases data were subsumed into other, different, themes.
| Source of data | Theme |
|---|---|
| Data from patients | Positive factors relating to behavioural modification interventions as reported by patients/what did you gain from being referred to a behavioural modification intervention? |
| Factors reported as important, wanted or expected in behavioural modification interventions | |
| What patients did not like about (being referred to) behavioural interventions | |
| Barriers – why patients did not want behavioural interventions/or could not engage with them | |
| Neutral effects (neither benefits or harms) | |
| Data from HPs | Positive factors relating to behavioural modification interventions as reported by HPs/what did you gain from being trained to deliver and/or delivering a behavioural modification intervention? |
| Factors reported as important, wanted or expected for training in and delivery of behavioural modification interventions | |
| What did not help or was detrimental to the patients or delivery of the intervention | |
| Barriers – from the perspective of HPs as to why patients did not want behavioural interventions/or could not engage with them, or barriers to delivery of the intervention | |
| Neutral effects (neither benefits or harms) |
Data synthesis strategy
Qualitative evidence synthesis was undertaken to provide added value to the quantitative analysis by indicating patient and service provider issues around the acceptability of interventions. Specifically, thematic synthesis was used to aggregate the findings. 176 The framework developed for data extraction was used to shape the synthesis of the findings. Themes were then developed within the framework elements.
Combining the quantitative and qualitative data
Methodological work to date has been unable to establish the superiority of conducting the qualitative and quantitative synthesis in parallel or of conducting quantitative followed by qualitative, qualitative followed by quantitative or some more iterative approach. Our choice of method of combining data was determined by the needs of this particular review, in which the quantitative data were the main focus and the qualitative data were used for their explanatory potential. We, therefore, employ methods similar to those described by Noyes et al. 177 to explore the effectiveness review in the light of supporting qualitative research data.
Results of the qualitative review
Included studies: qualitative review
From the 1735 citations identified from the initial searches, 42 remained after title and abstract sift and these citations were considered at full-paper sift for the qualitative review. Figure 57 shows the flow chart of studies included in the qualitative review. The sifting process resulted in the inclusion of 10 studies at full paper. These 10 papers reported evidence from eight studies. Two papers reported evidence from the same study but from some of the same participants; of these one reported evidence from both the patients and HPs178 and the other reported evidence only from patients. 179 A further two papers reported on the same study, with one paper reporting evidence from patients180 and the other reporting evidence from HPs. 181 Details of studies excluded at full-paper sift together with reasons are shown in Appendix 5. All included full papers were published between 2007 and 2016. A summary of the included studies and their sample and study characteristics can be found in Table 6.
FIGURE 57.
The PRISMA flow diagram for the MUS qualitative review.
Study respondents
Eight studies assessed patients’ attitudes and three studies assessed HPs’ attitudes to the intervention (of these, two studies178,181 reported data from both patients and the HPs who had been involved in delivering the intervention). Altogether, the studies contained qualitative data from 130 patients and from 38 HPs. In terms of HPs, the data were specifically from 24 GPs, eight CBT therapists, three nurse therapists and three nurse therapist supervisors. A summary of HPs’ characteristics is shown in Table 7.
| First author and year of publication | Sample (contributing qualitative data) | Population being treated as described in the study | Data collection | Intervention |
|---|---|---|---|---|
| Burton, 2012157 | 11 patients | MUS | Interview | GP with special interest ‘symptoms’ clinic |
|
Chew-Graham, 2011179 Peters, 2011178 |
3 nurse therapists, 3 supervisors, 46 patients | CFS/myalgic encephalomyelitis | Semistructured interviews | Pragmatic rehabilitation and supportive listening |
|
Dowrick, 2008181 Peters, 2009180 (report different data from the study above) |
24 GPs 23 patients |
MUS | Semistructured interviews | Reattribution (by trained GPs) |
| Gerskowitch, 2015182 | 11 patients | MUPS | Semistructured interviews | CBT and mindfulness-based stress reduction |
| Graham, 2007183 | 6 patients | Somatisation | Self-reports – a series of open questions put to each patient in writing | Group counselling (humanistic) |
| Lewis, 2013184 | 8 CBT therapists | MUS | Semistructured interviews | Individual CBT |
| Morton, 2016185 | 17 patients | MUS | Semistructured interviews | GP with special interest ‘symptoms’ clinic |
| Payne, 2015186 | 16 patients | MUS | Case studies; qualitative data from a survey | The BodyMind Approach™ (Pathways2Wellbeing, University of Hertfordshire, Hatfield, UK) group intervention |
| First author and year of publication | Sample | Population being treated as described in the study | Sex | Age (years) |
|---|---|---|---|---|
| Peters, 2011178 | 3 nurse therapists, 3 supervisors | CFS/myalgic encephalomyelitis | Nurses = 3 women; NR for supervisors | NR for nurses and supervisors |
| Dowrick, 2008181 | 24 GPs | MUS | 16 female; 8 male | Three aged < 35; 14 between 35 and 50 and 7 > 50 |
| Lewis, 2013184 | 8 CBT therapists | MUS | Five female; three male | Mean 43.5 (SD = 9.04) (range 30–60) |
Respondent characteristics
Six studies focused on interventions for MUS/MUPS, one focused on interventions for CFS/myalgic encephalomyelitis and one focused on interventions for somatisation. The majority of patient participants met the criteria for MUS or MUPS (n = 78) as reported in the studies, with 46 patients diagnosed with CFS/myalgic encephalomyelitis and six who met the criteria for somatisation as reported in the studies. Reporting of patient characteristics was limited and incomplete in a number of studies. Details, where reported, are shown in Table 8. The general trend was that more women than men contributed data for the studies, and age varied widely, ranging from 19 to 84 years.
| First author and year of publication | Sample (contributing qualitative evidence) | Population being treated as described in the study | Sex | Age (years) |
|---|---|---|---|---|
| Burton, 2012157 | 11 patients | MUS | NR | NR |
|
Chew-Graham, 2011179 Peters, 2011178 |
46 patients | CFS/myalgic encephalomyelitis | 33 female; 13 male | Patients = mean 46.11 (range 20–73) |
| Gerskowitch, 2015182 | 11 patients | MUPS | 8 female; 3 male | Median age 50 (range 19–60) |
| Graham, 2007183 | 6 patients | Somatisation | NR | NR |
| Morton, 2016185 | 17 patients | MUS | NR | NR |
| Payne, 2015186 | 16 patients | MUS | 10 female; 6 male | Range 19–80 |
| Peters, 2009180 | 23 patients | MUS | 20 female; 3 male | Mean = 53 (range 32–84) |
For HP details, reporting of participant characteristics was also limited across the included primary studies. Again, more of the HPs who contributed data were women and the age of the participants ranged from 30 to 60 years, where reported.
Study setting
All interventions took place in primary care or in the community. One study reported on interventions that were delivered in patients’ homes,178,179 one in a primary care mental health trust,184 one in a primary care psychological therapies service,182 four in GP practices157,180,183,185,187 and one in a community setting. 186
Intervention description and facilitators
Three of the interventions described here were GP delivered; these included reattribution180,187 and special interest ‘symptoms’ clinic. 157,185 These interventions were delivered individually to patients and involved a series of structured consultations. Nurse-delivered interventions were reported in one study; these interventions were pragmatic rehabilitation, a therapist-facilitated self-management intervention178,179 and supportive listening,178 which were delivered individually to patients in their own home. Trained CBT therapists delivered one of the interventions investigated in two studies. 182,184 In one of these studies, the intervention that was delivered by a trained CBT therapist was described as high intensity182 and simply described as CBT in the other study. 184 A psychological well-being practitioner delivered a CBTLI intervention and a trained mindfulness-based stress reduction (MBSR) facilitator delivered a MBSR intervention in one of these studies. 182 All CBT interventions were delivered to patients individually, whereas the MBSR intervention was delivered in a group setting. The final two interventions included were also delivered in a group setting. 183,186 Group counselling (humanistic) was delivered by a facilitator described as a ‘group counsellor’ in one study,183 and the BodyMind Approach intervention was delivered by clinical psychologists together with facilitators who were psychotherapists or art therapists in the final study. 186
The reattribution intervention delivered by GPs180,187 was linked to the MUST trial,108 and the pragmatic rehabilitation and supportive listening interventions delivered by nurse therapists reported by Peters et al. 178 and Chew-Graham et al. 179 were linked to the Fatigue Intervention by Nurses Evaluation (FINE) trial. 111 Both of these trials are included in the quantitative review. A description of the interventions delivered in the included studies can be found in Table 9.
| First author and year of publication | Population being treated as described in the study | Intervention | Intervention provider | Intervention setting | Intervention duration | Individual or group |
|---|---|---|---|---|---|---|
| Burton, 2012157 | MUS | GP with special interest ‘symptoms’ clinic. The consultations were structured to first hear the patient’s experience of illness then to propose and negotiate constructive explanations of physical symptoms. These explanations were used as the basis for simple cognitive and behavioural actions to modify symptoms and their impact | GP | GP practice | Four appointments; the first was of 1-hour duration and the subsequent three lasted 20 minutes | Individual |
|
Chew-Graham, 2011179 Peters, 2011178 |
CFS/ME |
|
Primary care nurses trained to deliver the interventions | Patients’ homes | 90-minute session followed by 1-hour sessions on weeks 2, 4, 10 and 18. 30-minute telephone calls on weeks 3, 6, 8, 12 and 15 | Individual |
|
Dowrick, 2008181 Peters, 2009180 |
MUS | Reattribution – a structured intervention, designed to provide a simple explanation of the mechanism of a patient’s MUS, through negotiation and other features of patient-centred communication, and to be delivered during routine consultations | GP | GP practice | The time since the index consultation ranged from 8 to 55 weeks (mean 32 weeks) | Individual |
|
Gerskowitch, 2015182 Gerskowitch, 2015182 |
MUPS | CBT high intensity | HI-CBT therapist | Primary care psychological therapies service | Weekly 1-hour appointments. The median number of sessions attended was 17 (range 4–25) | Individual |
| CBT low intensity | Psychological well-being practitioner | Fortnightly with homework set between meetings | Individual | |||
| MBSR | Trained MBSR facilitator | An 8-week programme for 2 hours per week | Group | |||
| Graham, 2007183 | Somatisation | Group counselling (humanistic) | Group counsellor | GP practice | 1.5-hour weekly sessions for half a year | Group |
| Lewis, 2013184 | MUS | CBT | CBT therapists with a postgraduate diploma in CBT | Primary care mental health-care trust | Not specified | Individual |
| Morton, 2016185 | MUS | SCI – a structured series of consultations. The SCI comprises four key elements: recognition, explanation, action and learning | GP | GP practice | 3 or 4 consultations over a period of 6–8 weeks. The first lasts 50 minutes and subsequent consultations are shorter (15–20 minutes) | Individual |
| Payne, 2015186 | MUS | The BodyMind Approach, based on a biopsychosocial model derived from dance movement psychotherapy | Clinical psychologist/facilitators (psychotherapists or art therapists) | Community setting | 2 hours for 12 sessions over 8 weeks and other communication over a 12-month period. Total face-to-face contact is 27 hours | Group |
Quality of the included studies
The inclusion criteria for study design for the qualitative review were broad, in that any study presenting qualitative evidence was eligible. Therefore, it was evident at the outset that the majority of the included studies would not meet the quality criteria for qualitative research. Nevertheless, these data are important to assess the perceptions of patients and HPs who are receiving or delivering behavioural modification interventions for MUS. To this end, it is acknowledged that a number of the included studies did not have qualitative research aims as their primary aims. Furthermore, owing to journal word count limits, it may not have been possible to provide the level of detail relating to qualitative methods and analysis in the published papers that would be expected for qualitative research. Therefore, we have used an adaptation of the CASP assessment of study quality for qualitative studies to assess the included studies. This sought to provide some leeway in terms of reporting of the findings in order to account for studies with primary research designs that were not qualitative in nature. Studies were assessed in terms of the extent to which they met each of the seven CASP requirements. If the study met all seven requirements it was assessed as high quality, if it met six of the criteria it was assessed as moderate quality, and it was assessed as low quality if it met only five or fewer criteria. Results are presented in Table 10.
| Question | Yes/somewhat (N = 9 studies), % (n) | |
|---|---|---|
| 1 | Is the study qualitative research or does it provide qualitative data? | 100 (9) |
| 2 | Are the study context and aims clearly described? | 100 (9) |
| 3 | Is there evidence of research reflexivity? | 33 (3) |
| 4 | Are the sampling methods clearly described and appropriate for the research question? | 89 (8) |
| 5 | Are the methods of data collection clearly described and appropriate to the research question? | 89 (8) |
| 6 | Is the method of analysis clearly described and appropriate to the research question? | 67 (6) |
| 7 | Are the claims made supported by sufficient evidence (i.e. Did the data provide sufficient depth detail and richness)? | 56 (5) |
Certainty of the review findings: CERQual assessment
The CERQual assessment relied on the assessment of the methodological quality of each study contributing to the review finding, as assessed by CASP. The relevance of the individual studies contributing to the review finding was assessed by considering both the format of the intervention (whether it was individual or group based) and the facilitator who delivered the intervention (whether a GP, nurse therapist or psychological therapist). As the review inclusion criteria specified that the setting was primary care or community based, this criterion was not considered in the CERQual analysis of relevance. The coherence of each review finding was assessed by considering if all the data contributing to that finding supported the finding and whether or not there were any ambiguities or any plausible alternative explanations. Finally, the adequacy of the data was assessed by considering the richness and number of data supporting each review finding. The synthesis of evidence from patients yielded 22 findings overall. Only one finding was assessed as high confidence, six findings were assessed as moderate confidence, 13 were assessed as low confidence and two were assessed as very low confidence. The evidence from HPs yielded 16 findings overall. Four findings were assessed as high confidence, eight as moderate confidence and four as low confidence. The results of the CERQual assessment are set out alongside each review finding in Tables 11 and 12.
| Meta theme | Theme | Studies (first author) contributing to the finding | CERQual assessment of confidence in the evidence | Explanation CERQual assessment |
|---|---|---|---|---|
| Positive factors relating to behavioural modification interventions as reported by patients/what did you gain from being referred to a behavioural modification intervention. | ||||
| Support | Acceptance | Gerskowitch;182 Chew-Graham;179 Graham;183 Morton;185 Peters;180 Peters;178 Payne186 | High confidence | 6 studies, in general the studies were moderately well done. The finding was seen across most of the studies in the synthesis and across different intervention formats and facilitators. Included studies with thick data. Coherent data across the studies |
| Validation | ||||
| Empathy | ||||
| Being listened to | ||||
| Explanation leading to understanding | Accepting the treatment model | Gerskowitch;182 Chew-Graham179 | Moderate confidence | 2 studies, both of moderate quality, and including thick data. Evidence came from different intervention types, and different facilitators. Coherent data across the studies |
| Understanding the treatment model | Gerskowitch182 | |||
| Learning achieved/gains from the interventions | Self-management techniques | Gerskowitch;182 Graham;183 Morton;185 Payne186 | Moderate confidence | 4 studies of moderate to low quality. Evidence came from different settings. However, there were limited rich data. The data appeared to be coherent across the studies |
| Improved confidence | Graham;183 Payne186 | |||
| Improved communication with family and friends | Graham183 | |||
| Benefits from being part of a group (specific to group interventions) | Sharing experiences with other patients | Gerskowitch182 | Low confidence | 2 studies: one moderate quality with rich data and one low quality with thin data. Data were specific to group interventions. Coherent data across the studies |
| Learning from other patients | Gerskowitch;182 Graham183 | |||
| Factors reported as important, wanted or expected in behavioural modification interventions | ||||
| Balance between psychological and physical elements | Burton157 | Very low confidence | 1 one study of low quality with thin dataa | |
| Flexibility | Gerskowitch182 | Low confidence | 1 study of moderate quality with thick dataa | |
| Other factors such as diet/nutrition advice, the role of faith | Gerskowitch182 | Low confidence | 1 study of moderate quality with thick dataa | |
| Group-based treatment | Gerskowitch182 | Low confidence | 1 study of moderate quality with thick dataa | |
| More consultation time | Burton157 | Very low confidence | 1 one study of low quality with thin dataa | |
| More investigations | Peters180 | Low confidence | 1 study of moderate quality with thick dataa | |
| Explanation | Specialist knowledge of the symptoms | Gerskowitch;182 Peters180 | Moderate confidence | 2 studies, both of moderate quality with thick data. Data evident across different formats, and different facilitators. Coherent data across the studies |
| Explanation of the symptoms | ||||
| HPs’ understanding of the patient and symptoms | Reassurance | Peters180 | Low confidence | 1 study of moderate quality with thick dataa |
| Relationship with the GP | Peters180 | |||
| Learning skills to deal with the symptoms | Self-management techniques | Gerskowitch;182 Peters180 | Moderate confidence | 2 studies, both of moderate quality with thick data. Data evident across different formats, and different facilitators. Coherent data across the studies |
| Support for learning | ||||
| What patients did not like about (being referred to) behavioural interventions | ||||
| It made things worse/did not make any difference (adverse effects) | Frustration – things are not getting any better | Gerskowitch182 | Low confidence | 3 studies of moderate quality. Coherence is unclear as findings within the theme are diverse, thick data, consistent across different formats and facilitators |
| Negativity of the intervention | Gerskowitch182 | |||
| Made symptoms worse | Gerskowitch182 | |||
| Inflexibility of the model/not workable in everyday life | Chew-Graham;179 Peters180 | |||
| HPs’ lack of understanding | HP does not understand | Peters180 | Low confidence | 2 studies, both moderate quality, only one with thick data. Specific to, GP-delivered, individual interventions. Coherent data across the studies |
| Feeling of being blamed by the GP | Morton185 | |||
| It ended | A feeling of loss when the intervention ends/lost a friend | Peters;178 Chew-Graham179 | Low confidence |
1 study of moderate quality with thick dataa This finding was only seen in the one included study of patients with CFS and may only be specific to this group |
| Conflicts between HPs and patients | Intervention asserted as the only right answer | Chew-Graham179 | Low confidence |
1 study of moderate quality with thick dataa This finding was only seen in the one included study of patients with CFS and may only be specific to this group |
| Patients felt the intervention did not address their symptoms | Peters178 | |||
| A belief exercise is damaging despite HP advice | Chew-Graham179 | |||
| Barriers – why patients did not want behavioural interventions/or could not engage with them | ||||
| Lack of choice | Lack of information/choice given by referrers | Gerskowitch182 | Low confidence | 1 study of moderate quality with thick dataa |
| Sceptical of HPs | GPs aim is to assert problems are psychological | Burton157 | Moderate confidence | 4 studies, three of moderate quality, one low quality, three with thick data, across different intervention formats and facilitators. Coherent data across the studies |
| Mention of psychosocial problems diverts GPs attention from other problems (so should not mention them) | Peters180 | |||
| Simplistic explanations | Gerskowitch;182 Peters180 | |||
| Disagreements with HPs that illness is physical | Peters178 | |||
| Physical limitations | Cannot physically undertake the intervention | Gerskowitch182 | Low confidence | 1 study of moderate quality with thick dataa |
| HPs’ lack of knowledge/skill at treating the symptoms | GPs are unskilled | Peters180 | Moderate confidence | 2 studies, both moderate quality and with thick data, across different intervention formats and facilitators. Coherent data across the studies |
| Nurse therapists are novices | Peters178 | |||
| Patients beliefs that you should deal with it yourself | Patients should self-manage | Peters180 | Low confidence | 1 study of moderate quality with thick dataa Specific to a GP-delivered intervention |
| Inappropriate to discuss psychosocial problems (with GP) | Peters180 | |||
| Stigma related to psychosocial problems | Peters180 | |||
| Meta theme | Theme | Studies (first author) contributing to the finding | CERQual assessment of confidence in the evidence | Explanation CERQual assessment |
|---|---|---|---|---|
| Positive factors relating to behavioural modification interventions as reported by HPs/what did you gain from being trained to and/or delivering a behavioural modification intervention? | ||||
| Training and supervision was useful | Supervision in managing tensions | Peters178 | Moderate confidence | 2 studies, both of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Therapist peer support | Peters178 | |||
| Increased GPs’ awareness and altered perceptions | Dowrick181 | |||
| Addressed professional and/or training needs | Dowrick181 | |||
| Extended knowledge and skill of the HP | Learning and applying boundaries of their role | Peters178 | High confidence | 3 studies: one high quality and two of moderate quality. All three including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Flexibility of the therapist | Peters;178 Lewis184 | |||
| Increasing confidence in discussing MUS | Dowrick181 | |||
| Helped structure MUS consultations | Dowrick181 | |||
| The primary care/community setting was helpful | Secondary care may support a belief in a physical cause | Dowrick181 | Moderate confidence | 2 studies, both of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Tool to protect patients (from secondary care) | Dowrick181 | |||
| Primary care allowed a more tailored approach | Peters178 | |||
| Being in patients’ homes was helpful in building a therapeutic relationship | Peters178 | |||
| Rewarding experience | Rewarding when patients engaged | Peters;178 Lewis184 | High confidence | 2 studies, one high quality, one of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| What did not help or were detrimental to the patients or delivery of the intervention? | ||||
| When HPs felt that they were a novice or did not have the required skill levels | Being a novice therapist/lack of experience | Peters;178 Lewis184 | High confidence | 2 studies, one high quality, one of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Anxiety because of a lack of training/knowledge | Lewis184 | |||
| Unfamiliarity with MUS | Lewis184 | |||
| No specific MUS model to work with | Lewis184 | |||
| Not as comfortable with mental health aspects of the interventions (as physical health) | Peters178 | |||
| When HPs struggled to deal with their own or patient emotions | Not dealing well with failure | Peters178 | Moderate confidence | 2 studies, both of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Pessimism (from HP) about dealing with MUS | Lewis184 | |||
| Angry patients | Peters178 | |||
| HPs’ anxieties that the intervention may have detrimental consequences for the patients | Nurse therapists worries about the interventions consequences | Peters178 | Moderate confidence | 2 studies, both of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Could increase patient dependency | Dowrick181 | |||
| Too much complexity | Difficulty of applying when the patients have multiple symptoms/complaints | Dowrick181 | Low confidence | 1 study of moderate quality with thick dataa |
| The community setting was unhelpful | Difficulties of being in the patients’ home | Peters178 | Low confidence | 1 study of moderate quality with thick dataa |
| It’s nothing new (neutral effects) | Ride it out (when the intervention is not addressing needs and both parties know it) | Peters178 | Moderate confidence | 2 studies, both of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Nothing new (HPs felt that they were already doing it) | Dowrick181 | |||
| Barriers – from the perspective of HPs why patients did not want behavioural interventions/or could not engage with them, or barriers to delivery of the intervention | ||||
| Resource constraints | Not enough time | Peters;178 Dowrick181 | Moderate confidence | 3 studies, all of moderate quality, both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Other time constraints/competing pressures | Dowrick181 | |||
| Other pressures go against applying reattribution (medicolegal) | Dowrick181 | |||
| Lack of clarity on service provision for MUS | Lewis184 | |||
| Patient beliefs | Patients’ belief in a physical cause | Dowrick181 | High confidence | 3 studies: one high quality and two of moderate quality. All three including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Patients benefit from their symptoms (an agenda – secondary gain) | Dowrick181 | |||
| Patients have an agenda (e.g. target the locum)/patients’ pre-existing beliefs/an agenda | Dowrick;181 Lewis184 | |||
| Patients’ beliefs about not being able to do physical activity | Lewis184 | |||
| Patients’ resistance to intervention/patients did not want to engage with the model | Peters;178 Lewis184 | |||
| HP skill, beliefs and attitudes | Some GPs more skilled than others | Dowrick181 | Low confidence | 1 study of moderate quality with thick dataa |
| GPs’ prior expectations of patients | Dowrick181 | |||
| GPs’ mood | Dowrick181 | |||
| Personality clash (between HPs and patients) | Dowrick181 | |||
| Patient barriers | Patients not able/willing to share | Dowrick181 | Moderate confidence | 2 studies: one of high quality and one of moderate quality. Both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| Patients’ physical difficulties were barriers to attendance and adherence | Lewis184 | |||
| Other factors | Not knowing if it is actually MUS | Dowrick181 | Low confidence | 1 study of moderate quality with thick dataa |
| Factors reported as important, wanted or expected for training in and delivery of behavioural modification interventions | ||||
| More training and supervision required | Supervision | Dowrick;181 Lewis184 | Moderate confidence | 2 studies: one of high quality and one of moderate quality. Both including thick data. Evidence came from different intervention types with different facilitators. Coherent data across the studies |
| More training | Lewis184 | |||
Synthesis of patient and health professional evidence
Findings were synthesised across all intervention types and organised in accordance with the questions outlined in the data extraction framework items as detailed in the methods section. The findings of the synthesis have been summarised and are presented in Figure 58 to demonstrate the expectations, perceived barriers and facilitators, and potential outcomes of the interventions. Thus, the figure represents the experience of having received the interventions rather than the anticipation of them. Within these, a number of important themes emerged. Each metatheme, together with subthemes where applicable, with examples and an estimate of the strength of the evidence, is presented in Tables 11 and 12. The themes are synthesised further within the following narrative synthesis. As illustrated in Table 11, the synthesis of evidence from patients yielded 22 findings overall. Only one finding, ‘support’, was assessed as high confidence and was derived from six studies, in which the quality was judged to be moderate, and the finding was seen across most of the studies in the synthesis and across different intervention formats and facilitators. Six findings were assessed as moderate confidence, with these each including evidence from more than one study, with some thick data. In addition, the overall quality was at least moderate, with coherence across different studies. Thirteen findings were assessed as low confidence and two as very low confidence, with evidence coming from only one study for each finding and, therefore, this paucity of evidence should be taken into account when interpreting the synthesis findings. The evidence from HPs yielded 16 findings overall. Four findings were assessed as high confidence, with evidence emerging from two or more studies, including some thick data. They were of moderate quality, and had coherence across studies. Eight were assessed as moderate confidence, with evidence coming from at least two studies (of moderate quality), including thick data and with coherence across studies. Finally, four studies were assessed as low confidence, with evidence only coming from one study for each finding and, again, the paucity of the evidence for these findings should be taken into account during interpretation.
FIGURE 58.
Patients’ and HPs’ perceptions of behavioural interventions for MUS in primary care settings. Model to show expectations, perceived barriers and facilitators, and potential outcomes, based on evidence from included studies.
Factors identified as important in primary care-delivered behavioural modification interventions for medically unexplained symptoms from the perspective of patients who had received the interventions
Support
Support was defined by patients as being accepted and validated by HPs and fellow intervention participants, receiving empathy and being listened to. This was found across an array of intervention types and there was high confidence in this finding. 178,180,182,183,185,186 This type of support was described as the most positive part of the pragmatic rehabilitation intervention in one study, with ‘being believed and feeling understood by the therapist’ as a key part of the intervention. 179 Patients across various intervention types reported valuing a feeling of being understood and validated by the HP. 179,182 This was sometimes as a result of the knowledge that the HP had about the patient’s symptoms, which gave them a sense of having someone ‘on their side’. 180,182,185 In some instances, it appeared that no one else in their lives was able to provide such support, ‘It was just an understanding from her that I didn’t, haven’t had from anybody else’ (patient). 179 One patient reporting that the empathetic nature of the nurse delivering the supportive listening intervention was their most valued attribute. 178 This feeling of being understood also came from fellow participants with similar symptoms, as part of group interventions. 182,183,186 These factors were reported as key to patient engagement and contributed to whether or not the interventions were reported as acceptable to the participants. 179 Some participants thought that the ‘extra time’ afforded by the interventions allowed these aspects of support to be accessed. 185
Explanation
Patients also clearly valued explanations that they had been provided with as part of the interventions, and reported that this led to accepting and understanding the treatment model,179,182 with moderate confidence assessed in this finding. Patients felt that the pragmatic rehabilitation intervention helped them to come to terms with and accept a diagnosis (of CFS) and that having an explanation and understanding of their symptoms was key to this. ‘She explained all about CFS and the physiology of it really, which was the first time really that I understood why my energy was so low, so that made a lot of sense’ (patient). 179 Gaining new knowledge about their symptoms was reported to be reassuring to patients, enabling them to accept the idea that the pragmatic rehabilitation intervention might be appropriate for them and, therefore, facilitated the progress made with the intervention. 179 Similar factors were also identified in the data about what patients wanted from an intervention. Patients reported that they consulted their GP to seek an explanation for their symptoms180 and took part in interventions to see someone who had specialist knowledge and information about their symptoms. 182
Important intervention elements for the delivery and success of primary care-delivered behavioural modification interventions for MUS from the perspective of patients and health professionals
Intervention elements
An important feature of the evidence from patients was the reported benefit that they had gained from the behavioural self-management techniques that they had learned. This was evident across four of the studies that included the following interventions, with participants commenting on their improved confidence and improved communication with significant others: CBT, mindfulness, group counselling, GP interventions and the BodyMind Approach intervention. 182,183,185,186 This also emerged in the evidence around what was wanted or needed as part of any future intervention, with patients saying they needed to learn skills such as self-management techniques to deal with the symptoms and to have the support for learning such techniques. 180,182 Specific to group interventions, patients found sharing their experiences with other patients182 and learning skills from other patients182,183 valuable as part of the intervention.
A number of additional elements were suggested by patients in single studies as potentially important as part of any intervention for MUS. These elements included having a balance between the psychological and physical elements of an intervention in one GP-delivered intervention study,157 and incorporating other factors, such as diet and nutrition advice and the role of faith, in the management of symptoms in another study. 182 Patients also reported that flexibility in accessing the therapy was important to them in one study,182 as was the availability of group-based treatment. 182 Specific to a GP-delivered study, patients wanted more consultation time,157 which, in comparison with standard care, most of the behavioural interventions detailed here provided. Again specific to a GP-delivered intervention, some patients simply wanted to have further investigations. 180
Health professional training and supervision
From the point of view of the HPs, training and supervision in applying behaviour modification interventions was reported as helpful in two studies178,187 and was also recommended as useful for future training in two studies. 184,187 Supervision was reported as ‘fundamental’ and as key to resolving difficulties that might arise between the deliverer of the intervention and the patient receiving the intervention in a nurse therapist-delivered intervention. 178 Having the support of peers who were delivering the same intervention was also valued, ‘If we were having a really difficult time with a certain patient, then we would sort of pool ideas, and ask advice how they would cope with it, the other nurse therapists or what do they think is going on’ (nurse). 178 GPs delivering a reattribution intervention found the training useful in that it increased their awareness of MUS and suggested it ‘altered my perception a bit, it’s easy to get stale and view that group of patients as difficult or troublesome or irksome at times because we’re not always at our best every time (GP). 187 GPs also reported that it addressed their training needs and allowed them the opportunity it compare consultation skills with colleagues. 187
Primary care and community setting
The primary care or community setting was reported to be a helpful factor by the HPs in two studies. In one study, the reasons for this were that the primary care setting provided a tailored approach, and the setting (e.g. being in patients’ homes) was useful when trying to develop a therapeutic relationship. 178 In another case, this contrasted with the perceived failing of the secondary care setting. 187 ‘You see these people getting referred to the hospital with back pain and the next thing you know some bright spark is going to operate on them and you think ‘What!’ . . . Maybe we’re here in a way as a gateway to try and prevent harm as well as anything else’ (GP). 187 This appears to indicate that primary care practitioners see part of their role as protecting their patient from entering secondary care when they believe this is unnecessary. Although the evidence pointed towards favourable views regarding the primary care and community setting, it should be considered that the home setting could also be a source of difficulties (e.g. issues with privacy and interference from family members), but this finding was specific to the intervention for CFS/myalgic encephalomyelitis. 178
Important facilitating factors for intervention success by patients and health professionals
Good relationships
It was important that the HPs delivering an intervention had an understanding of the patients and their symptoms. Specifically, the patients in one study reported needing reassurance and a good relationship with the person delivering the intervention. 180 Patients liked the fact that they had a longstanding relationship with the HP. In a reattribution intervention, this gave them a feeling that they knew what type of treatment, if any, would be suitable for them: ‘He knows I like to keep myself to myself, knows I believe in self-help . . . knows I’m somebody that likes to work it out for myself’ (patient). 180 Others simply appreciated the understanding and reassurance they received from the HP and just wanted an opportunity to share their difficulties. 180
Health professional knowledge and skill
Health professionals across three different interventions reported that they found being involved in delivering behaviour modification interventions helped them to develop their knowledge and skills in the area of MUS. 178,184,187 This training helped them learn about the boundaries of their own role178 and to be flexible when delivering interventions. 178,184 GPs reported being able to structure consultations in an appropriate way for MUS and that the intervention, in this case reattribution, provided this structure and increased their confidence in discussing MUS with patients. 187 Although acknowledging that there can be difficulties in delivering MUS interventions, CBT therapists184 and nurse therapists178 both reported that it can be rewarding when it is evident that patients have gained benefit from the intervention.
Factors identified as barriers to intervention success by patients and health professionals
Difficult relationships
At the same time as good relationships facilitating intervention success, difficult relationships between patients and HPs appeared to undermine progress, and this was reported across five studies. Some patients, taking part in a reattribution intervention, felt that there was a lack of understanding on the part of the HP, with simplistic causes and resolutions to problems being proffered. 180 Sometimes, reported specifically in one GP-delivered intervention study, this extended to patients feeling that they were being blamed by the GP. 185 Specific to the intervention study for patients with CFS/myalgic encephalomyelitis, the relationship between HPs and patients also sometimes led to conflict, in some cases patients felt the HPs were asserting the intervention (in this case pragmatic rehabilitation) as the only right answer, and patients disagreed with this point of view, ‘I think my main reason is the fundamental theory behind it [the treatment model offered] just disregards it as illness’ (patient). 179 There was also a feeling from patients that some heath professionals were unskilled or lacked the knowledge to deal with their symptoms or provide appropriate intervention. 178,180
Patients in a GP-delivered intervention study also reported being sceptical of HPs, believing that the aim was to assert that their problems were psychological,157 and some patients reported having disagreements with their HP regarding the physical nature of their symptoms in an intervention for patients with CFS. 178 Patients were also sceptical that they would only be provided with simplistic explanations for their symptoms,182 and that if they were to mention any psychosocial problems this would divert the HP’s attention away from other problems associated with their symptoms. 180 Specific to a GP-delivered intervention, GPs reported that there might be a personality clash between them and their patients, meaning that insufficient rapport is built for reattribution to be successfully delivered. 187 Other factors reported by HPs included patients not being able or willing to communicate with their GP. 187
Attitudes and beliefs
In one study, some patients simply felt that the intervention did not address their symptoms and that they could not believe in the intervention,178 contrary to the HP view. It was also reported that several patients held a belief that an activity (as part of the pragmatic rehabilitation intervention) could be damaging to them, despite the fact that it was described in the intervention manual and recommended by the HP delivering the intervention. 179 Patients reported that the lack of information given to them by those who referred them meant that they did not know the remit of the intervention, and it was also reported that patients felt that the referring GP did not know what the intervention entailed. Patients were therefore surprised about the content of the intervention and found it difficult to see why they had been referred to a psychological therapy-based intervention when their problems were physical. 182 Furthermore, evidence suggested that some patients may hold the belief that they should deal with the symptoms themselves, and that they should self-manage without the input of HPs. 180 Patients had their own ideas around what they should be doing to self-manage: ‘I’ve started knitting – gets my mind occupied’ (patient). 180 Further barriers came from an opinion that it is inappropriate to discuss psychosocial problems (with the GP)180 and the idea of there being a stigma related to reporting psychosocial problems. 180 In addition to these beliefs, patients also reported physical limitations that represented barriers to physically undertaking the intervention. 182
Health professionals also reported that they felt patient beliefs could be significant barriers to engagement with behavioural modification interventions. 178,184,187 HPs described, and some appeared to expect, patients to have a belief in a physical cause. 187 HPs were reported in one reattribution study to suggest that this belief was a learned behaviour to deal with unhappiness. 187 HPs also reported in two studies that patients could be resistant to the intervention or the treatment model itself. 178,184 Furthermore, CBT therapists reported that patients sometimes held a belief that they were not able to take part in aspects of interventions that required physical activity, despite HPs’ reassurances otherwise. 184 Finally, it was suggested by HPs that patients might be benefiting materially from their MUS symptoms, in terms of family support and state benefits, which might have an impact on their willingness to engage with an intervention. 187
Health professionals’ lack of experience and skills
Nurse therapists and CBT therapists who had been trained to deliver MUS-specific interventions felt that they were novices and lacked the experience to deliver the interventions in two separate studies. 178,184 It was reported that they did not have a specific MUS model to work with, and they were unfamiliar with the symptoms. 184 Furthermore, they experienced anxiety because of this lack of training and experience in MUS184 and were not comfortable dealing with the mental health aspects as compared with the physical health aspects of the interventions they were delivering. 178 This also emerged as a barrier to service provision, with GPs acknowledging that the successful application of reattribution may be dependent on variations in the communication skills of the practitioner delivering the intervention. 187
Difficulty of dealing with emotions
Patient and HP emotions emerged as a source of difficulty in delivering MUS interventions in two studies. 178,184 Nurse therapists found at times that they did not deal well with failure, ‘One common theme I think with has come up is the difficult of accepting that you can’t get it right all the time’ (supervisor). 178 Sometimes patients became angry when an intervention could not address their needs, and this was also a source of difficulty for nurse therapists. 178 Conversely, some CBT therapists reported a level of pessimism when faced with patients with MUS. 184
Constraining factors
Health professionals reported that resource constraints, such as time constraints, and impositions about what data needed to be collected during a consultation presented barriers to delivering interventions in two studies. 178,187 The concerns regarding time constraints were associated with each consultation187 and also with the length of the course of therapy, with concerns expressed that too few sessions were available to deal with deep-seated issues. 178 Other constraining factors reported by HPs included physical difficulties preventing patients attending and adhering to interventions. 184
Problems of diagnosis
The GPs were also concerned about medicolegal issues in one study, such as overdiagnosing and overtreating patients. 187 It was also reported that a lack of information about service provision may have lowered the referral rate in another study. 184 Other constraining factors included situations in which HPs were unsure that the symptoms were actually medically unexplained,187 or in which there was too much symptom complexity. 187
Potential unwanted or adverse effects on patients as reported by patients and health professionals
The HPs in two studies reported that they had concerns that the intervention might be detrimental to patients. 184,187 These worries took the form in one case with a reattribution intervention that patients might form a dependent relationship with the GP and become reliant on them. 187 In another case, these worries took the form of the impact of the intervention (pragmatic rehabilitation and supportive listening) possibly being negatively felt in wider aspects of the patients’ lives. 184
Some patients considered that the psychological interventions made things worse, or were not making things any better in one study. 182 This manifested itself in the form of frustration in some cases,182 whereas, in others, patients found the intervention model negative in nature. 182 Some patients with CFS found the pragmatic rehabilitation intervention model inflexible and, therefore, were unable to make it work in everyday life. 179,180
Where good relationships were established between the two parties, as in a study of an intervention for CFS/myalgic encephalomyelitis, patients reported finding it difficult when the intervention came to an end, experiencing a feeling of loss when support was withdrawn. 178,179
In some cases, it was felt that the intervention was not suited to the patient and that both the HP delivering the intervention and the patient receiving the intervention had come to this realisation but both parties followed through with it despite this. 178 In another case, HPs felt that the intervention training on reattribution offered them nothing new and that they were already providing it in their consultations. 187
Table 13 shows a summary of the key findings.
| Patients | HPs | |
|---|---|---|
| Valuable results from intervention participation or training | Gaining support being validated | Developing knowledge and skills around MUS |
| Explanation | ||
| Support for self-management | Rewarding when patients engaged | |
| Facilitators | Good relationships between patients and HPs | Training and supervision |
| Primary care or community setting | ||
| Barriers | Patient attitudes and beliefs | Lack of confidence in their own skills and abilities to deal with MUS |
| Poor relationship with HPs | Patient attitudes and beliefs | |
| HP attitudes and beliefs | ||
| Resources constraints | ||
| Adverse effects | It ended | Detrimental intervention consequences for patients |
| Conflicts between HPs and patients, and HPs’ lack of understanding | Patient and HP emotions |
Discussion of the qualitative review
This review presents patients’ and HPs’ perspectives on the acceptability, relative benefits and potential harms of primary care or community-based behaviour modification interventions for MUS. The findings offer insight into what aspects of the interventions patients and HPs found helpful, what was unhelpful and what were the barriers to participation or intervention success, which could all contribute towards an understanding of what might make a more successful intervention.
Although the quality of the study reports included in the review was not universally high, studies were in general suitable for the purpose of the review. Although the quality of the studies would have been increased by limiting our review to studies that could be identified as qualitative research (i.e. using both accepted methods of qualitative data collection and data analysis), important data may have been missed had this criterion been applied.
The findings of the synthesis showed that, across all interventions, what patients thought was helpful was gaining support. Patients found being accepted and validated by the HP (and also other intervention participants, in the case of group interventions), together with receiving empathy and being listened to, helpful as part of an intervention. Patients also valued receiving an explanation for their symptoms and appeared to find interventions that offered this to be helpful. In terms of practical intervention elements, an important feature was the gains made from the self-management techniques they had learned, and patients reported that support for learning such techniques was important. The progress and perceived helpfulness of the intervention appeared to be facilitated by a good relationship between patients and the HPs delivering the intervention. In terms of findings relating exclusively to GP-delivered interventions, patients reported that helpful elements of an intervention included having a balance between psychological and physical elements,157 more consultation time157 and, in one study, further investigations. 180
Patients’ own attitudes and beliefs appeared to be potential barriers to their participation in the intervention and its success, with some patients reporting that they did not believe in the treatment model or found it unworkable, or had a belief in dealing with problems themselves. It also appeared that conflicts between HPs and patients could impede success. A small number of findings related exclusively to the intervention study for patients with CFS/ME. Patients reported finding it difficult when the intervention came to an end, experiencing a feeling of loss when support was withdrawn. 178,179 They also reported that the relationship between HPs and patients could sometimes lead to conflict – in some cases patients felt that the HPs were asserting the intervention (in this case pragmatic rehabilitation) as the only right answer and disregarded the patient’s point of view. 179 Specifically relating to GP-delivered interventions, patients reported that they felt that there was a lack of understanding from the GP,180 and they also reported a feeling of being blamed by the GP. 185 Again relating to GP-delivered interventions, the evidence suggested that some patients may hold a belief that you should deal with the symptoms yourself and that they should self-manage without the input of HPs. 180
Evidence from HPs described how they found being trained or delivering the interventions helped to develop their own knowledge and skills around MUS, and they reported that the experience was rewarding when patients engaged with them. They found important facilitators were training and supervision for the interventions and found primary care or the community an appropriate and helpful setting for the delivery of the intervention. Barriers to successful delivery of the intervention(s) were reported to include a lack of confidence in their own skills and abilities to deal with MUS, patient attitudes and beliefs that went against the intervention, their own attitudes and beliefs and resource constraints. HPs were also worried that the interventions might have inadvertent detrimental consequences for patients, and that they might be ill-equipped to deal with their own and the patient’s emotions. Although the evidence indicated favourable views regarding the primary care and community setting, it should be considered that the home setting could also be a source of difficulties (e.g. issues with privacy and interference from family members), but this finding was specific to the intervention for CFS/myalgic encephalomyelitis. 178
Barriers reported specifically from GP-delivered interventions included GPs feeling that there was too much condition complexity,187 GPs’ skills, beliefs and attitudes sometimes not helping,187 and GPs, on some occasions, not knowing if it was actually MUS did not help in the delivery of the intervention. 187
The implications of the findings suggest that, although a number of patients found the interventions helpful, with a number of key helpful factors that could inform the development of a future intervention, a minority did not find the intervention helpful or did not want to take part in the intervention at all. This links to one of the barriers to taking part in or engaging with interventions, which was reported as a lack of information and choice given by referrers. If patients do not have the information that they need to make an informed choice, it is less likely that the intervention will be suitable for them. Therefore, this requires consideration of the matching of intervention type to patients where possible. Other barriers that would need to be addressed in any future intervention include patients’ scepticism of HPs, with patients reporting disagreements with HPs regarding the nature of their symptoms, and an expectation that simplistic explanations will be provided. Patients also reported in some cases that they were not convinced that the HPs delivering the intervention had an appropriate level of knowledge and skill. Further barriers included physical limitations, problems of stigma around asking for help and a belief in dealing with problems oneself. A further consideration for all interventions, but specifically evidenced here from the patient evidence on the intervention for CFS/myalgic encephalomyelitis,178,179 is continuity of care from the same HP or team and that interventions should not end suddenly or without adequate follow-up.
Limitations
Limitations include the fact that some of the included studies were of moderate or low quality, as assessed using CASP,174 and that confidence in some of the findings was assessed as moderate or low using the CERQual assessment. 169 Word limits imposed by journals may have contributed to this. Such limitations may result in a lack of rich data consistently across all studies, limiting (to some degree) the interpretations that can be made, particularly for some of the minor themes identified. Although the metathemes identified were supported by a number of studies and were supported by the CERQual assessment, a number of other findings reported here did not offer the same strength of evidence. In addition, although the validity of these findings should not be understated, it may be the case that further research is required to ascertain their generalisability and importance in the development of future interventions. A further limitation concerns the inclusion criteria for the qualitative review. The scope of the qualitative review was to present evidence about the experience of the interventions rather than the anticipation of them; therefore, the data included were in the main from patients who had agreed to take part in an intervention. Evidence about the general management of MUS was not included in the qualitative review but these data did meet the inclusion criteria for the realist synthesis.
Summary of the findings of the qualitative review
-
Both patients and HPs reported positive gains from taking part or delivering interventions.
-
A number of reported barriers appeared to be underpinned by the relationship between the patients and HPs delivering the intervention, and by beliefs and attitudes held by both parties.
-
Both positive and negative experiences of the primary care setting were reported.
-
Detrimental effects of the intervention were associated with an abrupt end to the intervention and frustration attributable to symptoms not improving.
-
Some patients and HPs did not feel it could add anything for them.
Combining the quantitative and qualitative findings
The quantitative review did not consistently reveal one intervention as more successful than any other across the various outcomes, with the exception of GP-delivered interventions, which were rarely successful. Therefore, as suggested in the qualitative evidence, considered matching of intervention type to patients, taking into account patient preferences, should take place. Further to this, the qualitative evidence showed that patients’ own attitudes and beliefs may be barriers to intervention participation, with some patients reporting that they did not believe in the treatment model, found it unworkable or had a belief in dealing with problems themselves, again giving support for careful matching of intervention to patient. Although not frequently reported in the trials of clinical effectiveness, this theme was sometimes reflected in the numbers of eligible participants declining to take part, with reasons including beliefs that the interventions would not work, or would exacerbate symptoms.
Although the quantitative findings did not show that the GP-delivered interventions were generally effective, the qualitative review demonstrated that patients found that the GPs’ understanding of them and their symptoms and a good relationship with them was an important part of a successful intervention. However, the qualitative review also demonstrated that a key barrier for patients taking part in interventions, including those that were GP delivered, was a scepticism regarding the deliverer. Specifically, some patients reported disagreements with GPs regarding the physical or psychological nature of their symptoms and also that sometimes simplistic explanations were proffered by GPs.
Evidence outlined in the review of cost-effectiveness demonstrated a large variation in the number of patients recruited by GPs between studies. Evidence from the qualitative review showed that some HPs found delivering the interventions to be a rewarding experience, and it might therefore be assumed that such HPs would be more likely to recruit patients. Others had a contrasting view, that the training they were given for MUS interventions was nothing new, and some reported that they found it difficult to deal with failure and were ill-equipped to deal with angry patients. Some HPs worried that it could increase dependency and that it would be difficult to apply when symptoms were complex. Other factors that may prevent HPs from recruiting to interventions include resource constraints, medicolegal concerns and a lack of clarity around service provision. HPs were also concerned that it was not always clear that it was actually MUS that the patients had, and they wanted more training and supervision.
Findings from the cost-effectiveness review also demonstrated a large variation in the number of sessions the patients attended both within and between studies. Evidence from the qualitative review may provide explanations for this; for example, there were a number of things that patients felt that they did not like about the intervention or were barriers to them taking part or continuing with it. These factors included frustration that things were not getting any better after attending and finding the rationale of the intervention negative and that they could not work it into their everyday life. They also reported feeling that the HP did not understand, or that the HP was blaming them. There were also conflicts reported between HPs and patients, and a reported lack of understanding from the HP. This may include patients losing faith in the intervention when it was asserted as the only right answer to their problems and that their needs were not addressed.
Chapter 5 The contribution of contextual factors to the success or failure of behaviour modification interventions for medically unexplained symptoms: realist synthesis
This chapter aims to provide an overview and analysis of the evidence for the contribution of contextual factors associated with the ongoing primary care consultation and the patient’s interaction with primary care professionals to the success or failure of behaviour modification interventions (‘behavioural interventions’) for MUS. Given the complexity of the phenomenon and the interaction between HPs for a variety of purposes and at multiple levels, the review team has undertaken a review utilising realist principles. Realist reviews particularly offer a lens or strategy by which complex interventions can be examined and analysed.
This realist synthesis on behaviour modification interventions for MUS draws on three main sources:
-
study ‘clusters’ of primary studies, linking UK quantitative and qualitative reports, identified from the effectiveness review and from the qualitative synthesis, to all other published accounts that share a study context
-
the qualitative synthesis conducted for this HTA, which focused on UK intervention studies only
-
published narrative and systematic reviews, particularly qualitative evidence syntheses, covering the entire primary care pathway from making an appointment through to potential referral to secondary care.
In addition, further systematic reviews, qualitative research and conceptual papers were examined as referenced by any of the above, offering productive ‘forays’ into the relevant literature.
Evidence identification
Study ‘clusters’ of primary studies
When exploring complex interventions, a review team must seek to gain as full an understanding as possible of the essential characteristics of the context and the intervention. One possible way of acquiring such an understanding is by examining papers and additional studies that have been conducted alongside an effectiveness study, either as part of an integrated mixed-methods study or as a ‘sibling study’. Sibling studies may include qualitative research studies, economic evaluations or process evaluations associated with specific RCTs, whereas additional papers may offer commentaries on the programme theory or contextual issues. Sibling and associated studies are particularly valuable because they offer additional insights beyond the reporting limitations of the key index paper ‘in explaining contextual factors that influence implementation and/or outcomes’. 58 The systematic searches reported in Chapter 3 identified 59 RCTs, of which 15 were UK-based trials and 10 were UK-based qualitative intervention studies. From this initial yield of 24 citations, including one mixed-methods paper that featured in both sets, 19 ‘clusters’ were formed to reflect a proportion of overlap of three RCTs (five papers) with qualitative papers. Each of the 24 citations was searched using the ‘cited by’ facility on Google Scholar (Google Inc., Mountain View, CA, USA) via Publish or Perish software (Harzing.com, London, UK)188 to retrieve a total of 1182 citations. These were manually checked to identify associated study reports. Other methods of identification included author searching, checking of websites, and trial name and number searching. A total of 80 additional papers were identified to augment the original 24 papers, making a total of 104 papers spread across 19 different clusters (see Appendix 2, Table 69).
Qualitative synthesis
Findings from the qualitative synthesis conducted for this HTA (see Chapter 4), which focused on UK intervention studies only, were examined in addition to details of the interventions and contexts with additional data being extracted to templates for reporting interventions (see Appendix 6).
Review-level evidence
Qualitative syntheses on MUS and associated conditions were identified from the study register of the Cochrane Qualitative and Implementation Methods Group together with systematic reviews and narrative reviews (see Appendix 2, Table 68) identified from the PubMed Reviews subset.
Formulate review question
The realist review was framed within the overall study objectives for the HTA as a whole. However, within this broader aim it sought to address the following specific question:
To what extent do contextual factors associated with the ongoing primary care consultation between the PCP (GP) and the patient, or the patient’s interaction with other primary care professionals, collectively contribute to the success or failure of behavioural interventions for MUS?
Scope the primary literature
The team started by using a large number of narrative reviews and qualitative syntheses (see Appendix 2, Table 68) as key documents to be harvested as a source of potential programme theory. Once a generic pathway had been developed to map the patient’s journey, and challenging junctures throughout this process had been identified, the team moved on to interrogate ‘clusters’ of primary studies. According to previously published work, clusters represent multiple related outputs associated with a single project or, at the very least, with investigation within the same context. 58 These documents, labelled ‘sibling studies’, are considered particularly important because they remove variation in context among the set of studies while contributing collective detail and interpretation. By extension, a further group of related studies, identified as ‘kinship studies’, relate to the original cluster at a more conceptual level, not sharing the same context but sharing conceptual links. For example, kinship papers may share an underpinning theory of change or may seek to replicate, or indeed may have preceded the intervention or study design that are the focus of the index study.
Previous research, by ourselves and others, has found that these sets of cluster documents are often thickly populated; expenditure on a trial and the range of individual collaborators on a project act as catalysts for substantive collateral research activity. 135,189,190 Given the large number of trials identified, and anticipating multiple related reports per trial, the review team took a decision to focus on trials and/or qualitative intervention studies conducted in a UK context. To this extent our focus was more co-terminous with the qualitative synthesis for this project, rather than the review of effectiveness. Nevertheless, a major difference was that the realist synthesis extended its purview, not only by covering the entire primary care pathway, and not just the intervention itself, but by accessing non-intervention-based qualitative sibling studies and metasyntheses, narrative reviews and systematic reviews recording experiences from a diverse range of countries and settings.
The review team identified six UK-based potential clusters (Table 14). 157,182,184,185,191 However, ultimately only three of these represented viable clusters with at least one trial report and one qualitative study from within the same study context. The remaining interventions were centred on qualitative intervention studies already covered by the qualitative synthesis. Two independent studies examined the IAPT Long-Term Condition/MUPS intervention and were therefore linked to broaden their explanatory power. Similarly, studies by Burton et al. ,157 Morton et al. 185 and Morton et al. 191 related to the same intervention in the same context, but were not formally linked to the same project. For included studies, see Chapter 4.
| Cluster identifier | RCT | Other quantitative | Qualitative | Economic | Other evidence |
|---|---|---|---|---|---|
| The BodyMind Approach | Payne192,193 | Payne194–197 | Payne186 | Payne192,198 |
Lin202 Samaritter205 |
| FINE (2010) | Wearden111 | Wearden111,206–208 |
Chew-Graham179 [I]; Peters178 [I] Hannon214 Bayliss215 |
Bayliss165 (QES) |
|
| Humanistic group counselling for somatisation84 | Graham183 [I] | ||||
| IAPT long-term condition/MUPS85 |
Gerskowitch182 [I] Lewis184 [I] |
||||
| MUST reattribution training (2007) | Morriss63,108 |
Morriss223 |
Morriss220 [NI] Dowrick181 [I] Peters180 [I] Salmon226 [NI] |
Morriss221 |
Salmon222 Salmon226 [C] Gask85 [NR] |
| (Primary) SCI (2012/16) |
Burton157 [I] |
Decide on the scope of the review
In contrast to the qualitative systematic review, which focused on the direct treatment context (see Chapter 4), the realist review adopted a broad pathway-based perspective. Essentially this followed an input–processes–output logic, with input being the decision of the patient to present with symptoms to their GP, the processes covering all interactions with primary health-care services, including initial and repeat consultations and the treatment itself, and output occurring at the resolution of the initial situation through relief, recovery, resignation to the situation or referral (Table 15). McGowan et al. 227 describe a typical pathway in connection with chronic pelvic pain:
| Why will patients be more or less likely to present to their GP with MUS? | What will make the initial consultation more or less successful from the patient’s viewpoint? | What will make a patient more or less likely to benefit from being assigned the MUS label? | What will make a patient more or less likely to benefit from being assigned a diagnostic label? | How do patients/doctors benefit from more having a disease model? | What will make the initial consultation more or less successful from the GP’s viewpoint? | What will make the ongoing relationship between patient and GP/HP more or less successful? | What will make a patient likely to attend? | What will make a patient more or less likely to receive/accept behavioural treatment? | What will make a patient more or less likely to continue behavioural treatment? | What will make a patient more or less likely to respond to behavioural treatment? | What will make a patient more or less likely to be able to cope with unsuccessful behavioural treatment? | What will make a patient more or less likely to use/be referred to another service? | What will make the patient more or less likely to view the overall therapeutic process as successful? | What will make the GP more or less likely to view the overall therapeutic process as successful? |
The process of seeking medical advice followed a typical cycle of events: consultation with GP (initial investigations and treatment) – diagnosis – no resolution of symptoms – referral (further investigations and treatment) – diagnosis – back to the GP – diagnosis – (treatment and management). 227
Table 15 necessarily presents a simplified primary care pathway; the pathway to see a HP other than the GP for treatment involves various significant additional steps that have been omitted from the pathway as described.
Despite this broad perspective, which extends beyond that of the accompanying syntheses, this model might still be considered simplistic, in that therapeutic scenarios can include diagnosis and treatment in primary care, diagnosis in primary care and treatment within secondary or tertiary services, and referral to secondary care and subsequent treatment in primary care. The other observation to help unravel this complexity is that important issues such as trust and the continuity of the therapeutic relationship inform the context even prior to presentation to the GP (e.g. past experiences of the GP, practice or health services in general by either the patient or their friends and family). 103,104 Indeed, the decision as to whether or not to attend a GP appointment depends on the patient’s anticipation of the reception from the GP and practice to their particular presentation.
Mapping the programme theory/theories
Initially, seven programme theory statements were developed for testing from engagement with conceptual and qualitative literature. Such theory-driven evaluation aims to offer plausible explanations, not probabilistic statements:228
-
Belief (in the patient, doctor and symptoms) is key.
-
Focus on symptom management not labels.
-
A healthy professional–patient relationship is fundamental.
-
Congruity between disease models facilitates treatment.
-
Working together, professional and patient can exit the diagnostic cycle.
-
Psychosocial talk only to be cued by the patient.
-
Honesty helps more than certainty.
Following discussion within the team, and advice from a clinical expert, one of these statements was modified and another was added. The initial programme theory on ‘honesty’ was felt to imply some absolute value; in fact, what the patient responds to is a belief that the professional is offering an explanation for their symptoms that they (the professional) believe is congruent with the facts and uncertainties at that particular point in time, while acknowledging that this may change as new data or experience become available. This ‘explanation’ may include not being able to offer an explanation at all, and openly acknowledging this. The more relative statement ‘a useful explanation can be good enough’ was felt within the team to more accurately reflect the data in capturing the dynamic context within which the explanation is negotiated between GP and patient, therefore requiring that the prevalent explanation be more contingent, flexible and modifiable as new clinically relevant information or additional patient experience becomes apparent.
In addition, it was felt important to try to surface the specific impact of setting, given the exclusive focus of this project on primary care interventions. The team therefore explored the proposition that ‘a contextually sensitive therapeutic response (i.e. a setting within which a patient feels that their concerns are being addressed) facilitates resolution’. Three of the programme theory statements were prioritised for a fuller exploration [indicated below with a double asterisk (**)]. See the finalised list of exploratory programme statements in Box 1.
PT1: belief (in the patient, doctor and symptoms) is key.
PT2: focus on symptom management versus labels.
PT3: a healthy professional–patient relationship is fundamental.
PT4: a contextually sensitive therapeutic response facilitates resolution (**).
PT5: congruity between disease models facilitates treatment.
PT6: working together, professional and patient can exit the diagnostic cycle.
PT7: psychosocial talk only to be cued by the patient (**).
PT8: a useful explanation can be good enough (**).
Prioritised statements are indicated with a double asterisk (**). A fuller explanation of each programme statement follows below (see programme therories 1–8). PT, programme theory.
Finally, at a relatively late stage of the review, the wording of programme theory 2 was amended to read ‘focus on symptom management versus labels’. This was in direct response to the equivocation within the literature, and surfaced among stakeholders, regarding the utility of a label. In some contexts, a label was seen to offer validation to a patient or entry on a pathway, and there was some evidence of perceived differences in the value between two labels (e.g. CFS and myalgic encephalomyelitis). The revised wording offered a choice rather than a dogmatic course of action.
Propositional statements
A series of propositional statements were devised linking characteristics with key outcomes in the form of ‘if–then’ statements.
Programme theory 1: belief (in the patient, doctor and symptoms) is key
How might this work?
IF HPs show that they take the patient and their concerns seriously, THEN the patient will engage with treatment. 1
A consistent response from patients within qualitative studies, qualitative syntheses and the intervention clusters is that patients want their concerns to be taken seriously. Patients are more likely to feel that they are being taken seriously when the GP:
-
pays empathic attention to them as individuals, meaning that the GP knows their personal circumstances and has an open and empathic approach105–107
-
ensures a good conversation by treating the patient as an equal partner105,108
-
is attentive to their symptoms by exploring these symptoms in depth and by acting on them. 108,109
These findings cross-cut the foci of other programme theories, especially programme theory 2 (focus on symptom management vs. labels) and programme theory 3 (a healthy professional–patient relationship is fundamental). Furthermore, within a healthy climate for communication, there is increased likelihood that a further programme theory will be accessed [i.e. that doctor and patient will negotiate and reach a congruent understanding of the disease (programme theory 5)]. Locating or co-ordinating treatment within primary care is also described as being likely to activate a further programme theory, not as a location but in recognition of the GP role, namely that of better knowledge of personal circumstances, which is achieved through continuity of care.
Two further observations can be made. First, the trend away from the personal GP to more of a team-based approach, with the associated difficulties in making an appointment with the same GP, is likely to fragment this knowledge. Paradoxically, it may heighten the likelihood of meeting at least one GP with whom one has a rapport. Larger practices also extend the prospect of more flexible appointments and individual therapists. 110 Second, the need to be believed is often highlighted as key at the beginning of the consultation episode. However, patients also express a need for the GP to keep believing in them, even when negative test results or other modifiers challenge this therapeutic relationship. In group interventions, this feeling of being understood could alternatively derive from fellow participants with similar problems (see Chapter 465,84,85).
Conversely, not being believed may have a significant impact on individuals with myalgic encephalomyelitis/CFS. These patients have reported feelings of shame associated with invalidation of their symptoms,111 with patients sometimes attributing their feelings of shame to the attitudes encountered from medical professionals and others expressing the idea that they were made to feel that myalgic encephalomyelitis/CFS was ‘all in their heads’. 112
Contextually, there is reason to believe that the impact of non-belief by the HP may differ according to the amount of social support being received from friends and family by the individual. Kendrick229 cites the mechanism of the Stress Buffering Model in achieving an ameliorative role of social support so that any potentially shaming effects of illness invalidation are diminished. However, ‘illness invalidation’ must not only be considered in the context of the direct relationship between HP and patient; the GP is also a source of diagnosis that offers social identity, removing the patient from perceived isolation. 113 Such labels may be accessible for fibromyalgia or CFS, but are typically lacking for MUS/somatoform disorders, which are, in fact, characterised by an absence of diagnostic labels. In this context, ‘diagnosis’ lacks the precise use favoured by the biomedical model, instead referring to a number of unexplained symptoms making it challenging to pass these on to the patient. Recent commentators have remarked on the risk that the new somatic symptom disorder in DSM-5 may mislabel many people as mentally ill. 20 The absence of a label or diagnosis does not necessarily throw into question the GP’s belief in the validity of the patient’s symptoms. However, were this to translate into an expression of non-belief, verbal or non-verbal, by the HP this could indirectly undermine the patient’s credibility should the patient decide to disclose this non-belief to family members or significant others. 111
How the disease label is viewed within the medical profession and within society more generally is another source of potential variation. Doctors may be more likely to maintain their belief if they are able to map the patient’s symptoms to a disorder that they recognise or understand. Labels may also offer patients a communicative mechanism by which they can validate their experience with their family members, serving as a form of ‘shorthand’. The corollary is that where the label is perceived negatively, being associated with ‘skivers’73 or those who are ‘hysterical’,114,115 the label operates in a negative way, causing frustration, stigma and shame. Patients with unexplained neurological symptoms are reported to prefer the term ‘functional’ to MUS as it appears to offer a diagnostic category with which they can identify. 114 As Jutel230 observes, diagnosis, as a form of classification, valorises some points of view, and silences others.
How might this be explained?
Many studies confirm the need for a patient to feel heard, believed, accepted,118 and this aligns well with the Health Belief Model,119 which emphasises the importance of the patient feeling that the GP is gaining a true understanding of where they are coming from. According to the Health Belief Model, three factors explain whether or not a patient will decide to consult their GP:
-
the extent to which a person perceives a threat to his or her health
-
the degree to which a person believes that consultation with the GP will help to reduce that threat
-
‘cues to action’, prompting a person to consult the GP, either internal (symptoms) or external (e.g. mass media communications or interpersonal interactions). 231
When a patient feels that their GP understands them, this may pave the way for a healthy relationship between patient and professional that helps when subsequent challenges need to be ‘smoothed out’. As a corollary, the literature also reveals the unhelpfulness of the patient not being listened to and the psychological impact that this may have. 120 For example, Horton-Salway232 describes how a member of a myalgic encephalomyelitis support group attributed much of the associated depression with the challenge of trying to convince doctors that there is something wrong with them.
Therefore, a major role of the initial conversation may be in serving to validate the patient’s experience. 122,123 Numerous studies describe how this is critical to the receptivity and subsequent engagement of the patient within the treatment pathway. Hoedeman et al. 233 describe the contents of a ‘consultation letter’, which states that patients are best helped by having their symptoms taken seriously. Other content of this consultation letter implicitly confirms that the GP symbolises their belief by not telling the patient that their symptoms are ‘all in your head’; conducting a physical examination at each visit; and arranging to see the patient at regular intervals. However, this study does not address whether or not repeated examinations might have unintended consequences in contributing to patient concern that the doctor had previously missed significant physical pathology. Potentially, a pattern where further physical examination is directly triggered either by new symptoms or by exacerbation of existing symptoms may be more acceptable and more feasible within existing time and resource constraints.
What are the implications?
A key factor in whether or not a patient will engage with, and ultimately benefit from, a treatment is whether or not they believe that their concerns are being taken seriously and that their feelings are being respected. It is important that a GP works continuously on maintaining this belief beyond the initial consultation, making sure that any referrals and tests are justified123 and explained so that they are not misconstrued. However, the reality of demands of GP workload must understandably recognise instances where GP and patient negotiate an agreed explanation for and understanding of the patient’s symptoms (e.g. programme theory 8). Given the absence of a link between the patient’s symptoms and a serious pathology, both may agree that there is no need for ongoing review by the GP. In other more complex cases, regular review every few weeks by the same GP with whom the patient maintains a good relationship may be helpful in managing the patient’s concerns and may help to avoid superfluous investigations and referrals. 126 However, such complex cases probably constitute only a minority of patients presenting with MUS, and so such regular reviews need not place a substantial burden on already-pressed services. Some evidence suggests that preparing the patient for a negative or inconclusive outcome, in cases where the tests and investigations are predominantly for the purpose of ‘reassurance’, can be considered helpful. 127–129
Programme theory 2: focus on symptom management versus labels
How might this work?
IF HPs focus on the patient’s symptoms and how to manage them, THEN the patient feels that something is being done about their symptoms and that their symptoms are not being dismissed as all in the mind. 130
Several commentators have explored the importance of the label assigned to the patient’s symptoms. 113,114,125,131 Stone et al. 4 has reported on the potential of particular diagnostic labels to offend patients, particularly ‘when they imply a psychological rather than a physical explanation’. He observes that ‘Although “medically unexplained” is scientifically neutral, it had surprisingly negative connotations for patients’4 and proposes use of a preferred term ‘functional’. Jutel proposes reorientation from clinical diagnosis to clinical interpretation as a route to ‘reconcile medicine and the individual, the unexplained symptom and the patient distress’. 230 The objective becomes not accurate diagnosis but satisfactory interpretation (e.g. programme theory 8). More recently, Sharpe et al. 234 highlights the erosion of the uncertain boundaries between medically explained and MUS, with the latter having previously been seen as the traditional domain of the psychiatrist. 131 In doing so, he advocates a closer working relationship between psychiatrist and clinician consonant with the current professional movement towards a more integrated approach to physical and mental symptoms in primary care. 130
How might this be explained?
Medical diagnosis explains, legitimises and normalises. 113 The absence of diagnosis denies the patient an immediate explanatory framework, a treatment, access to the sick role and legitimisation of the complaint. 113 This factor surfaced during the project’s public involvement consultations (for details see Chapter 2, Patient and public involvement). Diagnosis of MUS is recognised as particularly challenging, with HPs demonstrating considerable uncertainty, particularly where symptoms are seen as particularly complex (see the qualitative review in Chapter 491). Jutel230 sees this as inevitable given that both practice and the literature have sought to reduce a diversity of symptoms into a single unitary term: ‘MUS’. In this context we can identify a demonstrable difference between IBS, fibromyalgia and CFS, where the clinician is able to pass on a definition in their transaction with the patient, and ‘MUS’, where a clinician seeks to explain that the patient has one or more symptoms for which the clinician does not have an explanation.
Jutel230 highlights circumstances where, when a patient presents with multiple unexplained causes, their doctor may translate these into ‘a unitary condition under an informal diagnostic category linked to psychiatric dysfunction’. Whether the new DSM-5 classification114 is practically useful or whether the label simply operates as a ‘secret’ label that facilitates coding of patients on administrative systems or for academic research purposes requires further exploration. However, this technical consideration should not be allowed to subvert recognition of overall resistance to the use of labels. Labels may be perceived as consigning patients to a metaphorical ‘scrapheap’132 and so it has been suggested that metaphors, models, informal terms or stories could productively be used as an alternative to labels. 132 Stone235 also describes how taking physical symptoms seriously helps to establish trust between the patient and the professional. 132 Some clinicians report not giving the patient a formal ‘label’, suggesting that there are circumstances when either patient or clinician, or even both, do not feel it productive to ‘trigger’ a label.
What are the implications?
Given that evidence on the value of either condition-specific labels or of the general MUS label is equivocal, GPs should seek to focus the consultation and subsequent treatment on symptoms. Use of labels is located firmly within a biomedical disease model which is not generally helpful within the context of MUS. If it becomes clear that the patient would value assignation of a label, then a GP can consider this possibility. In some cases, diagnostic labels will be required in order to access a particular treatment pathway.
Programme theory 3: a healthy professional–patient relationship is fundamental
How might this work?
IF HPs maintain a positive relationship with the patient, THEN professional and patient develop a shared understanding of the patient’s experience.
Many papers identify the importance of initiating and maintaining a healthy relationship between professional and patient as a key factor in successful resolution of MUS. Deale and Wessely29 confirmed this programme theory via a questionnaire survey of 68 patients with CFS. They found that dissatisfied patients were more likely to describe delay, dispute or confusion over diagnosis; to have received and rejected a psychiatric diagnosis; and to perceive doctors as dismissive, sceptical or lacking knowledge about CFS. Dissatisfied patients were also more likely to feel that the advice given was inadequate or conflicting. In contrast, satisfied patients were more likely to perceive doctors as caring, supportive and interested in their illness; to state that they did not expect their doctors to cure CFS; and to perceive their GP or hospital doctor as the source of greatest help during their illness. These findings support the view that medical care is evaluated less on the ability of doctors to treat CFS, and more on the interpersonal and informational skills of those delivering care. One apparent mechanism for improved satisfaction with care is ‘engagement’. 133 Such engagement uses a ‘five areas approach’ that ‘allows patient and therapist to develop a deeper understanding of the patient’s physical symptoms by integrating the full range of the patient’s experiences in areas of (1) situation, relationships, resources, and practical problems; (2) symptoms; (3) behaviour; (4) thinking; and (5) feeling’. 131 Extra time spent on such engagement is seen as critical. 86 However, this provides particular challenges within the pressured time and resource context of primary care.
How might this be explained?
Substantial numbers of studies have identified the importance of trust to the success of the primary care consultation. Although trust is key to most medical contexts, it becomes even more important within a context where epistemological incongruence and communication challenges are likely to occur. In this context, trust has both a supportive function and an enabling function, and it affects not only what is said but how it is said and interpreted.
Continuity of care is also related to the formation of trust. GPs feel that trust makes patients more likely to disclose problems and, in return, GPs feel more able to openly challenge patients. 160 In contrast, referral to other types of HP may subvert the trusting doctor–patient relationship; for example, referral to an orthopaedic surgeon for neck pain was seen to act contrary to a provisional psychosomatic cause. 141 Under other circumstances, however, referral to an orthopaedic surgeon offered reassurance and backup to the GP diagnosis. Collectively, this emphasises the importance of preparing the ground with clear communication of expectations of both GP and patient from the consultation.
Central to trust is empathy; this may be challenging if patients are felt to be ‘heartsink’132 or demanding of time and resources. 131 Wileman et al. 236 report how GPs, when interviewed, felt that showing an empathy with the patient, and taking an interest in them, enabled the patients to gain personal trust in the doctor.
Another sign of a healthy relationship is avoidance of destructive and unhealthy relationship behaviours. Commentators have highlighted the risk of what they have labelled the ‘duet of escalating antagonism’. 138 A similar challenge is posed by ‘looping effects’ (vicious cycles where the emotion of one person causes escalation within the relationship). 149
A contrary type of dysfunctional relationship is the ‘dependency relationship’. Salmon et al. 218 highlight the importance of ‘getting the balance right’ between meeting the patient’s support needs and encouraging dependency. It should also not be assumed that all patients experience an ongoing relationship with their GP72 or that they may not be simultaneously managed by multiple practitioners. 91,92
What are the implications?
The state of the professional–patient relationship is critical to the success of the primary care model, both in initiating a consultation and in maintaining engagement. Where a patient’s interaction persists beyond initial consultation and investigation, a GP must utilise communication behaviours that allow continual renegotiation of the patient’s understanding of their symptoms in the light of that patient’s ongoing experience and as new information becomes available. 98 A patient will be more likely to volunteer useful information within a climate of mutual trust. Increasingly, the general public appears to be becoming aware of the potentially debilitating effects of stress-related symptoms that do not have a clear organic basis. 237 At the same time, a GP must work hard to prevent the creation of a model of dependency.
Programme theory 4: a contextually sensitive therapeutic response facilitates resolution
How might this work?
IF diagnosis, management and treatment takes place within a primary care setting, THEN the patient experiences continuity of care that may lead to progress and resolution.
A key issue in assessing the value of primary care-based behavioural approaches relates to the perceived relative advantage of delivering treatment within a primary care setting. Within the literature the discourse revolves around two types of potential advantage, namely the operational and the conceptual. 139 It is noteworthy that many operational issues, relating to the logistics and practicality of a primary care setting, are stated explicitly, whereas perceived inherent advantages of the setting from a therapeutic perspective tend to be understated or even implicit.
Operational benefits of a primary care setting
Commentators express many potential benefits in favour of treating MUS in primary care. For example, Kroenke and Swindle238 suggest that primary care may be considered an appropriate care setting, highlighting how patients who do not wish to explore psychological attributions for their symptoms, and are therefore unwilling to accept psychological/psychiatric referral elsewhere, may find primary care more acceptable. 86 However, such a conclusion fails to acknowledge that a large proportion of MUS patients do not have underlying psychological causes or associations. 83 Foremost among identified benefits is the opportunity to understand the patient’s circumstances and context and to bring this understanding to bear when interpreting their symptoms. Personalised care is a fundamental value for general practice. Knowledge of the patient and his/her history and context is considered to facilitate recognition of MUS early in the consultation. 142,151,160 Other clues used by GPs to aid recognition of MUS in familiar patients include frequent and often prolonged consultations, with many frequent requests for referrals, and engendering subjective feelings such as irritability and frustration.
Huibers156 advances several reasons why GPs may be particularly suited to performing psychosocial interventions in primary care. First, GPs are able to exercise the ‘stepped-care’ principle so that all appropriate options are explored prior to referral to specialist care. Second, many GPs already provide general support to patients with stress, anxiety or depression. Offering GPs a behavioural toolkit to manage patients with MUS as well as those with more generalised complaints may simply be making better use of their time with a stronger likelihood of benefit or resolution. However, introducing additional demands on an already pressurised frontline service, with the associated demands on time, skills and motivation that such an approach might bring, may not be considered feasible. One alternative may lie with the availability of psychological practitioners based in primary care (e.g. through IAPT workers based in primary care), which is becoming an increasingly common model. These interventions offer an opportunity to build on an existing relationship of trust as an alternative to an unfamiliar provider in secondary care. Finally, provision in general practice could potentially disrupt the strongly entrenched split between psychological and physical forms of primary care.
Conceptual benefits of a primary care setting
Continuity of care is frequently cited as a benefit of locating treatments in a primary care setting239 and is particularly endorsed by the Royal Colleges (Psychiatrists, GPs and Physicians) which espouse the value of co-locating physical and mental health services. 240 It is frequently assumed in the literature that being seen repeatedly by the same professional is advantageous. Although this may be true from a service perspective (e.g. in perpetuating stable expectations and a consistent standard of care), there may also be negative consequences that become particularly important during the breakdown of the therapeutic relationship. Different personnel may be able to offer an alternative perspective, establish a different type of relationship or draw on a different set of intellectual and emotional ‘resources’ in seeking to address the symptoms. By way of a contrast, Peters et al. 180 report how patients may gravitate towards the doctor with whom they have had a longstanding relationship in the knowledge that they would ‘avoid challenging their current ways of managing their problem’. The long-term nature of the relationship, ‘time is on our side’, was seen as an opportunity to work together on resolving patient symptoms.
Although it is difficult to draw robust conclusions about continuity of care, given such variation within the professional–patient relationships, it is certainly worth recording that developments within UK general practice (e.g. larger practices, practice teamworking rather than personal GP provision, increasing front-line nurse roles, etc., to support improved access and out-of-hours coverage) may subvert this as a claimed advantage. 160
Contrary evidence
This realist review found some evidence that patients believed that primary care is an inappropriate setting for exploration of psychosocial issues. Murray et al. 241 offer multiple reasons why this might be the case, citing patient beliefs concerning the superiority of self-management of symptoms and assumptions that limited treatment options would be available. Past experiences of primary care may also be influential. 104 Patients may feel that their primary care provider does not have the ability to resolve such health problems. 92,137,228 Advocates of the BodyMind Approach, one suggested therapeutic intervention for disrupting the expression of MUS, attributed benefit from delivering the intervention in a setting that does not hold the negative connotations of health premises. 65
Murray et al. 241 further highlights other, practical, constraints that may negatively impact on diagnosis in primary care. Of particular resonance are the short consultation time available and a correspondingly heavy patient caseload. Patients may ‘second guess’ the receptivity of the primary care professional, based on prior familiarity with the practice context. Other considerations may relate to financial factors and a lack of local mental health resources.
How might this be explained?
Many factors used to endorse location of services in primary care are generic and seem determined by current models of service delivery, not by the specific requirements of these treatments. This ongoing discourse is currently taking place against a backdrop of service reconfiguration, new models of care and migration of services from secondary to primary care. A further argument relates to integration of physical and mental health services, for which MUS seems to have a particular case.
What are the implications?
A GP must seek to establish a role not as an expert on the patient’s symptoms, but as a co-ordinator and facilitator of the patient’s overall care, regardless of setting. Referral must not be used simply for respite, possibly signalling that the doctor has run out of ideas or patience. Conversely, demand for referral may be initiated by the patient, who requires some sort of ‘reassurance’ that nothing more sinister underlies their symptoms. Care therefore needs to be taken that referral is being initiated in response to need and that entry into the referral pathway will benefit the individual patient in the long run.
The main strength of a primary care setting appears to be the opportunity to offer continuity of care, contextualised within an understanding of a patient’s individual circumstances. However, recourse to external expertise or resources may help in exiting unproductive loops and in advancing the negotiation. Such a finding fits well with the collective body of effectiveness studies that cover a wide variety of primary care-based service models from those that deliver services on primary care premises through to those where the GP acts as initiator and co-ordinator of care but where the services are actually delivered by another professional.
Programme theory 5: congruity between disease models facilitates treatment
How might this work?
IF the disease model advanced by the HP matches, or is compatible with, that of the patient, THEN the patient will accept treatment. 20
Qualitative research has revealed that a dissonance frequently exists between the mental conception, or models, of disease held by the patient and those held by the GP. 242 In a qualitative study that accompanied a trial of training family practitioners in reattribution to manage patients with MUS (the MUST trial), Peters et al. 180 describe how patient explanatory models were ‘multifaceted, simultaneously incorporating disease and non-disease causes and the interaction between them’. This complexity was portrayed in fragmented and chaotic narratives ‘presenting multiple, seemingly unconnected and incoherent problems with no clear beginning or end’. In contrast, some patients viewed GPs as holding ‘more unidimensional, dualistic and hence simplistic models about cause and resolution of problems than their own’. 180
Other studies note how patients remark on the insufficiency of doctor’s illness models. 243 In their metasynthesis of 32 quantitative and qualitative studies of CFS/myalgic encephalomyelitis, Bayliss et al. 165 report that working within the biomedical model has led some GPs to be sceptical about the existence of CFS/myalgic encephalomyelitis. In contrast, GPs who provide a diagnosis were likely to hold a broader, multifactorial biopsychosocial model of CFS/myalgic encephalomyelitis and also to possess more positive attitudes to its symptoms.
Achieving greater congruity between disease models,244 reached through a process of negotiation as modelled by medical reattribution interventions, may thus contribute to improved communication and, potentially, to a healthier patient–professional relationship. 98 However, some evidence suggests that these improvements are not as far-reaching as intended, contributing only to improved patient satisfaction but not to more effective treatment.
van Ravenzwaaij et al. 245 identified nine different potential explanatory models for MUS. This examination concluded by identifying the superiority of a cognitive–behavioural metamodel, principally because it was the only model to include all four domains (i.e. somatic causes, perception, illness behaviour and predisposition) inhabited by the alternative models. 11 This metamodel proposes that the cause of MUS is a self-perpetuating multifactorial cycle, with interaction of different factors across the four domains. This model provides a framework to incorporate patients’ own personal perpetuating factors as well as predisposing and precipitating factors. These three categories offer a framework for understanding individual variation in MUS. Predisposing factors are individual characteristics that make people more vulnerable to distress, physiological arousal and bothersome physical symptoms. Precipitating factors are events or factors that initiate physically symptomatic episodes within vulnerable asymptomatic individuals or that increase disability or distress among those who are already symptomatic. Perpetuating factors maintain symptoms, distress and disability, and extend their period of duration. 246 Each factor can result in physical symptoms and/or distress. The doctor and patient collaborate to identify those elements of a patient’s personal circumstances that might contribute to the patient’s distress. The metamodel incorporates at least five different theories: sensitivity, sensitisation, somatosensory amplification, endocrine dysregulation and the illness behaviour model. The authors conclude that these explanatory models should be ‘integrated in the educational programs of all medical doctors in order to improve the quality of care for patients with persistent MUS’. 247
How might this be explained?
The conceptualisation of the ‘unexplained symptom’ has been seen ‘to shift responsibility for the inability to explain the symptom from the doctor to the patient’. Hadler248 describes a ‘contest of diagnosis’, with medicine ‘not [being] likely to accept blame for subjecting the patient to months of an exercise that turn[s] out to be flawed in design and iatrogenic in execution’. 248 Johansen and Risor242 identify what they term ‘epistemological incongruence’ between dominant disease models and the realities of encountering patients with persistent symptoms. When facing cognitive incongruence, GPs should strive to achieve relational or emotional congruence, through the establishment of good relations, alliances and partnerships with patients. 242 This is seen to contribute to positive experiences of mutual trust and validation.
Achieving congruity of models is also a key factor in treatment success. If a patient feels that a proposed treatment is compatible with their illness model and with their preferences/lifestyle (e.g. dance classes vs. exercise) then they are more likely to initiate the treatment and to persist with it. 249
What are the implications?
Successful resolution of, not necessarily recovery from, challenging MUS is largely dependent on a shared understanding of the problem. 250 There is clear evidence that the different mental models used by GPs and patients to understand the symptoms may contribute to communication difficulties. Ideally, professional and patient should arrive at a negotiated and necessarily contingent understanding; a biopsychosocial model, as proposed by Engel > 40 years ago,251 may offer a meeting point between initially conflicting or incompatible models.
Programme theory 6: working together, professional and patient can exit the diagnostic cycle99,102
How might this work?
IF patient and professional can avoid perpetuating extensive and unproductive diagnostic tests, THEN patient and professional achieve a sense of progress. 252
Patients appear to attach great importance to gaining a diagnosis, which was reported in one study of CFS patients as the single most helpful event. 253 However, the same study reveals this as another example of dissonance between patients and GPs, with GPs seeing a definitive diagnosis for CFS as disabling, potentially becoming a self-fulfilling prophecy. 253 Patients also find it frustrating if they have to wait for years before receiving a diagnosis. Diagnostic tests are viewed as a way of securing a diagnosis, but at the risk of increased frustration if a definitive diagnosis is not actually realised. Frustrations also accompany receiving the ‘wrong’ diagnosis, with some patients expressing a preference of myalgic encephalomyelitis over CFS. 253 Misdiagnosis also carries frustrations and risks. GPs therefore find themselves in a paradoxical situation where the ordering of diagnostic tests may be variously seen as a response to patient demand, a signal to the patient that they are unable to resolve the situation within their more immediate resources, an abdication of interest or commitment to the therapeutic relationship and a potential cause of additional frustration when faced with a subsequent negative result. Such frustration can be mediated if the GP or secondary care clinician has created clear expectations, with the patient being advised of the potential implications of a negative test. Avoidance of unnecessary tests carries further benefits for the patient and health-care system; it reduces costs, patient anxiety and the possibility of identifying ‘incidentalomas’, which then need further investigation but may not be of serious significance. 254,255
How might this be explained?
Several authors use the concept of the ‘diagnostic cycle’ in connection with MUS. Figure 59 shows a pictorial representation of diagnostic loops potentially leading to an unproductive diagnostic cycle. 257 This concept figures in the title of a study of chronic pelvic pain. 102 The authors describe the underpinning biomedical programme theory that diagnosis will, in turn, lead to a ‘suitable resolution of the problem’. However, the reality as revealed by the women’s narratives is that ‘many women do not complete this cycle, they become stuck at a certain point, or re-enter the cycle repeatedly. Women can only opt out of the cycle if the problem is resolved, or by choosing to disengage with medical care’. 227 Women enter the diagnostic cycle actively seeking an explanation for their pain and are consequently frustrated if this is not attained.
In a subsequent study, McGowan et al. 258 revisit this phenomenon, again in the context of chronic pelvic pain, describing how failure to receive a ‘medical explanation’ for their pain can lead women to enter ‘a cycle of re-investigation and re-referral’. Burton et al. 157 extend this concept to a wider group of symptoms to be addressed by a primary symptom clinic. The Symptom Clinic model ‘aims to negotiate a “medical” explanation for symptoms involving physiological processes’. 157 The Serene health organisation in Plymouth uses a diagram of the unproductive diagnostic cycle to illustrate the need to ensure that patients progress along the pathway and to not become becalmed within a cycle of successive tests and/or referrals. 257
Stone235 extends this analogy, using the words of a GP informant, to that of a spiral or cascade effect, where ‘someone sees a specialist, and because the thing is not within the specialty for which they are trained [they] don’t feel able to exclude organic pathology, and will therefore either make a referral or intimate that a referral would be required . . .’.
The consultation letter intervention described by Hoedeman et al. 259 specifically seeks to avoid unproductive tests, with the psychiatrist providing detailed recommendations to the GP concerning management, that is, to avoid unnecessary diagnostic procedures and hospitalisation.
Embarking on such a diagnostic cascade, where the clinician feels out of control of the decision-making process, has previously been identified as a source of particular stress and anxiety, both for the GP and for those experiencing the symptoms. 260 This further explains why non-resolution of MUS can be perceived as threatening to the professional–patient relationship. However, a counterargument relates to the persistent ‘niggling biomedical doubt’ that may be held by a GP, or a prevalent fear of litigation, requiring them to rule out symptoms related to such conditions as cancer, thus precipitating further tests or referral. 132
The literature reflects an ambiguous attitude towards referral to a specialist and/or for further tests. Use of referral appears to differ according to whether the GP seeks resolution, reassurance or respite. In the first context, the GP may genuinely feel, operating within a culture of ‘medical certainty’, that the patient simply needs a label to attach to their symptoms and then this will serve as a passport to their legitimate entry into a treatment pathway. If they communicate this prospect to the patient, then the result of a negative test can prove frustrating to patient and professional alike. On the other hand, it may be valuable to forewarn, and thereby prepare, the patient when a negative result from investigations may be expected. 261 In the second context, the GP is seeking a second opinion, either to confirm or to disconfirm a putative diagnosis or to offer an additional ‘resource’ that they would otherwise not be able to access. Confirmation or negation of a GP’s diagnosis may reassure the GP that they are not missing or overlooking something, but may not necessarily advance the patient’s agenda. 129 In the final context, the GP feels unable to sustain their interaction with the patient in a healthy manner and so looks for respite during which they are able to reorganise their resources and plan a future strategy. Shattock et al. 262 describes circumstances under which frustration and hopelessness may precipitate referral.
The verdict resulting from the external referral either to a specialist or for diagnostic tests may or may not provide useful additional information, but in a way that is only a secondary objective. Being referred to a specialist or for further diagnostic tests holds considerable potential for frustration – from a negative test result, from a narrowing down of possibilities or from being assigned a label that is either unhelpful or unwelcome.
If the patient seeks a referral, in the quest to receive a clear and ambiguous diagnosis, they may find any persistent medical uncertainty unfulfilling and unrewarding. 263 From the patient’s perspective, there are potential unintended consequences. Referral from the GP to an external source may undermine the GP’s credibility, creating the impression that the GP is unable to address the patient’s symptoms within their own expertise or resources. Alternatively, the patient may perceive that the GP is trying to obviate responsibility, whether or not this is actually the case, or is beginning to find the professional–patient relationship too demanding and thus requires respite.
What are the implications?
Being tied into the diagnostic cycle can prove frustrating for GP and patient alike and can result in unwarranted utilisation of referral and/or tests. Ongoing referral from an inconclusive consultation with a first specialist to subsequent additional specialists may perpetuate and exacerbate this problem. When tests or referral are being ordered, GP and patient should share an understanding of how they will potentially advance the patient’s circumstances. 123 Tests should not be used in a symbolic way, nor as a manifestation of GP, or even patient, power.
Programme theory 7: psychosocial talk only to be cued by the patient
How might this work?
IF HPs introduce psychosocial explanations, THEN the patient feels that they are being problematised.
If GPs conduct a physical examination for any new symptoms at each consultation, this may signal to the patient that the GP is resisting a tendency for early closure by offering an expeditious psychosocial explanation. 264 Some GP trainees describe how they try to initiate discussion as early as possible about underlying psychological factors. 265 This opens the way for future consultations to be able to legitimately return to these possibilities without appearing a course of last recourse. This illustrates an important contextual determinant in how programme theory 7 is seen to operate. 141
Doctors have also described how CFS/myalgic encephalomyelitis patients resisted any attempt to attribute their disorder to psychological factors. Patients may feel that a physical explanation for a symptom does not imply that the doctor is attributing personal responsibility for its onset. Conversely, patients may feel that, when a doctor provides a psychological explanation, they are implying that the patient might be able to control, or even reverse, the physical symptoms. 113 Given that the patient has sought help for their distress in the first place, such an interpretation may seem impossible to the sufferer, and several commentators85,158 observe that this may precipitate stigma and shame. Recent literature on the causes of CFS reveals a vigorous and ongoing debate about the extent to which the condition may be explained by physiological findings. The extent to which physiological explanations might offer a challenge to the stigma associated with specific labels is not yet clear and may, in fact, vary across the conditions associated with MUS.
In particular, patients may show themselves as reluctant to engage with follow-up that attributes a psychological explanation to their symptoms. 266 Lewis184 describes how patients were less likely to engage with CBT if they had strong ideas about what was causing their symptoms, perceiving that their problem was purely a physical one. The literature contrasts this response to that for other applications of CBT where patients accept that there is a psychological element to their symptoms. 184
Murray et al. 241 describe how some GPs are reluctant to suggest a significant influence of psychosocial factors fearing that this will have adverse consequences. For example, patients may feel that a psychosocial-oriented discussion will deflect attention away from some serious underlying pathology. 180 Indeed, some patients may avoid discussing psychosocial factors for this reason.
Evidence suggests that patients often actively resist reattribution. 267 Even when they experience anxiety and depression, patients with MUS may see these as associated with, rather than causal to, their physical symptoms. 268 Patients’ reasons for not wanting to share psychosocial information may be sensitive to context; for example, some patients may want to protect their GP from burden. 92 However, such caveats do not, in themselves, represent sufficient reason for not asking the patients about possible anxiety or depression symptoms when taking a medical history.
The GPs may be reluctant to hand the onus of initiating discussion of psychosocial cues to patients, feeling that the patient is unlikely to volunteer such information within the time-limited constraints of a brief consultation. 269 However, qualitative analysis of audiotaped consultations between patients and GPs revealed that in > 95% of the consultations with patients with MUS, patients presented signals or cues of one or more psychosocial problems. 270 Many patients with MUS presented psychosocial cues that their GPs tended to ignore until they encountered a physical symptom that offered a rationale for exploring or investigating. Salmon et al. 271 found that patients with MUS appeared to be seeking emotional reassurance more than patients with clear organic problems.
Thus, GPs have cause to reflect on their own communication skills with such patients. 272
How might this be explained?
Traditionally, GPs have sought to rule out an organic illness before exploring a possible psychological origin. This focus on organic illness has been suggested as a potential reason for low detection rates for mental health problems in primary care generally. 273 It has also been suggested that this approach can reinforce the patient’s view that the cause is physical, which could exacerbate the symptoms – especially if the doctor and patient have not previously discussed the possibility of something other than an organic cause. 274,275 By contrast, current GP training places emphasis on taking a proper biopsychosocial history, acknowledging that even when a patient appears to have a physical problem this may be associated with accompanying psychological symptoms or exacerbating social difficulties. 130 This approach is further underpinned by the philosophy of ‘parity of esteem’ between physical and mental health problems currently endorsed by the same Royal Colleges. 276
Promising initial results in reattribution training from poorly designed studies have probably been unduly influential in supporting the hypothesis that patients will achieve better outcomes if the GP and patient are able to establish a clear link between psychological distress and physical symptoms. 277 This remains an area where the literature and the theoretical base underpinning particular interventions demonstrate largely unresolved dissonance. Rosendal77 highlights potential explanations for these conflicting results, citing the heterogeneity of both patients and their disorders and the differential effects that psychosocial interpretations and interventions may have in the short term. 278
What are the implications?
Most patients presenting to HPs with MUS believe that they have a physical health problem, rather than a mental health problem. They therefore feel more able to trust a physician to respond appropriately to their ‘physical’ health problems. 279 In a sizeable proportion of cases, patients with MUS may in fact have an underlying psychological cause for their physical symptoms. What is less clear is the extent to which psychological problems are the consequences of a patient having unexplained physical symptoms. In the light of this, doctors may choose to respond to patient cues of emotional distress rather than directly asking patients about their feelings, unless doing so in the context of understanding how their symptoms have affected their mood. The increased risk of suicide reported among patients with MUS suggests that elucidation of mood and associated risks, by one means or another, is important in any assessment of such patients. However, GPs need to ensure that the approaches that they typically use in consultation and diagnosis do not in themselves generate or maintain MUS. 132,280
Programme theory 8: a useful explanation can be good enough
How might this work?
IF HPs offer an explanation that they believe is congruent with current facts and uncertainties AND this explanation makes sense to, and is accepted by, the patient, THEN the patient believes that it is possible to maintain the therapeutic relationship. 274
Patients frequently express dissatisfaction with explanations that are provided for their symptoms. 20,129 However, where explanations are effective and empowering, making sense to both patient and clinician, patients feel a corresponding sense of satisfaction. 20 Salmon et al. 281 describe how empowering explanations are tangible, exculpating and involving. 20 Conversely, primary care providers sometimes feel unable to explain symptoms. 272,282 Numerous reasons are advanced to seek to explain this inability. 139 These include feeling unsure or inadequate. Doctors frequently resort to the use of multiple tests in the pursuit of certainty or may refer to a specialist to compensate for their own perceived uncertainties or inadequacies in relation to this complex range of symptoms and explanations. Kirmayer et al. 283 observe that lack of explanation reflects ‘the limits of medical knowledge, available technology and the epistemological difficulties of assigning a clear cause to subjective complaints like pain and fatigue, which may have no objectively measurable correlates and may change rapidly over time in quality and intensity’. Additionally, clinicians, including GPs, are unlikely to have received any training in providing effective explanations for MUS. 159,265
This uncertainty is seen to persist to subsequent stages of the pathway, namely to access to treatment or referral. 139 In the intervention studies, patients clearly valued explanations that helped them to come to terms with and accept a diagnosis (see Chapter 4). Where this explanation is not forthcoming within a primary care setting, patients valued being able to access specialist knowledge and information on their symptoms and what they might mean. 85
The converse is that if the GP and patient feel able to accommodate uncertain or ‘good enough’ explanation at the stage of diagnosis, they may jointly feel able to carry this forward, given their ongoing relationship, to the subsequent stage of action, making that equally contingent and pragmatic. 20 For example, Lewis184 describes how therapists were prepared for ‘having a go’ rather than feeling sure about their approach, acknowledging, however, that working in such an ad hoc way can feel uncomfortable. Under such contingency, patients may modify their expectations and may seek to self-manage their symptoms rather than necessarily resolve them. In a trial of self-management for chronic fatigue in primary care, Friedberg et al. 151 contrast the theory of illness with a more common physical deconditioning and avoidance model, particularly emphasising that their model makes no assumptions about recovery or cure.
How might this be explained?
Although GPs may feel a compulsion to seek a definitive diagnosis and thus achieve ‘closure’ of the consultation, several commentators describe the unhelpfulness of this ‘myth of certainty’. 284,285 Howman286 characterises the prevailing need for GP trainees to want to ‘fix things’ as evidenced in contemporary medical training. Where distrust is initially present, either of doctors or of HPs more generally, providing a frank and bounded explanation, even if acknowledged as contingent or incomplete, may be seen as a form of propitiation.
Some commentators have located that doctors, including GPs, experience difficulty with admitting uncertainty within a prevalent discourse of power. 236 Medical training and the role of the GP assert the importance of the GP appearing competent and knowledgeable168,287 as a way of ensuring their medical authority. 241 Jutel230 highlights that, specifically, diagnosis is the foundation of that social authority. MUS, and their associated uncertainty, can pose a direct challenge to this authority. 181,288
Furthermore, some commentators describe how truthfulness about the limitations of medical options can enable patients to discard the likelihood of medical resolution of their symptoms. This can serve to reduce their dependence on their GP and encourage self-reliance. 288 In this way, patients are facilitated in building up their resilience, adaptability and defiance. 187 However, it may also have the unintended consequence of steering the patient to seek resolution from complementary therapies, often at great personal expense and with limited evidence of effectiveness. 289
The value of a contingent ‘good enough’ explanation, rather than a more definite conclusive verdict, is that it keeps open the possibility of medical options. Clinicians may be concerned about the possibility of missing a serious diagnosis, and this risk-averse culture may influence their test-ordering behaviour. Lines of communication between GP and patient remain open without offering a solid prospect of immediate resolution by medical means.
Caution must be expressed, however, against the presumption that it is simply the apparent agreement of doctor and patient with a ‘shared’ explanation that underpins satisfactory resolution. In a qualitative analysis, Stanley et al. 290 demonstrated that agreement between patients and their doctors in assigning symptoms to broadly defined ‘syndromes’ appears to reflect collaboration that is largely expedient, implying that something more pervasive than superficial compromise is required.
What are the implications?
While the evidence relating to the value of the GP’s explanation for the patient’s symptoms is broad and diverse, it offers a strong theme relating to providing a contingent and sincere explanation. This explanation needs to be seen, by both GP and patient, to be compatible with available facts and experiences but to remain flexible to future adaptation and modification as further information and experience becomes available. Above all, the explanation must keep open possible physical explanations as a potentially viable line of inquiry. The explanation should be jointly owned and the product of ongoing negotiation between GP and patient. 98
Specific evidence exists for the value of specific management techniques in improving clinician communication skills for patients with MUS with the aim of reducing investigations and health-care costs. 152 For example, in a multicentre RCT, Weiland et al. 291 describe how medical specialists received a 14-hour training programme based around experiential learning and feedback. Using techniques from CBT, those in the intervention group were stimulated to ‘seek interrelating factors (symptoms, cognitions, emotions, behaviour, and social environment) that reinforced a patient’s symptoms’. 291 The specialists were taught to explain MUPS understandably, reassure patients effectively and avoid unnecessary diagnostic testing.
Rief et al. 102 describe a training package entitled ‘How to Manage Patients With Unexplained Physical Symptoms’, which utilises guidelines that embody much of the programme theory elicited by this realist review (Table 16 maps programme theory components to the guidelines). This 1-day workshop included how to communicate with MUPS patients, when to start and stop medical examinations, and treatment options. GPs valued this workshop as highly relevant to their daily practice. Patients in the study reported a small reduction in general psychopathology between the index visit and the 6-month follow-up. Data suggesting better outcomes for the patients of trained GPs were inconclusive, given that the attribution of this improvement to the GP training is less clear and could potentially be confounded by time effects as GPs become more experienced or attuned to patients with unexplained physical symptoms. 29
| Stage | Approach |
|---|---|
| General aspects |
Show empathy and understanding for the complaints and frustrating experiences the patient has had so far (PT1) Develop a good patient–physician relationship (PT3) |
| Diagnosis |
Explore not only history of complaints and former treatments, but impairment, (health) anxiety, psychosocial issues Use symptom diaries to assess course and influencing factors on symptoms (PT7) When patient presents with a new symptom, examine the relevant organ system (PT2) Show results of investigations to explain absence of pathology and to give clear reassurance that there is no serious physical disease (PT8) Avoid unnecessary diagnostic tests or surgical procedures (PT6) |
| Treatment |
Provide regularly scheduled visits (e.g. every 4–6 weeks), especially in cases of very frequent health-care utilisation (PT1) Explain that treatment is coping, not curing (when pathology cannot be found or does not explain degree of complaints) (PT8) Suggest coping strategies (e.g. regular physical activity, relaxation, distraction) |
| Referral | If referral is necessary to start psychotherapy or psychopharmacotherapy, prepare the patient for the treatment. Reassure him/her that you will continue to be his/her ‘doctor’ (PT3) |
Programme components
Generically speaking, behavioural intervention programmes follow a similar process:
-
sensitisation of the HP to the patient’s symptoms and circumstances
-
selection of resources or strategies
-
initiation of the treatment
-
review and evaluation of the treatment
-
continuation or discontinuation of the treatment.
A previous meta-analysis found that psychological interventions can be effective for patients with MUS, but the effects were modest and of probable short duration. 78 Analysis demonstrated that psychotherapists achieved larger effects on unexplained physical symptoms than GPs. However, it was unclear whether or not this advantage is attributable to a larger number of psychotherapist-led sessions. Evidence for the effectiveness of teaching GPs skills in brief CBT for depression suggests that CBT training has little effect on GPs’ knowledge and attitudes or on overall treatment outcomes at 6 months. 292 This may equally be the case for MUS. The authors conclude that GPs may require more training and support than a basic educational package on brief CBT to acquire requisite skills. 292 Current training does not equip junior doctors with the necessary knowledge and skills to effectively and confidently manage patients with MUS. 293
Providing patients with a convincing explanation for their symptoms (programme theory 8) is seen as a key component of the FINE Supportive Listening Intervention. 206 This initial session also involves indicating how symptoms may be targeted for treatment (programme theory 2).
Interventions may differ in the extent to which stages 1 and 2 of the behavioural components listed occur within or outside a typical primary care consultation. Furthermore, where treatment is being delivered by a HP other than the patient’s GP, it seems inevitable that there will be some overlap or repetition of these two stages across both HPs. In other cases, the gatekeeper role of the GP is simply to channel the patient towards a particular pathway and then the selection of resources is fundamental to the actual treatment. This personalised tailoring of resources could contribute to a feeling of individualised care and of being taken seriously. This element of personalisation is commonly believed to be more likely to be achieved if the HP is able to gain an understanding of the patient’s individualised personal circumstances.
What is absent from this generic five-stage model for behavioural intervention programmes is a formal stage of ‘labelling’ to achieve matching of strategy to label in stage 2. As itemised above, Burton et al. 157 point out the shared limitation of approaches that ‘concentrate on making the link between physical symptoms and underlying psychological cause’ (cp. Hubley et al. 277). Instead, professionals are encouraged to base the selection of resources and strategies on the patient’s symptoms (see Programme theory 2). Burton suggests that there is value in initially providing conventional biomedical (i.e. contingent ‘good enough’) explanations for MUS (see Programme theory 8). 157 However, it remains unclear if such a biomedical model offers advantages over a biopsychosocial model as endorsed by other commentators. Similarly, the suggestion to embark on psychosocial talk only when cued by the patient (see Programme theory 7)157 receives an equivocal response from the literature and from practitioners. For example, initiating psychosocial talk only when cued by the patient contrasts with reattribution approaches that specifically aim to ‘encourage people to reattribute their MUS to physiological or psychosocial causes rather than to somatic causes’. 294
Another CBT-based approach to MUS is problem-solving treatment. The aim is to reduce complaints associated with unresolved problems in daily life by enhancing a person’s problem-solving capacities in a step-by-step manner. This approach is not explicitly addressed by any of our focal programme theories. Nevertheless, it is firmly grounded implicitly in several of these programme theory assumptions. For example, maintenance of a healthy professional–patient relationship (programme theory 3) is essential in creating a climate in which the patient is prepared to share their problems. Similarly, being provided with a ‘good enough’ explanation may allow a patient to metaphorically draw a line under a potentially futile quest for a definitive explanation and to move on to adopt self-management and problem-solving strategies.
Neither does this staged approach acknowledge a recursive and unproductive loop, with the patient requiring legitimation of their symptoms through a definitive result from a diagnostic test in order to be admitted to the treatment pathway (see programme theory 6, below). Where the diagnosis cycle is repeated, particularly with non-productive results, then not only does the patient feel frustrated, but concerns about the professional–patient relationship (programme theory 3) may be exacerbated. The patient may start to question the skills, abilities and resources possessed by their GP or may feel that they are becoming a ‘guinea pig’295 for diagnostic testing or that they are being ‘fobbed off’. 296 The ‘consultation letter’ intervention259 includes three specific provisions that can be seen as tackling some of these concerns. Unnecessary tests are to be avoided (programme theory 6), the patient is to be monitored regularly and a physical examination is to be conducted at each consultation. This last provision may help to confirm that the professional is still taking physical symptoms seriously (programme theory 1). 297 However, it is recognised that regular physical examination, outside the specific context of new symptoms, may be prohibitive in terms of time and may add little additional useful information. Collectively, avoidance of unnecessary tests and physical examination for new symptoms may serve to maintain a healthy relationship between GP and patient (programme theory 3) although, in this particular study, intervention by an external specialist (i.e. a psychiatrist) precipitated this series of measures. Focusing entry into the treatment pathway on symptoms, not on labels or test results, appears to be key to making progress.
The consultation letter approach259 includes three components that may contribute to the effect of the intervention and that could explain how the intervention might work:
-
confirmation of the persistent diagnosis MUPS
-
management rules for communication
-
management rules for case management.
It should be emphasised that it is the mechanisms that underpin this intervention that may be generalisable to primary care and not the exact intervention that was targeted at secondary care and delivered by psychiatrists. Hoedeman et al. 259 conclude that it is not known which of these components is effective or whether a synergistic effect exists.
Burton et al. 157 claim several advantages for components of his symptom clinic approach. Unlike the consultation letter approach, the symptom clinic model aims to negotiate a ‘medical’ explanation for symptoms involving physiological processes (programme theory 8), thus reducing uncertainty and permitting an exit from the diagnostic cycle (programme theory 6). They describe how this explanation is followed up over a series of shorter consultations, offering the prospect of the professional demonstrating belief in the patient (programme theory 1), of maintaining a healthy professional–patient relationship (programme theory 3) and, above all, of negotiating a shared mental model (programme theory 5). However, as demonstrated throughout this chapter, avoidance of psychosocial cues unless initiated by the patient is more equivocal and does not receive substantive support from within the literature as a whole.
In addition, Burton et al. ’s157 underlying epidemiological approach, identifying patients from a combination of their symptoms and health-care resource use, may be felt to negate recognition of the individualised experience of MUS. Individual symptoms may be common but may present in an almost infinite variety of combinations. Patients may therefore resist any attempt to pigeonhole their experience through strictly epidemiological means.
Another group of therapies offered to people with MUS, and examined in a Cochrane review,77 relates to ‘enhanced care’. Usual care, typically delivered by GPs, is enhanced by the addition of ‘participant education, structured counselling moments, a psychiatric interview, or a reattribution training of the doctor’. 77 These therapies offer no specific treatment agenda or structure. Instead, they aim to offer the person tools to assist them in the recovery process, stimulating self-management (programme theory 8).
Although validation of patient symptoms is primarily invoked within the context of the patient–professional consultation,298 other types of treatment offer alternative forms of validation. 179,182 Group-based sessions can lead to the discovery of a shared experience thereby legitimating patients’ symptoms. Such findings have been reported for fibromyalgia,299 low back pain,300 CFS,301 pelvic pain227 and MUS. 302 However, group sessions do not privilege a particular worldview and, therefore, dissonant explanations and disease models across a group of patients may prove threatening or divisive. As with other treatments, the patient often requires some form of ‘label’ to access a pathway leading to group treatment, again emphasising the potential value of employing a symptom-based approach (programme theory 2) rather than assigning patients by label or test result. Notably, the BodyMind Approach focuses only on symptoms and does not utilise labels. 186 In doing so it acknowledges the personalised and individualised nature of the patient experience, harnessing an almost exclusively somatic frame of reference when offering potential physical explanations (negating programme theory 7). Searches of the literature identified no RCTs of this intervention.
Summary statements
Programme theory 1: take concerns seriously
There is consistent evidence across multiple studies and intervention packages that it is necessary for the GP to take patients’ concerns seriously both at the time of initial consultation and at subsequent visits.
Programme theory 2: focus on symptom management
Many interventions focus on symptoms rather than labels. Evidence for the effect of patients being given a label in general, or a specific diagnostic label, is equivocal and dependent on the needs of the patient, public perceptions of particular labels and the link with subsequent action/inaction.
Programme theory 3: maintain positive relationship
Irrespective of outcome, a GP must seek to maintain a positive relationship characterised by the presence of empathy, trust and good communication.
Programme theory 4: provide a context-sensitive response
Claimed benefits for treating MUS in a primary care setting include enhanced knowledge of the patient and their particular circumstances. However, evidence for these advantages of the primary care setting is equivocal, with some patients feeling that a GP does not have the skills, expertise or resources to manage their symptoms. In turn, doctors may feel that the practical constraints of the consultation, against a backdrop of moves to extended opening and a consequent reduction in access to a personal GP, may run counter to such a direction of travel. Access to psychological practitioners in primary care, as under the IAPT programme, may help to ameliorate such concerns.
Programme theory 5: share congruent disease models
The literature is consistent in highlighting the shortcomings of the biomedical model for MUS. A cognitive–behavioural metamodel136 offers an alternative way of explaining what is happening that may be useful in bridging the incompatibility of professional and patient models.
Programme theory 6: avoid futile diagnostic cycle
There is widespread agreement in the literature, and incorporated within the components of particular interventions, that the HP should seek to avoid perpetuating an unhelpful cycle of tests. However, professional and patient attitudes towards tests or referral are complex and may be construed positively or negatively. In some cases, a diagnostic label can function as a pass key to a particular treatment pathway.
Programme theory 7: consider the appropriateness of initiating psychosocial cues
We were unable to identify definitive evidence in support of this initial programme theory. The literature is equivocal regarding the value of the GP making a connection between the physical symptoms and psychosocial factors. Some treatments are predicated on this connection whereas others purposefully circumvent it. It is likely that this factor is very dependent on the context in which these cues operate. It can be helpful not to exclude this possibility too prematurely. Patient-initiated psychosocial cues have been shown to be more prevalent than doctors think. Further research is required to explore those contexts in which initiation of psychosocial cues by the GP is helpful and those where it might be more appropriate to have such issues raised by the patient. At present, recommendations for each approach in the literature appear determined by the prevailing philosophy of the intervention and not by more salient contextual characteristics of the doctor–patient consultation and ongoing interaction.
Programme theory 8: offer useful explanations
A GP must seek a balance between their expected role of medical authority and the considerable uncertainty that accompanies MUS. Plausible and believable explanations, constructed through negotiation between GP and patient, can offer a contingent way forward, particularly in encouraging an expectation of coping rather than curing.
Discussion of the realist review
The therapeutic relationship is key to a satisfactory interaction between patient and HP and to successful engagement with the treatment. Establishing trust, believing and valuing the patient and developing an empathetic style of communication may all contribute to a positive experience, irrespective of treatment outcome.
Patient and professionals must feel that they are not entering into a destructive diagnostic cycle that may lead to unnecessary or uninformative tests or to referral to secondary care. Such a cycle may lead to excess health-care costs, unnecessary anxiety and frustration and, ultimately, may threaten the therapeutic relationship.
It is challenging to try to extricate the differential effects of a productive consultation, interaction according to behavioural principles and a manualised approach to behavioural modification. Attention must be directed to all three if one is to optimise clinician–patient interaction.
A CBT metamodel136 may offer a useful frame for analysis of an individual presentation and a tool for communication and a constructive dialogue. This may challenge existing mental models, relieve epistemological incongruence and acknowledge individual differences in MUPS.
Limitations
The realist review served a subsidiary, explanatory function for the overall intervention-focused HTA. It focused on a UK context and accessed large bodies of qualitative data through existing rich metasyntheses (Table 17), including the tailored qualitative systematic review reported in Chapter 4. It was not possible to engage with all available primary qualitative studies (Table 18) and the focus was necessarily limited to interventions trialled within a UK setting. Descriptions of interventions were variable, making it challenging to identify intervention components and putative mechanisms of effect. Lewis184 describes how many published RCTs give only brief descriptions of the techniques used, with treatment manuals being rarely published.
| Study (first author and year of publication) | Patient group | Number of included studies/references | Principal themes |
|---|---|---|---|
| Anderson, 2012303 | Myalgic encephalomyelitis/CFS | 34 qualitative studies | Three substantive thematic areas:
|
| Bayliss, 2014165 | CFS/myalgic encephalomyelitis | 21 studies | HPs report a limited understanding of CFS/myalgic encephalomyelitis. Working within the biomedical model has led some GPs to be sceptical about the existence of CFS/myalgic encephalomyelitis. GPs who provide a diagnosis tend to have a broader, multifactorial, model of CFS/myalgic encephalomyelitis and more positive attitudes towards CFS/myalgic encephalomyelitis. GPs collaborate with patients to reach agreement on symptom management, and use their therapeutic skills to promote self-care |
| Burton, 2003304 | MUPS | 77 references | Neither somatised mental distress nor somatisation disorders, based on symptom counts, account for most MUPS patients. With substantial overlap between symptoms/syndromes, suggesting much commonality, patients with MUPS may best be viewed as having complex adaptive systems in which cognitive and physiological processes interact with each other and with their environment. CBT and antidepressant drugs can be effective. Their effects may be greatest when the patient feels empowered by their doctor to tackle their problem |
| Drachler, 2009305 | CFS/myalgic encephalomyelitis | 32 quantitative and qualitative studies | Identified support needs:
|
| Edwards, 201031 | MUS | 109 references | No single approach will be effective in all patients. Chronic, high-utilising patients with MUS need care that is patient centred and attentive to their biopsychosocial needs. Requires careful assessment of the following: psychological concerns, family and cultural issues, a history of a dysfunctional childhood and symptoms of depression, anxiety and post-traumatic stress disorder. To be followed with confirmation that symptoms are real even when linked to psychosocial stress. GPs should clarify appropriate options for individual patients [e.g. reattribution, progressive muscle relaxation/related techniques, CBT (group/individual, by GP/mental health clinician) or medication]. Clinicians require patience and empathic communication. Multiple methods available but territory is unfamiliar/uncharted. Time taken to develop one’s clinical approach can contribute to welfare of patients and their family and to GPs’ personal and professional growth |
| Gask, 201185 | MUS | 25 publications from 13 studies | The reattribution model is too simplistic in its current form to address the needs of many people presenting with MUS in primary care. Reattribution of physical symptoms to psychological causes is often unnecessary. Further research is required into the effectiveness of stepped and collaborative care models that include the education of primary care practitioners. The consultation process is best seen as a conversation and ongoing negotiation between doctor and patient in which there are no certainties about the presence/absence of organic pathology |
| Hubley, 2016277 | MUS | 62 references | Challenges include complex symptom presentations, strained patient–physician relationships and treatment-resistant symptoms that challenge clinician competency. Emphasises importance of co-creating plausible explanations for MUS, understanding the pitfalls of consultations involving MUS and developing multimodal treatment plans |
| Johansen, 2017242 | MUS | 13 studies | Epistemological incongruence between dominant disease models and reality of meeting patients suffering from persistent illness. GPs have used flexible approaches to manage the situation, yet patients and doctors have parallel negative experiences of being stuck, untrustworthy and helpless. When facing cognitive incongruence, GPs strive to achieve relational congruence with patients. This has led to parallel positive experiences of mutual trust and validation. With more experience, some GPs overcome incongruences; later studies point towards reframing of problem |
| Larun, 2007301 | CFS | 20 studies | Symptom experiences and responses from significant others could jeopardise the patient’s sense of identity; feeling severely ill, yet blamed and dismissed. Patients’ beliefs and causal attributions oppose the doctor’s understanding of CFS. For the patient, getting a diagnosis and knowing more was necessary for recovery. Doctors were reluctant towards the diagnosis, and struggle to maintain professional authority. For patients, experience of discreditation could lead to withdrawal and behavioural disengagement |
| MacNeela, 2013187 | Chronic low back pain | 38 studies | Proposes four themes:
|
| Murray, 2016241 | Somatoform disorders | 42 studies | Identified 379 barriers within 77 barrier-level codes, 16 thematic categories and five overarching themes (i.e. patient-related, primary care practitioner-related, doctor–patient interactional, situational, and conceptual and operational barriers |
| Pinxsterhuis, 2015306 | CFS | 15 studies | Coping strategies, including activity management and the use of cognitive and emotional strategies, and psychological processes, such as acceptance and the rebuilding of identities and lives, may promote coping with CFS. Use of adequate coping strategies facilitated by progress in these psychological processes. Coping facilitated mainly by self-management, occasionally complemented by treatments and social support |
| Toye, 2013307–309 | Chronic non-malignant musculoskeletal pain; fibromyalgia | 77 studies of musculoskeletal pain including 28 on fibromyalgia | Concept of adversarial struggle (includes struggle to affirm self and construct self over time; find an explanation for pain; negotiate the health-care system while feeling compelled to stay in it; be valued and believed; and find the right balance between sick/well and hiding/showing pain). Some people move forward by listening to their body rather than fighting it, letting go of old self and finding a new self, becoming part of a community and not feeling like the only one, telling others about pain and redefining relationships, realising that pain is here to stay rather than focusing on diagnosis and cure, and becoming the expert and making choices |
| van Ravenzwaaij, 2011247 | MUS | 24 papers | Identified nine specific explanatory models of MUS: somatosensory amplification, sensitisation, sensitivity, immune system sensitisation, endocrine dysregulation, signal filter model, illness behaviour model, autonomous nervous system dysfunction and abnormal proprioception. Explanatory models focus on four domains: somatic causes, perception, illness behaviour and predisposition. One metamodel, incorporates all four domains: the CBT model. May help GPs in providing explanations to patients |
| Study (first author and year of publication) | Publication type | Contribution |
|---|---|---|
| Aiarzaguena, 2007136 | RCT | Primary care-based intervention |
| Aiarzaguena, 2009310 | RCT | Primary care-based intervention |
| Arnold, 2009311 | Controlled trial | Primary care-based intervention |
| Blane, 2014312 | ||
| Chew-Graham, 2017313 | Editorial | |
| Corbett, 2007300 | ||
| Deary, 200744 | Conceptual model | |
| den Boeft, 2017314 | Doctor–patient communication | |
| den Boeft, 2016315 | ||
| den Boeft, 2017316 | ||
| Dimsdale, 201319 | Diagnostic background | |
| Finset, 2009317 | Doctor–patient communication | |
| Gerger, 201578 | Meta-analysis | |
| Graugaard, 2003318 | Doctor–patient communication | |
| Hoedeman, 2010259 | Systematic review | Cochrane review |
| Howman, 2016286 | GP trainee perspectives | |
| Huibers, 2007319 | Systematic review | Cochrane review |
| Jutel, 2010230 | Literature review | Effects of ‘labelling’ |
| Kromme, 2016320 | Doctor perspective | |
| Layard, 2006321 | Editorial | Trends in psychological treatment centres |
| Malins, 2016322 | ||
| McDermott, 2011323 | Qualitative research | Patient perspective |
| Picariello, 2017324 | Qualitative research | Patient perspective |
| Rask, 2014325 | Mixed methods | GP perspective |
| Reid, 2001261 | ||
| Rosendal, 201377 | Systematic review | Cochrane review |
| Schweickhardt, 2005326 | ||
| Shattock, 2013262 | Qualitative research | Trainee perspective |
| Sirri, 2017327 | Qualitative research | Doctor perspective |
| Stone, 2014235 | Qualitative research | Trainers’ and trainees’ perspective |
| van Dessel, 2014294 | Systematic review | Cochrane review |
| Weiland, 2015291 | RCT | Doctor training programme |
| Ziadni, 2018328 | RCT |
Nevertheless, the realist review has expanded the analytical lens beyond the UK intervention studies included in the qualitative systematic review and extends beyond delivery of the intervention to include the entire pathway from initiation of the consultation through to resolution or referral. In particular, the realist review has engaged at a higher level of theorising with a view to helping to highlight promising candidate interventions or components. A valuable contribution of the realist review is the completion of the TIDieR templates for each of the focal interventions150 (see Appendix 6). The TIDieR checklist and guide was developed to improve the completeness of reporting, and ultimately the replicability, of interventions and can be used by reviewers to assess completeness of intervention descriptions and by readers who want to access more detailed information on the focal interventions.
Conclusions
This realist review was able to identify eight key programme theories (Table 19) that underpin the likely success of current UK-based interventions as explored in clinical trials (see Table 70). Confirmatory evidence was found for the majority of these programme theories, primarily in qualitative research and expert commentary. Few of these component theories have been explored empirically and, indeed, many of them would challenge current trial design. The two remaining programme theories, regarding the role of labels (programme theory 2) and avoidance of psychosocial cues unless initiated by the patient (programme theory 7), present more equivocal evidence – with the literature and our clinical experts equally being split on both the theoretical underpinnings and the practical usefulness of such theories. Although it seems that the value of such strategies would ultimately be determined by the characteristics of the patient, the nature of their symptoms and the context of the clinician–patient interaction, it remains unclear how these strategies might be differentially selected according to patient need. Reasons for this include the fact that these strategies are inextricably tied up with particular types of intervention (e.g. reattribution or the primary care symptom clinic) and not to some underlying theory-informed differentiation. Nevertheless, this chapter has been able to identify where and to what extent current intervention components seem to engage with our prioritised programme theories (Table 20). In doing this, we extend the usefulness of the realist review beyond existing UK trials to include all interventions covered by the effectiveness review and inform possible future intervention design.
| Cluster identifier | PT1 | PT2 | PT3 | PT4 | PT5 | PT6 | PT7 | PT8 |
|---|---|---|---|---|---|---|---|---|
| IF HPs show that they take the patient and their concerns seriously, THEN the patient will engage with treatment | IF HPs focus on patient’s symptoms and how to manage them, THEN patient feels something is being done about their symptoms and that their symptoms are not being dismissed as all in the mind | IF HPs maintain a positive relationship with the patient, THEN professional and patient develop a shared understanding of the patient’s experience | IF diagnosis, management and treatment takes place within a primary care setting, THEN the patient experiences continuity of care that may lead to progress and resolution | IF the disease model advanced by the HP matches, or is compatible with, that of the patient, THEN the patient will accept treatment | IF patient and professional can avoid perpetuating extensive and unproductive diagnostic tests, THEN patient and professional achieve a sense of progress | IF HPs introduce psychosocial explanations, THEN the patient feels that they are being problematised | IF HPs offer an explanation that they believe is congruent with current facts and uncertainties THEN the patient, believes that it is possible to maintain the therapeutic relationship | |
| The BodyMind Approach | By validating the symptom helps patients to feel believed. Utilises patient’s strengths and resources rather than focusing on remediation and deficits (cp. other mental health approaches) | Validates the symptom [cp. other approaches that invalidate and/or negate symptoms (e.g. terms such as psychological therapies/psychosomatic conditions that result in patients not feeling believed)]. By starting where patient is, sensory, physical, bodily symptom is acknowledged and worked with as an ally, promoting positive reassociation with the body, which is often viewed as the ‘enemy’ | Alternative: relationship with other members of the group is important in combating isolation | Group sessions conducted in a non-stigmatising, non-medical venue in the community, designed to be more anonymous than GP setting or a room within psychological services setting | Frames sessions as social rather than medical model, promoting inclusion | Case study states that frequent referrals for tests and scans increase patient’s belief that there would be a physical, medical explanation | Negating: biopsychosocial model is fundamental. Relies on somatic awareness, a normal part of consciousness, to resolve body–mind dualism inherent in conventional multidisciplinary approaches | Works on principle of recovery, giving hope to people to live well with their symptom rather than promising a cure or having to learn to live with it. Empowers patient to self-manage their symptoms |
| FINE (2010) (pragmatic rehabilitation) | No details | Nurses give patients detailed physiological explanation of symptom patterns, followed by treatment programme focusing on graded exercise, sleep and relaxation | No details | No details | No details | No details | Patients keep a diary of their progress on the exercise programme, together with a note of their daily activities, rest and sleep, and any problems encountered and how they were dealt with. Diaries are reviewed at each contact with the nurse | Initial session, which provides patients with convincing explanations for their symptoms and also indicates how they can be targeted in treatment, is crucial to the success of the intervention |
| FINE (2010) (supportive listening) | Patients may improve if they feel that the therapist empathises with them and takes their concerns seriously. Aims to provide emotional support and validation for the patient | No details | Aims to develop a collaborative relationship in which the patient is held in unconditional positive regard, and encouraged to talk about his or her experience of CFS/myalgic encephalomyelitis and problems which he or she has in dealing with it | No details | No details | No details | No details | Patients encouraged to find their own approach to addressing their symptoms |
| IAPT long-term condition/MUPS85 | No details | Seeks to elicit patient beliefs and concerns about symptoms, provide positive diagnosis if possible, biopsychosocial explanation of symptoms | No details | Describes importance of centring on setting and integration with existing services. Physical and mental health-care provision should be co-located to address:
|
No details | Proposes that MUPS are caused by a self-perpetuating multifactorial cycle, based on interaction of different factors in several domains, including somatic (physical) aspects, cognitions (thoughts), behaviour, emotions and environment. Important not to continue to look for possible disease once rigorous diagnostic workup has been completed. Ongoing referral and testing increases patient anxiety and can be iatrogenic preventing patients from moving forward into appropriate treatment | GPs have an important role in ensuring people experiencing FSS engage with IAPT services. GPs are usually first point of contact for people with FSS. They may need to prepare patients who present with FSS for psychological therapies by explaining the biopsychosocial model of FSS | No details |
| MUST reattribution training (2007) | No details | No details | No details | No details | Aims to generate information to provide a simple three-stage psychological explanation (symptom, psychosocial problem, physiological or temporal mechanism linking symptom to psychosocial problem) for the patient’s MUS through negotiation between the GP and patient | No details | Explores social and family factors. explores possibility that physical symptoms might be linked to psychosocial factors. Links physical symptom to an underlying psychosocial or lifestyle issue using normalising physiological, temporal or social link | Three elements to explanation: the symptom (e.g. backache), the psychosocial issue (e.g. stress at home) and mechanism (e.g. muscles held tight by stress for long time start to ache) |
| (Primary) symptoms clinic intervention (2012/16) | No details | No specific attempt is made to screen for common mental disorders; however, patients are encouraged to describe their emotional responses to symptoms and other events, and diagnostic labels such as depression were discussed collaboratively with the patient rather than imposed by the doctor | No details | No details | Consultations are structured to first hear the patient’s experience of illness then to propose and negotiate constructive explanations of physical symptoms. Aims to negotiate a ‘medical’ explanation for symptoms involving physiological processes | Aims to reduce uncertainty and thereby permit an exit from the diagnostic cycle | Focuses on biological (including neurological and cognitive) mechanisms rather than psychological cause | GP explores acceptable explanations for symptoms in terms of biological (including neurological and cognitive) mechanisms rather than psychological cause |
| Intervention | PT1 | PT2 | PT3 | PT4 | PT5 | PT6 | PT7 | PT8 |
|---|---|---|---|---|---|---|---|---|
| The BodyMind Approach | ✓✓ | ✓✓ | ✓ | ✓✓ | ✓ | ✓ | ✗✗ | ✗ |
| FINE (pragmatic rehabilitation) | ? | ✓ | ? | ? | ✓ | ? | ✓ | ✓✓ |
| FINE (supportive listening) | ✓✓ | ? | ✓✓ | ? | ? | ? | ? | ✓ |
| IAPT | ✓✓ | ✓✓ | ✓ | ✓ | ? | ✓✓ | ✗✗ | ✓ |
| MUST | ✓✓ | ✓✓ | ✓ | ✓ | ✓ | ? | ✗✗ | ✓ |
| Symptom clinic intervention | ✓✓ | ✓✓ | ✓✓ | ? | ✓✓ | ✓✓ | ✓✓ | ✓✓ |
Chapter 6 Assessment of cost-effectiveness
Systematic review of existing cost-effectiveness evidence
Objective
A systematic review was conducted to identify published economic evaluations examining the cost-effectiveness of behavioural modification interventions in patients with MUS.
Inclusion/exclusion criteria
Populations, interventions and comparators were defined as per the clinical effectiveness review (see Chapter 3) with the additional requirement that the study population was restricted to patients treated in the UK NHS as estimates of resource use, and costs may not be transferable between different health-care settings. In terms of study design, we restricted the review to cost-effectiveness studies. In terms of outcomes, we restricted the review to studies that measured benefits using quality-adjusted life-years (QALYs), as the QALY has been defined by the National Institute for Health and Care Excellence (NICE) as the reference case measure for benefit for UK cost-effectiveness studies and the NICE methods guide provides guidance on the range of cost per QALY values that can be considered to represent good value for money within the UK NHS. 244 Cost–consequences studies that reported incremental costs but did not report benefits as QALYs were excluded from the cost-effectiveness review, but are narratively summarised in Narrative summary of cost-consequences studies as these studies provide evidence on whether behavioural modification interventions have a negative or positive impact on net health-care costs.
Search strategy
A systematic search strategy was developed in consultation with the review team, to identify economic evaluations relating to the defined population. The focus was on identifying studies in primary care or community-based settings; therefore, population terms were combined with terms to define the setting. A combination of free-text terms and thesaurus searching was used. Published methodological search filters329 to limit to study type (economic evaluation) were used where available. No other search limits were applied.
Searches were conducted in the following sources:
-
MEDLINE via OvidSP (1946–15 August 2016)
-
MEDLINE In-Process & Other Non-Indexed Citations and Epub Ahead of Print & MEDLINE® without Revisions via OvidSP (2013–15 August 2016)
-
EMBASE via Ovid SP (1974–25 August 2016)
-
CINAHL via EBSCOhost (1981–25 August 2016)
-
PsycINFO via OvidSP (1967–25 August 2016)
-
NHS EED via the Cochrane Library (1968–April 2015 – no longer updated, archive only searched 25 August 2015)
-
Science Citation Index via Web of Science (1900–25 August 2015)
-
Social Sciences Citation Index via Web of Science (1956–25 August 2015).
Searches for economic evaluations were conducted between 15 and 25 August 2016. The search results were imported into EndNote [Clarivate Analytics (formerly Thomson Reuters), Philadelphia, PA, USA] and subsequently filtered to identify UK studies, using terms from line 29 of the EU economies search filter,249 to search the EndNote library for potentially relevant references. Detailed search strategies are provided in Appendix 1.
Methods
The citations were sifted by a single reviewer. Full-text articles were examined for those studies that could not be excluded based on the title or the abstract and the reason for exclusion was recorded for all articles for which the full text was examined. Those studies that were also identified in the search for relevant RCTs were cross-checked to ensure that the population, intervention and comparator inclusion criteria were being consistently applied across the clinical effectiveness and cost-effectiveness reviews. The applicability of the studies to the NICE reference case and their methodological quality was addressed using the checklist provided in the NICE guidelines manual. 250
Results
The search identified 1606 unique citations, of which 1541 were excluded at either the title or the abstract searching stage. The exclusion of papers at each stage is summarised in Figure 60. One paper reported clinical outcomes for a study that reported the economic analysis in a separate paper. 112 Three papers were excluded because they reported a population not recruited from primary care. 252,330 Six papers were excluded because they reported a population that did not meet the requirements to be classified as MUS253–256,331,332 and 14 papers were excluded because they reported non-UK populations. 109,138,260,263,265–270,272,273,333,334 One study was excluded because it was not considered to be a behavioural modification intervention. 274 Six papers were classified as cost–consequence studies131,275,276,278–280 and 30 papers were classified as providing only costing or resource use data. Reasons for exclusion for each paper examined at full text are provided in Appendix 7.
FIGURE 60.
The PRISMA flow diagram for identification of cost-effectiveness studies.
Four papers met the criteria for inclusion in the review. 125,126,282,335 Two papers125,126 described results from the MUSICIAN trial at different periods of follow-up, and one paper335 was a conference abstract for the FINE study that was reported in full in another included paper. 282 Clinical outcomes for the FINE study were reported by Wearden et al. 111 and clinical outcomes for the MUSICIAN trial were reported in the same two papers that reported the economic evaluation. Therefore, our review identified two unique cost-effectiveness studies (referred to in this review as Beasley et al. 126 and Richardson et al. 282). The characteristics of the two included studies are summarised in Table 21.
| First author and year of publication | Population | Interventions and comparators | Study design | Incremental costs (£) (mean, 95% CI) | Incremental QALYs (mean, 95% CI) | Cost-effectiveness | Uncertainty | Applicability and limitationsa |
|---|---|---|---|---|---|---|---|---|
| Richardson, 2013282 | CFS/ME |
|
Within-trial analysis |
1 vs. 3 = 218 (–474 to 911) 2 vs. 3 = 460 (–250 to 1169) |
1 vs. 3 = –0.012 (–0.09 to 0.07) 2 vs. 3 = –0.042 (–0.12 to 0.04) |
1 and 2 were dominated by 3 |
Probability that 3 was cost-effective was 0.645 at £20,000 per QALY and 0.626 at £30,000 per QALY 1 not dominated by 3 for sensitivity analysis using complete cases (ICER for 1 vs. 3 = £39,583), but 2 remained dominated |
Directly applicable. Minor limitations |
| Beasley, 2015126 (short-term outcomes in McBeth, 2012125) | Chronic pain |
|
Within-trial analysisb |
1 vs. 4 = 574 (–441 to 1554) 2 vs. 4 = 1924 (782 to 3295) 3 vs. 4 = 1778 (690 to 3009) |
1 vs. 4 = 0.097 (–0.05 to 0.24) 2 vs. 4 = 0.025 (–0.10 to 0.15) 3 vs. 4 = 0.047 (–0.09 to 0.18) |
1 vs. 4 = £5917 2 dominated by 1 3 dominated by 1 |
1 had approximately a 75% chance of being optimal at £20,000 per QALY Sensitivity analysis using multiple imputation to handle missing data gave similar results with 1 vs. 4 having an ICER of £3957 and 2 and 3 remaining dominated by 1 |
Directly applicable. Minor limitations |
Each of the included analyses examined a different population, with Richardson et al. 282 examining patients with chronic fatigue and Beasley et al. 126 examining patients with chronic pain. Both of the studies included treatment as usual as a comparator and both evaluated more than one behavioural modification intervention. In terms of the interventions examined, there were some common elements, with both studies including a psychological intervention and both including an intervention that involved some form of activity or exercise. However, the interventions examined by Richardson et al. 282 have been categorised in the clinical review as GA and other psychotherapy, whereas the interventions examined by Beasley et al. 126 have been categorised as CBTHI and SES. In addition, Beasley et al. 126 examined a multimodal intervention combining both CBTHI and exercise, whereas Richardson et al. 282 did not examine any multimodal interventions. The two studies were not judged to be sufficiently similar in their population, interventions and comparators to allow a direct comparison of outcomes but similar issues were identified when assessing the applicability and quality of the studies. Therefore, a narrative summary of methodological quality across both studies is provided followed by a summary of results from each individual study.
There were significant similarities in the studies in terms of the methods employed. Both were within-trial economic analyses. The duration of the analysis period was 70 weeks for Richardson et al. 282 and 2 years for Beasley et al. ,126 which are both relatively short time horizons for a population with chronic symptoms. Both studies applied the discount rate recommended in the NICE methods guide (3.5% per annum). 244 Owing to the inclusion criteria of the review, both were conducted in the UK and valued outcomes using QALYs. Both measured HRQoL using the EQ-5D and restricted benefits to those accrued by patients. Both studies used patient self-report to assess resource use, which may be subject to recall bias.
Richardson et al. 282 stated that they took an NHS and PSS perspective. The approach taken by Beasley et al. 126 is consistent with an NHS perspective. In terms of the assessment of broader costs falling outside the stated perspective, Richardson et al. 282 examined the costs to patients and families, including lost working days, but excluded these from the cost-effectiveness analysis. Both studies evaluated the uncertainty surrounding the cost-effectiveness estimates by using bootstrapping to generate a cost-effectiveness acceptability curve (CEAC), which shows the probability that the intervention is cost-effective for various willingness-to-pay thresholds. Both studies used multiple imputation to handle missing data either in their base-case analysis or in a sensitivity analysis, but no other sensitivity analyses were reported. Overall, both of the studies were assessed to be directly applicable and to have minor limitations.
The MUSICIAN trial compared CBTHI alone, SES alone and CBTHI combined with SES against treatment as usual in patients with chronic widespread pain. This study was originally reported by McBeth et al. 125 using data from a 9-month follow-up, but here we focus on the results at 24 months post intervention, which were reported by Beasley et al. 126 When using the complete-case analysis, all three active strategies (i.e. CBTHI alone, SES alone and CBTHI with SES) provided more benefits than usual care but at an additional cost. However, SES alone and CBTHI with SES were both dominated by CBTHI alone, which had an incremental cost-effectiveness ratio (ICER) of £5917 compared with treatment as usual and approximately a 75% chance of being optimal at £20,000 per QALY. Sensitivity analysis using multiple imputations to handle missing data resulted in qualitatively similar results but with an ICER of £3957 for CBTHI compared with treatment as usual. These results were more favourable than the short-term results reported by McBeth et al. ,125 suggesting that the cost-effectiveness is improved by considering a longer time frame.
The FINE study282 compared pragmatic rehabilitation (GA), supportive listening (other psychotherapy) and treatment as usual in patients with CFS or myalgic encephalomyelitis/encephalitis. Richardson et al. 282 reported that both pragmatic rehabilitation (GA) and supportive listening (other psychotherapy) provided less benefit than treatment as usual but at an additional cost and, therefore, both strategies were dominated by treatment as usual, which had a 65.5% probability of being cost-effective at £20,000 per QALY. In the complete-case analysis, pragmatic rehabilitation (GA) was no longer dominated by treatment as usual but the ICER versus treatment as usual was £39,583. Supportive listening (other psychotherapy) remained dominated in the complete-case analysis.
Conclusions
The review found relatively few UK-based cost-effectiveness analyses in comparison with the number of RCTs included in the clinical effectiveness review. The studies that were identified both used a within-trial analysis to examine the short-term cost-effectiveness of behavioural modification interventions. Although both studies assessed the cost-effectiveness of behavioural modification interventions in patients with MUS, the two studies examined different behavioural modification interventions in different populations. Richardson et al. 282 found that neither GA nor other psychotherapy provided more benefit than usual care in patients with CFS/myalgic encephalomyelitis. Beasley et al. 126 found that CBTHI was cost-effective compared with usual care when valuing a QALY at £20,000 in patients with chronic pain. In addition Beasley et al. 126 found that CBTHI provided more benefit than both SES and a multimodal intervention combining both CBTHI and SES. The studies were all generally applicable to the stated aims of this review and were mostly of reasonable quality, although their short-term nature is a potential limitation.
Narrative summary of cost–consequences studies
Six of the papers131,275,276,278–280 excluded from the review of cost-effectiveness were classified as cost–consequence studies, as they reported the impact of behavioural modification interventions on both costs and benefits but did not measure benefits using QALYs and, therefore, did not meet the inclusion criteria for the review of cost-effectiveness studies. However, these studies do provide evidence regarding whether behavioural modification interventions result in an overall reduction or increase in health-care costs. Two of these papers276,278 reported results from the same study. This resulted in a total of five unique cost–consequence studies being included.
We applied relevant aspects of the applicability and quality criteria used in the cost-effectiveness review250 to the cost element of these studies. The findings for each study are provided in Appendix 8 and are narratively summarised below. The incremental costs reported by each study are summarised in Table 22 and discussed in turn for each study.
| First author and year of publication | Population | Interventions and comparators | Study design | Incremental costs (£) (mean, 95% CI) | Applicability and limitationsa |
|---|---|---|---|---|---|
| Chisholm, 2001275 | Chronic fatigue |
|
Within-trial analysis | 1 vs. 2 = –63 (–258 to 42) (health care only) | Directly applicable for health service cost outcomes. Minor limitations |
| Kennedy, 2005129 | IBS (not responded to drug therapy) |
|
Within-trial analysis | 1 vs. 2 = –64b (–201 to 83) at 12 months [does not include intervention costs, which were 308 (SD 202)] | Directly applicable for health service cost outcomes. Minor limitations |
| McCrone, 2004279 (clinical paper is Ridsdale, 2004113) | Chronic fatigue |
|
Within-trial analysis |
1 vs. 2 = 193 (–946 to 458), 0.589 probability of 2 being cost-saving vs. 1 Therapy (1 or 2) vs. 3 = 149 (–708 to 1001)c |
Partly applicable. Potentially serious limitations for the comparison against 3. Minor limitations for 1 vs. 2 |
| Robinson, 2006131 | IBS |
|
Within-trial analysis |
Effect of guidebook = –72.74 (–102.63 to –42.84); p = 0.000d Effect of self-help group (over and above effect of guidebook) = 7.53 (–20.41 to 35.48); p = 0.591 |
Directly applicable for health service cost outcomes. Potentially serious limitations |
| Sabes-Figuera, 2012280 (clinical paper is Ridsdale, 2012115) | Chronic fatigue |
|
Within-trial analysis |
1 vs. 3 = 261 (141 to 382) 2 vs. 3 = 423 (288 to 559) 1 vs. 2 = –202 (–362 to –43) |
Directly applicable for health service cost outcomes. Minor limitations |
All of the studies had a UK setting as this was one of the inclusion criteria. All five were economic evaluations conducted alongside a RCT, although one paper279 reported a non-randomised comparison against usual care in addition to the randomised comparison of two behavioural modification interventions. All five studies had a time horizon of ≤ 1 year and, therefore, costs were not discounted. The perspectives were often not reported explicitly. Robinson et al. 131 included only NHS costs, but the remaining four studies275,276,279,280 included costs for alternative/complementary therapies within health service costs. Three papers reported the broader costs falling on patients or their carers including lost employment275,276 or informal care-giving275,279 and one paper275 did not report health service costs separately from these societal costs. Four papers relied on patient self-report for resource using some form of the Client Service Receipt Inventory275,276,279,280 Robinson et al. 131 used GP records for primary care resource use but relied on patient self-report for secondary care consultations. Kennedy et al. 129 reported the intervention costs separately from the costs for other resource use. All five used bootstrapping to estimate the uncertainty around the difference in costs between trial arms. Four papers131,276,279,280 used regression analysis to estimate differences in costs and three papers131,279,280 adjusted for baseline costs or resource use in their regression analysis. All five studies reported using standard sources for unit costs, such as NHS Reference Costs336 or Personal Social Services Research Unit (PSSRU)337 costs, but two papers279,280 reporting using a nominal cost for the self-help booklet provided alongside usual care.
One study279 was considered to be partly applicable because, although it did report informal care separately for each treatment arm, it did not report incremental costs falling on the health service separately from total incremental costs, which included costs of informal care. This study was also considered to have potentially serious limitations, but only for the comparison against usual care with self-help booklet as this comparison was non-randomised. However, the randomised comparison of two behavioural modification interventions was considered to have only minor limitations. The other four studies131,275,276,280 were considered to be directly applicable because they reported health service costs separately from broader societal costs. Three of these four papers were considered to have minor limitations. 275,276,280 The remaining paper, by Robinsons et al. ,131 was found to have potentially serious limitations because it did not appear to include the costs of the intervention in the analysis.
Chisholm et al. 275 report results from a RCT that compared CBTHI with counselling (classified as other psychotherapy) for patients with chronic fatigue. The clinical results are reported by Ridsdale et al. 112 and have been included in the clinical effectiveness review. Costs at baseline are compared with costs at 6 months to estimate the change in costs for each therapy and the difference between arms is calculated for change from baseline. CBTHI resulted in a statistically significant increase in health-care costs from baseline (mean change in cost from baseline £129, 95% CI £23 to £242). Other psychotherapy (counselling) resulted in a non-significant increase from baseline (mean change in cost from baseline £65, 95% CI –£6 to £146). The difference in change from baseline for health-care costs was not statistically significantly different between the two active interventions (mean change in cost from baseline –£63, 95% CI –£258 to £42).
Kennedy et al. 129 reported the results of a RCT that compared CBTLI with mebeverine to mebeverine alone in patients who continue to report moderate to severe IBS symptoms after 4 weeks of drug therapy with mebeverine. Mean costs for resource use (other than that associated with CBT) were lower in the CBTLI with mebeverine arm at 12 months than in the mebeverine-alone arm, but the difference was not statistically significant. The cost of CBT was reported to be £308 (SD £202), but these costs do not appear to have been combined with the costs of other resource use to estimate the overall incremental cost for CBT with mebeverine versus mebeverine alone.
McCrone et al. 279 report results from a RCT that compared GA with CBTLI. A non-randomised comparison against usual care plus (self-help booklet) is also reported but was considered to have potentially serious limitations. The clinical results are reported by Ridsdale et al. 113 and the results from the randomised comparison have been included in the clinical effectiveness review. GA was found to have a lower mean cost than CBTLI but the difference was not statistically significant. Behavioural modification interventions (CBTLI or GA) were found to have higher costs than usual care plus (self-help booklet) when results from the two therapies were combined but the comparison against usual care plus was non-randomised.
Robinson et al. 131 reported a comparison of a self-help guidebook (guided self-help) and a self-help group in addition to a guidebook (multimodal combining guided self-help and RSSE) with usual care in patients with IBS. In addition to reporting the mean costs across each of the arms, regression was also used to estimate the individual impact on costs of the guidebook and the self-help group. The guidebook was found to have statistically significantly lower costs by £72.74 (a 40% reduction) but the addition of the self-help group was found not to have a statistically significant impact on costs. The results were not sensitive to whether or not the patients met the Rome criteria for IBS.
Sabes-Figuera et al. 280 report results from a RCT that compared two behavioural modification interventions [i.e. GA and counselling (classified as other psychotherapy)] against usual care plus self-help booklet (classified as guided self-help) for patients with chronic fatigue. The clinical results are reported by Ridsdale et al. 115 and have been included in the clinical effectiveness review. Both GA and other psychotherapy (counselling) resulted in additional costs relative to guided self-help. GA resulted in significantly lower costs than other psychotherapy (counselling). The authors conclude that GA dominates (has lower cost and better outcomes) other psychotherapy (counselling), but this conclusion is made on the basis of a single clinical effectiveness outcome measure and, therefore, should be interpreted with caution.
Overall, the impact of behavioural modification interventions on net health-care costs was found to vary between studies. Only two studies131,280 reported a statistically significant difference in health-care costs between trial arms. From the study by Robinson et al. ,131 we can conclude that the self-help guidebook used in this study is likely to be cost-saving for patients with IBS provided that it has a low cost per patient (i.e. < £72). From the study by Sabes-Figuera et al. ,280 we can conclude that the choice of behavioural modification intervention is important as some interventions may achieve similar effectiveness results while having a lower impact on overall costs than other interventions. However, two other studies275,279 were unable to show a statistically significant difference in overall health-care costs when comparing one behavioural modification against another. From the study by Kennedy et al. ,129 we can conclude that CBT does not lead to a statistically significant increase in health service costs over and above those associated with the intervention.
Independent economic assessment
The general approach taken in our independent economic assessment was to generate a within-study estimate of cost-effectiveness for individual trials included in the clinical effectiveness review. The four main steps required were to:
-
estimate the incremental costs relative to usual care of delivering the behavioural modification interventions (and any active comparator interventions) reported in the individual studies
-
estimate health-utility outcomes from the clinical outcomes reported in the individual studies
-
estimate the difference in benefits between trial arms in terms of the incremental QALYs gained
-
combine these to estimate the cost-effectiveness.
The methods for each of these steps are described in Inclusion/exclusion criteria to Methods, respectively, and the results are described in sections Results: intervention and comparator costs to Results: cost-effectiveness analysis. Information on the impact of interventions on other NHS resource use (i.e. resource not directly related to the delivery of the intervention) has also been incorporated in the estimates of cost-effectiveness described in Results: cost-effectiveness analysis, where this was identified within the economic studies described in Systematic review of existing cost-effectiveness evidence.
Our decision to focus on within-trial estimates of cost-effectiveness rather than using the outcomes of the NMA to conduct a single incremental analysis covering all of the different categories of behavioural modification interventions is discussed further in Chapter 7, Cost-effectiveness.
Methods used to estimate intervention costs
Each of the included studies was examined to determine the resources required to deliver the intervention and any active comparators (i.e. comparators that were not usual care/treatment as usual or waiting list control). We have assumed that usual care (or treatment as usual/waiting list control) has zero cost so that any intervention or active comparator is costed relative to usual care. Six studies100,113,118,151,159,160 had comparator arms that were categorised as ‘usual care plus’, and these have been costed where possible. Information was extracted on the number of sessions, duration of sessions and the health-care professionals involved in delivering or facilitating the intervention. For interventions delivered to groups, information was also extracted on the number of individuals who started the intervention and either the group size or the number of groups to allow an estimation of the average cost per patient. Unit costs were taken from the most recent edition of the PSSRU Unit Costs of Health and Social Care 2016338 with three exceptions: the unit costs for clinical psychologists, counsellors and mindfulness trainers were reported in the 2014 edition of PSSRU unit costs (Unit Costs of Health and Social Care 2014339) but not in the 2016 edition, so the costs from the 2014 edition have been uplifted to 2016 prices using the Hospital and Community Health Service Pay and Prices index. 338 We have used unit costs including qualification costs where these are reported. The ratio of direct to indirect time is the ratio of time spent in direct contact with patients relative to time spent on other activities. As the resource use in our studies is based on face-to-face time with patients, we have used the cost per minute of face-to-face time where this has been reported. Where this has not been reported we have used the ratio of direct to indirect time to estimate the cost per minute of face-to-face time. For example, the ratio of direct to indirect time for a nurse in general practice is 1 : 0.20, so for each hour spent in contact with a patient an additional 12 minutes is spent on other activities. Therefore, if the cost per working hour is £43 then the cost per minute of face-to-face time is £0.86 (= 1.2 × £43/60). An exception was made for the time that GPs spent training in the GP interventions, which included a training element. For these interventions, the training time was based on the cost per minute across all GP activity, but the time spent with patients was still based on the cost per minute of face-to-face time with patients. Unit costs are summarised in Table 23.
| Health-care staff | Unit cost | Ratio of direct to indirect time | Cost per minute (£)a | Source |
|---|---|---|---|---|
| CBT | £97 per 55 minutes of contact time | NAb | 1.76 | Unit Costs of Health and Social Care 2016 338 |
| Clinical psychologist | £138 per hour of client contact | NAb | 2.35c | Unit Costs of Health and Social Care 2014 339 |
| Counsellor | £50 per hour | Not reported | 0.85c | Unit Costs of Health and Social Care 2014 339 |
| Consultant, medicine | £135 per working hour | 1 : 0.61d | 3.62 | Unit Costs of Health and Social Care 2016 338 |
| Consultant, psychiatry | £138 per working hour | 1 : 0.61d | 3.70 | Unit Costs of Health and Social Care 2016 338 |
| GP (training) | £147 per hour of GMS activity | NA | 2.45 | Unit Costs of Health and Social Care 2016 338 |
| GP (treating) | £3.90 per minute of patient contact time | 1 : 0.61 | 3.90 | Unit Costs of Health and Social Care 2016 338 |
| Mindfulness-based cognitive therapy | £172 per 2-hour session | NAb | 1.47c | Unit Costs of Health and Social Care 2014 339 |
| Nurse practitioners | £61 per hour | 1 : 0.33 | 1.35 | Unit Costs of Health and Social Care 2016 338 |
| Physiotherapist | £45 per hour | 1 : 0.37 | 1.03 | Unit Costs of Health and Social Care 2016 338 |
| Primary care nurse | £43 per hour | 1 : 0.20 | 0.86 | Unit Costs of Health and Social Care 2016 338 |
| Social worker adult services | £98 per hour of client-related work | NAb | 1.63 | Unit Costs of Health and Social Care 2016 338 |
Interventions delivered to groups and interventions delivered to individuals have been split into separate categories as interventions delivered to groups are usually cheaper on a cost per patient basis. A list of the detailed assumptions used in the costing analysis is provided in Appendix 9.
Methods used to estimate utility values from study outcomes
The method for estimating treatment benefit recommended by NICE is to estimate the difference in QALYs between groups receiving different interventions. 244 Health utility is a measure of HRQoL on a scale of zero to one; one represents full health and zero represents a state equivalent to death. When calculating QALYs, the patient’s estimated survival is adjusted to reflect their health utility during that period of survival. Therefore, an estimate of health utility is needed to estimate treatment benefit using QALYs. Some generic measures of HRQoL, such as the EQ-5D, come with a tariff that allows utility to be estimated from an individual’s response to the EQ-5D using preferences obtained from a UK general population sample. The EQ-5D is NICE’s preferred measure of health utility for estimating QALYs. 244 When health-utility values based on the EQ-5D are not available, NICE recommends that these are estimated by using mapping functions that estimate EQ-5D utility values from the other HRQoL measure or health-related benefits observed in the trials. 244
It was our aim to measure health benefits using QALYs based on EQ-5D outcomes valued with UK preferences whenever this was feasible from the data collected, as this is consistent both with the NICE methods guide and with the data reported in those studies included in our systematic review of published cost-effectiveness studies. We examined the generic preference and non-preference-based measures of HRQoL reported in the included studies. For studies that did not report EQ-5D directly, we looked to identify mapping algorithms that would allow us to estimate EQ-5D utility values from the other measures of HRQoL reported. Studies that reported EQ-5D utility values obtained with a non-UK preference set were considered acceptable where it was not possible to obtain EQ-5D values obtained using the UK preference set. Estimates obtained from alternative generic quality-of-life instruments were also considered acceptable if they met NICE’s requirement that health-utility values should be based on a valuation of public preferences from a representative sample of the UK population using a choice-based method.
Summary of generic preference and non-preference-based measures reported
Both of the published cost-effectiveness analyses identified in our systematic review126,282 measured QALYs using the EQ-5D with UK preferences. A further six studies108,115,137,138,143,340 reported EQ-5D outcomes in their clinical or economic papers but did not meet the inclusion criteria for our systematic review of published cost-effectiveness analyses. Morriss et al. 108 reported both the EQ-5D utility index and resource use but did not conduct a full economic evaluation. Furthermore, the scale on which the EQ-5D score is reported does not appear to be the standard zero to one scale and, therefore, it is unclear if the value reported is based on the UK tariff or some other method for scoring the EQ-5D. Luciano et al. 138 (the economic paper for Alda et al. )137 present a full economic evaluation, but the setting was the USA rather than the UK. In addition, Luciano et al. 138 used a Spanish tariff rather than the UK tariff to calculate utility values from the EQ-5D. van Ravesteijn et al. 143 reported only the VAS part of the EQ-5D, which is not a preference-based measure of health utility and, therefore, cannot be used to calculate QALYs. The related economic paper341 reported QALY gains, but these were obtained using a different measure of health utility rather than the EQ-5D. A second study by Luciano et al. 138 also reported only the VAS part of the EQ-5D. van der Roer et al. 340 reported QALY gains calculated using the EQ-5D in their within-trial economic evaluation,340 but the utility values were estimated using the Dutch rather than the UK tariff. Ridsdale et al. 115 reported EQ-5D values at baseline only within the related economic paper,280 stating that there was an ‘unexpected unavailability of EQ-5D questionnaire data for the six-month follow-up’. Therefore, although eight studies reported some aspect of the EQ-5D, only the two studies included in the systematic review of cost-effectiveness analyses126,282 reported EQ-5D values that can be used to calculate QALYs based on UK preferences, although a further two studies reported EQ-5D utility values based on non-UK preferences. 137,340
Twenty-five of the studies included in the systematic review of clinical effectiveness reported some aspect of the Medical Outcomes Survey (MOS) Short Form questionnaire-36 items (SF-36). 97,103,106,107,111,112,114,119,125,131,136,140,141,143–147,149,151–153,156,342,343 This is a generic measure of HRQoL that is usually reported as either eight domain scores (i.e. physical functioning, role physical, role emotional, vitality, mental health, social functioning, bodily pain and general health) or a pair of higher-level summary scores described as the mental component summary (MCS) score and the physical component summary (PCS) score. An additional five studies reported outcomes from the MOS SF-12. 118,133,150,157,158 This is a shortened version of the SF-36 that is able to produce MCS and PCS scores similar to those reported by the SF-36 with less respondent burden. 344 The Short Form questionnaire-6 Dimensions (SF-6D) is a shortened version of the SF-36 that contains only six items. 345 A preference-based measure of utility exists for the SF-6D, which was derived using preferences expressed by a sample representative of the UK general population. Although this is not NICE’s preferred method for estimating utility, it has similarities with the EQ-5D in that it provides a measure of HRQoL that is based on public preferences from a representative sample of the UK population using a choice-based method. The study by van Ravesteijn et al. 143 used the SF-6D to estimate QALYs for the within-trial economic analysis, and one further study by Aiarzaguena et al. 136 reported outcomes using the SF-6D (although the utility values appear to have been reported on a 0 to 100 scale rather than a 0 to 1 scale).
Identification and selection of mapping algorithms
‘Mapping’ or ‘cross-walking’ algorithms exist that allow the expected outcomes from the EQ-5D instrument to be calculated from other generic or disease-specific HRQoL instruments. These have been systematically reviewed by Dakin et al. 346,347 and a database of instruments that map to the EQ-5D is provided on the website of the Health Economic Research Centre (HERC) at the University of Oxford (version 5.0, last updated 16 May 2016). We examined this database to identify algorithms that would allow us to map trial outcomes measured using the SF-36 to utility values measured using the EQ-5D with the UK tariff. We also examined the list of titles identified during a systematic review of mapping studies that was ongoing at ScHARR at the time to find any studies not picked up in the HERC database.
Six studies that mapped from the SF-36 to the EQ-5D were identified from the HERC database and no additional studies were identified from the list of titles taken from the ongoing ScHARR systematic review. One of these studies was discounted because it used preferences from a Korean population rather than a UK population. 348 Three others were discounted because they used a data set for a specific condition, which may limit the generalisability of the results to patients with a more diverse range of health conditions. 349–351 Of the two remaining papers, the paper by Ara and Brazier352 was selected in preference to the paper by Rowen et al. 353 because it specifically validated the algorithm for use in predicting mean EQ-5D scores from mean SF-36 domain scores in situations where the patient-level data are not available. It was also noted that the data sets used by Ara and Brazier352 covered a range of conditions, including some conditions, such as IBS, which would be managed in primary care, whereas the data set used by Rowen et al. 353 was limited to secondary care inpatients and outpatients and, therefore, it might be less generalisable to our target primary care population. Ara and Brazier352 examined seven different ordinary least squares (OLS) models and concluded the simplest model (referred to as eq1) was the preferred model for estimating group EQ-5D scores from group domain scores, whereas the model referred to as eq4, which contained second-order terms, was preferred for estimating between-group differences and changes over time. However, after discussion with the corresponding author, we decided that as there was little difference between the performances of these two models, it was best to use the simplest model, which contained only linear functions of the domain scores, as this allowed us to calculate changes from baseline and differences between groups where the absolute scores were not reported.
Eight studies that mapped from SF-12 to EQ-5D were identified from the HERC database and two additional studies were identified from the list of titles taken from the ongoing ScHARR systematic review. Of these 10 studies, five354–358 used US preferences for the EQ-5D and three359–361 mapped to the actual EQ-5D responses rather than to the preference-based EQ-5D index. The remaining three papers362–364 used US samples to derive the mapping algorithm, but they used UK preferences for the EQ-5D index as they were developed before the US preference set was published. All three studies used a similar methodology of OLS regression to map from the MCS and PCS scores of the SF-12 to the UK preference-based index for EQ-5D. One study used a sample (n = 240) from a single health centre to examine the validity of the EQ-5D in a low-income minority population. 364 The other two papers362,363 used larger general population samples and these were considered more generalisable to our target population. The algorithm reported by Lawrence and Fleishman362 was selected in preference to the one reported by Franks et al. 363 because Lawrence and Fleishman362 explicitly validated whether or not the algorithm could predict the mean EQ-5D score across a group from the mean PCS and MCS score for that group. This was carried out for groups of patients stratified by age and for groups stratified according to the presence of six different conditions. 362 Although the algorithm developed by Franks et al. 363 was validated using an external data set, the validation did not explicitly examine whether or not the algorithm could predict group mean EQ-5D scores for different groups of patients. 363 Lawrence and Fleishman362 examined two-, three- and six-variable algorithms but found that two- and three-variable models performed better than the six-variable model and there was not much difference in performance between the two- and three-variable models. 362 We therefore decided to use the two-variable model, which predicted the EQ-5D score as a linear function of MCS and PCS. This had the added advantage that differences in EQ-5D scores (over time or between groups) could be calculated from the differences in MCS and PCS when the absolute scores are not reported.
Twelve of the included studies reported outcomes on only one or two domains of the SF-36. 97,103,107,111,112,114,149,151–153,156,343 We considered whether or not it was feasible to use these outcomes to estimate changes in EQ-5D based on changes in a single SF-36 domain by using the linear algorithm presented by Ara and Brazier352 and assuming no changes on any other domains. The study by Wearden et al. 111 provided an opportunity to validate if this approach provided reasonable estimates as it reported only a single domain of the SF-36 in the clinical paper, but also reported EQ-5D scores in the accompanying economic analysis by Richardson et al. 282 We found that the changes in utility from baseline based on changes in the single domain of SF-36 (physical functioning in this case) were a factor of 10 smaller than the changes in utility measured on the EQ-5D. This discrepancy may be due to the intervention having an effect on multiple aspects of HRQoL and these effects not being constrained to the single domain reported, as changes in other domains may influence the overall EQ-5D score. For these reasons, the approach of mapping EQ-5D changes from changes in a single domain of the SF-36 was not considered to be valid and we did not explore applying this mapping approach any further.
Three studies reported SF-36 MCS and PCS scores but did not report the scores for all eight domains. A study by Maund et al. 351 used OLS regression to estimate a mapping algorithm for SF-36 MCS and PCS to EQ-5D in a population with shoulder pain. The median duration of shoulder pain in the population used to derive the algorithm was < 12 weeks, with population being acute pain and, therefore, this population would not have been categorised as having chronic pain according to the definition used in our clinical effectiveness review. Although other studies were chosen in preference to the study by Maund et al. 351 for mapping between eight domains of SF-36 to EQ-5D, no alternatives were found in the HERC database that mapped SF-36 MCS and PCS to EQ-5D. We used the data from McBeth et al. 125 to validate whether or not using the algorithm presented by Maund et al. 351 generated QALY gains similar to those estimated directly using the EQ-5D in the McBeth et al. 125 study. We found that the utility values estimated indirectly from the SF-36 MCS and PCS scores varied from those estimated directly by up to 20% with lower absolute utility values predicted. The QALY gains between active intervention and the treatment-as-usual arm were up to 70% smaller than those estimated directly and the ordering of interventions in terms of the QALY gain was not consistent, with combined CBTHI and GA having the largest QALY gain when measuring utility indirectly from SF-36 MCS and PCS scores and CBTHI alone having the largest QALY gain based on direct EQ-5D scores. For these reasons, we did not explore applying this mapping approach any further.
The paper by Franks et al. 363 suggested that algorithms that map MCS and PCS scores from the SF-12 to the EQ-5D may be suitable for use with summary scores from the SF-36, but Franks et al. 363 recommended that this approach be validated further. Given that the algorithm by Lawrence and Fleishman362 for mapping between SF-12 MCS and PCS scores and EQ-5D has been developed and validated in a data set covering a more diverse population than the algorithm reported by Maund et al. ,351 we decided to explore the validity of using the SF-36 MCS and PCS scores in the algorithm intended for SF-12 MCS and PCS scores developed by Lawrence and Fleishman. 362 Again, we validated this approach using the data from McBeth et al. ,125 which reported both EQ-5D and SF-36 MCS and PCS scores. We found that the utility values calculated from the SF-36 MCS and PCS scores using the algorithm published by Lawrence and Fleishman362 were not similar to those measured directly using the EQ-5D with differences of 34% to 38% in the absolute utility values predicted. Furthermore, the incremental QALY gains differed by up to 80%, and again the ordering of interventions in terms of the greatest QALY gain versus treatment as usual was not consistent with the directly measured EQ-5D values. Therefore, this mapping approach was not considered to be valid and was not explored further.
Overall, there were 19 studies that reported either EQ-5D (four studies reported utility using UK or non-UK preferences),111,125,137,340 the SF-6D (two studies),136,143 all eight domains of the SF-36 (eight studies)106,131,141,144–147,342 or both the MCS and PCS of the SF-12 (five studies). 118,133,150,157,158 There were 41 studies remaining either that did not report any generic measures of HRQoL or for which we were unable to estimate preference-based utilities from the generic measures of HRQoL reported using a mapping algorithm. We considered if it was feasible to estimate QALY gains for these remaining studies by mapping from a single key symptom measure, such as fatigue, pain or disability. However, this was considered to potentially have limitations similar to mapping from a single domain of the SF-36; an improvement in one symptom may underestimate the overall benefit if other symptoms improve but the opposite could be true if an intervention improved one symptom but caused others to worsen. We therefore decided that this was not a valid approach as it could provide misleading results.
In summary, utility values over time have been extracted for those studies reporting utility values directly from the EQ-5D (using either UK preferences or non-UK preferences and the SF-6D (using UK preferences). For those studies reporting all eight domains of the SF-36 or the MCS and PCS of the SF-12, utility values have been estimated using algorithms provided by Ara and Brazier352 and Lawrence and Fleishman,362 respectively.
For studies that reported both SF-6D outcomes and all eight domains of the SF-36, we have conducted a sensitivity analysis to determine whether or not the estimates of utilities differed substantially when using utility values based on mapping to the EQ-5D from the SF-36 rather than those obtained directly from the SF-6D.
Methods used to estimate quality-adjusted life-years from utility values
The QALY gains have been estimated by assuming a linear change in utility values between each reported value and using an area under the curve approach to estimate QALYs for each arm. Where the absolute utility values were not reported, we have used data on the change in utility values from baseline or data on the difference in utilities between arms to estimate the QALY gains relative to usual care (or the lowest cost comparator arm if no usual-care arm was included).
We calculated the QALY gains using both the raw utility values and the utility values adjusted for baseline differences. We adjusted for utility differences at baseline using one of the following methods. Where the difference between arms has been reported after adjustment for differences in baseline characteristics, these have been used in preference to the raw values. Where these data are not available, the baseline value in all arms has been set to that observed in the comparator arm and then values thereafter have been determined using the difference from baseline for each study arm.
The time frame for estimating QALYs has been restricted to the last time point with comparative utility data for each study. Given the differences in populations and interventions across the studies being considered, it did not seem reasonable to use data from studies with a longer follow-up period to make predictions about the long-term persistence of treatment effects beyond the study period for studies with shorter follow-up times. Only two studies reported utility values beyond 12 months. Both of these were within-trial economic evaluations that reported QALY gains after discounting at 3.5% per annum. 111,125,126,282 Therefore, no further adjustments for discounting were required.
A full probabilistic assessment of the uncertainty around the estimates of QALY gains has not been conducted. This is because to do so for those studies where utility values have been estimated by applying a mapping algorithm to the mean SF-36 or SF-12 scores would have required the covariance matrix for the regression coefficients, which we did not have. Instead, we have used the estimates of variance for the SF-36 and SF-12 scores to derive the IQR for each score used in the mapping algorithm. We have then estimated the output of the mapping algorithm when all of the SF-36 (or both SF-12) scores are at their lower IQR and the output when they are all at their upper IQR. These utility values have been used as the upper and lower IQR for the utility values. The utility values in the comparator arm were kept fixed at their mean values and the lower and upper IQR for the QALY gain versus the comparator were calculated by setting the utility values in the intervention arm to their upper and lower IQR. (Note: if the lower IQR for both intervention and comparator are compared then the incremental QALY gains remain broadly similar as both the intervention and the comparator move a similar amount in a similar direction.) The IQR was used instead of the 95% CI because our method assumes that all eight domains of the SF-36 (or both summary component scores of the SF-12) are perfectly correlated (i.e. change in the same direction all of the time), which is unlikely to be true. So while the variability in the QALY gains is presented as an IQR, it is possible that the variance for the QALY gains is overestimated. Using the broader 95% CI as inputs to the mapping algorithm would have further exacerbated this potential overestimation of variance. Where studies have reported the 95% CI for QALYs directly, these have been used in preference to our IQR estimate.
Methods used to estimate cost-effectiveness
We sought to obtain a within-trial estimate of cost-effectiveness for each of those studies for which we were able to obtain an estimate of incremental QALYs, either from utilities/QALYs reported directly in the study or from mapping from SF-36/SF-12 outcomes to utility values (described previously in this chapter). The estimates of incremental QALYs described in Results: estimates of quality-adjusted life-year gains have been combined with the estimates of intervention costs described in Results: comparator costs. Where there were data from the published cost-effectiveness studies and cost–consequence studies, described in Systematic review of existing cost-effectiveness evidence, to indicate that an intervention increased or decreased other NHS resource use (e.g. GP consultations, hospital visits, etc. not related directly to delivery of the intervention), these have also been incorporated in the incremental cost used to calculate the ICER. However, the estimates of intervention costs from earlier in this chapter have been used, rather than any estimates of intervention costs reported in the studies themselves, for two reasons: first, it allowed a consistent costing methodology to be employed across all studies and, second, it allowed us to apply up-to-date unit costs. A fully probabilistic assessment of uncertainty was not feasible as the QALY estimates were based on a mapping algorithm and no information was available on the correlation between different quality-of-life domains used as inputs to the mapping algorithm (see Results: estimates of health utility). Instead, the lower and upper IQR range for the incremental QALYs, estimated as described in Results: estimates of quality-adjusted life-year gains, were used to calculate an IQR for the ICERs. For studies involving more than two treatment options, the IQR for the ICER was calculated only versus the lowest cost treatment option (usual care or lowest cost active comparator if no usual-care arm).
Results: intervention and comparator costs
Tables of costs estimated for each behavioural modification intervention (or active comparator, e.g. medication) by intervention types are provided in Tables 77–96 of Appendix 11. Costs could not be estimated for three studies105,146,149 because no information was provided in the publications on the duration of sessions with health-care providers. In addition, costs for the training element of two GP interventions101,159,365 could not be estimated because of a lack of information on the nature of the training, so costs for these interventions exclude training costs and total costs are therefore underestimated. For three studies that were classified as guided self-help,115,131,133 one of the study arms consisted of providing materials to patients without any sessions with health-care providers. One of these studies reported a nominal cost for a CBT booklet of £5 in its economic analysis. 115 No costs were reported for the materials provided in the other two studies. 131,133 In the study by Smith et al. ,100 GPs were provided with a consultation letter that gave advice on how to manage patients with somatisation disorder, but no cost was provided for this consultation letter. Consultation letters were also provided to GPs as part of ‘usual care plus’ in four other studies but the associated costs were not provided and could not be estimated from the papers. 143,144,156,161 Two studies137,138 had active comparator arms that consisted of recommended pharmacological management, which was costed as £217.60 and £214.03 respectively for the 6-month trial period (see Appendix 11, Table 96, for details). In Kennedy et al. ,129 both the intervention and the comparator arms received medication, but this was not costed as the medication was considered to form part of usual care. Overall, a cost was estimated for at least one arm of 55 studies that included a total of 77 intervention or active comparator arms. The median costs according to intervention type are summarised in Table 24. For GP interventions, the costs are summarised both by the type of intervention (e.g. CBT, reattribution) and by the mode of delivery (e.g. whether or not there was GP training and whether the intervention involved any interventions delivered to individual patients or groups of patients). We have summarised the other interventions according to whether they are delivered to a group of patients or to individuals.
| Intervention type | Group or individual intervention for patients | Number of study armsb | Median cost (£) |
|---|---|---|---|
| CBT low intensity | Group | 1 | 123 |
| Individual | 5 | 317 | |
| CBT high intensity | Group | 4 | 216 |
| Individual | 4 | 794 | |
| Other psychotherapy | Group | 2 | 228 |
| Individual | 8 | 462 | |
| GA | Individual | 7 | 564 |
| SES | Group | 1 | 128 |
| Individual | 6 | 740 | |
| RSSE | Group | 7 | 62 |
| Individual | 1 | 212 | |
| Guided self-help | Group | 1 | 143 |
| Individual | 2 | 240 | |
| Multimodal | Group | 3 | 193 |
| Individual | 5 | 892 | |
| Pharmacological | Individual | 2 | 216 |
| GP interventions summarised by delivery mode | |||
| GP training | NA | 7 | 337 |
| GP training with group intervention for patients | Group | 1 | 1299 |
| GP interventions for individuals | Individual | 2 | 505 |
| GP training with individual GP intervention | Individual | 8 | 871 |
| GP interventions summarised by intervention type | |||
| CBT | One individual | 1 | 939 |
| Reattribution (including modified) | Individual | 5 | 793 |
| Multimodal | Three individual and one group | 4 | 1288 |
| MUS management | Individual | 7 | 337 |
| Other psychotherapy | Individual | 1 | 802 |
| All GP interventions | NA | 18 | 546 |
The costs for interventions are summarised in Figures 61–64 as box-and-whisker plots. The green box for each intervention type shows the IQR, with the line dividing it indicating the median value. The upper and lower whiskers show the maximum and minimum values for interventions of that type. Means are shown on the plot as individual points sitting on the line between the maximum and minimum values.
FIGURE 61.
Distribution of costs for interventions delivered to individuals.
FIGURE 62.
Distribution of costs for interventions delivered to groups.
FIGURE 63.
Distribution of costs for GP interventions by delivery model.
FIGURE 64.
Distribution of costs for GP interventions by type of intervention.
The costs for interventions delivered to individuals are summarised in Figure 61. It can be seen that within each intervention type there is typically a broad range of costs, with the greatest range being for other psychotherapy because of one very high-cost outlier. 96 There was also a wide IQR for the SES intervention type because of the results from two of the six interventions, which came from the same study,140 having a very high cost. However, the median values were less variable and ranged from £212 for RSSE to £892 for multimodal interventions.
A similar picture is seen for interventions delivered to groups in Figure 62. The median costs for group interventions range from £62 for RSSE to £227 for other (i.e. not CBT) psychotherapy interventions. The multimodal group intervention had a broad range of costs because of the results from one study, which had a very high cost (£2088), possibly because it combined both group and individual sessions. In general, the mean and median costs per patient for each intervention type are lower for interventions delivered to groups than for interventions delivered to individuals.
The costs for GP interventions by delivery mode are presented in Figure 63. The most common mode of delivery was one that involved training for GPs followed by a specific intervention for individual patients (eight study arms),101,136,153,154,159,160,311,343 which had a median cost £871 but a wide range of costs (£500–2561). It should be noted that training costs were not estimable for two of the studies included so the costs of this delivery model may have been underestimated. The next most common delivery mode was GP training with no specific intervention for patients (seven study arms),97,102,108,146,147,158,160 which had a median cost of £337 (range £160–821). The highest median cost was for GP training followed by a group intervention for patients, but the estimate was based on a single study. 147,365
When summarising the GP interventions according to the type of intervention (see Figure 64) rather than the mode of delivery, the most common intervention type was MUS management with a median cost of £337 and a range of £160–550. The other intervention types generally had higher costs and a greater spread of cost estimates, ranging from approximately £450 to £2500.
When considering all GP interventions (18 study arms), the median cost was £546 and the range was £160–2561. Part of the reason for the large variation in cost per patient is due to significant variation in the number of patients treated by each GP who was trained. In the studies that included a training element, the cost per patient is based on the number of patients treated by the GP within the study. However, this may overestimate the long-term cost per patient if GPs continue to use the training to improve their management of additional patients who were not recruited to the trial. Variation in the number of patients treated per GP trained is an important factor driving variation in the cost per patient. For example, van der Feltz-Cornelis et al. 160 included 18 GPs in the arm that received GP training but only 23 patients were enrolled in this arm. This was partly explained by a fall-off in recruitment once GPs discovered that they had been allocated to the control arm. For this reason, we used the number recruited in the intervention arm (n = 58) to estimate the cost per patient. However, even when using this latter figure, the cost per patient was very high in the control arm (i.e. £821). Conversely, Rosendal et al. 146 had a slightly higher duration of training but 506 patients were treated by 22 GPs, resulting in a much lower cost per patient (i.e. £160).
Results: estimates of health utility
Figures 65–72 show the profile of utility values over time for the 14 studies for which we were able to estimate absolute utility values either directly from the studies or via mapping from the SF-36 or the SF-12 outcomes reported. It can be seen from these figures that the utility at baseline is not always equal between arms taken from the same study, with clear baseline differences in 8 out of the 14 studies. 111,118,125,133,144,147,157,158
FIGURE 65.
Utility values for studies comparing CBTHI with usual care.
FIGURE 66.
Utility values for studies comparing GA with usual care or guided self-help.
FIGURE 67.
Utility values for studies comparing other psychotherapy with usual care.
FIGURE 68.
Utility values for studies comparing RSSE with usual care.
FIGURE 69.
Utility values for studies comparing guided self-help with usual care.
FIGURE 70.
Utility values for studies comparing SES with usual care.
FIGURE 71.
Utility values for studies comparing multimodal treatments with single modality treatments.
FIGURE 72.
Utility values for studies comparing GP interventions with usual care or other GP interventions.
In general, those interventions that achieved a difference in utility at short-term follow-up were able to maintain that utility gain in the long term. 118,125,147,150 However, given that the populations and interventions varied considerably across the studies, it was not considered reasonable to assume that a similar maintenance of treatment effects would be seen across other behavioural modification interventions where long-term data were lacking.
We were unable to estimate absolute utility values for 5 of the 19 studies identified as including EQ-5D, SF-6D, SF-12 or SF-36 data, so these studies are not included in Figures 65–72. 106,131,145,146,340 Three of these studies131,145,146 provided data that allowed us to estimate the utility difference between trial arms, by assuming a common utility at baseline. Pols and Battersby145 reported the difference from baseline for each study arm. Robinson et al. 131 reported the difference between trial arms at 12 months, and Rosendal et al. 146 reported the difference from baseline for each arm and the adjusted difference between trial arms, with the adjusted difference between trial arms being used in the analysis. For one study,106 which reported SF-36 outcomes, we were unable to apply the utility mapping algorithm from SF-36 to EQ-5D as a mixture of mean and median SF-36 domain scores. Therefore, no utility values or QALYs could be estimated for this one study. The remaining study reported incremental QALYs directly, in the accompanying economic paper,340 so it was not necessary to calculate absolute or incremental utility values.
A sensitivity analysis was conducted to explore whether or not the utility values differed substantially when using EQ-5D-based utility values obtained via mapping from the SF-36 compared with using utility values based directly on the SF-6D. The results for the studies by Aiarzaguena et al. 136 and van Ravesteijn et al. 143 are shown in Figures 73 and 74, respectively.
FIGURE 73.
Sensitivity analysis comparing utility values mapped from SF-36 to EQ-5D with those obtained directly from SF-6D for Aiarzaguena et al. 136
FIGURE 74.
Sensitivity analysis comparing utility values mapped from SF-36 to EQ-5D with those obtained directly from SF-6D for van Ravesteijn et al. 143
In the study by Aiarzaguena et al. ,136 the utility values obtained by mapping from the SF-36 to the EQ-5D and those obtained directly from the SF-6D are similar in the reattribution arm. In the modified reattribution arm, the values at baseline are lower when obtained by mapping from the SF-36 and the gain in utility values is slightly larger.
In the study by van Ravesteijn et al. ,143 the absolute utility values are higher in both arms when using the values obtained by mapping from the SF-36 to the EQ-5D. The ordering of the intervention and comparator arms in terms of absolute utility is different at final follow-up between the two methods for estimating health utility, with other psychotherapy having greater utility at final follow-up when using the values obtained directly from the SF-6D and usual care having greater utility at final follow-up when using the values mapped from SF-36 to EQ-5D. However, in both cases the differences between trial arms are small and the utility profiles cross during the duration of the study.
Results: estimates of quality-adjusted life-year gains
The QALY gains estimated for each study are provided in Table 25. It can be seen that the mean estimates of QALY gains are generally small and are consistently under 0.1 QALYs. In the five studies that directly reported the QALY gains and their 95% CI,111,125,126,137,143,282,340 it was found that there was no statistically significant difference in QALYs between trial arms. In the studies where the IQRs for the QALY gains were estimated from the IQRs for the SF-36/SF-12 outcomes, we found that the lower IQR was above zero for all but three comparisons. 144,146,147
| First author and year of publication | Treatment strategies | Utility measurement method | Baseline utility | Utility at last follow-up | ΔQALYs using unadjusted data | ΔQALYs using adjustment for baseline differences |
|---|---|---|---|---|---|---|
| Aiarzaguena, 2007136 |
|
SF-6D at baseline and 1 year |
|
|
1 vs. 2: 0.036 |
1 vs. 2: 0.019a (95% CI 0.002 to 0.036)a (mid-point of 0.028 when using EQ-5D mapped from SF-36) |
| Alda, 2011137 |
|
EQ-5D with Spanish tariff at baseline and 6 months and QALYs reported directly |
|
|
1 vs. 3: 0.02 2 vs. 3: 0.00 |
1 vs. 3: 0.02 (95% CI –0.00 to 0.03)b 2 vs. 3: 0.00 (95% CI –0.01 to 0.02)b |
| Burton, 2012157 |
|
EQ-5D mapped from SF-12 using Lawrence and Fleishman’s362 algorithm |
|
|
1 vs. 2: 0.008 | 1 vs. 2: 0.004 (IQR –0.015 to 0.023) |
| Cuesta-Vargas, 2012118 |
|
EQ-5D mapped from SF-12 using Lawrence and Fleishman’s362 algorithm |
|
|
NR | 1 vs. 2: 0.074c (IQR 0.049 to 0.100)d |
| Gili, 2014144 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 |
|
|
1 vs. 3: 0.164 2 vs. 3: 0.025 |
|
| Ho, 2012150 |
|
EQ-5D mapped from SF-12 using Lawrence and Fleishman’s362 algorithm |
|
|
1 vs. 2: 0.029 | 1 vs. 3: 0.027e (IQR 0.020 to 0.033)d |
| Larisch, 2004158 |
|
EQ-5D mapped from SF-12 using Lawrence and Fleishman’s362 algorithm |
|
|
1 vs. 2: –0.015 | 1 vs. 2: 0.041f (IQR 0.01 to 0.068)d |
| LeFort, 1998342 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 |
|
|
1 vs. 2: 0.009 | 1 vs. 2: 0.007e (IQR 0.005 to 0.009)d |
| Marques, 2015133 |
|
EQ-5D mapped from SF-12 using Lawrence and Fleishman’s362 algorithm |
|
|
1 vs. 2: 0.052 | 1 vs. 2: 0.020e (IQR 0.010 to 0.031)d |
| McBeth, 2012125 |
|
EQ-5D measured in study |
|
|
1 vs. 4: 0.1272 vs. 4: 0.2293 vs. 4: 0.151 |
1 vs. 4: 0.047 (95% CI –0.086 to 0.182)b 2 vs. 4: 0.097 (95% CI –0.048 to 0.240)b 3 vs. 4: 0.025 (95% CI –0.099 to 0.154)b (Discounted) |
| Peters, 2002106 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 | NEg | NEg | NEg | NEg |
| Pols, 2008145 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 | NEh | NEh | NEh | 1 vs. 2: –0.003i (IQR NE as no measure of variance presented) |
| Robinson, 2006131 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 | NEj | NEj | NEj | 2 vs. 3: 0.013i (IQR 0.006 to 0.020)d |
| Rosendal, 2007146 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 | NEj | NEj | NEj | 1 vs. 2: –0.018i (IQR –0.023 to –0.003)d |
| Schaefert, 2013147 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 |
|
|
1 vs. 2: 0.039 | 1 vs. 2: 0.008d (IQR –0.004 to 0.020)d |
| van der Roer, 2008340 |
|
QALY gains estimated from EQ-5D with Dutch tariff reported but not utility values | NE | NE | NE | 1 vs. 2: 0.03 (95% CI –0.06 to 0.12)b |
| van Ravesteijn, 2013143 |
|
SF-6D utility values reported in study |
|
|
1 vs. 2: 0.010k |
1 vs. 2: 0.012 (95% CI −0.019 to 0.041)b (mid-point of 0.015 when using EQ-5D mapped from SF-36) |
| Wearden, 2010111 |
|
EQ-5D reported in study |
|
|
1 vs. 3: 0.016 2 vs. 3: –0.019 |
1 vs. 3 = –0.012 (95% CI –0.088 to 0.065)b 2 vs. 3 = –0.042 (95% CI –0.122 to 0.038)b (Discounted) |
| Zonneveld, 2012141 |
|
EQ-5D mapped from SF-36 using Ara and Brazier352 |
|
2 vs. 1: 0.006l | 2 vs. 1: 0.007l (IQR 0.005 to 0.008)d |
A sensitivity analyses was conducted in which the utility values obtained by mapping from the SF-36 to the EQ-5D were used to calculate QALY gains for the two studies that reported SF-6D utility values. 136,143 For Aiarzaguena et al. 136 the QALY using the utility value obtained by mapping from SF-36 to EQ-5D was 0.028, after adjusting for baseline differences, whereas the equivalent estimate when using the SF-6D utility values reported directly in the paper was 0.019. For the study by van Ravesteijn et al. ,143 the QALY gain was calculated to be 0.015, after adjusting for baseline differences, whereas the equivalent figure when using the SF-6D was reported to be 0.012 in the paper. Therefore, the choice of method to obtain utility estimates does appear to be important and the mid-point ICERs for these two studies may be overestimated in the base-case analysis because of the use of utility values measured using the SF-6D rather than the EQ-5D.
Results: cost-effectiveness analysis
A summary of the cost-effectiveness evidence is provided in Table 26, with the detailed results for each study reported in Table 27. In the summary table, for each comparison we have recorded whether the mid-point ICER is under or over £30,000 per QALY (denoted by ‘< £30K’ or ‘> £30K’, respectively). This cut-off point has been chosen as it is the upper limit of the range considered to be cost-effective by NICE (£20,000 to £30,000 per QALY)244 so interventions with a mid-point ICER above this range are unlikely to be considered cost-effective when applying the NICE threshold. Where the mid-point ICER is under £30,000 but the upper range of the ICER is above £30,000, then the intervention may not be considered cost-effective because of the degree of uncertainty in the ICER. For this reason, where the IQR crosses £30,000, we have added ‘W’ to indicate that there is a wide IQR and there may be uncertainty regarding the cost-effectiveness. Comparisons where one intervention was dominated are also indicated (denoted by ‘Dom’); an intervention is dominated by another if it has higher costs but lower QALY gains.
| Intervention | UC | UC+ | CBTHI | GSH | SES | Graded activity | GP reattribution | GP MUS management |
|---|---|---|---|---|---|---|---|---|
| CBTHI |
< £30KW137 < £30KW125 > £30K141 |
< £30K144 Dom144 |
< £30K144a | |||||
| Other psychotherapy | Dom111 | < £30KW143 | Dom111 | |||||
| GA | Dom111 | < £30KW133 | ||||||
| SES | > £30K125 | Dom125 | ||||||
| GSH |
< £30KW131 < £30KW342 |
|||||||
| RSSE | < £30KW150 | |||||||
| GP modified reattribution | < £30KW158 | > £30KW136 | ||||||
| Pharmaceutical | Dom137 | |||||||
| Multimodal |
> £30K125 Dom145 |
Dom125 | < £30K118 |
< £30KW125 > £30KW340 |
||||
| GP MUS |
Dom146 > £30KW157 |
|||||||
| GP multimodal | > £30K147 |
| First author and year of publication | Treatment strategies | ΔQALYs using adjustment for baseline differences | Intervention costsa | Other costs | ICER | IQR for ICER |
|---|---|---|---|---|---|---|
| Aiarzaguena, 2007136 |
|
1 vs. 2: 0.019b (95% CI 0.002 to 0.036)b |
|
NR | 1 vs. 2: £39,142 (£26, 638 when using EQ-5D mapped from SF-36 instead of SF-6D) | 1 vs. 2: £20,719 to £353,214 |
| Alda, 2011137 |
|
1 vs. 3: 0.02 (95% CI 0.00 to 0.03)c 2 vs. 3: 0.00 (95% CI –0.01 to 0.02)c |
|
Non-UK resource use |
1 vs. 3: £10,025 2 vs. 1: 2 is dominated by 1 |
1 vs. 3: £6683 to > £40,000 |
| Burton, 2012157 |
|
1 vs. 2: 0.004 (IQR –0.103 to 0.018) |
|
NR | 1 vs. 2: £129,267 | 1 vs. 2: £20,658 to dominated |
| Cuesta-Vargas, 2012118 |
|
1 vs. 2: 0.074d (IQR 0.049 to 0.100)e |
|
NR | 2 vs. 1: £18,641 | 2 vs. 1: £13,883 to £28,261 |
| Gili, 2014144 |
|
|
NR |
1 vs. 3: £14,146 2 is dominated by 3 1 vs. 2: £9502 |
1 vs. 3: £10,928 to £20,049 2 vs. 3: £23,199 to dominated |
|
| Ho, 2012150 |
|
1 vs. 2: 0.027f (IQR 0.020 to 0.033)e |
|
NR | 1 vs. 2: £1397 (£23,048 if sessions are one to one) | 1 vs. 2: £1124 to £1843 (£18,558 to £30,406 if group sessions are two to one) |
| Larisch, 2004158 |
|
1 vs. 2: 0.041g (IQR 0.01 to 0.068)e |
|
Non-UK resource use | 1 vs. 2: £11,863 | 1 vs. 2: £7,115 to £47,022 |
| LeFort, 1998342 |
|
1 vs. 2: 0.007f (IQR 0.005 to 0.009)e |
|
NR | 1 vs. 2: £21,755 | 1 vs. 2: £16,632 to £31,437 |
| Marques, 2015133 |
|
1 vs. 2: 0.020f (IQR 0.010 to 0.031)e |
|
NR | 1 vs. 2: £27,894 | 1 vs. 2: £18.358 to £58,052 |
| McBeth, 2012125,126 |
|
1 vs. 4: 0.047 (95% CI –0.086 to 0.182)c 2 vs. 4: 0.097 (95% CI –0.048 to 0.240)c 3 vs. 4: 0.025 (95% CI –0.099 to 0.154)c (Discounted) |
Estimated from resource use:
|
2 vs. 4: £9797 3 and 1 are dominated by 2 3 vs. 4: £56,064 1 vs. 4: £39,387 1 vs. 3: £20,436 |
2 vs. 4: £3,959 to dominated | |
| Peters, 2002106 |
|
NEg |
|
No significant difference in resource use between arms (costs not reported) | NE | NE |
| Pols, 2008145 |
|
1 vs. 2: –0.003i (IQR NE as no measure of variance presented) |
|
Some resource use reported but non-UK | 1 dominated by 2 | NE |
| Robinson, 2006131 |
|
2 vs. 3: 0.013i (IQR 0.006 to 0.020)e 1 vs. 2: NE |
|
2 vs. 3: –£72.741 vs. 2: £7.53 |
2 vs. 3: NE (Guidebook would dominate UC if it cost under £72.74) 1 vs. 2: NE |
NE |
| Rosendal, 2007146 |
|
1 vs. 2: –0.018i (IQR –0.023 to –0.003)e |
|
NR | 1 is dominated by 2 | 1 vs. 2: dominated to dominated |
| Schaefert, 2013147 |
|
1 vs. 2: 0.008f (IQR –0.004 to 0.020)e |
|
Some resource use reported but non-UK | 1 vs. 2: £124,535 | 1 vs. 2: £51,184 to dominated |
| van der Roer, 2008340 |
|
1 vs. 2: 0.03 (95% CI –0.06 to 0.12)c |
|
Some resource use reported but non-UK | 1 vs. 2: £42,887 | 1 vs. 2: £10,722 to dominated |
| van Ravesteijn, 2013143 |
|
1 vs. 2: 0.012 (95% CI −0.019 to 0.041)c |
|
Some resource use reported but non-UK | 1 vs. 2: £13,825 (£11,171 when using EQ-5D mapped from SF-36 instead of SF-6D) | £4046 to dominated |
| Wearden, 2010111 |
|
1 vs. 3 = –0.012 (95% CI –0.088 to 0.065)c 2 vs. 3 = –0.042 (95% CI –0.122 to 0.038)c (Discounted) |
Estimated from resource use
|
1 vs. 3: dominated 2 vs. 3: dominated |
1 vs. 3: £2182 to dominated 2 vs. 3: £8653 to dominated |
|
| Zonneveld, 2012141 |
|
2 vs. 1: 0.007j (IQR 0.005 to 0.008)e |
|
NR | 1 vs. 2: £56,792 | 2 vs. 1: £45,400 to £76,816 |
There is a large degree of heterogeneity in the estimates of cost-effectiveness for individual studies, with some interventions having mid-point ICERs in the range that would normally be considered cost-effective by NICE (i.e. below £20,000 to £30,000 per QALY) and others having ICERs well above this range. Some interventions were also found to be dominated by usual care or dominated by other behavioural modification interventions.
Where there was more than one study providing an estimate of cost-effectiveness for a particular comparison, the evidence was often inconsistent. For example, the ICERs for two studies125,137 that compared CBTHI with usual care were below £20,000 with an upper IQR above £30,000, suggesting that CBTHI is potentially cost-effective. However, a third study141 had an ICER of £56,792 (IQR £45,400–£76,816). This was despite the Zonneveld et al. 141 study having a higher intervention cost than one of the other two studies, indicating a higher intensity of CBT. These inconsistencies may be driven by differences in the response of different study populations to CBT or differences in the delivery of the CBT that are not reflected in the cost estimates. In this case, the two studies with favourable mid-point ICERs125,137 were both studies in patients with chronic pain (fibromyalgia in both cases), whereas the study141 with the higher ICER had a broader population that included both patients with chronic pain and those with somatoform disorder.
High-intensity CBT was also found to have inconsistent cost-effectiveness compared with usual care plus, although it should be noted that this was based on a single study144 that found that individual CBTHI had an ICER < £30,000 compared with usual care plus, whereas group CBTHI was dominated by usual care plus. 144 This suggests that, although these interventions have been grouped together under the heading of CBTHI, the particular mode of delivery (i.e. group vs. individual) is capable of producing important differences in the estimates of benefits and the estimates of cost-effectiveness. Therefore, caution should be exercised when comparing interventions that have been grouped together but that may differ significantly.
The two studies that provide ICERs for guided self-help versus usual care suggest that this intervention has the potential to be cost-effective but there remains some uncertainty regarding its cost-effectiveness as the CIs crossed £30,000 per QALY. The study by Robinson et al. 131 found that guided self-help, delivered as a guidebook, was cost-saving and produced a QALY gain and, therefore, guided self-help would dominate usual care provided the guidebook cost less than the cost-saving (£72.74). It seems likely, therefore, that this intervention would be cost-effective but this cannot be confirmed without an estimate of the cost of the guidebook (the ICER has been summarised in Table 26 as ‘< £30KW’). The other study comparing guided self-help with usual care, by LeFort et al. ,342 gave a mid-point ICER of £21,775 with an IQR of £16,632–£31,437. These two studies also had different populations (i.e. IBS and chronic pain), suggesting that guided self-help has the potential to be cost-effective across a number of different subsets of the MUS population, but the cost-effectiveness remains uncertain.
The ICER for RSSE versus usual care based on the study by Ho et al. 150 was £1397 with a narrow IQR (£1124–£1843), but there was additional uncertainty related to this study because the intervention was described as being a group intervention but the group size was not indicated. The mid-point estimate was calculated assuming that the intervention was delivered to all patients at the same time in a single group but a sensitivity analysis exploring the impact of lower group sizes (two patients per group) gave a mid-point ICER of £23,048 (IQR £18,558–£30,406). For this reason, this study has been denoted as having wide uncertainty in Table 26.
Two studies provided estimates of the cost-effectiveness of GA in patients with chronic fatigue, but the comparator arms differed. The analysis based on Wearden et al. 111 found that GA was dominated by usual care. Conversely, the analysis based on Marques et al. 133 found a positive QALY gain but an ICER at the upper limit of the NICE cost-effectiveness range (£27,894) for GA compared with guided self-help, which in this study consisted of usual care plus the provision of information about physical activity. These two studies therefore appear to give inconsistent results. However, it should be noted that, in the study by Wearden et al. ,111 the QALY gains were estimated from directly measured EQ-5D utility values, whereas, in the study by Marques et al. ,133 the QALY gains were estimated via mapping from SF-36 MCS and PCS scores to EQ-5D utilities, which is associated with additional uncertainty.
The analysis based on the study by Wearden et al. 111 also found that GA dominated other psychotherapy, but other psychotherapy itself was also dominated by usual care. The only other study to examine other psychotherapy, by van Ravesteijn et al. ,143 had usual care plus as the comparator, and for this study we estimated a low mid-point ICER but a broad IQR, with other psychotherapy being dominated at the upper limit. The study by van Ravesteijn et al. 143 also had a broader MUS population than the study by Wearden et al,111 which was restricted to patients with chronic fatigue.
The evidence of cost-effectiveness for SES versus usual care was limited to a single study by McBeth et al. ,125 from which we estimated a high ICER versus usual care in patients with chronic pain. The same study also found that SES was dominated by CBTHI.
Pharmaceutical therapy was only included in one study, by Alda et al. ,137 and was found to be dominated by CBTHI in patients with chronic pain.
In general, the GP interventions were not particularly cost-effective. Modified reattribution had a high ICER versus reattribution in patients with MUS based on the study by Aiarzaguena et al. ,136 although the mid-point ICER was reduced to < £30,000 when EQ-5D utility values obtained by mapping from the SF-36 to the EQ-5D were used instead of using SF-6D utility values. GP MUS management was dominated by usual care in patients with somatoform disorder based on the study by Rosendal et al. ,146 and GP MUS management had a higher ICER (£129,000) with a wide IQR based on the pilot study by Burton et al. ,157 also in patients with MUS. GP multimodal intervention had a high ICER (£124,000) versus GP MUS management in patients with MUS, based on the study by Schaefert et al. 147 However, there was an exception for GP-modified reattribution compared with usual care that was found to have an ICER of £11,863 (IQR £7115–47,022) in patients with somatoform disorder based on the paper by Larisch et al. 158
Only one study118 compared a multimodal intervention combining guided self-help and SES with guided self-help alone in patients with back pain. The mid-point ICER was < £20,000 per QALY and the upper limit of the IQR for the ICER was < £30,000 per QALY, suggesting reasonable confidence in the cost-effectiveness of the multimodal intervention.
Two studies compared a multimodal intervention with SES in patients with chronic pain. van der Roer et al. 340 compared a multimodal intervention combining sport, education and a behavioural programme to SES in patients with back pain, but the ICER estimated for this comparison was above the range usually considered cost-effective. McBeth et al. 125 compared a combination of CBTHI with SES against SES alone in patients with chronic widespread pain. Multimodal intervention versus SES alone had an ICER in the range usually considered cost-effectiveness by NICE. However, the same study also included a CBTHI arm and multimodal therapy was dominated by CBTHI alone. Therefore, CBTHI would be the optimal intervention based on this study in preference to SES.
Conclusions and discussion
The behavioural modification interventions for patients with MUS identified in our review were found to have a very broad range of interventions costs. Interventions delivered to groups were generally found to have a cost per patient that is lower than the cost per patient for interventions of the same type delivered to individuals. For interventions delivered to groups, the cost per patient is dependent on the group size. In studies where the group size or number of groups was not reported, we have assumed that all patients received the intervention as part of a single group that may have underestimated the cost per patient if smaller groups were in fact used. The cost per patient for interventions that involve training GPs to manage MUS in a particular way is very dependent on the number of patients treated by each GP who is trained. The broad range of intervention costs is indicative of the high degree of heterogeneity in the behavioural modification interventions included in our review. In future research, it would be helpful to investigate whether or not more intensive interventions with higher costs generate greater QALY gains.
Utility estimates were often not equal at baseline, making adjustment for baseline differences necessary. Where outcomes adjusted for baseline differences were not reported in the study paper, adjustment for baseline differences was achieved by setting the starting utility of the intervention arm equal to that of the control arm and using either the difference from baseline or the difference between arms to estimate utility thereafter. This is a fairly crude method for adjusting for baseline differences and may have introduced bias, particularly if the utility changes that can be achieved in an individual are not independent of their baseline utility.
In general, those interventions that achieved a difference in utility at short-term follow-up were able to maintain that utility gain in the long term. However, as different interventions may have different capacities to achieve treatment effects that persist, we decided not to extrapolate the maintenance of treatment effects from those studies with long-term data to other studies with only short-term data. For this reason, our analysis of QALYs gained was limited to the last study time point reporting a randomised comparison of outcomes for each individual study.
Using a different time frame for each study was considered appropriate when the focus was on the within-trial estimates of cost-effectiveness for each study. However, the use of differing time frames for each study makes the estimates of QALY gains not comparable across studies. It would therefore not have been appropriate to have used these estimates of QALY gain to conduct an incremental analysis that compared interventions across multiple studies.
In general, the QALY gains estimated for behavioural modification interventions from individual studies were small, but our analysis may have underestimated the true long-term QALY gains if the differences between trial arms observed during the trial period persist in the long term beyond the trial follow-up.
The uncertainty in the QALY estimates has been captured in a fairly crude manner by using the upper and lower IQR for the HRQoL scores going into the utility mapping algorithm to explore the uncertainty in the incremental QALYs gained. The IQR values presented for the QALY gains are intended to give only an indication of the potential variation in the QALY gain that arises from uncertainty in the SF-36/SF-12 scores used within the mapping algorithm to calculate utilities. A more complex approach would have been necessary to allow a full probabilistic assessment of the uncertainty in the QALY gains but we felt that this may not provide a more unbiased estimate because of a lack of information on the correlation between changes in the various domain scores used in the mapping algorithm. Although our method assumes that changes in these domains are perfectly correlated and may therefore overestimate the variance, the use of the IQR rather than the full distribution for each input to the algorithm should mitigate any overestimation of the uncertainty. However, the true uncertainty around the QALY gains may be greater as our approach does not quantify uncertainty around the mapping algorithm itself.
This set of within-trial estimates of cost-effectiveness has several important limitations. The estimates of QALYs are based on the data and time points reported for each individual study rather than the outputs of the NMA. We have therefore not been able to benefit from pooling clinical efficacy data across studies, but this was considered to be a fairly minor limitation given that, in general, there were only one or two studies for each comparison in the NMA. The conclusions that can be drawn are also limited to the direct comparisons presented in the individual studies whereas the NMA allows indirect comparisons to be made regarding the clinical effectiveness of interventions not compared directly in a single study.
The included studies cover many different populations classified as having MUS including those with chronic pain, chronic fatigue, IBS and somatoform disorders. It is not clear from the evidence that we have identified if interventions that are cost-effective in one population would be similarly cost-effective in another population with MUS. Many of the interventions that have received the same intervention classification differ considerably in terms of their intensity or in terms of their content, owing to the broad intervention groupings. Although differences in intensity are easier to identify, as these usually translate into differences in cost, it may be harder to identify where differences in the content are responsible for inconsistencies in the cost-effectiveness evidence.
Very few of the studies reported UK estimates of resource use so the cost-effectiveness of the interventions may be overestimated or underestimated depending on whether the interventions reduced or increased health service resource use not directly related to the delivery of the intervention. Although several studies reported non-UK resource use or costs, we did not convert these to UK NHS costs as we were concerned that changes in the number of health-care contacts in one health-care system may not translate into an equivalent change in a different health-care system.
Finally, the estimates of uncertainty around the ICER should be considered to be providing only an indication of the potential uncertainty around the ICERs in a manner similar to a univariate sensitivity analysis because of the limitations in method used to estimate the uncertainty in the QALY gains, as described above.
We had originally planned to use the outcomes of the NMA to conduct an incremental cost-effectiveness analysis for each of the intervention categories included in the NMA. However, there were several issues with this approach. First, the individual interventions grouped within the same categories for the purposes of the NMA were often very different in terms of the resources that would be required to deliver them, as demonstrated in the costing analysis. Therefore, an individual estimate of cost-effectiveness would be required even if we assumed a consistent treatment effect across each category; an assumption that may not itself be justified given that we were not able to explore whether or not differences in the interventions or the population led to differences in effectiveness within the NMA because of the sparsity of the data informing the networks. Second, it was unclear how to generate a single estimate of health utility from the diverse range of outcomes analysed, particularly given that all of the outcomes have the potential to influence quality of life and their influence may not be independent of each other. This complication was avoided in the within-study estimates of cost-effectiveness by limiting the analysis to those studies that provided either a direct estimate of health utility (or incremental QALYs) or a generic measure of HRQoL that could be mapped to provide an estimate of health utility. The main disadvantage of the approach that we have used is that we have not been able to provide estimates of cost-effectiveness for all of the comparisons represented within the full set of clinical trials, as not all of the studies reported a generic measure of HRQoL.
Chapter 7 Discussion
Statement of principal findings
A total of 59 studies were identified that met the inclusion criteria for the clinical effectiveness review. These studies covered a broad and diverse range of populations, interventions and outcomes. A series of network meta-analyses of clinical effectiveness compared intervention types with usual care. There was moderate to high heterogeneity of intervention effects between studies. Networks were sparse and the results of the meta-analysis must therefore be evaluated with caution. In terms of symptom severity outcomes, results indicated some beneficial effects for behavioural interventions targeting specific physical symptoms (e.g. pain, fatigue, bowel symptoms), but not somatisation or generic physical symptom measures. Results also indicated beneficial effects of some behavioural interventions on physical functioning, impact of symptoms on daily activities and psychological symptoms (depression, anxiety and emotional distress). Across outcomes, significant beneficial effects were most frequently seen for CBTHI and multimodal therapy interventions. There were no significant effects for any of the GP interventions (e.g. reattribution or GP MUS management).
Pain
At the end of treatment, only CBTHI and multimodal interventions were statistically more beneficial than usual care, with medium effect sizes. At short-term follow-up, only CBTHI was statistically more beneficial than usual care, with large effect size. By long-term follow-up, no interventions were statistically more beneficial than usual care. The number of studies included at the end of treatment was 10, and by long-term follow-up this had reduced to seven.
Fatigue
At the end of treatment, CBTLI (large effect size), multimodal interventions (large effect size), GA (medium effect size) and RSSE interventions (medium effect size) were statistically more beneficial than usual care. At short-term follow-up, only CBTLI (medium effect size) and RSSE (small effect size) were statistically more beneficial than usual care. By long-term follow-up, only CBTLI was statistically more beneficial than usual care. The number of studies included at the end of treatment was nine and by long-term follow-up this had reduced to five.
Bowel symptoms
There were insufficient data to conduct meta-analyses for end-of-treatment or short-term follow-up. At long-term follow-up, only CBTLI had a statistically significant beneficial effect compared with usual care.
Somatisation and generic physical symptoms
There were no significant beneficial effects for any intervention types for somatisation at end-of-treatment, short-term or long-term follow-up. There were no significant beneficial effects found for generic physical symptoms at any time point.
Physical functioning
At the end of treatment, only the multimodal therapy intervention type was statistically more beneficial than usual care, with a small effect size. This effect was also found at short-term follow-up, with medium effect size, but no intervention types were statistically more beneficial than usual care at long-term follow-up. This outcome was well represented, with 13 out of 14 studies remaining in the network at long-term follow-up.
Impact of symptoms on daily activities
At the end of treatment, only CBTHI was statistically more beneficial than usual care (large effect size). This effect was also found at short-term follow-up (large effect size), but by long-term follow-up there were no statistically significant results. The number of studies informing the network reduced from nine at the end of treatment to four at long-term follow-up.
Anxiety
At the end of treatment, only CBTHI was statistically more beneficial than usual care (medium effect size). This was repeated at both short- and long-term follow-up (large effect, 14 studies informed the network at the end of treatment, reducing to 11 by long-term follow-up).
Depression
At the end of treatment, only CBTHI was statistically more beneficial than usual care (large effect size). At short-term follow-up, only CBTHI (large effect size) was statistically more beneficial than usual care. At long-term follow-up, only the multimodal intervention type was statistically significant (small effect size). Depression was well represented in the network, with 13 studies at the end of treatment and 14 studies by long-term follow-up.
Emotional distress
At the end of treatment, CBTHI, other psychotherapy, RSSE and SES intervention types were statistically more beneficial than usual care (small to medium effect sizes). Only RSSE and multimodal interventions were statistically more beneficial at short-term follow-up (large and small effect sizes, respectively). By long-term follow-up, no intervention was statistically more beneficial than usual care. The number of studies in the network reduced from 15 at the end of treatment to only nine at long-term follow-up.
Results were not consistent across intervention type and outcome. Across outcomes, significant beneficial effects were most frequently seen for CBTHI and multimodal interventions. The outcome on which positive benefits were demonstrated across behavioural interventions was emotional distress, where four different intervention types were seen to have significant beneficial effects at the end of treatment. The outcome that appeared least affected by the interventions was somatisation, with no statistically significant treatment effects, despite there being 13 studies that measured this outcome at the end of treatment. No effects were found for generic physical symptoms, although this outcome was poorly represented in the network (three studies). It was not possible to explore the observed moderate to high heterogeneity seen across all outcomes with metaregression because of insufficient replication of each treatment effect across studies. There were no significant effects for any of the GP interventions (i.e. reattribution) or GP MUS management.
Although there was variation in the study population, these were all patients with chronic conditions. The nature of these conditions has implications for the interpretation of results. Long-term follow-up time periods were generally no longer than 1 year, which is relatively short within the context of a chronic condition. Symptoms may wax and wane over time. A number of individual studies reported significant improvements for both the intervention and the control groups, and this was seen in studies with active control arms as well as studies with passive control arms (i.e. usual care and waiting list control). Improvement in passive control groups over time may indicate natural remission of symptoms. Many of the observed effects seen at the end of treatment and at short-term follow-up were no longer apparent at long-term follow-up. This may in part be influenced by a lack of long-term follow-up data for studies for which the final follow-up was < 6 months after the end of intervention. However, it may also indicate a decline in the beneficial effects of behavioural interventions once treatment has ceased. A review of CBT for CFS366 found a slight trend to suggest that effect size increased with longer follow-up. This could be explained by the possibility that patients in waiting list control groups were able to access the trial interventions once the study period was completed; however, this was not tested in the current review. The NMA allowed active control arms to be categorised into intervention types, where appropriate, and evaluated alongside all other interventions. These active controls were often classified into the RSSE intervention type or guided self-help, and the RSSE category of interventions was found to have a small effect size for fatigue, which suggests that, when interventions are compared against active controls with a RSSE component, this may influence the relative effect size measured.
Behavioural interventions appeared to be more beneficial for specific physical symptoms compared with outcomes where multiple symptoms were measured (i.e. somatisation and generic physical symptoms). This finding is consistent with other meta-analyses that evaluated the effectiveness of psychotherapies for specific physical symptoms. Kleinstäuber et al. 76 highlight that meta-analyses evaluating specific physical symptoms (e.g. Kisely et al. 367 for non-cardiac chest pain) report larger effect sizes than those that evaluate psychotherapies for multiple MUS. They suggest that psychotherapies for somatoform disorders are more likely to focus on coping strategies rather than on curing specific symptoms. These results may be an artefact of either classification of condition or type of protocol. In terms of condition, measures of somatisation were more frequently used in populations classified as having ‘MUS/somatoform disorder’ rather than IBS, chronic pain or chronic fatigue. Studies of ‘MUS/somatoform disorder’ populations may have also measured individual symptoms, but it is not clear which symptoms were dominant at that time. Studies were also statistically powered based on their primary outcome, which was often the predominant symptom in the target population (e.g. fatigue in chronic fatigue populations); therefore, it is possible that they were insufficiently powered to detect differences in secondary outcomes. Alternatively, it may be that treatment protocols that focus on a specific diagnosis of a functional condition or specific symptom group are more effective than interventions that are designed to focus more broadly on the concept of somatisation or MUS. There is a substantive literature showing that the term MUS is offensive to many patients so these interventions may also seem less acceptable. 4,368 The observed statistical lack of effectiveness of GP interventions may in part be due to these interventions mainly targeting MUS/somatoform patients, with somatisation perhaps more complex and appearing to be more difficult to treat.
Although the current review is, to our knowledge, the first to look at a broad range of behavioural interventions for diverse populations of patients meeting our broad inclusion criteria for ‘MUS’, specifically within a primary care setting, results are consistent with other reviews in this area. The current review found no significant beneficial effects for GP-delivered interventions included in the meta-analysis. Although there were some significant effects for GP interventions in individual studies, this is consistent with the review by Rosendal et al. ,77 who found no conclusive evidence to show any beneficial effects for enhanced generalist care, and Gerger et al. ,78 who found psychological therapy to be effective only when delivered by psychologists rather than GPs. Price et al. 66 found significant effects for CBT for CFS at the end of treatment, not only when compared with usual care, but also when compared with other psychological therapies (which included relaxation, counselling, educational support – which would be defined as RSSE/guided self-help in the current review). However, by follow-up, effects were heterogeneous and inconsistent. For IBS, Pajak et al. 369 found minimal contact CBT to show promise in symptom management. The results of the current review are not consistent with van Dessel et al. ,75 who reported less severe somatic symptoms for all psychotherapies (with no superiority of CBT over other psychotherapies) at the end of treatment. However, their review was not restricted to primary care and it may be that differences in setting (i.e. inclusion of tertiary care) contribute to different treatment effects. A review of GA and graded exposure for non-specific low back pain370 found GA to be slightly more effective than minimal intervention, but the effect was no greater than other types of exercise for pain. The most populated networks were for psychological symptoms (i.e. depression, anxiety and emotional distress). Examination of baseline scores showed some variation in anxiety and depression, with, for example, HADS-A baseline scores ranging from 6 to 12 (normal to moderate), whereas HADS-D scores generally fell within the range from 5 to 9 (normal to mild), although Escobar et al. 156 reported severe baseline depression. When baseline psychological symptoms are low, one would not generally expect to find large effect sizes.
Although the behavioural interventions reviewed in the current review showed some effectiveness for improvement of specific physical symptoms compared with usual care, there was less success at improvement in subjective ratings of physical functioning. Furthermore, few studies reported objective measures of functioning, such as return to work. The variability in range and severity of symptoms means that, although some patients with MUS may be able to work, others may not. The Action for M.E. survey371 found that only 10% of respondents were in full-time work, education or training, and a further 14% were in part-time work, education or training. Rask et al. 372 found both patients with MUS of recent onset and those with persistent somatoform disorders to be at increased risk of sick leave and work disability over a 10-year period. It was not possible to explore severity of symptoms as a moderator of treatment effects because of the sparsity of the networks; however, it is unlikely that the interventions explored in this study were trialled on patients at the more severe end of the spectrum, who would be more likely to be treated in tertiary care. It is unlikely that patients in the studies reviewed here would, for example, be non-ambulatory and, therefore, the results should not be generalised to these patients. Furthermore, few of the included studies framed their effectiveness in terms of ‘cure’ or ‘recovery’ and, therefore, the findings should not be interpreted as such.
The qualitative review focused on the perceptions of patients who had received a specific behavioural intervention for MUS. It did not include the perspectives of patients about their general care, or of those patients who had not received such an intervention.
Prior perceptions of behavioural interventions are important as they may influence patient acceptance/uptake. The current review summarised information on acceptance/uptake from the included trials. There was wide variability in the number of patients declining to take part in the studies, and in the number of participants dropping out of the studies. Where reasons were given, these often included lack of time, a lack of faith in therapy,113,133,143,149 or an exacerbation in symptoms. 121 In the study by Robinson et al. ,131 most participants did not attend the group meeting, some of whom stated that they were unwilling to discuss bowel symptoms with strangers. Some participants dropped out because they felt that the intervention was not helping120 or that their symptoms had reduced. 109 Some of these themes were reflected in the qualitative review, with patients reporting that they did not believe that behavioural interventions would address their symptoms178 or that they felt that the intervention would do harm or make things worse. 179,182 The issue of acceptability of behavioural interventions is further highlighted by a recent feasibility study of an early intervention for patients with a history of fatigue lasting 1 to 4 months. 373,374 The intervention consisted of an information booklet and support from a fatigue specialist. The study was able to recruit fewer than half of the participants it aimed to and the authors therefore concluded that this type of study was not feasible and that the type of intervention was not acceptable to many of the participants.
Few of the studies in the clinical effectiveness review reported adverse events, and it is not clear if this is a result of there being no adverse events or if adverse events were not reported. A concern identified in the realist synthesis was that of potential misdiagnosis. Few of the trials included in the clinical effectiveness review reported on whether participants were misdiagnosed with MUS or one of the functional somatic disorders. Tummers et al. 149 reported that 12 participants had received an incorrect diagnosis of CFS, with four having an identified somatic explanation for their fatigue and eight found to have a psychiatric disorder. The potential for a wrong diagnosis is a concern for both GPs and for patients. In general, adverse events have not been reported systematically in psychological therapy trials and deterioration rates are rarely reported. 375,376 This should be addressed in future trials.
A pragmatic approach to the treatment of MUS has been suggested that takes account of patient preferences. Studies included in the current review offered behavioural interventions to primary care patients; however, there was a large range of variability in the nature of the primary care settings. At one end of the spectrum, GP practices were randomised and participating GPs were trained to manage patients with MUS within the course of their everyday practice. Other GP-delivered interventions were more structured, with GPs delivering a specific intervention to patients over a prescribed number of sessions. At the opposite end of the spectrum, participants were recruited in primary care but received an intervention from a specialist HP, sometimes on the GPs’ premises, but alternatively the intervention may have been delivered elsewhere – in community health centres, or gymnasiums or swimming pools. These differences are one of many sources of heterogeneity within the review, and they offer an opportunity to explore issues around pragmatism in clinical trials. There has been increasing interest in how well the results of clinical trials can be applied to everyday clinical practice. Explanatory trials test the efficacy of interventions under conditions strictly controlled for errors and biases. At the opposite end of the continuum, pragmatic trials are designed to test the effectiveness of an intervention when applied to real-world conditions. 377 Although the GP interventions were not shown to be effective compared with usual care in this review, it is possible that these trials represent the most pragmatic end of the continuum. In particular, this review may represent those trials that involved the cluster randomisation of GP practices and where the GPs used MUS training in their everyday practice, rather than with specific patients considered to need to be treated in a prescribed way at a prescribed time. In this review, trials of interventions delivered to primary care patients by other professionals showed more of a trend towards reducing physical symptoms. It is possible that these trials were, in fact, less pragmatic and, therefore, may be less likely to be effective when delivered in a less controlled manner. That having been said, we have no definitive evidence for or against this view. Thorpe et al. 378 developed the PRagmatic Explanatory Continuum Indicator Summary (PRECIS) tool to evaluate trials according to relevant domains, including flexibility in the experimental and comparator interventions, practitioner expertise, the intensity of follow-up, clinical meaningfulness of outcomes, flexibility in practitioner adherence to the protocol, and strictness of eligibility criteria. For example, many of the trials included in this review required the application of diagnostic criteria to their patients to screen for eligibility. In practice, general practice ‘Read’ codes for MUS may not always be applied. A future review could apply the PRECIS criteria to included trials to assess to what extent they are pragmatic and more likely to be effective when applied to real-world clinical practice.
Cost-effectiveness
A range of methods were utilised to evaluate the cost-effectiveness of behavioural modification interventions for MUS, including a systematic literature review of cost-effectiveness studies using QALYs as the measure of benefit, an evaluation of cost–consequence studies and an independent economic assessment. With only two relevant studies,131,284 findings from UK studies of cost-effectiveness were heterogeneous, although differences in the findings may be explained by differences in the populations and interventions studied. For cost–consequence studies, the net impact on health-care costs varied between studies, with some but not all the interventions being found to have a favourable impact on either overall costs or costs unrelated to delivery of the intervention. From this we can conclude that future studies should measure the impact of behavioural modification interventions on health-care costs. Although exploring possible approaches for estimating QALY gains from study outcomes, we found that estimating QALY differences based on mapping from changes on a single domain of the SF-36 to changes on the EQ-5D was unlikely to provide a robust estimate of QALY differences. This may be because the interventions may affect more than one domain of quality of life (e.g. vitality and physical functioning) such that using a single domain would overestimate or underestimate the QALY gain depending on the correlation between changes in different domains. We therefore conclude that future studies should include either a direct measure of utility such as the EQ-5D or a generic measure of HRQoL that covers all relevant domains and that allows utility values to be estimated via mapping, such as the SF-36. Cost analysis of all studies included in the clinical effectiveness review revealed a broad range of costs per patient within each intervention type. Although the ranges for each type of intervention were wide, the median values across different types of interventions were less variable, with median costs of intervention influenced by group versus individual mode of delivery. The median costs ranged from £62 for seven studies106,110,122,123,150,161 that examined RSSE delivered to groups (see Table 87), to £892 for five studies107,118,121,125,145 of multimodal interventions delivered to individuals. GP interventions had a wide range of costs, from £160 to £2560 (median £546) (see Table 91). Variation in costs was influenced by significant variation in the number of patients treated for each GP trained.
In general, the QALY gains estimated for behavioural modification interventions from individual studies were small, with mid-point estimates consistently < 0.1. None of the studies that reported 95% CIs for QALY gains measured directly in the study found a statistically significant difference between trial arms. The method used to estimate the uncertainty around the QALY gains for studies where utility values were mapped from the SF-36 or SF-12 was crude, so although some studies had a positive lower IQR for the QALY gain, this does not allow us to conclude that a statistically significant difference in QALYs was observed.
The independent economic assessment found a large degree of heterogeneity in estimates of cost-effectiveness. Some studies had mid-point ICERs in the range that NICE considers cost-effective, whereas others had ICERs that were above the NICE threshold. Two studies of CBTHI suggest that this intervention is cost-effective, although a third study had an ICER above the NICE threshold. The cost-effective ICERs were for populations with chronic pain, whereas the intervention with the higher ICER had a population that included both patients with chronic pain and those with somatoform disorders. Two studies suggested that guided self-help has the potential to be cost-effective, but one of these studies131 was associated with some uncertainty as there was no estimate of the cost of the self-help guidebook. One study suggested that RSSE has the potential to be cost-effective; however, additional uncertainty arose from the fact that the intervention was delivered to groups, but there was no indication of group size, which is an important determinant of cost per person. Evidence for GA interventions was inconsistent, with one study showing a positive QALY gain compared with guided self-help, albeit with an ICER at the upper limit of the NICE threshold, and the other finding that GA was dominated by usual care. CBTHI dominated medication in one study of a population with chronic pain. CBTHI was also shown to dominate SES in one study. GP interventions were generally not found to be cost-effective, with the exception of modified reattribution compared with usual care in patients with somatoform disorder, where the mid-point ICER was < £30,000 per QALY but was associated with significant uncertainty. Multimodal interventions were inconsistent in patients with chronic pain, with one study showing an estimated ICER above the NICE limit and another within the range considered cost-effective when compared with SES. However, the same study found that the multimodal intervention was dominated by CBTHI, which was estimated to be the optimal intervention based on that study, as seen in Table 27.
Discussion of qualitative findings
In identifying qualitative intervention studies conducted in a UK setting, we were able to explore the perspectives of patients and health-care providers of interventions delivered within a UK setting. However, we acknowledge that this perspective is necessarily limited by including only intervention-based studies and by operating within a tight geographical definition. The value of qualitative studies examining the experience of a condition and of studies from a broader range of geographical contexts and settings could prove valuable. Generally, the qualitative evidence base has identified features considered important by those receiving interventions. However, our ability to identify the most useful, acceptable or efficacious components of intervention packages was constrained by the limited detail of reporting of each intervention. Nevertheless, we believe that the qualitative synthesis, informed and enhanced by the wider scope of the realist synthesis that extends beyond intervention-based studies, offers a potentially useful contribution to the design of future primary care-based interventions.
Implications for individual interventions
Several hypothesised features for successful inclusion in a primary care-based intervention for MUS are presented within the realist synthesis. Many of these features receive clear support, with the exception of the value of ‘labelling’ and the ongoing controversy regarding initiation of psychosocial cues. At a minimum, an intervention should allow patient and care provider to maintain a relationship in which the patient feels supported and believed. Avoidance of an unproductive diagnostic cycle and a seemingly endless pattern of referral are key. The important role of non-intervention-specific factors has been recognised in a variety of contexts for treatment of chronic disease more generally in primary care, and this has been confirmed in this review. Above all, patients should be encouraged to pursue opportunities for self-management while drawing on the resources and skills available via the primary care provider.
Strengths and limitations of the assessment
We undertook a rigorous systematic review following accepted standards for the conduct and reporting of an effects review and using comprehensive search strategies to identify relevant trials evaluating the clinical effectiveness of interventions to treat MUS in primary care. Although we appraised and summarised a relatively large number of trials, much of the evidence was inconclusive because of heterogeneity within interventions and across groups of interventions. Pragmatically, we have employed an approach that differs from that typically used in Cochrane reviews and other reviews by not distinguishing between interventions within trials in terms of the control or comparator interventions. A further strength was the use of multiple review strategies to address diverse questions relating to the overall review aim. Hence, the qualitative data were able to explore the acceptability of interventions to patients, and the economic evaluation studies were able to explore cost implications while the realist synthesis examined features of the patient–clinician consultation in primary care.
Inclusion criteria were kept broad for many elements of the review. This was because of variation in the use of labels for MUS; inconsistencies in diagnostic criteria; a range of interventions meeting our a priori definition of ‘behavioural modification’; a range of ways of incorporating the primary care setting into the intervention; and multiple and diverse outcomes measures employed in the studies. The use of broad criteria has allowed a comprehensive presentation of the potential sources of heterogeneity both within and between studies and these are apparent from the narrative synthesis. The NMA also identified moderate to high heterogeneity of intervention effects between studies. Insufficient replication of treatment effects across studies made it impossible to perform a metaregression to statistically explain the heterogeneity of effects, and this represents a key limitation of the analysis of clinical effectiveness. Differences in point estimates for treatments (vs. usual care), however, indicate that the effectiveness of the interventions may differ by condition grouping, although overlapping CIs demonstrate uncertainty around this finding.
Despite the large number of studies, the network was sparse because of the considerable heterogeneity between interventions and the resulting large number of intervention types. Individual studies also varied in follow-up time periods and this meant that data from many studies were not available for all three time points. Additional reasons why it was not possible to include data from some of the included studies included both arms being categorised into the same intervention type, incomplete reporting of data for non-significant outcomes, only graphical data presented, or no reported variance. Sparsity of the networks meant that the observed effects were often informed by only one study in each intervention type for each outcome at each time point. Results from the network meta-analyses can therefore only give an indication of effects and should be considered within the context of the narrative summaries of individual studies, with consideration to the considerable heterogeneity identified. Evidence from both the network meta-analyses and the narrative summaries of all individual studies suggests the presence of differences in effects between intervention types. However, scrutiny of the individual studies shows variability in their design and conduct, to such an extent that interpretation of the results is not straightforward and, therefore, firm conclusions cannot be drawn. Although combining all intervention types into one ‘behavioural modification’ intervention type would have potentially allowed a metaregression to identify moderators of effects, it was considered that this would be inappropriate because of these differences.
The diversity in the interventions meant that to synthesise the data, individual interventions needed to be categorised into broader intervention groups. This process necessitated subjective judgements. Although every effort was made to reach decisions on intervention groupings by expert consensus, the prevalence of borderline cases resulted in the inclusion within one category of interventions that could have arguably met the criteria of a different category. With a sparse network, it is possible that such borderline cases could influence the results of the meta-analysis. However, while this is to our knowledge the first review to address this specific research question, the results are generally consistent with existing reviews that have addressed similar questions. We were unable to investigate publication bias via funnel plots owing to there being < 10 trials per comparison (of the same interventions) (Cochrane handbook)87 for all outcomes, and we are therefore unable to address the possibility of small study effects.
Owing to the diversity of interventions and sparsity of the network, we were unable to address the question as to which components of effective interventions were most influential. Results from the NMA suggest that multimodal therapy interventions have beneficial effects across outcomes and time points, but there was considerable diversity in the individual components across multimodal interventions and it has not been possible to identify if specific components were driving the effectiveness or if the effects were due to the combination of components or increased intensity of the treatment.
There was limited information available from published cost-effectiveness analyses that reported outcomes using QALYs. The main strength of our approach to the independent economic assessment is that we have been able to provide an indication of within-study cost-effectiveness for a large number of studies covering a diverse range of interventions and populations. The main disadvantages of the approach that we have used is that we have not been able to provide estimates of cost-effectiveness for all of the comparisons represented within the full set of clinical trials, as not all of the studies reported a generic measure of HRQoL and we have been unable to conduct a full incremental analysis to identify the optimal behavioural modification intervention using evidence from multiple studies.
The review is further limited by the dates of the searches, which were conducted between November 2015 and August 2016. Studies published since this time may therefore offer more evidence to inform the research questions addressed in this review.
Conclusions
Evidence from the current review offers some support for the value of behavioural interventions in primary care for improvement in specific symptoms (pain, fatigue and bowel symptoms), with both CBTLI and CBTHI showing some beneficial effects in different conditions, and multimodal therapy supported across a number of outcomes. No one intervention was found to be effective across all MUS, suggesting that there is no generic treatment that can be applied across all MUS. Only CBTHI had support over usual care for anxiety, but a range of therapies showed effects in emotional distress including other psychotherapies. The beneficial effects shown for chronic pain were short term. There was no clear evidence for these interventions in somatisation disorders and multiple symptoms. GP-led interventions, such as reattribution, were not supported by the evidence. Further interrogation of the results may indicate whether or not this is because of a lack of effectiveness of the more pragmatic trials. Developing a clear referral pathway in primary care to therapist-supported CBT or multimodal therapy for specific symptoms or FSSs may be most indicated at this point. The synthesis of qualitative evidence on process suggests that the quality of relationship between the service user and their GP is vital, and this ‘therapeutic alliance’ is the determining factor for how successfully these interventions can be implemented. These conclusions must be read with caution because of the large degree of heterogeneity of studies. There was also a low level of replication of intervention types within the networks, which coupled with the observed between-study heterogeneity of effects means that conclusions on specific therapies could be altered by addition or exclusion of one study. These findings were mirrored in the economic analysis, which found that where there was more than one study providing an estimate of cost-effectiveness for a particular comparison, the evidence was often inconsistent.
Suggested research priorities
The following research priorities are suggested, based on the findings of the review:
-
Explanation of observed between-study differences in effects within the same intervention type. This may be addressed by –
-
More detailed reporting of information regarding the defined mechanisms of the behavioural interventions under study, and how these map onto a theoretical and empirical understanding of the conditions.
-
More research on potentially influencing factors such as effective dosage and therapist competency within the more promising behavioural interventions.
-
Within-trial comparisons of interventions targeting specific syndromes with those targeting general somatic symptoms.
-
-
Testing the therapeutic effect of the GP–patient relationship. This may be addressed by –
-
Increased awareness of likely GP effects by researchers conducting trials of behavioural interventions for MUS, with planned assessment of these as potential confounders.
-
More research aimed at better understanding the therapeutic elements behind a successful therapeutic GP–patient alliance, which are key to a successful outcome, and how these elements can be formalised as GP (and health-care practitioner) skills.
-
Development of standardised measures of adverse effects in trials of behavioural interventions for MUS.
-
Acknowledgements
The authors would like to thank the advisors who have contributed to the project Expert Advisory Group:
-
Professor Else Guthrie, Professor of Psychological Medicine, University of Leeds
-
Professor Chris Dowrick, Professor of Primary Medical Care, University of Liverpool
-
Dr Nick Read, Psychoanalytical Psychotherapist. Previously, Professor of Gastrointestinal Physiology, Human Nutrition and Integrated Medicine, University of Sheffield.
Three experts by experience also contributed to all stages of the project and were members of the Expert Advisory Group, and Professor Chris Burton, ScHARR, University of Sheffield, for his comments on the realist review.
The authors would also like to thank Matt Stevenson, Professor of Health Technology Assessment, ScHARR, for providing advice on the cost-effectiveness modelling; Andrea Shippam, Programme Manager, ScHARR, for her help in formatting the report; and Carmen Galvan de la Cruz for her help with data extraction and protocol searching.
Contributions of authors
Joanna Leaviss (https://orcid.org/0000-0002-5632-6021) (Research Fellow in Evidence Synthesis) was the principal investigator and undertook the quantitative review.
Sarah Davis (https://orcid.org/0000-0002-6609-4287) (Senior Lecturer in Health Economics) conducted the cost-effectiveness modelling.
Shijie Ren (https://orcid.org/0000-0003-3568-7124) (Research Fellow in Bayesian Statistics, NMA and Survival Analysis) and Jean Hamilton (https://orcid.org/0000-0003-3326-9842) (Research Associate in Statistics) conducted the network meta-analyses.
Alison Scope (Research Fellow in Evidence Synthesis) undertook the qualitative review.
Andrew Booth (https://orcid.org/0000-0003-4808-3880) (Reader in Evidence-based Information Practice and Director of Information) conducted the realist synthesis.
Anthea Sutton (https://orcid.org/0000-0003-2449-2516) (Information Resources Group Manager) designed and conducted the literature searches.
Glenys Parry (https://orcid.org/0000-0002-4339-0577) (Professor of Applied Psychological Therapies), Marta Buszewicz (https://orcid.org/0000-0003-4016-5857) (Reader in Primary Care), Rona Moss-Morris (https://orcid.org/0000-0002-2927-3446) (Professor of Health Psychology) and Peter White (https://orcid.org/0000-0002-8193-5355) (Emeritus Professor of Psychological Medicine) provided subject expertise and were involved in writing the final report.
Data-sharing statement
Data are archived at ScHARR. All data requests should be submitted to the corresponding author for consideration. Access to anonymised data may be granted following review.
Disclaimers
This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.
References
- Chitnis A, Dowrick C, Byng R, Turner P, Shiers D. Guidance for Health Professionals on Medically Unexplained Symptoms – 2011. London: Royal College of General Practitioners and Royal College of Psychiatrists; 2014.
- Henningsen P, Jakobsen T, Schiltenwolf M, Weiss MG. Somatization revisited: diagnosis and perceived causes of common mental disorders. J Nerv Ment Dis 2005;193:85-92. https://doi.org/10.1097/01.nmd.0000152796.07788.b6.
- Wessely S, Nimnuan C, Sharpe M. Functional somatic syndromes: one or many?. Lancet 1999;354:936-9. https://doi.org/10.1016/S0140-6736(98)08320-2.
- Stone J, Wojcik W, Durrance D, Carson A, Lewis S, MacKenzie L, et al. What should we say to patients with symptoms unexplained by disease? The ‘number needed to offend’. BMJ 2002;325:1449-50. https://doi.org/10.1136/bmj.325.7378.1449.
- Creed F, Guthrie E, Fink P, Henningsen P, Rief W, Sharpe P, et al. Is there a better term than ‘medically unexplained symptoms’?. J Psychosom Res 2010;68:5-8. https://doi.org/10.1016/j.jpsychores.2009.09.004.
- Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A. The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Int Med 1994;121:953-9. https://doi.org/10.7326/0003-4819-121-12-199412150-00009.
- Drossman DA, Dumitrascu DL. Rome III: new standard for functional gastrointestinal disorders. J Gastrointestinal Liver Dis 2006;15.
- Wolfe F, Clauw DJ, Fitzcharles MA, Goldenberg DL, Katz RS, Mease P, et al. The American College of Rheumatology preliminary diagnostic criteria for fibromyalgia and measurement of symptom severity. Arthritis Care Res 2010;62:600-10. https://doi.org/10.1002/acr.20140.
- Kroenke K, Spitzer RL, Williams JB. The PHQ-15: validity of a new measure for evaluating the severity of somatic symptoms. Psychosom Med 2002;64:258-66. https://doi.org/10.1097/00006842-200203000-00008.
- Rief W, Heuser J, Hiller W. SOMS: das Screening für Somatoforme Störungen. Manual zum Fragebogen; mit zusätzlichen Informationen zur Hypochondrie-Messung und zum Einsatz von Tagesprotokollen. Bern: Verlag Hans Huber; 1997.
- Mumford DB, Bavington JT, Bhatnagar KS, Hussain Y, Mirza S, Naraghi MM. The Bradford Somatic Inventory. A multi-ethnic inventory of somatic symptoms reported by anxious and depressed patients in Britain and the Indo-Pakistan subcontinent. Br J Psychiatry 1991;158:379-86. https://doi.org/10.1192/bjp.158.3.379.
- Hoedeman R, Blankenstein AH, Krol B, Koopmans PC, Groothoff JW. The contribution of high levels of somatic symptom severity to sickness absence duration, disability and discharge. J Occup Rehabil 2010;20:264-73. https://doi.org/10.1007/s10926-010-9239-3.
- Fink P, Rosendal M, Olesen F. Classification of somatization and functional somatic symptoms in primary care. Aust N Z J Psychiatry 2005;39:772-81. https://doi.org/10.1080/j.1440-1614.2005.01682.x.
- American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders 2000.
- American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders 2013. https://doi.org/10.1176/appi.books.9780890425596.
- World Health Organization . International Classification of Diseases 1978.
- World Health Organization . International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) 1992.
- Newby JM, Hobbs MJ, Mahoney AEJ, Wong SK, Andrews G. DSM-5 illness anxiety disorder and somatic symptom disorder: Comorbidity, correlates, and overlap with DSM-IV hypochondriasis. J Psychosom Res 2017;101:31-7. https://doi.org/10.1016/j.jpsychores.2017.07.010.
- Dimsdale JE, Creed F, Escobar J, Sharpe M, Wulsin L, Barsky A, et al. Somatic symptom disorder: an important change in DSM. J Psychosom Res 2013;75:223-8. https://doi.org/10.1016/j.jpsychores.2013.06.033.
- Frances A. The new somatic symptom disorder in DSM-5 risks mislabeling many people as mentally ill. BMJ 2013;346. https://doi.org/10.1136/bmj.f1580.
- Ghoshal UC, Gwee KA. Post-infectious IBS, tropical sprue and small intestinal bacterial overgrowth: the missing link. Nat Rev Gastroenterol Hepatol 2017;14:435-41. https://doi.org/10.1038/nrgastro.2017.37.
- Drossman DA, Hasler WL. Rome IV-Functional GI disorders: disorders of gut–brain interaction. Gastroenterology 2016;150:1257-61. https://doi.org/10.1053/j.gastro.2016.03.035.
- Smith RC, Dwamena FC. Classification and diagnosis of patients with medically unexplained symptoms. J Gen Intern Med 2007;22:685-91. https://doi.org/10.1007/s11606-006-0067-2.
- Fink P, Schröder A. One single diagnosis, bodily distress syndrome, succeeded to capture 10 diagnostic categories of functional somatic syndromes and somatoform disorders. J Psychosom Res 2010;68:415-26. https://doi.org/10.1016/j.jpsychores.2010.02.004.
- Joint Commissioning Panel for Mental Health . Guidance for Commissioners of Services for People With Medically Unexplained Symptoms 2017. www.jcpmh.info/wp-content/uploads/jcpmh-mus-guide.pdf (accessed 5 November 2019).
- Geraghty KJ, Esmail A. Chronic fatigue syndrome: is the biopsychosocial model responsible for patient dissatisfaction and harm?. Br J Gen Pract 2016;66:437-8. https://doi.org/10.3399/bjgp16X686473.
- de Vega WC, Herrera S, Vernon SD, McGowan PO. Epigenetic modifications and glucocorticoid sensitivity in myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). BMC Med Genomics 2017;10. https://doi.org/10.1186/s12920-017-0248-3.
- Gur A, Oktayoglu P. Status of immune mediators in fibromyalgia. Curr Pain Headache Rep 2008;12:175-81. https://doi.org/10.1007/s11916-008-0031-4.
- Deale A, Wessely S. Patients’ perceptions of medical care in chronic fatigue syndrome. Soc Sci Med 2001;52:1859-64. https://doi.org/10.1016/S0277-9536(00)00302-6.
- White P. Biopsychosocial Medicine: An Integrated Approach to Understanding Illness. Oxford: Oxford Uniersity Press; 2005.
- Edwards TM, Stern A, Clarke DD, Ivbijaro G, Kasney LM. The treatment of patients with medically unexplained symptoms in primary care: a review of the literature. Ment Health Fam Med 2010;7:209-21.
- Taylor RE, Marshall T, Mann A, Goldberg DP. Insecure attachment and frequent attendance in primary care: a longitudinal cohort study of medically unexplained symptom presentations in ten UK general practices. Psychol Med 2012;42:855-64. https://doi.org/10.1017/S0033291711001589.
- Bermingham SL, Cohen A, Hague J, Parsonage M. The cost of somatisation among the working-age population in England for the year 2008–2009. Ment Health Fam Med 2010;7:71-84.
- Barsky AJ, Orav EJ, Bates DW. Somatization increases medical utilization and costs independent of psychiatric and medical comorbidity. Arch Gen Psychiatry 2005;62:903-10. https://doi.org/10.1001/archpsyc.62.8.903.
- Li CT, Chou YH, Yang KC, Yang CH, Lee YC, Su TP. Medically unexplained symptoms and somatoform disorders: diagnostic challenges to psychiatrists. J Chin Med Assoc 2009;72:251-6. https://doi.org/10.1016/S1726-4901(09)70065-6.
- Collin SM, Sterne JA, Hollingworth W, May MT, Crawley E. Equity of access to specialist chronic fatigue syndrome (CFS/ME) services in England (2008–2010): a national survey and cross-sectional study. BMJ Open 2012;2. https://doi.org/10.1136/bmjopen-2012-001417.
- Hughes G, Martinez C, Myon E, Taïeb C, Wessely S. The impact of a diagnosis of fibromyalgia on health care resource use by primary care patients in the UK: an observational study based on clinical practice. Arthritis Rheum 2006;54:177-83. https://doi.org/10.1002/art.21545.
- Akehurst RL, Brazier JE, Mathers N, O’Keefe C, Kaltenthaler E, Morgan A, et al. Health-related quality of life and cost impact of irritable bowel syndrome in a UK primary care setting. PharmacoEconomics 2002;20:455-62. https://doi.org/10.2165/00019053-200220070-00003.
- olde Hartman TC, Borghuis MS, Lucassen PL, van de Laar FA, Speckens AE, van Weel C. Medically unexplained symptoms, somatisation disorder and hypochondriasis: course and prognosis. A systematic review. J Psychosom Res 2009;66:363-77. https://doi.org/10.1016/j.jpsychores.2008.09.018.
- Creed FH, Davies I, Jackson J, Littlewood A, Chew-Graham C, Tomenson B, et al. The epidemiology of multiple somatic symptoms. J Psychosom Res 2012;72:311-17. https://doi.org/10.1016/j.jpsychores.2012.01.009.
- Kroenke K. Efficacy of treatment for somatoform disorders: a review of randomized controlled trials. Psychosom Med 2007;69:881-8. https://doi.org/10.1097/PSY.0b013e31815b00c4.
- Ford AC, Talley NJ, Schoenfeld PS, Quigley EM, Moayyedi P. Efficacy of antidepressants and psychological therapies in irritable bowel syndrome: systematic review and meta-analysis. Gut 2009;58:367-78. https://doi.org/10.1136/gut.2008.163162.
- Beck AT, Therapy C. The Emotional Disorders. New York, NY: International University Press; 1976.
- Deary V, Chalder T, Sharpe M. The cognitive behavioural model of medically unexplained symptoms: a theoretical and empirical review. Clin Psychol Rev 2007;27:781-97. https://doi.org/10.1016/j.cpr.2007.07.002.
- Goldberg D, Gask L, O’Dowd T. The treatment of somatization: teaching techniques of reattribution. J Psychosom Res 1989;33:689-95. https://doi.org/10.1016/0022-3999(89)90084-6.
- Nezu AM, Nezu CM, Lombardo ER. Cognitive-behavior therapy for medically unexplained symptoms: a critical review of the treatment literature. Behavior Therapy 2001;32:537-83. https://doi.org/10.1016/S0005-7894(01)80035-6.
- Buckelew SP, Conway R, Parker J, Deuser WE, Read J, Witty TE, et al. Biofeedback/relaxation training and exercise interventions for fibromyalgia: a prospective trial. Arthritis Care Res 1998;11:196-209. https://doi.org/10.1002/art.1790110307.
- Keefer L, Blanchard EB. The effects of relaxation response meditation on the symptoms of irritable bowel syndrome: results of a controlled treatment study. Behav Res Ther 2001;39:801-11. https://doi.org/10.1016/S0005-7967(00)00078-4.
- Creamer P, Singh BB, Hochberg MC, Berman BM. Sustained improvement produced by nonpharmacologic intervention in fibromyalgia: results of a pilot study. Arthritis Care Res 2000;13:198-204. https://doi.org/10.1002/1529-0131(200008)13:4<198::AID-ANR4>3.0.CO;2-P.
- van Ravesteijn HJ, Suijkerbuijk YB, Langbroek JA, Muskens E, Lucassen PL, van Weel C, et al. Mindfulness-based cognitive therapy (MBCT) for patients with medically unexplained symptoms: process of change. J Psychosom Res 2014;77:27-33. https://doi.org/10.1016/j.jpsychores.2014.04.010.
- Abbass A, Kisely S, Kroenke K. Short-term psychodynamic psychotherapy for somatic disorders. Systematic review and meta-analysis of clinical trials. Psychother Psychosom 2009;78:265-74. https://doi.org/10.1159/000228247.
- de Jong JR, Vlaeyen JW, Onghena P, Cuypers C, den Hollander M, Ruijgrok J. Reduction of pain-related fear in complex regional pain syndrome type I: the application of graded exposure in vivo. Pain 2005;116:264-75. https://doi.org/10.1016/j.pain.2005.04.019.
- Heymans MW, de Vet HC, Bongers PM, Knol DL, Koes BW, van Mechelen W. The effectiveness of high-intensity versus low-intensity back schools in an occupational setting: a pragmatic randomized controlled trial. Spine 2006;31:1075-82. https://doi.org/10.1097/01.brs.0000216443.46783.4d.
- Smeets RJ, Vlaeyen JW, Hidding A, Kester AD, van der Heijden GJ, van Geel AC, et al. Active rehabilitation for chronic low back pain: cognitive-behavioral, physical, or both? First direct post-treatment results from a randomized controlled trial. BMC Musculoskelet Disord 2006;7. https://doi.org/10.1186/1471-2474-7-5.
- White P, Chalder T, Sharpe M. The PACE trial: results of a large trial of nonpharmacological treatments. J Psychosom Res 2011;70.
- Jason LA, Torres-Harding S, Friedberg F, Corradi K, Njoku MG, Donalek J, et al. Non-pharmacologic interventions for CFS: a randomized trial. J Clin Psychol Med Settings 2007;14:275-96. https://doi.org/10.1007/s10880-007-9090-7.
- Gowans SE, deHueck A, Voss S, Silaj A, Abbey SE, Reynolds WJ. Effect of a randomized, controlled trial of exercise on mood and physical function in individuals with fibromyalgia. Arthritis Rheum 2001;45:519-29. https://doi.org/10.1002/1529-0131(200112)45:6<519::AID-ART377>3.0.CO;2-3.
- Jentoft ES, Kvalvik AG, Mengshoel AM. Effects of pool-based and land-based aerobic exercise on women with fibromyalgia/chronic widespread muscle pain. Arthritis Rheum 2001;45:42-7. https://doi.org/10.1002/1529-0131(200102)45:1<42::AID-ANR82>3.0.CO;2-A.
- Jones AC, Coulson L, Muir K, Tolley K, Lophatananon A, Everitt L, et al. A nurse-delivered advice intervention can reduce chronic non-steroidal anti-inflammatory drug use in general practice: a randomized controlled trial. Rheumatology 2002;41:14-21. https://doi.org/10.1093/rheumatology/41.1.14.
- Witthöft M, Hiller W. Psychological approaches to origins and treatments of somatoform disorders. Annu Rev Clin Psychol 2010;6:257-83. https://doi.org/10.1146/annurev.clinpsy.121208.131505.
- Paterson C, Taylor RS, Griffiths P, Britten N, Rugg S, Bridges J, et al. Acupuncture for ‘frequent attenders’ with medically unexplained symptoms: a randomised controlled trial (CACTUS study). Br J Gen Pract 2011;61:e295-305. https://doi.org/10.3399/bjgp11X572689.
- Bagnall AM, Whiting P, Richardson R, Sowden AJ. Interventions for the treatment and management of chronic fatigue syndrome/myalgic encephalomyelitis. Qual Saf Health Care 2002;11:284-8. https://doi.org/10.1136/qhc.11.3.284.
- Morriss R, Gask L, Dowrick C, Dunn G, Peters S, Ring A, et al. Randomized trial of reattribution on psychosocial talk between doctors and patients with medically unexplained symptoms. Psychol Med 2010;40:325-33. https://doi.org/10.1017/S0033291709990353.
- Bagnall A-M, Hempel S, Chambers D, Forbes C. The Treatment and Management of Chronic Fatigue Syndrome (CFS) Myalgic Encephalomyelitis (ME) in Adults and Children. CRD Report 35 2007.
- Clark MR, McCarberg BH, Mease PJ. Fibromyalgia in primary care: what you need to know now! Practical applications for patient care. Johns Hopkins Adv Stud Med 2011;11:4-22.
- Price JR, Mitchell E, Tidy E, Hunot V. Cognitive behaviour therapy for chronic fatigue syndrome in adults. Cochrane Database Syst Rev 2008;3. https://doi.org/10.1002/14651858.CD001027.pub2.
- Chambers D, Bagnall AM, Hempel S, Forbes C. Interventions for the treatment, management and rehabilitation of patients with chronic fatigue syndrome/myalgic encephalomyelitis: an updated systematic review. J R Soc Med 2006;99:506-20. https://doi.org/10.1177/014107680609901012.
- Arnold B, Häuser W, Arnold M, Bernateck M, Bernardy K, Brückle W, et al. Multicomponent therapy of fibromyalgia syndrome. Systematic review, meta-analysis and guideline. Schmerz 2012;26:287-90. https://doi.org/10.1007/s00482-012-1173-1.
- Marcus DA. Fibromyalgia: diagnosis and treatment options. Gend Med 2009;6:139-51. https://doi.org/10.1016/j.genm.2009.01.004.
- Busch AJ, Barber KA, Overend TJ, Peloso PM, Schachter CL. Exercise for treating fibromyalgia syndrome. Cochrane Database Syst Rev 2007;4. https://doi.org/10.1002/14651858.CD003786.pub2.
- Häuser W, Bernardy K, Arnold B, Offenbächer M, Schiltenwolf M. Efficacy of multicomponent treatment in fibromyalgia syndrome: a meta-analysis of randomized controlled clinical trials. Arthritis Rheum 2009;61:216-24. https://doi.org/10.1002/art.24276.
- Zijdenbos IL, de Wit NJ, van der Heijden GJ, Rubin G, Quartero AO. Psychological treatments for the management of irritable bowel syndrome. Cochrane Database Syst Rev 2009;1. https://doi.org/10.1002/14651858.CD006442.pub2.
- Aggarwal VR, Lovell K, Peters S, Javidi H, Joughin A, Goldthorpe J. Psychosocial interventions for the management of chronic orofacial pain. Cochrane Database Syst Rev 2011;11. https://doi.org/10.1002/14651858.CD008456.pub2.
- Champaneria R, Daniels JP, Raza A, Pattison HM, Khan KS. Psychological therapies for chronic pelvic pain: systematic review of randomized controlled trials. Acta Obstet Gynecol Scand 2012;91:281-6. https://doi.org/10.1111/j.1600-0412.2011.01314.x.
- van Dessel N, den Boeft M, van der Wouden JC, Kleinstäuber M, Leone SS, Terluin B, et al. Non-pharmacological interventions for somatoform disorders and medically unexplained physical symptoms (MUPS) in adults. Cochrane Database Syst Rev 2014;11. https://doi.org/10.1002/14651858.CD011142.pub2.
- Kleinstäuber M, Witthöft M, Hiller W. Efficacy of short-term psychotherapy for multiple medically unexplained physical symptoms: a meta-analysis. Clin Psychol Rev 2011;31:146-60. https://doi.org/10.1016/j.cpr.2010.09.001.
- Rosendal M, Blankenstein AH, Morriss R, Fink P, Sharpe M, Burton C. Enhanced care by generalists for functional somatic symptoms and disorders in primary care. Cochrane Database Syst Rev 2013;10. https://doi.org/10.1002/14651858.CD008142.pub2.
- Gerger H, Hlavica M, Gaab J, Munder T, Barth J. Does it matter who provides psychological interventions for medically unexplained symptoms? A meta-analysis. Psychother Psychosom 2015;84:217-26. https://doi.org/10.1159/000380914.
- Garcia-Campayo J, Magdalena J, Magallón R, Fernández-García E, Salas M, Andrés E. A meta-analysis of the efficacy of fibromyalgia treatment according to level of care. Arthritis Res Ther 2008;10. https://doi.org/10.1186/ar2455.
- Castonguay LG, Hayes AM, Goldfried MR, DeRubeis RJ. The focus of therapist interventions in cognitive therapy for depression. Cog Ther Res 1995;19:485-503. https://doi.org/10.1007/BF02230510.
- Cho HJ, Hotopf M, Wessely S. The placebo response in the treatment of chronic fatigue syndrome: a systematic review and meta-analysis. Psychosom Med 2005;67:301-13. https://doi.org/10.1097/01.psy.0000156969.76986.e0.
- Williams AC, Eccleston C, Morley S. Psychological therapies for the management of chronic pain (excluding headache) in adults. Cochrane Database Syst Rev 2012;11. https://doi.org/10.1002/14651858.CD007407.pub3.
- Smith RC, Lein C, Collins C, Lyles JS, Given B, Dwamena FC, et al. Treating patients with medically unexplained symptoms in primary care. J Gen Intern Med 2003;18:478-89. https://doi.org/10.1046/j.1525-1497.2003.20815.x.
- Heijmans M, Olde Hartman TC, van Weel-Baumgarten E, Dowrick C, Lucassen PL, van Weel C. Experts’ opinions on the management of medically unexplained symptoms in primary care. A qualitative analysis of narrative reviews and scientific editorials. Fam Pract 2011;28:444-55. https://doi.org/10.1093/fampra/cmr004.
- Gask L, Dowrick C, Salmon P, Peters S, Morriss R. Reattribution reconsidered: narrative review and reflections on an educational intervention for medically unexplained symptoms in primary care settings. J Psychosom Res 2011;71:325-34. https://doi.org/10.1016/j.jpsychores.2011.05.008.
- Broom DH, Woodward RV. Medicalisation reconsidered: toward a collaborative approach to care. Sociol Health Illness 1996;18:357-78. https://doi.org/10.1111/1467-9566.ep10934730.
- Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of Interventions 2011. http://handbook-5-1.cochrane.org (accessed 27 March 2020).
- Dias S, Sutton AJ, Ades AE, Welton NJ. Evidence synthesis for decision making 2: a generalized linear modeling framework for pairwise and network meta-analysis of randomized controlled trials. Med Decis Making 2013;33:607-17. https://doi.org/10.1177/0272989X12458724.
- Rhodes KM, Turner RM, Higgins JP. Predictive distributions were developed for the extent of heterogeneity in meta-analyses of continuous outcome data. J Clin Epidemiol 2015;68:52-60. https://doi.org/10.1016/j.jclinepi.2014.08.012.
- Ren S, Oakley JE, Stevens JW. Incorporating genuine prior information about between-study heterogeneity in random effects pairwise and network meta-analyses. Med Decis Making 2018;38:531-42. https://doi.org/10.1177/0272989X18759488.
- Turner RM, Davey J, Clarke MJ, Thompson SG, Higgins JP. Predicting the extent of heterogeneity in meta-analysis, using empirical data from the Cochrane Database of Systematic Reviews. Int J Epidemiol 2012;41:818-27. https://doi.org/10.1093/ije/dys041.
- Dias S, Welton NJ, Sutton AJ, Caldwell DM, Lu G, Ades AE. NICE DSU Technical Support Document 4: Inconsistency in Networks of Evidence Based on Randomised Controlled Trials. Bristol: Decision Support Unit, University of Bristol and University of Sheffield; 2011.
- Lunn DJ, Thomas A, Best N, Spiegelhalter D. WinBUGS – a Bayesian modelling framework: concepts, structure, and extensibility. Stat Comput 2000;10:325-37. https://doi.org/10.1023/A:1008929526011.
- Stutrtz S, Ligges U, Gelman A. R2WinBUGS: a package for running WinBUGS from R. J Stat Software 2005;12:1-16. https://doi.org/10.18637/jss.v012.i03.
- Brooks S, Gelman A. Alternative methods for monitoring convergence of iterative simulations. J Comput Graph Stat 1998;7:434-55. https://doi.org/10.1080/10618600.1998.10474787.
- Posse M. Jungian psychotherapy for somatization. Eur J Psychiatr 2004;18:23-9.
- Toft T, Rosendal M, Ørnbøl E, Olesen F, Frostholm L, Fink P. Training general practitioners in the treatment of functional somatic symptoms: effects on patient health in a cluster-randomised controlled trial (the Functional Illness in Primary Care study). Psychother Psychosom 2010;79:227-37. https://doi.org/10.1159/000313691.
- Escobar JI, Gara M, Waitzkin H, Silver RC, Holman A, Compton W. DSM-IV hypochondriasis in primary care. Gen Hosp Psychiatry 1998;20:155-9. https://doi.org/10.1016/S0163-8343(98)00018-8.
- Martin A, Rauh E, Fichter M, Rief W. A one-session treatment for patients suffering from medically unexplained symptoms in primary care: a randomized clinical trial. Psychosomatics 2007;48:294-303. https://doi.org/10.1176/appi.psy.48.4.294.
- Smith GR, Rost K, Kashner TM. A trial of the effect of a standardized psychiatric consultation on health outcomes and costs in somatizing patients. Arch Gen Psychiatry 1995;52:238-43. https://doi.org/10.1001/archpsyc.1995.03950150070012.
- Sumathipala A, Siribaddana S, Hewege S, Sumathipala K, Prince M, Mann A. Understanding the explanatory model of the patient on their medically unexplained symptoms and its implication on treatment development research: a Sri Lanka Study. BMC Psychiatry 2008;8. https://doi.org/10.1186/1471-244X-8-54.
- Rief W, Martin A, Rauh E, Zech T, Bender A. Evaluation of general practitioners’ training: how to manage patients with unexplained physical symptoms. Psychosomatics 2006;47:304-11. https://doi.org/10.1176/appi.psy.47.4.304.
- Kocken PL, Zwanenburg EJ, de Hoop T. Effects of health education for migrant females with psychosomatic complaints treated by general practitioners. A randomised controlled evaluation study. Patient Educ Couns 2008;70:25-30. https://doi.org/10.1016/j.pec.2007.09.016.
- Kolk AM, Schagen S, Hanewald GJ. Multiple medically unexplained physical symptoms and health care utilization: outcome of psychological intervention and patient-related predictors of change. J Psychosom Res 2004;57:379-89. https://doi.org/10.1016/S0022-3999(04)00036-4.
- McLeod CC, Budd MA, McClelland DC. Treatment of somatization in primary care. Gen Hosp Psychiatry 1997;19:251-8. https://doi.org/10.1016/S0163-8343(97)00047-9.
- Peters S, Stanley I, Rose M, Kaney S, Salmon P. A randomized controlled trial of group aerobic exercise in primary care patients with persistent, unexplained physical symptoms. Fam Pract 2002;19:665-74. https://doi.org/10.1093/fampra/19.6.665.
- Smith RC, Lyles JS, Gardiner JC, Sirbu C, Hodges A, Collins C, et al. Primary care clinicians treat patients with medically unexplained symptoms: a randomized controlled trial. J Gen Intern Med 2006;21:671-7. https://doi.org/10.1111/j.1525-1497.2006.00460.x.
- Morriss R, Dowrick C, Salmon P, Peters S, Dunn G, Rogers A, et al. Cluster randomised controlled trial of training practices in reattribution for medically unexplained symptoms. Br J Psychiatry 2007;191:536-42. https://doi.org/10.1192/bjp.bp.107.040683.
- Ryan M, Gevirtz R. Biofeedback-based psychophysiological treatment in a primary care setting: an initial feasibility study. Appl Psychophysiol Biofeedback 2004;29:79-93. https://doi.org/10.1023/B:APBI.0000026635.03016.ef.
- Kobeissi L, Mahfoud Z, Khoury B, El Kak F, Ghantous Z, Khawaja M, et al. The Relaxation Exercise and Social Support Trial (RESST): a community-based randomized controlled trial to alleviate medically unexplained vaginal discharge symptoms. BMC Psychiatry 2012;12. https://doi.org/10.1186/1471-244X-12-195.
- Wearden AJ, Dowrick C, Chew-Graham C, Bentall RP, Morriss RK, Peters S, et al. Nurse led, home based self help treatment for patients in primary care with chronic fatigue syndrome: randomised controlled trial. BMJ 2010;340. https://doi.org/10.1136/bmj.c1777.
- Ridsdale L, Godfrey E, Chalder T, Seed P, King M, Wallace P, et al. Fatigue Trialists’ Group . Chronic fatigue in general practice: is counselling as good as cognitive behaviour therapy? A UK randomised trial. Br J Gen Pract 2001;51:19-24.
- Ridsdale L, Darbishire L, Seed PT. Is graded exercise better than cognitive behaviour therapy for fatigue? A UK randomized trial in primary care. Psychol Med 2004;34:37-49. https://doi.org/10.1017/S0033291703001247.
- Chalder T, Wallace P, Wessely S. Self-help treatment of chronic fatigue in the community: a randomized controlled trial. Br J Health Psychol 1997;2:189-97. https://doi.org/10.1111/j.2044-8287.1997.tb00535.x.
- Ridsdale L, Hurley M, King M, McCrone P, Donaldson N. The effect of counselling, graded exercise and usual care for people with chronic fatigue in primary care: a randomized trial. Psychol Med 2012;42:2217-24. https://doi.org/10.1017/S0033291712000256.
- Sumathipala A, Hewege S, Hanwella R, Mann AH. Randomized controlled trial of cognitive behaviour therapy for repeated consultations for medically unexplained complaints: a feasibility study in Sri Lanka. Psychol Med 2000;30:747-57. https://doi.org/10.1017/S0033291799002160.
- Sumathipala A, Siribaddana S, Abeysingha MR, De Silva P, Dewey M, Prince M, et al. Cognitive-behavioural therapy v. structured care for medically unexplained symptoms: randomised controlled trial. Br J Psychiatry 2008;193:51-9. https://doi.org/10.1192/bjp.bp.107.043190.
- Cuesta-Vargas AI, Adams N, Salazar JA, Belles A, Hazañas S, Arroyo-Morales M. Deep water running and general practice in primary care for non-specific low back pain versus general practice alone: randomized controlled trial. Clin Rheumatol 2012;31:1073-8. https://doi.org/10.1007/s10067-012-1977-5.
- Macedo LG, Latimer J, Maher CG, Hodges PW, McAuley JH, Nicholas MK, et al. Effect of motor control exercises versus graded activity in patients with chronic nonspecific low back pain: a randomized controlled trial. Phys Ther 2012;92:363-77. https://doi.org/10.2522/ptj.20110290.
- van der Roer N, van Tulder M, Barendse J, Knol D, van Mechelen W, de Vet H. Intensive group training protocol versus guideline physiotherapy for patients with chronic low back pain: a randomised controlled trial. Eur Spine J 2008;17:1193-200. https://doi.org/10.1007/s00586-008-0718-6.
- Walti P, Kool J, Luomajoki H. Short-term effect on pain and function of neurophysiological education and sensorimotor retraining compared to usual physiotherapy in patients with chronic or recurrent non-specific low back pain, a pilot randomized controlled trial. BMC Musc Dis 2015;16. https://doi.org/10.1186/s12891-015-0533-2.
- Loew TH, Sohn R, Martus P, Tritt K, Rechlin T. Functional relaxation as a somatopsychotherapeutic intervention: a prospective controlled study. Altern Ther Health Med 2000;6:70-5.
- Lansinger B, Larsson E, Persson LC, Carlsson JY. Qigong and exercise therapy in patients with long-term neck pain: a prospective randomized trial. Spine 2007;32:2415-22. https://doi.org/10.1097/BRS.0b013e3181573b4b.
- Wolfe F, Smythe HA, Yunus MB, Bennett RM, Bombardier C, Goldenberg DL, et al. The American College of Rheumatology 1990 criteria for the classification of fibromyalgia: report of the Multicenter Criteria Committee. Arthritis Rheum 1990;33:160-72. https://doi.org/10.1002/art.1780330203.
- McBeth J, Prescott G, Scotland G, Lovell K, Keeley P, Hannaford P, et al. Cognitive behavior therapy, exercise, or both for treating chronic widespread pain. Arch Intern Med 2012;172:48-57. https://doi.org/10.1001/archinternmed.2011.555.
- Beasley M, Prescott GJ, Scotland G, McBeth J, Lovell K, Keeley P, et al. Patient-reported improvements in health are maintained 2 years after completing a short course of cognitive behaviour therapy, exercise or both treatments for chronic widespread pain: long-term results from the MUSICIAN randomised controlled trial. RMD Open 2015;1. https://doi.org/10.1136/rmdopen-2014-000026.
- Rothman MG, Ortendahl M, Rosenblad A, Johansson AC. Improved quality of life, working ability, and patient satisfaction after a pretreatment multimodal assessment method in patients with mixed chronic muscular pain: a randomized-controlled study. Clin J Pain 2013;29:195-204. https://doi.org/10.1097/AJP.0b013e318250e544.
- LeFort SM. Randomized Controlled Trial of a Community-Based Nursing Intervention for Those Experiencing Chronic Non-Malignant Pain 2000.
- Kennedy T, Jones R, Darnley S, Seed P, Wessely S, Chalder T. Cognitive behaviour therapy in addition to antispasmodic treatment for irritable bowel syndrome in primary care: randomised controlled trial. BMJ 2005;331. https://doi.org/10.1136/bmj.38545.505764.06.
- Moss-Morris R, McAlpine L, Didsbury LP, Spence MJ. A randomized controlled trial of a cognitive behavioural therapy-based self-management intervention for irritable bowel syndrome in primary care. Psychol Med 2010;40:85-94. https://doi.org/10.1017/S0033291709990195.
- Robinson A, Lee V, Kennedy A, Middleton L, Rogers A, Thompson DG, et al. A randomised controlled trial of self-help interventions in patients with a primary care diagnosis of irritable bowel syndrome. Gut 2006;55:643-8. https://doi.org/10.1136/gut.2004.062901.
- Schaefert R, Kaufmann C, Wild B, Schellberg D, Boelter R, Faber R, et al. Specific collaborative group intervention for patients with medically unexplained symptoms in general practice: a cluster randomized controlled trial. Psychother Psychosom 2013;82:106-19. https://doi.org/10.1159/000343652.
- Marques M, De Gucht V, Leal I, Maes S. Effects of a self-regulation based physical activity program (the ‘4-STEPS’) for unexplained chronic fatigue: a randomized controlled trial. Int J Behav Med 2015;22:187-96. https://doi.org/10.1007/s12529-014-9432-4.
- Schulz KF, Grimes DA. Sample size slippages in randomised trials: exclusions and the lost and wayward. Lancet 2002;359:781-5. https://doi.org/10.1016/S0140-6736(02)07882-0.
- Sutton A, de la Cruz MDCG, Leaviss J, Booth A. Searching for trial protocols: a comparison of methods. Res Synth Meth 2018;9:551-60. https://doi.org/10.1002/jrsm.1281.
- Aiarzaguena JM, Grandes G, Gaminde I, Salazar A, Sánchez A, Ariño J. A randomized controlled clinical trial of a psychosocial and communication intervention carried out by GPs for patients with medically unexplained symptoms. Psychol Med 2007;37:283-94. https://doi.org/10.1017/S0033291706009536.
- Alda M, Luciano JV, Andrés E, Serrano-Blanco A, Rodero B, del Hoyo YL, et al. Effectiveness of cognitive behaviour therapy for the treatment of catastrophisation in patients with fibromyalgia: a randomised controlled trial. Arthritis Res Ther 2011;13. https://doi.org/10.1186/ar3496.
- Luciano JV, Guallar JA, Aguado J, López-Del-Hoyo Y, Olivan B, Magallón R, et al. Effectiveness of group acceptance and commitment therapy for fibromyalgia: a 6-month randomized controlled trial (EFFIGACT study). Pain 2014;155:693-702. https://doi.org/10.1016/j.pain.2013.12.029.
- Luciano JV, Martinez N, Penarrubia-Maria MT. Effectiveness of a psychoeducational treatment program implemented in general practice for fibromyalgia patients: a randomized controlled trial. Clin J Pain 2011;27. https://doi.org/10.1097/AJP.0b013e31820b131c.
- Sañudo B, Galiano D, Carrasco L, Blagojevic M, de Hoyo M, Saxton J. Aerobic exercise versus combined exercise therapy in women with fibromyalgia syndrome: a randomized controlled trial. Arch Phys Med Rehabil 2010;91:1838-43. https://doi.org/10.1016/j.apmr.2010.09.006.
- Zonneveld LN, van Rood YR, Timman R, Kooiman CG, Van’t Spijker A, Busschbach JJ. Effective group training for patients with unexplained physical symptoms: a randomized controlled trial with a non-randomized one-year follow-up. PLOS ONE 2012;7. https://doi.org/10.1371/journal.pone.0042629.
- Smith RC, Gardiner JC, Luo Z, Schooley S, Lamerato L, Rost K. Primary care physicians treat somatization. J Gen Intern Med 2009;24:829-32. https://doi.org/10.1007/s11606-009-0992-y.
- van Ravesteijn H, Lucassen P, Bor H, van Weel C, Speckens A. Mindfulness-based cognitive therapy for patients with medically unexplained symptoms: a randomized controlled trial. Psychother Psychosom 2013;82:299-310. https://doi.org/10.1159/000348588.
- Gili M, Magallón R, López-Navarro E, Roca M, Moreno S, Bauzá N, et al. Health related quality of life changes in somatising patients after individual versus group cognitive behavioural therapy: a randomized clinical trial. J Psychosom Res 2014;76:89-93. https://doi.org/10.1016/j.jpsychores.2013.10.018.
- Pols RG, Battersby MW. Coordinated care in the management of patients with unexplained physical symptoms: depression is a key issue. Med J Aust 2008;188:133-7. https://doi.org/10.5694/j.1326-5377.2008.tb01877.x.
- Rosendal M, Olesen F, Fink P, Toft T, Sokolowski I, Bro F. A randomized controlled trial of brief training in the assessment and treatment of somatization in primary care: effects on patient outcome. Gen Hosp Psychiatry 2007;29:364-73. https://doi.org/10.1016/j.genhosppsych.2007.03.005.
- Schaefert R, Kirsch G, Kaufmann C, Hermann K, Szecsenyi J, Herzog W, et al. New ways in caring for patients with functional syndromes – interdisciplinary group intervention in the primary care setting: a qualitative study. Psychother Psychosom Med Psychol 2013;63:167-75. https://doi.org/10.1055/s-0032-1323790.
- Moreno S, Gili M, Magallón R, Bauzá N, Roca M, Del Hoyo YL, et al. Effectiveness of group versus individual cognitive-behavioral therapy in patients with abridged somatization disorder: a randomized controlled trial. Psychosom Med 2013;75:600-8. https://doi.org/10.1097/PSY.0b013e31829a8904.
- Tummers M, Knoop H, van Dam A, Bleijenberg G. Implementing a minimal intervention for chronic fatigue syndrome in a mental health centre: a randomized controlled trial. Psychol Med 2012;42:2205-15. https://doi.org/10.1017/S0033291712000232.
- Ho RTH, Chan JSM, Wang CW, Lau BW, So KF, Yuen LP, et al. A randomized controlled trial of qigong exercise on fatigue symptoms, functioning, and telomerase activity in persons with chronic fatigue or chronic fatigue syndrome. Annal Behav Med 2012;44:160-70. https://doi.org/10.1007/s12160-012-9381-6.
- Friedberg F, Napoli A, Coronel J, Adamowicz J, Seva V, Caikauskaite I, et al. Chronic fatigue self-management in primary care: a randomized trial. Psychosom Med 2013;75:650-7. https://doi.org/10.1097/PSY.0b013e31829dbed4.
- Moss-Morris R, Sharon C, Tobin R, Baldi JC. A randomized controlled graded exercise trial for chronic fatigue syndrome: outcomes and mechanisms of change. J Health Psychol 2005;10:245-59. https://doi.org/10.1177/1359105305049774.
- Huibers MJ, Beurskens AJ, Van Schayck CP, Bazelmans E, Metsemakers JF, Knottnerus JA, et al. Efficacy of cognitive-behavioural therapy by general practitioners for unexplained fatigue among employees: randomised controlled trial. Br J Psychiatry 2004;184:240-6. https://doi.org/10.1192/bjp.184.3.240.
- Whitehead L, Campion P. Can general practitioners manage chronic fatigue syndrome? A controlled trial. J Chronic Fatigue Syndr 2002;10:55-64. https://doi.org/10.1300/J092v10n01_05.
- Leone SS, Huibers MJ, Kant I, van Amelsvoort LG, van Schayck CP, Bleijenberg G, et al. Long-term efficacy of cognitive-behavioral therapy by general practitioners for fatigue: a 4-year follow-up study. J Psychosom Res 2006;61:601-7. https://doi.org/10.1016/j.jpsychores.2006.04.010.
- Escobar JI, Gara MA, Diaz-Martinez AM, Interian A, Warman M, Allen LA, et al. Effectiveness of a time-limited cognitive behavior therapy type intervention among primary care patients with medically unexplained symptoms. Ann Fam Med 2007;5:328-35. https://doi.org/10.1370/afm.702.
- Burton C, Weller D, Marsden W, Worth A, Sharpe M. A primary care Symptoms Clinic for patients with medically unexplained symptoms: pilot randomised trial. BMJ Open 2012;2. https://doi.org/10.1136/bmjopen-2011-000513.
- Larisch A, Schweickhardt A, Wirsching M, Fritzsche K. Psychosocial interventions for somatizing patients by the general practitioner: a randomized controlled trial. J Psychosom Res 2004;57:507-14. https://doi.org/10.1016/j.jpsychores.2004.04.372.
- Schilte AF, Portegijs PJ, Blankenstein AH, van der Horst HE, Latour MB, van Eijk JT, et al. Randomised controlled trial of disclosure of emotionally important events in somatisation in primary care. BMJ 2001;323. https://doi.org/10.1136/bmj.323.7304.86.
- van der Feltz-Cornelis CM, van Oppen P, Adèr HJ, van Dyck R. Randomised controlled trial of a collaborative care model with psychiatric consultation for persistent medically unexplained symptoms in general practice. Psychother Psychosom 2006;75:282-9. https://doi.org/10.1159/000093949.
- Kashner TM, Rost K, Cohen B, Anderson M, Smith GR. Enhancing the health of somatization disorder patients. Effectiveness of short-term group therapy. Psychosomatics 1995;36:462-70. https://doi.org/10.1016/S0033-3182(95)71627-9.
- Marks I. Behavioural Psychotherapy: Maudsley Pocket Book of Clinical Management. Bristol: Wright; 1986.
- Cohen J. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers; 1988.
- Ax S, Gregg VH, Jones D. Chronic fatigue syndrome: sufferers’ evaluation of medical support. J Royal Soc Med 1997;90:250-4. https://doi.org/10.1177/014107689709000505.
- Bayliss K, Goodall M, Chisholm A, Fordham B, Chew-Graham C, Riste L, et al. Overcoming the barriers to the diagnosis and management of chronic fatigue syndrome/ME in primary care: a meta synthesis of qualitative studies. BMC Family Pract 2014;15. https://doi.org/10.1186/1471-2296-15-44.
- Wallace LS, Wexler RK, McDougle L, Miser WF, Haddox JD. Voices that may not otherwise be heard: a qualitative exploration into the perspectives of primary care patients living with chronic pain. J Pain Res 2014;7:291-9. https://doi.org/10.2147/JPR.S62317.
- Deale A, Wessely S. Diagnosis of psychiatric disorder in clinical evaluation of chronic fatigue syndrome. J Royal Soc Med 2000;93:310-12. https://doi.org/10.1177/014107680009300608.
- Raine R, Carter S, Sensky T, Black N. General practitioners’ perceptions of chronic fatigue syndrome and beliefs about its management, compared with irritable bowel syndrome: qualitative study. BMJ 2004;328:1354-7. https://doi.org/10.1136/bmj.38078.503819.EE.
- Glenton C, Colvin CJ, Carlsen B, Swartz A, Lewin S, Noyes J, et al. Barriers and facilitators to the implementation of lay health worker programmes to improve access to maternal and child health: qualitative evidence synthesis. Cochrane Database Syst Rev 2013. https://doi.org/10.1002/14651858.CD010414.
- Rashidian A, Shakibazadeh E, Karimi-Shahanjarini A, Glenton C, Noyes J, Lewin S, et al. Barriers and facilitators to the implementation of doctor-nurse substitution strategies in primary care: qualitative evidence synthesis (Protocol). Cochrane Database Syst Rev 2013;2. https://doi.org/10.1002/14651858.CD010412.
- Colvin CJ, de Heer J, Winterton L, Mellenkamp M, Glenton C, Noyes J, et al. A systematic review of qualitative evidence on barriers and facilitators to the implementation of task-shifting in midwifery services. Midwifery 2013;29:1211-21. https://doi.org/10.1016/j.midw.2013.05.001.
- Munthe-Kaas HM, Hammerstrøm K, Kurtze N, Nordlund KR. Effekt Av Og Erfaringer Med Kontinuitetsfremmende Tiltak I Barnevernsinstitusjoner. Rapport Fra Kunnskapssenteret No. 4–2013 2013.
- Bohren MA, Hunter EC, Munthe-Kaas HM, Souza JP, Vogel JP, Gülmezoglu AM. Facilitators and barriers to facility-based delivery in low- and middle-income countries: a qualitative evidence synthesis. Reprod Health 2014;11. https://doi.org/10.1186/1742-4755-11-71.
- Critical Appraisal Skills Programme . CASP Qualitative Checklist 2018. https://casp-uk.net/wp-content/uploads/2018/03/CASP-Qualitative-Checklist-2018_fillable_form.pdf (accessed 1 June 2019).
- Noyes J, Lewin S, Noyes J, Booth A, Hannes K, Harden A, et al. Supplementary Guidance for Inclusion of Qualitative Research in Cochrane Systematic Reviews of Interventions 2011. http://methods.cochrane.org/qi/supplemental-handbook-guidance (accessed 1 June 2019).
- Thomas J, Harden A. Methods for the thematic synthesis of qualitative research in systematic reviews. BMC Med Res Methodol 2008;8. https://doi.org/10.1186/1471-2288-8-45.
- Noyes J, Popay J, Pearson A, Hannes K, Booth A, Higgins JPT, et al. Cochrane Handbook for Systematic Reviews of Interventions. The Cochrane Collaboration; 2008.
- Peters S, Wearden A, Morriss R, Dowrick CF, Lovell K, Brooks G, et al. Challenges of nurse delivery of psychological interventions for long-term conditions in primary care: a qualitative exploration of the case of chronic fatigue syndrome/myalgic encephalitis. Implement Sci 2011;6. https://doi.org/10.1186/1748-5908-6-132.
- Chew-Graham C, Brooks J, Wearden A, Dowrick C, Peters S. Factors influencing engagement of patients in a novel intervention for CFS/ME: a qualitative study. Prim Health Care Res Dev 2011;12:112-22. https://doi.org/10.1017/S146342361000037X.
- Peters S, Rogers A, Salmon P, Gask L, Dowrick C, Towey M, et al. What do patients choose to tell their doctors? Qualitative analysis of potential barriers to reattributing medically unexplained symptoms. J Gen Int Med 2009;24. https://doi.org/10.1007/s11606-008-0872-x.
- Dowrick C, Gask L, Hughes JG, Charles-Jones H, Hogg JA, Peters S, et al. General practitioners’ views on reattribution for patients with medically unexplained symptoms: a questionnaire and qualitative study. BMC Fam Pract 2008;9. https://doi.org/10.1186/1471-2296-9-46.
- Gerskowitch C, Norman I, Rimes KA. Patients with medically unexplained physical symptoms experience of receiving treatment in a primary-care psychological therapies service: a qualitative study. Cogn Behav Therap 2015;8. https://doi.org/10.1017/S1352465815000235.
- Graham J, Manor O, Wiseman S. Introducing humanistic group counselling for somatisation in a primary care practice: Preliminary findings. Couns Psychother Res 2007;7:220-6. https://doi.org/10.1080/14733140701718420.
- Lewis H. An exploratory study of primary-care therapists’ perceived competence in providing cognitive behavioural therapy to people with medically unexplained symptoms. Cogn Behav Therap 2013;6. https://doi.org/10.1017/S1754470X13000202.
- Morton L, Elliott A, Thomas R, Cleland J, Deary V, Burton C. Developmental study of treatment fidelity, safety and acceptability of a Symptoms Clinic intervention delivered by General Practitioners to patients with multiple medically unexplained symptoms. J Psychosom Res 2016;84:37-43. https://doi.org/10.1016/j.jpsychores.2016.03.008.
- Payne H. The body speaks its mind: the BodyMind Approach for patients with medically unexplained symptoms in primary care in England. Arts Psychoth 2015;42:19-27. https://doi.org/10.1016/j.aip.2014.12.011.
- MacNeela P, Doyle C, O’Gorman D, Ruane N, McGuire BE. Experiences of chronic low back pain: a meta-ethnography of qualitative research. Health Psychol Rev 2015;9:63-82. https://doi.org/10.1080/17437199.2013.840951.
- Harzing AW. Publish or Perish 6.0. London: Harzing.com; 2017.
- Sutton A, de la Cruz MDCG, Booth A, Leaviss J. VP198 efficient retrieval of trial protocols: an empirical study. Int J Technol Assess Health Care 2017;33. https://doi.org/10.1017/S0266462317004251.
- Booth A, Sutton A, Scope A, Leaviss J. OP105 systematically reconstructing trial context-role for CLUSTER searches. Int J Technol Assess Health Care 2017;33. https://doi.org/10.1017/S026646231700174X.
- Morton L, Elliott A, Cleland J, Deary V, Burton C. A taxonomy of explanations in a general practitioner clinic for patients with persistent ‘medically unexplained’ physical symptoms. Patient Educ Counsel 2017;100:224-30. https://doi.org/10.1016/j.pec.2016.08.015.
- Payne H, Brooks SDM. Clinical outcomes from The BodyMind Approach™ in the treatment of patients with medically unexplained symptoms in primary health care in England: practice-based evidence. Arts Psychother 2016;47:55-6. https://doi.org/10.1016/j.aip.2015.12.001.
- Payne H, Warnecke T, Karkou V, Westland G. A comparative analysis of body psychotherapy and dance movement psychotherapy from a European perspective. Body Mov Dance Psychother 2016;11:144-66. https://doi.org/10.1080/17432979.2016.1165291.
- Payne H. Pilot study to evaluate Dance Movement Psychotherapy (the BodyMind Approach) in patients with medically unexplained symptoms: participant and facilitator perceptions and a summary discussion. Body Mov Dance Psychother 2009;4:77-94. https://doi.org/10.1080/17432970902918008.
- Payne H. The BodyMind Approach (BMA) to psychotherapeutic groupwork with patients with medically unexplained symptoms (MUS): a review of the literature, description of approach and methodology for a pilot study. Eur J Psychother Couns 2009;11:287-310. https://doi.org/10.1080/13642530903230392.
- Payne H. Medically unexplained conditions and the bodymind approach. Couns Prim Care Rev 2009;10:6-8.
- Payne H, Stott D. Change in the moving bodymind: quantitative results from a pilot study on the use of the BodyMind approach (BMA) to psychotherapeutic group work with patients with medically unexplained symptoms (MUSs). Couns Psychother Res 2010;10:295-306. https://doi.org/10.1080/14733140903551645.
- Payne H, Fordham R. A Group Bodymind Approach to Medically Unexplained Symptoms: Proof of Concept and Potential Cost Savings 2009.
- Payne H, Eskioglou M, Story J. Medically Unexplained Symptoms: A Holistic BodyMind Approach to Group Therapy for Depression and Anxiety in Primary Care. Newport: East of England Development Agency; 2009.
- Payne H. Psychosomatic conditions in primary care: change within the moving bodymind. E-Motion 2010;20:20-4.
- Payne H. Arts Therapies Conference ‘The Space Between – The Potential For change’. Conference Handbook and Abstract Book. Ottersberg: University of Applied Sciences; 2011.
- Lin Y, Payne H. The BodyMind Approach™, medically unexplained symptoms and personal construct psychology. Body Mov Dance Psychother 2014;9:154-66. https://doi.org/10.1080/17432979.2014.918563.
- Payne H, Payne H, Brooks SD, Brooks SD. Moving on: the BodyMind Approach™ for medically unexplained symptoms. J Pub Ment Health 2017;16:63-71. https://doi.org/10.1108/JPMH-10-2016-0052.
- Payne H, Karkou V, Oliver S, Lycouris S. The Oxford Handbook of Dance and Wellbeing. Oxford: Oxford University Press; 2017.
- Samaritter R, Payne H. Being moved: Kinaesthetic reciprocities in psychotherapeutic interaction and the development of enactive intersubjectivity. Eur Psychother 2016;13.
- Wearden AJ, Riste L, Dowrick C, Chew-Graham C, Bentall RP, Morriss RK, et al. Fatigue Intervention by Nurses Evaluation – the FINE Trial. A randomised controlled trial of nurse led self-help treatment for patients in primary care with chronic fatigue syndrome: study protocol. BMC Med 2006;4. https://doi.org/10.1186/1741-7015-4-9.
- Wearden AJ, Dunn G, Dowrick C, Morriss RK. Depressive symptoms and pragmatic rehabilitation for chronic fatigue syndrome. Br J Psychiatry 2012;201:227-32. https://doi.org/10.1192/bjp.bp.111.107474.
- Wearden AJ, Emsley R. Mediators of the effects on fatigue of pragmatic rehabilitation for chronic fatigue syndrome. J Consult Clin Psychol 2013;81:831-8. https://doi.org/10.1037/a0033561.
- Chew-Graham CA, Cahill G, Dowrick C, Wearden A, Peters S. Using multiple sources of knowledge to reach clinical understanding of chronic fatigue syndrome. Ann Fam Med 2008;6:340-8. https://doi.org/10.1370/afm.867.
- Chew-Graham C, Dixon R, Shaw JW, Smyth N, Lovell K, Peters S. Practice nurses’ views of their role in the management of chronic fatigue syndrome/myalagic encephalitis: a qualitative study. BMC Nurs 2009;8. https://doi.org/10.1186/1472-6955-8-2.
- Chew-Graham C, Dowrick C, Wearden A, Richardson V, Peters S. Making the diagnosis of chronic fatigue syndrome/myalgic encephalitis in primary care: a qualitative study. BMC Fam Pract 2010;11. https://doi.org/10.1186/1471-2296-11-16.
- Wearden AJ, Chew-Graham C. Managing chronic fatigue syndrome in U.K. primary care: challenges and opportunities. Chronic Illn 2006;2:143-53. https://doi.org/10.1177/17423953060020020901.
- Band R, Wearden A, Barrowclough C. Patient outcomes in association with significant other responses to chronic fatigue syndrome: a systematic review of the literature. Clin Psychol 2015;22:29-46. https://doi.org/10.1111/cpsp.12093.
- Hannon K, Peters S, Fisher L, Riste L, Wearden A, Lovell K, et al. Developing resources to support the diagnosis and management of chronic fatigue syndrome/myalgic encephalitis (CFS/ME) in primary care: a qualitative study. BMC Fam Pract 2012;13. https://doi.org/10.1186/1471-2296-13-93.
- Bayliss K, Riste L, Fisher L, Wearden A, Peters S, Lovell K, et al. Diagnosis and management of chronic fatigue syndrome/myalgic encephalitis in black and minority ethnic people: a qualitative study. Prim Health Care Res Dev 2014;15:143-55. https://doi.org/10.1017/S1463423613000145.
- Brooks JM, Daglish J, Wearden AJ. Attributions, distress and behavioural responses in the significant others of people with chronic fatigue syndrome. J Health Psychol 2013;18:1288-95. https://doi.org/10.1177/1359105312464670.
- Brooks J, King N, Wearden A. Couples’ experiences of interacting with outside others in chronic fatigue syndrome: a qualitative study. Chronic Illn 2014;10:5-17. https://doi.org/10.1177/1742395312474478.
- Salmon P, Peters S, Clifford R, Iredale W, Gask L, Rogers A, et al. Why do general practitioners decline training to improve management of medically unexplained symptoms?. J Gen Int Med 2007;22:565-71. https://doi.org/10.1007/s11606-006-0094-z.
- Salmon P, Peters S, Rogers A, Gask L, Clifford R, Iredale W, et al. Peering through the barriers in GPs’ explanations for declining to participate in research: the role of professional autonomy and the economy of time. Fam Pract 2007;24:269-75. https://doi.org/10.1093/fampra/cmm015.
- Morriss R, Dowrick C, Salmon P, Peters S, Rogers A, Dunn G, et al. Turning theory into practice: rationale, feasibility and external validity of an exploratory randomized controlled trial of training family practitioners in reattribution to manage patients with medically unexplained symptoms (the MUST). Gen Hosp Psychiatry 2006;28:343-51. https://doi.org/10.1016/j.genhosppsych.2006.03.012.
- Morriss R, Kai J, Atha C, Avery A, Bayes S, Franklin M, et al. Persistent frequent attenders in primary care: costs, reasons for attendance, organisation of care and potential for cognitive behavioural therapeutic intervention. BMC Fam Pract 2012;13. https://doi.org/10.1186/1471-2296-13-39.
- Salmon P, Wissow L, Carroll J, Ring A, Humphris GM, Davies JC, et al. Doctors’ attachment style and their inclination to propose somatic interventions for medically unexplained symptoms. Gen Hosp Psychiatry 2008;30:104-11. https://doi.org/10.1016/j.genhosppsych.2007.12.002.
- Morriss R, Lindson N, Coupland C, Dex G, Avery A. Estimating the prevalence of medically unexplained symptoms from primary care records. Public Health 2012;126:846-54. https://doi.org/10.1016/j.puhe.2012.05.008.
- Morriss R, Dowrick C, Salmon P, Peters S, Dunn G, Rogers A, et al. ‘Reattribution for medically unexplained symptoms’: authors’ reply. Br J Psychiatry 2008;192. https://doi.org/10.1192/bjp.192.4.315.
- Morriss R, Gask L. Assessment and immediate management of patients with medically unexplained symptoms in primary care and general hospital settings. Found Years 2008;4:59-63. https://doi.org/10.1016/j.mpfou.2008.02.003.
- Salmon P, Ring A, Humphris GM, Davies JC, Dowrick CF. Primary care consultations about medically unexplained symptoms: how do patients indicate what they want?. J Gen Intern Med 2009;24:450-6. https://doi.org/10.1007/s11606-008-0898-0.
- McGowan L, Luker K, Creed F, Chew-Graham CA. ‘How do you explain a pain that can’t be seen?’: the narratives of women with chronic pelvic pain and their disengagement with the diagnostic cycle. Br J Health Psychol 2007;12:261-74. https://doi.org/10.1348/135910706X104076.
- Pawson R, Greenhalgh T, Harvey G, Walshe K. Realist review – a new method of systematic review designed for complex policy interventions. J Health Serv Res Policy 2005;10:21-34. https://doi.org/10.1258/1355819054308530.
- Kendrick E. Impact of Social Support on the Relationship Between Illness Invalidation and Shame Among Individuals with Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Norman, OK: University of Oklahoma; 2015.
- Jutel A. Medically unexplained symptoms and the disease label. Social Theory Health 2010;8:229-45. https://doi.org/10.1057/sth.2009.21.
- van der Kar AG. Determinants of Consulting the General Practitioner. Maastricht: Maastricht University; 1992.
- Horton-Salway M. The local production of knowledge: disease labels, identities and category entitlements in ME support group talk. Health 2004;8:351-71. https://doi.org/10.1177/1363459304043474.
- Hoedeman R, Blankenstein AH, van der Feltz-Cornelis CM, Krol B, Stewart R, Groothoff JW. Consultation letters for medically unexplained physical symptoms in primary care. Cochrane Database Syst Rev 2010;12. https://doi.org/10.1002/14651858.CD006524.pub2.
- Sharpe M, Peveler R, Mayou R. The psychological treatment of patients with functional somatic symptoms: a practical guide. J Psychosom Res 1992;36:515-29. https://doi.org/10.1016/0022-3999(92)90037-3.
- Stone L. Managing the consultation with patients with medically unexplained symptoms: a grounded theory study of supervisors and registrars in general practice. BMC Family Pract 2014;15. https://doi.org/10.1186/s12875-014-0192-7.
- Wileman L, May C, Chew-Graham CA. Medically unexplained symptoms and the problem of power in the primary care consultation: a qualitative study. Family Pract 2002;19:178-82. https://doi.org/10.1093/fampra/19.2.178.
- Sobel RK. In the heart, not the head. Chronic fatigue is real. US News World Rep 2000;129.
- Kroenke K, Swindle R. Cognitive-behavioral therapy for somatization and symptom syndromes: a critical review of controlled clinical trials. Psychother Psychosom 2000;69:205-15. https://doi.org/10.1159/000012395.
- Ridd M, Shaw A, Salisbury C. ’Two sides of the coin’ – the value of personal continuity to GPs: a qualitative interview study. Fam Pract 2006;23:461-8. https://doi.org/10.1093/fampra/cml010.
- Royal College of Psychiatrists . Whole-Person Care: From Rhetoric to Reality (Achieving Parity Between Mental and Physical Health) (OP88) 2013.
- Murray AM, Toussaint A, Althaus A, Lowe B. The challenge of diagnosing non-specific, functional, and somatoform disorders: a systematic review of barriers to diagnosis in primary care. J Psychosom Res 2016;80:1-10. https://doi.org/10.1016/j.jpsychores.2015.11.002.
- Johansen ML, Risor MB. What is the problem with medically unexplained symptoms for GPs? A meta-synthesis of qualitative studies. Patient Educ Couns 2017;100:647-54. https://doi.org/10.1016/j.pec.2016.11.015.
- Denz-Penhey H, Murdoch JC. Service delivery for people with chronic fatigue syndrome: a pilot action research study. Fam Pract 1993;10:14-8. https://doi.org/10.1093/fampra/10.1.14.
- National Institute for Health and Care Excellence (NICE) . Guide to the Methods of Technology Appraisal 2013.
- van Ravenzwaaij J, Olde Hartman TC, van Ravesteijn H, Eveleigh R, van Rijswijk E, Lucassen PLBJ. Explanatory models of medically unexplained symptoms: a qualitative analysis of the literature. Ment Health Fam Med 2011;7:223-31.
- Engel CC, Katon W. Commissioned Paper: Unexplained Physical Symptoms in Primary Care and the Community: What Might We Learn for Prevention in the Military? Paper Prepared for the Strategies to Protect the Health of Deployed U.S. Forces: Medical Surveillance, Medical Record Keeping, and Risk Reduction. Washington, DC: Institute of Medicine; 1999.
- van Ravenzwaaij J, Olde Hartman TC, van Ravesteijn H, Eveleigh R, van Rijswijk E, Lucassen PLBJ. Explanatory models of medically unexplained symptoms: a qualitative analysis of the literature. Ment Health Fam Med 2010;7:223-31.
- Hadler NM. If you have to prove you are ill, you can’t get well. The object lesson of fibromyalgia. Spine 1996;21. https://doi.org/10.1097/00007632-199610150-00021.
- Allen A. EU Economies Filter: ISSG Search Filters Resource n.d. https://sites.google.com/a/york.ac.uk/issg-search-filters-resource/other-filters/filters-to-find-studies-of-geographic-locations/eu-economies (accessed 16 May 2017).
- National Institute for Health and Care Excellence (NICE) . Developing NICE Guidelines: The Manual Appendix H 2015.
- Engel GL. The need for a new medical model: a challenge for biomedicine. Science 1977;196:129-36. https://doi.org/10.1126/science.847460.
- O’Dowd H, Gladwell P, Rogers CA, Hollinghurst S, Gregory A. Cognitive behavioural therapy in chronic fatigue syndrome: a randomised controlled trial of an outpatient group programme. Health Technol Assess 2006;10. https://doi.org/10.3310/hta10370.
- Friedberg MW. Group cognitive behavioral treatment improves chronic low back pain in a cost-effective manner. J Clin Outcomes Manage 2010;17:7-9.
- Hill JC, Whitehurst DG, Lewis M, Bryan S, Dunn KM, Foster NE, et al. Comparison of stratified primary care management for low back pain with current best practice (STarT Back): a randomised controlled trial. Lancet 2011;378:1560-71. https://doi.org/10.1016/S0140-6736(11)60937-9.
- Johnson RE, Jones GT, Wiles NJ, Chaddock C, Potter RG, Roberts C, et al. Active exercise, education, and cognitive behavioral therapy for persistent disabling low back pain: a randomized controlled trial. Spine 2007;32:1578-85. https://doi.org/10.1097/BRS.0b013e318074f890.
- Taylor SJ, Carnes D, Homer K, Kahan BC, Hounsome N, Eldridge S, et al. Novel three-day, Community-Based, Nonpharmacological Group Intervention for Chronic Musculoskeletal Pain (COPERS): a randomised clinical trial. PLOS Med 2016;13. https://doi.org/10.1371/journal.pmed.1002040.
- NHS Plymouth . Medically Unexplained Symptoms (MUS): A Whole Systems Approach in Plymouth 2009. www.iapt.nhs.uk/silo/files/medically-unexplained-symptoms-mus-a-wholesystems-approach-in-plymouth.pdf (accessed June 2020).
- McGowan L, Escott D, Luker K, Creed F, Chew-Graham C. Is chronic pelvic pain a comfortable diagnosis for primary care practitioners: a qualitative study. BMC Fam Pract 2010;11. https://doi.org/10.1186/1471-2296-11-7.
- Hoedeman R, Krol B, Blankenstein AH, Koopmans PC, Groothoff JW. Sick-listed employees with severe medically unexplained physical symptoms: burden or routine for the occupational health physician? A cross sectional study. BMC Health Serv Res 2010;10. https://doi.org/10.1186/1472-6963-10-305.
- Geraets JJ, Goossens ME, van Haastregt JC, de Groot IJ, de Bruijn CP, de Bie RA, et al. Implications of process evaluation for clinical effectiveness and clinical practice in a trial on chronic shoulder complaints. Patient Educ Couns 2006;61:117-25. https://doi.org/10.1016/j.pec.2005.02.017.
- Reid S, Whooley D, Crayford T, Hotopf M. Medically unexplained symptoms – GPs’ attitudes towards their cause and management. Family Pract 2001;18:519-23. https://doi.org/10.1093/fampra/18.5.519.
- Shattock L, Williamson H, Caldwell K, Anderson K, Peters S. ‘They’ve just got symptoms without science’: Medical trainees’ acquisition of negative attitudes towards patients with medically unexplained symptoms. Patient Educ Couns 2013;91:249-54. https://doi.org/10.1016/j.pec.2012.12.015.
- Gerhards SA, de Graaf LE, Jacobs LE, Severens JL, Huibers MJ, Arntz A, et al. Economic evaluation of online computerised cognitive-behavioural therapy without support for depression in primary care: randomised trial. Br J Psychiatry 2010;196:310-18. https://doi.org/10.1192/bjp.bp.109.065748.
- Jones R, Barraclough K, Dowrick C. When no diagnostic label is applied. BMJ 2010;340. https://doi.org/10.1136/bmj.c2683.
- Konnopka A, Konig HH, Kaufmann C, Stuhldreher N, Wild B, Szecsenyi J, et al. Cost-effectiveness of a collaborative group intervention for patients with medically unexplained symptoms. J Ment Health Policy Econ 2015;18:S25-6.
- Luciano JV, Sabes-Figuera R, Cardeñosa E, T Peñarrubia-María M, Fernández-Vergel R, García-Campayo J, et al. Cost-utility of a psychoeducational intervention in fibromyalgia patients compared with usual care: an economic evaluation alongside a 12-month randomized controlled trial. Clin J Pain 2013;29:702-11. https://doi.org/10.1097/AJP.0b013e318270f99a.
- Luo Z, Goddeeris J, Gardiner JC, Smith RC. Costs of an intervention for primary care patients with medically unexplained symptoms: a randomized controlled trial. Psychiatr Serv 2007;58:1079-86. https://doi.org/10.1176/ps.2007.58.8.1079.
- Margalit AP, El-Ad A. Costly patients with unexplained medical symptoms: a high-risk population. Patient Educ Couns 2008;70:173-8. https://doi.org/10.1016/j.pec.2007.09.020.
- Meng H, Friedberg F, Castora-Binkley M. Cost-effectiveness of chronic fatigue self-management versus usual care: a pilot randomized controlled trial. BMC Fam Pract 2014;15. https://doi.org/10.1186/s12875-014-0184-7.
- Oliver K, Cronan TA, Walen HR, Tomita M. Effects of social support and education on health care costs for patients with fibromyalgia. J Rheumatol 2001;28:2711-19.
- Salmon P, Sharma N, Valori R, Bellenger N. Patients’ intentions in primary care: relationship to physical and psychological symptoms, and their perception by general practitioners. Social Sci Med 1994;38:585-92. https://doi.org/10.1016/0277-9536(94)90255-0.
- Schade R, Dieleman JP, Sturkenboom MC. Trends in the prescription of antiparkinsonian drugs among patients with Parkinson’s disease in the United Kingdom: a population-based analysis. Mov Disord 2011;26.
- van der Roer N, van Tulder MW, Barendse JM, van Mechelen W, Franken WK, Ooms AC, et al. Cost-effectiveness of an intensive group training protocol compared to physiotherapy guideline care for sub-acute and chronic low back pain: design of a randomised controlled trial with an economic evaluation. BMC Musculoskelet Disord 2004;5. https://doi.org/10.1186/1471-2474-5-45.
- Morriss R, Gask L, Ronalds C, Downes-Grainger E, Thompson H, Leese B, et al. Cost-effectiveness of a new treatment for somatized mental disorder taught to GPs. Fam Pract 1998;15:119-25. https://doi.org/10.1093/fampra/15.2.119.
- Chisholm D, Godfrey E, Ridsdale L, Chalder T, King M, Seed P, et al. Chronic fatigue in general practice: economic evaluation of counselling versus cognitive behaviour therapy. Br J Gen Pract 2001;51:15-8.
- Kennedy TM, Chalder T, McCrone P, Darnley S, Knapp M, Jones RH, et al. Cognitive behavioural therapy in addition to antispasmodic therapy for irritable bowel syndrome in primary care: randomised controlled trial. Health Technol Assess 2006;10. https://doi.org/10.3310/hta10190.
- Hubley S, Uebelacker L, Eaton C. Managing medically unexplained symptoms in primary care: a narrative review and treatment recommendations. Am J Lifestyle Med 2016;10:109-19. https://doi.org/10.1177/1559827614536865.
- McCrone P, Knapp M, Kennedy T, Seed P, Jones R, Darnley S, et al. Cost-effectiveness of cognitive behaviour therapy in addition to mebeverine for irritable bowel syndrome. Eur J Gastroenterol Hepatol 2008;20:255-63. https://doi.org/10.1097/MEG.0b013e3282f2519d.
- McCrone P, Ridsdale L, Darbishire L, Seed P. Cost-effectiveness of cognitive behavioural therapy, graded exercise and usual care for patients with chronic fatigue in primary care. Psychol Med 2004;34:991-9. https://doi.org/10.1017/S0033291704001928.
- Sabes-Figuera R, McCrone P, Hurley M, King M, Donaldson AN, Ridsdale L. Cost-effectiveness of counselling, graded-exercise and usual care for chronic fatigue: evidence from a randomised trial in primary care. BMC Health Serv Res 2012;12. https://doi.org/10.1186/1472-6963-12-264.
- Salmon P, Peters S, Stanley I. Patients’ perceptions of medical explanations for somatisation disorders: qualitative analysis. BMJ 1999;318:372-6. https://doi.org/10.1136/bmj.318.7180.372.
- Richardson G, Epstein D, Chew-Graham C, Dowrick C, Bentall RP, Morriss RK, et al. Cost-effectiveness of supported self-management for CFS/ME patients in primary care. BMC Fam Pract 2013;14. https://doi.org/10.1186/1471-2296-14-12.
- Kirmayer L, Groleau D, Looper K, Dao M. Explaining medically unexplained symptoms. Can J Psychiatr 2004;49:663-71. https://doi.org/10.1177/070674370404901003.
- Fox R, Bendelow G, Carpenter M, Vautier C, Williams S. Gender, Health and Healing: The Public/Private Divide. London: Routledge; 2002.
- Griffiths F, Green E, Tsouroufli M. The nature of medical evidence and its inherent uncertainty for the clinical consultation: qualitative study. BMJ 2005;330. https://doi.org/10.1136/bmj.38336.482720.8F.
- Howman M, Walters K, Rosenthal J, Ajjawi R, Buszewicz M. ‘You kind of want to fix it don’t you?’ Exploring general practice trainees’ experiences of managing patients with medically unexplained symptoms. BMC Med Educ 2016;16. https://doi.org/10.1186/s12909-015-0523-y.
- Wainwright D, Calnan M, O’Neil C, Winterbottom A, Watkins C. When pain in the arm is ‘all in the head’: the management of medically unexplained suffering in primary care. Health Risk Soc 2006;8:71-88. https://doi.org/10.1080/13698570500532512.
- May C, Allison G, Chapple A, Chew-Graham C, Dixon C, Gask L, et al. Framing the doctor-patient relationship in chronic illness: a comparative study of general practitioners’ accounts. Sociol Health Illn 2004;26:135-58. https://doi.org/10.1111/j.1467-9566.2004.00384.x.
- Artus M, Croft P, Lewis M. The use of CAM and conventional treatments among primary care consulters with chronic musculoskeletal pain. BMC Fam Pract 2007;8. https://doi.org/10.1186/1471-2296-8-26.
- Stanley IM, Peters S, Salmon P. A primary care perspective on prevailing assumptions about persistent medically unexplained physical symptoms. Int J Psychiatry Med 2002;32:125-40. https://doi.org/10.2190/AVM3-8GU8-JW70-5RX5.
- Weiland A, Blankenstein AH, Van Saase JL, Van der Molen HT, Jacobs ME, Abels DC, et al. Training medical specialists to communicate better with patients with medically unexplained physical symptoms (MUPS). A randomized, controlled trial. PLOS ONE 2015;10. https://doi.org/10.1371/journal.pone.0138342.
- King M, Davidson O, Taylor F, Haines A, Sharp D, Turner R. Effectiveness of teaching general practitioners skills in brief cognitive behaviour therapy to treat patients with depression: randomised controlled trial. BMJ 2002;324:947-51B. https://doi.org/10.1136/bmj.324.7343.947.
- Yon K, Nettleton S, Walters K, Lamahewa K, Buszewicz M. Junior doctors’ experiences of managing patients with medically unexplained symptoms: a qualitative study. BMJ Open 2015;5. https://doi.org/10.1136/bmjopen-2015-009593.
- van Dessel N, Leone SS, van der Wouden JC, Dekker J, van der Horst HE. The PROSPECTS study: design of a prospective cohort study on prognosis and perpetuating factors of medically unexplained physical symptoms (MUPS). J Psychosom Res 2014;76:200-6. https://doi.org/10.1016/j.jpsychores.2013.12.011.
- Paulson M, Norberg A, Danielson E. Men living with fibromyalgia-type pain: experiences as patients in the Swedish health care system. J Adv Nurs 2002;40:87-95. https://doi.org/10.1046/j.1365-2648.2002.02343.x.
- Toye F, Barker K. Persistent non-specific low back pain and patients’ experience of general practice: a qualitative study. Prim Health Care Res Dev 2012;13:72-84. https://doi.org/10.1017/S1463423611000387.
- Stone L. Blame, shame and hopelessness: medically unexplained symptoms and the ‘heartsink’ experience. Aust Fam Physician 2014;43:191-5.
- Hubley S, Uebelacker LA, Nash J, Eaton CB. Open trial of integrated primary care consultation for medically unexplained symptoms. J Behav Health Serv Res 2017;44:590-601. https://doi.org/10.1007/s11414-016-9528-5.
- Sallinen M, Kukkurainen ML, Peltokallio L. Finally heard, believed and accepted – peer support in the narratives of women with fibromyalgia. Patient Educ Couns 2011;85:e126-30. https://doi.org/10.1016/j.pec.2011.02.011.
- Corbett M, Foster NE, Ong BN. Living with low back pain – stories of hope and despair. Soc Sci Med 2007;65:1584-94. https://doi.org/10.1016/j.socscimed.2007.06.008.
- Larun L, Malterud K. Identity and coping experiences in chronic fatigue syndrome: a synthesis of qualitative studies. Patient Educ Couns 2007;69:20-8. https://doi.org/10.1016/j.pec.2007.06.008.
- Nettleton S. ‘I just want permission to be ill’: towards a sociology of medically unexplained symptoms. Soc Sci Med 2006;62:1167-78. https://doi.org/10.1016/j.socscimed.2005.07.030.
- Anderson VR, Jason LA, Hlavaty LE, Porter N, Cudia J. A review and meta-synthesis of qualitative studies on myalgic encephalomyelitis/chronic fatigue syndrome. Patient Educ Couns 2012;86:147-55. https://doi.org/10.1016/j.pec.2011.04.016.
- Burton C. Beyond somatisation: a review of the understanding and treatment of medically unexplained physical symptoms (MUPS). Br J Gen Pract 2003;53:231-9.
- Drachler Mde L, Leite JC, Hooper L, Hong CS, Pheby D, Nacul L, et al. The expressed needs of people with chronic fatigue syndrome/myalgic encephalomyelitis: a systematic review. BMC Public Health 2009;9. https://doi.org/10.1186/1471-2458-9-458.
- Pinxsterhuis I, Strand EB, Sveen U. Coping with chronic fatigue syndrome: a review and synthesis of qualitative studies. Fatig Biomed Health Behav 2015;3:173-88. https://doi.org/10.1080/21641846.2015.1035519.
- Toye F, Seers K, Allcock N, Briggs M, Carr E, Andrews J, et al. Patients’ experiences of chronic non-malignant musculoskeletal pain: a qualitative systematic review. Br J Gen Pract 2013;63:e829-41. https://doi.org/10.3399/bjgp13X675412.
- Toye F, Seers K, Allcock N, Briggs M, Carr E, Andrews J, et al. A meta-ethnography of patients’ experience of chronic non-malignant musculoskeletal pain. Osteoarthr Cartilage 2013;21:S259-S260. https://doi.org/10.1016/j.joca.2013.02.539.
- Toye F, Seers K, Barker K. A meta-ethnography of patients’ experiences of chronic pelvic pain: struggling to construct chronic pelvic pain as ‘real’. J Adv Nurs 2014;70:2713-27. https://doi.org/10.1111/jan.12485.
- Aiarzaguena JM, Gaminde I, Grandes G, Salazar A, Alonso I, Sánchez A. Somatisation in primary care: experiences of primary care physicians involved in a training program and in a randomised controlled trial. BMC Fam Pract 2009;10. https://doi.org/10.1186/1471-2296-10-73.
- Arnold IA, de Waal MW, Eekhof JA, Assendelft WJ, Spinhoven P, van Hemert AM. Medically unexplained physical symptoms in primary care: a controlled study on the effectiveness of cognitive-behavioral treatment by the family physician. Psychosomatics 2009;50:515-24. https://doi.org/10.1176/appi.psy.50.5.515.
- Blane DN, Williams C, Morrison J, Wilson A, Mercer S. Psychological therapies in primary care: a progress report. Br J Gen Pract 2014;64:556-7. https://doi.org/10.3399/bjgp14X682117.
- Chew-Graham CA. Focus on shared decision-making. Health Expect 2017;20:1181-2. https://doi.org/10.1111/hex.12654.
- den Boeft M, Huisman D, Morton L, Lucassen P, van der Wouden JC, Westerman MJ, et al. Negotiating explanations: doctor-patient communication with patients with medically unexplained symptoms – a qualitative analysis. Fam Pract 2017;34:107-13. https://doi.org/10.1093/fampra/cmw113.
- den Boeft M, Huisman D, van der Wouden JC, Numans ME, van der Horst HE, Lucassen PL, et al. Recognition of patients with medically unexplained physical symptoms by family physicians: results of a focus group study. BMC Fam Pract 2016;17. https://doi.org/10.1186/s12875-016-0451-x.
- den Boeft M, Claassen-van Dessel N, van der Wouden JC. How should we manage adults with persistent unexplained physical symptoms?. BMJ 2017;356. https://doi.org/10.1136/bmj.j268.
- Finset A, Mjaaland TA. The medical consultation viewed as a value chain: a neurobehavioral approach to emotion regulation in doctor–patient interaction. Patient Educ Couns 2009;74:323-30. https://doi.org/10.1016/j.pec.2008.12.007.
- Graugaard PK, Eide H, Finset A. Interaction analysis of physician–patient communication: the influence of trait anxiety on communication and outcome. Patient Educ Couns 2003;49:149-56. https://doi.org/10.1016/S0738-3991(02)00074-5.
- Huibers MJ, Beurskens AJ, Bleijenberg G, van Schayck CP. Psychosocial interventions by general practitioners. Cochrane Database Syst Rev 2007;3. https://doi.org/10.1002/14651858.CD003494.pub2.
- Kromme NMH, Ahaus CTB, Gans ROB, van de Wiel HBM. ‘It just has to click’: internists’ views of: what constitutes productive interactions with chronically ill patients. BMC Health Serv Res 2016;16. https://doi.org/10.1186/s12913-016-1430-6.
- Layard R. The case for psychological treatment centres. BMJ 2006;332:1030-2. https://doi.org/10.1136/bmj.332.7548.1030.
- Malins S, Kai J, Atha C, Avery A, Guo B, James M, et al. Cognitive behaviour therapy for long-term frequent attenders in primary care: a feasibility case series and treatment development study. Br J Gen Pract 2016;66:e729-36. https://doi.org/10.3399/bjgp16X686569.
- McDermott C, Lynch J, Leydon GM. Patients’ hopes and expectations of a specialist chronic fatigue syndrome/ME service: a qualitative study. Fam Pract 2011;28:572-8. https://doi.org/10.1093/fampra/cmr016.
- Picariello F, Ali S, Foubister C, Chalder T. ‘It feels sometimes like my house has burnt down, but I can see the sky’: a qualitative study exploring patients’ views of cognitive behavioural therapy for chronic fatigue syndrome. Br J Health Psychol 2017;22:383-41. https://doi.org/10.1111/bjhp.12235.
- Rask MT, Andersen RS, Bro F, Fink P, Rosendal M. Towards a clinically useful diagnosis for mild-to-moderate conditions of medically unexplained symptoms in general practice: a mixed methods study. BMC Fam Pract 2014;15. https://doi.org/10.1186/1471-2296-15-118.
- Schweickhardt A, Larisch A, Fritzsche K. Differentiation of somatizing patients in primary care: why the effects of treatment are always moderate. J Nerv Ment Dis 2005;193:813-19. https://doi.org/10.1097/01.nmd.0000188975.22144.3d.
- Sirri L, Grandi S, Tossani E. Medically unexplained symptoms and general practitioners: a comprehensive survey about their attitudes, experiences and management strategies. Fam Pract 2017;34:201-5. https://doi.org/10.1093/fampra/cmw130.
- Ziadni MS, Carty JN, Doherty HK, Porcerelli JH, Rapport LJ, Schubiner H, et al. A life-stress, emotional awareness, and expression interview for primary care patients with medically unexplained symptoms: a randomized controlled trial. Health Psychol 2018;37:282-90. https://doi.org/10.1037/hea0000566.
- InterTASC Information Specialists’ Sub-Group . Search Filter Resource n.d. https://sites.google.com/a/york.ac.uk/issg-search-filters-resource/ (accessed 15 May 2017).
- Creed F, Fernandes L, Guthrie E, Palmer S, Ratcliffe J, Read N, et al. The cost-effectiveness of psychotherapy and paroxetine for severe irritable bowel syndrome. Gastroenterology 2003;124:303-17. https://doi.org/10.1053/gast.2003.50055.
- Yardley L, Barker F, Muller I, Turner D, Kirby S, Mullee M, et al. Clinical and cost effectiveness of booklet based vestibular rehabilitation for chronic dizziness in primary care: single blind, parallel group, pragmatic, randomised controlled trial. BMJ 2012;344. https://doi.org/10.1136/bmj.e2237.
- Lamb SE, Hansen Z, Lall R, Castelnuovo E, Withers EJ, Nichols V, et al. Group cognitive behavioural treatment for low-back pain in primary care: a randomised controlled trial and cost-effectiveness analysis. Lancet 2010;375:916-23. https://doi.org/10.1016/S0140-6736(09)62164-4.
- Cherkin DC, Deyo RA, Battié M, Street J, Barlow W. A comparison of physical therapy, chiropractic manipulation, and provision of an educational booklet for the treatment of patients with low back pain. N Engl J Med 1998;339:1021-9. https://doi.org/10.1056/NEJM199810083391502.
- Ritzwoller DP, Crounse L, Shetterly S, Rublee D. The association of comorbidities, utilization and costs for patients identified with low back pain. BMC Musculoskelet Disord 2006;7. https://doi.org/10.1186/1471-2474-7-72.
- Richardson G, Epstein D, Wearden A. PRM23 Cost-effectiveness versus patient acceptability: the exemplar of CFS/ME. Value Health 2012;15. https://doi.org/10.1016/j.jval.2012.08.1487.
- Department of Health and Social Care . NHS Reference Costs 2002 2002.
- Netten A, Curtis L. Unit Costs of Health and Social Care 2003. Canterbury: Personal Social Services Research Unit, University of Kent; 2003.
- Curtis L. Unit Costs of Health and Social Care 2016. Canterbury: PSSRU, University of Kent; 2016.
- Curtis L. Unit Costs of Health and Social Care 2014. Canterbury: PSSRU, University of Kent; 2014.
- van der Roer N, van Tulder M, van Mechelen W, de Vet H. Economic evaluation of an intensive group training protocol compared with usual care physiotherapy in patients with chronic low back pain. Spine 2008;33:445-51. https://doi.org/10.1097/BRS.0b013e318163fa59.
- van Ravesteijn H, Grutters J, olde Hartman T, Lucassen P, Bor H, van Weel C, et al. Mindfulness-based cognitive therapy for patients with medically unexplained symptoms: a cost-effectiveness study. J Psychosom Res 2013;74:197-205. https://doi.org/10.1016/j.jpsychores.2013.01.001.
- LeFort SM, Gray-Donald K, Rowat KM, Jeans ME. Randomized controlled trial of a community-based psychoeducation program for the self-management of chronic pain. Pain 1998;74:297-306. https://doi.org/10.1016/S0304-3959(97)00190-5.
- Smith BJ, McGorm KJ, Weller D, Burton C, Sharpe M. The identification in primary care of patients who have been repeatedly referred to hospital for medically unexplained symptoms: a pilot study. J Psychosom Res 2009;67:207-11. https://doi.org/10.1016/j.jpsychores.2009.01.004.
- Jenkinson C, Layte R. Development and testing of the UK SF-12 (short form health survey). J Health Serv Res Policy 1997;2:14-8. https://doi.org/10.1177/135581969700200105.
- Brazier J, Roberts J, Deverill M. The estimation of a preference-based measure of health from the SF-36. J Health Econ 2002;21:271-92. https://doi.org/10.1016/S0167-6296(01)00130-8.
- Dakin H, Burns R, Yang Y. HERC Database of Mapping Studies, Version 5.0 (past Updated: 16 May 2016) 2016. www.herc.ox.ac.uk/downloads/herc-database-of-mapping-studies (accessed February 2017).
- Dakin HBR, Yang Y. HERC Database of Mapping Studies 2016.
- Kim SH, Kim SO, Lee SI, Jo MW. Deriving a mapping algorithm for converting SF-36 scores to EQ-5D utility score in a Korean population. Health Qual Life Outcomes 2014;12. https://doi.org/10.1186/s12955-014-0145-9.
- Longo M, Cohen D, Hood K, Robling M. Deriving an ‘Enhanced’ EuroQol from SF-36. Health Economists Studying Group Meeting (HESG) 2000.
- Longworth L. Estimating Quality Adjusted Life Years where Health-related Utility Data are Missing. London: Brunel University London; 2007.
- Maund E, Craig D, Suekarran S, Neilson A, Wright K, Brealey S, et al. Management of frozen shoulder: a systematic review and cost-effectiveness analysis. Health Technol Assess 2012;16. https://doi.org/10.3310/hta16110.
- Ara R, Brazier J. Deriving an algorithm to convert the eight mean SF-36 dimension scores into a mean EQ-5D preference-based score from published studies (where patient level data are not available). Value Health 2008;11:1131-43. https://doi.org/10.1111/j.1524-4733.2008.00352.x.
- Rowen D, Brazier J, Roberts J. Mapping SF-36 onto the EQ-5D index: how reliable is the relationship?. Health Qual Life Outcomes 2009;7. https://doi.org/10.1186/1477-7525-7-27.
- Chuang LH, Kind P. Converting the SF-12 into the EQ-5D: an empirical comparison of methodologies. PharmacoEconomics 2009;27:491-505. https://doi.org/10.2165/00019053-200927060-00005.
- Coca Perraillon M, Shih YC, Thisted RA. Predicting the EQ-5D-3L Preference index from the SF-12 health survey in a national US sample: a finite mixture approach. Med Decis Making 2015;35:888-901. https://doi.org/10.1177/0272989X15577362.
- Le QA. Probabilistic mapping of the health status measure SF-12 onto the health utility measure EQ-5D using the US-population-based scoring models. Qual Life Res 2014;23:459-66. https://doi.org/10.1007/s11136-013-0517-3.
- Sullivan PW, Ghushchyan V. Mapping the EQ-5D index from the SF-12: US general population preferences in a nationally representative sample. Med Decis Making 2006;26:401-9. https://doi.org/10.1177/0272989X06290496.
- Zarate V, Chuang L, Kind P. PMC41 mapping SF-12 to EuroQoL EQ-5D preference scores in the Spanish-speaking hispanic community in the United States. Value Health 2008;11. https://doi.org/10.1016/S1098-3015(10)70575-1.
- Conigliani C, Manca A, Tancredi A. Prediction of patient-reported outcome measures via multivariate ordered probit models. J Royal Statistical Soc 2015;178:567-91. https://doi.org/10.1111/rssa.12072.
- Gray AM, Rivero-Arias O, Clarke PM. Estimating the association between SF-12 responses and EQ-5D utility values by response mapping. Med Decis Making 2006;26:18-29. https://doi.org/10.1177/0272989X05284108.
- Rivero-Arias O, Gray A, Ramos-Goñi J. MA2 estimating the association between SF-12 responses to EQ-5D utility values using a response mapping and a parameter uncertainty approach. Value Health 2010;13. https://doi.org/10.1016/S1098-3015(11)71887-3.
- Lawrence WF, Fleishman JA. Predicting EuroQoL EQ-5D preference scores from the SF-12 Health Survey in a nationally representative sample. Med Decis Making 2004;24:160-9. https://doi.org/10.1177/0272989X04264015.
- Franks P, Lubetkin EI, Gold MR, Tancredi DJ, Jia H. Mapping the SF-12 to the EuroQol EQ-5D Index in a national US sample. Med Decis Making 2004;24:247-54. https://doi.org/10.1177/0272989X04265477.
- Franks P, Lubetkin EI, Gold MR, Tancredi DJ. Mapping the SF-12 to preference-based instruments: convergent validity in a low-income, minority population. Med Care 2003;41:1277-83. https://doi.org/10.1097/01.MLR.0000093480.58308.D8.
- Lidbeck J. Group therapy for somatization disorders in primary care: maintenance of treatment goals of short cognitive-behavioural treatment one-and-a-half-year follow-up. Acta Psychiatr Scand 2003;107:449-56. https://doi.org/10.1034/j.1600-0447.2003.00048.x.
- Malouff JM, Thorsteinsson EB, Rooke SE, Bhullar N, Schutte NS. Efficacy of cognitive behavioral therapy for chronic fatigue syndrome: a meta-analysis. Clin Psychol Rev 2008;28:736-45. https://doi.org/10.1016/j.cpr.2007.10.004.
- Kisely SR, Campbell LA, Skerritt P, Yelland MJ. Psychological interventions for symptomatic management of non-specific chest pain in patients with normal coronary anatomy. Cochrane Database Syst Rev 2010;1. https://doi.org/10.1002/14651858.CD004101.pub3.
- Picariello F, Ali S, Moss-Morris R, Chalder T. The most popular terms for medically unexplained symptoms: the views of CFS patients. J Psychosom Res 2015;78:420-6. https://doi.org/10.1016/j.jpsychores.2015.02.013.
- Pajak R, Lackner J, Kamboj SK. A systematic review of minimal-contact psychological treatments for symptom management in irritable bowel syndrome. J Psychosom Res 2013;75:103-12. https://doi.org/10.1016/j.jpsychores.2013.05.007.
- Macedo LG, Smeets RJ, Maher CG, Latimer J, McAuley JH. Graded activity and graded exposure for persistent nonspecific low back pain: a systematic review. Phys Ther 2010;90:860-79. https://doi.org/10.2522/ptj.20090303.
- Action for M.E . M.E. Time to Deliver 2014. www.actionforme.org.uk/uploads/pdfs/me-time-to-deliver-survey-report.pdf (accessed 5 November 2019).
- Rask MT, Rosendal M, Fenger-Grøn M, Bro F, Ørnbøl E, Fink P. Sick leave and work disability in primary care patients with recent-onset multiple medically unexplained symptoms and persistent somatoform disorders: a 10-year follow-up of the FIP study. Gen Hosp Psychiatry 2015;37:53-9. https://doi.org/10.1016/j.genhosppsych.2014.10.007.
- O’Dowd H. ISRCTN Registry n.d. https://doi.org/10.1186/ISRCTN72645894 (accessed 2 May 2019).
- O’Dowd H. WhatdotheyKnow.Com. Results for Trial ID ISRCTN72645894 2016. www.whatdotheyknow.com/request/results_for_trial_id_isrctn72645 (accessed 3 May 2019).
- Duggan C, Parry G, McMurran M, Davidson K, Dennis J. The recording of adverse events from psychological treatments in clinical trials: evidence from a review of NIHR-funded trials. Trials 2014;15. https://doi.org/10.1186/1745-6215-15-335.
- Parry GD, Crawford MJ, Duggan C. Iatrogenic harm from psychological therapies – time to move on. Br J Psychiatry 2016;208:210-12. https://doi.org/10.1192/bjp.bp.115.163618.
- Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutical trials. J Chronic Dis 1967;20:637-48. https://doi.org/10.1016/0021-9681(67)90041-0.
- Thorpe KE, Zwarenstein M, Oxman AD, Treweek S, Furberg CD, Altman DG, et al. A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers. J Clin Epidemiol 2009;62:464-75. https://doi.org/10.1016/j.jclinepi.2008.12.011.
- Lindström I, Öhlund C, Eek C, Wallin L, Peterson LE, Fordyce WE, Nachemson AL. The effect of graded activity on patients with subacute low back pain: a randomized prospective clinical study with an operant-conditioning behavioral approach. Phys Ther 1992;72:279-90. https://doi.org/10.1093/ptj/72.4.279.
- Hodges PW, Richardson CA. Delayed postural contraction of transversus abdominis in low back pain associated with movement of the lower limb. J Spin Disord 1998;11:46-5. https://doi.org/10.1097/00002517-199802000-00008.
- Swedish Council on Technology Assessment in Health Care . Methods of Treating Chronic Pain 2006.
- Salkovoskis PM, Hawton K, Salkovoskis PM, Kirk J, Clark DM. Cognitive Behaviour Therapy for Psychiatric Problems: A Practical Guide. Oxford: Oxford University Press; 1989.
- Gask L, Goldberg D, Porter R, Creed F. The treatment of somatization: evaluation of a teaching package with general practice trainees. J Psychosom Res 1989;33:697-703. https://doi.org/10.1016/0022-3999(89)90085-8.
- Ewing JA. Detecting alcoholism: the CAGE questionnaire. JAMA 1984;252:1905-7. https://doi.org/10.1001/jama.252.14.1905.
- Chalder T, Berelowitz G, Pawlikowska T, Watts L, Wessely S, Wright D, et al. Development of a fatigue scale. J Psychosom Res 1993;37:147-53. https://doi.org/10.1016/0022-3999(93)90081-P.
- Mersky H, Bogduk N. Classification of Chronic Pain. Washington, DC: IASP Press; 1994.
- Drossman DA, Richter JE, Talley NJ, Thompson WG, Corazziari E, Whitehead WE. The Functional Gastrointestinal Disorders: Diagnosis, Pathophysiology and Treatment. McLean, VA: Degnon Associates; 1994.
- Ware JE, Snow KK, Kosinski M. SF-36 Health Survey: Manual and Interpretation Guide, QualityMetric Inc. Lincoln, RI; 1993.
- Robins LN. Composing the CIDI: A Collaboration with the WHO/ADAMHA. St. Louis: Department of Psychiatry Research Seminar; 1988.
- Spitzer RL, Williams JB, Kroenke K, Linzer M, deGruy FV, Hahn SR, et al. Utility of a new procedure for diagnosing mental disorders in primary care. The PRIME-MD 1000 study. JAMA 1994;272:1749-56. https://doi.org/10.1001/jama.1994.03520220043029.
- Brugha TS, Cragg D. The List of Threatening Experiences: the reliability and validity of a brief life events questionnaire. Acta Psychiatr Scand 1990;82:77-81. https://doi.org/10.1111/j.1600-0447.1990.tb01360.x.
- Kressin NR, Spiro A, Skinner KM. Negative affectivity and health-related quality of life. Med Care 2000;38:858-67. https://doi.org/10.1097/00005650-200008000-00009.
- Sullivan MJ, Bishop SR, Pivik J. The pain catastrophizing scale: development and validation. Psychological Assessment 1995;7. https://doi.org/10.1037/1040-3590.7.4.524.
- Hamilton M. Rating depressive patients. J Clin Psychiatry 1980;41:21-4.
- Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol 1959;32:50-5. https://doi.org/10.1111/j.2044-8341.1959.tb00467.x.
- Melzak R. Pain Measurement and Assessment. New York, NY: Raven Press; 1983.
- Burckhardt CS, Clark SR, Bennett RM. The fibromyalgia impact questionnaire: development and validation. J Rheumatol 1991;18:728-33.
- Badia X, Roset M, Montserrat S, Herdman M, Segura A. The Spanish version of EuroQol: a description and its applications. European Quality of Life scale. Med Clin 1999;112:79-85.
- Spitzer RL, Kroenke K, Williams JB. Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. Primary Care Evaluation of Mental Disorders. Patient Health Questionnaire. JAMA 1999;282:1737-44. https://doi.org/10.1001/jama.282.18.1737.
- Spitzer RL, Kroenke K, Williams JB, Löwe B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006;166:1092-7. https://doi.org/10.1001/archinte.166.10.1092.
- Lewis G, Pelosi AJ, Araya R, Dunn G. Measuring psychiatric disorder in the community: a standardized assessment for use by lay interviewers. Psychol Med 1992;22:465-86. https://doi.org/10.1017/s0033291700030415.
- Goldberg D, Williams P. A User’s Guide to the General Health Questionnaire. Windsor: NFER-NELSON; 1988.
- Stewart AL, Hays R, Ware J. The medical outcomes study short-form general health survey. Medical Care 1988;26:724-35. https://doi.org/10.1097/00005650-198807000-00007.
- Huskisson EC. Measurement of pain. Lancet 1974;304:1127-31. https://doi.org/10.1016/S0140-6736(74)90884-8.
- Kovacs FM, Llobera J, Gil Del Real MT, Abraira V, Gestoso M, Fernández C, et al. Validation of the spanish version of the Roland-Morris questionnaire. Spine 2002;27:538-42. https://doi.org/10.1097/00007632-200203010-00016.
- Luo X, George ML, Kakouras I, Edwards CL, Pietrobon R, Richardson W, et al. Reliability, validity, and responsiveness of the short form 12-item survey (SF-12) in patients with back pain. Spine 2003;28:1739-45. https://doi.org/10.1097/01.BRS.0000083169.58671.96.
- Ware JE, Keller SD, Kosinski M. SF-12: How to Score the SF-12 Physical and Mental Health Summary Scales. Health Institute, New England Medical Center; 1995.
- Beck AT, Steer RA, Brown GK. Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation; 1996.
- Woolfolk RL, Allen LA, Tiu JE. New directions in the treatment of somatization. Psychiatric Clinics of North America 2007;30:621-44. https://doi.org/10.1016/j.psc.2007.07.001.
- Lobo A, Campos R, Pérez-Echeverría MJ, Izuzquiza J, García-Campayo J, Saz P, et al. A new interview for the multiaxial assessment of psychiatric morbidity in medical settings. Psychol Med 1993;23:505-10. https://doi.org/10.1017/s0033291700028592.
- Ware JE, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care 1992;30:473-83. https://doi.org/10.1097/00005650-199206000-00002.
- Lam CL, Tse EY, Gandek B. Is the standard SF-12 health survey valid and equivalent for a Chinese population?. Qual Life Res 2005;14:539-47. https://doi.org/10.1007/s11136-004-0704-3.
- Vercoulen JHMM, Alberts M, Bleijenberg G. De checklist individuele spankracht (CIS). Gedragstherapie 1999;32.
- Feinstein AR. Clinimetrics. New Haven, CT: Yale University Press; 1987.
- Derogatis LR, Lipman RS, Covi L. SCL-90: an outpatient psychiatric rating scale – preliminary report. Psychopharmacol Bull 1973;9:13-28.
- Brook RH, Ware JE, Davies-Avery A, Stewart AL, Donald CA, Rogers WH, et al. Overview of adult health status measures fielded in Rand’s Health Insurance Study. Med Care 1979;17:i-131.
- Francis CY, Morris J, Whorwell PJ. The irritable bowel severity scoring system: a simple method of monitoring irritable bowel syndrome and its progress. Aliment Pharmacol Ther 1997;11:395-402. https://doi.org/10.1046/j.1365-2036.1997.142318000.x.
- Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361-70. https://doi.org/10.1111/j.1600-0447.1983.tb09716.x.
- Mouanoutoua VL, Brown LG. Hopkins Symptom Checklist-25, Hmong version: a screening instrument for psychological distress. J Pers Assess 1995;64:376-83. https://doi.org/10.1207/s15327752jpa6402_16.
- Hosman CMH. Psychosociale problematiek en hulpzoeken Een socialepidemiologische studie t. b. v. de preventive geestelijke gezondheidszorg (Psychosocial problems and help seeking. A social-epidemiological study for the preventive mental health care). Amsterdam: Swets en Zeitlinger; 1983.
- Bridges KR, Sanderman R, Van Sonderen E. An English language version of the social support list: preliminary reliability. Psychological Reports 2002;90:1055-8. https://doi.org/10.2466/pr0.2002.90.3.1055.
- Aaronson NK, Muller M, Cohen PD, Essink-Bot ML, Fekkes M, Sanderman R, et al. Translation, validation, and norming of the Dutch language version of the SF-36 Health Survey in community and chronic disease populations. J Clin Epidemiol 1998;51:1055-68. https://doi.org/10.1016/S0895-4356(98)00097-3.
- Derogatis LR, Cleary PA. Factorial invariance across gender for the primary symptom dimensions of the SCL-90. Br J Soc Clin Psychol 1977;16:347-56. https://doi.org/10.1111/j.2044-8260.1977.tb00241.x.
- Cecile MT, van Wijk G, Kolk AM. Psychometric evaluation of symptom perception related measures. Personality and Individual Differences 1996;20:55-70. https://doi.org/10.1016/0191-8869(95)90023-N.
- Robbins JM, Kirmayer LJ. Attributions of common somatic symptoms. Psychol Med 1991;21:1029-45. https://doi.org/10.1017/S0033291700030026.
- Watson D, Clark LA, Tellegen A. Development and validation of brief measures of positive and negative affect: the PANAS scales. J Pers Soc Psychol 1988;54:1063-70. https://doi.org/10.1037//0022-3514.54.6.1063.
- Vernon H, Mior S. The Neck Disability Index: a study of reliability and validity. J Manipulative Physiol Ther 1991;14:409-15. https://doi.org/10.1037/t35122-000.
- Goldberg D, Paul W. A User’s Guide to the General Health Questionnaire. Windsor: Nfer-Nelson; 1991.
- Braden CJ. A test of the Self-Help Model: learned response to chronic illness experience. Nurs Res 1990;39:42-7. https://doi.org/10.1097/00006199-199001000-00009.
- Baker F, Curbow B, Wingard JR. Development of the satisfaction with life domains scale for cancer. J Psychosoc Oncol 1992;10:75-90. https://doi.org/10.1300/J077V10N03_05.
- Melzack R. The short-form McGill pain questionnaire. Pain 1987;30:191-7. https://doi.org/10.1016/0304-3959(87)91074-8.
- Beck AT, Beck RW. Screening depressed patients in family practice. A rapid technic. Postgrad Med 1972;52:81-5. https://doi.org/10.1080/00325481.1972.11713319.
- Jensen MP, Turner JA, Romano JM, Lawler BK. Relationship of pain-specific beliefs to chronic pain adjustment. Pain 1994;57:301-9. https://doi.org/10.1016/0304-3959(94)90005-1.
- Mishel MH. The measurement of uncertainty in illness. Nurs Res 1981;30:258-63. https://doi.org/10.1097/00006199-198109000-00002.
- Lorig K, Chastain RL, Ung E, Shoor S, Holman HR. Development and evaluation of a scale to measure perceived self-efficacy in people with arthritis. Arthritis Rheum 1989;32:37-44. https://doi.org/10.1002/anr.1780320107.
- Rosenbaum M. A schedule for assessing self-control behaviors: preliminary findings. Behavior Therapy 1980;11:109-21. https://doi.org/10.1016/S0005-7894(80)80040-2.
- Göbel H. Paper-pencil tests for retrospective and prospective evaluation of primary headaches on the basis of the IHS criteria. Headache 1994;34:564-8. https://doi.org/10.1111/j.1526-4610.1994.hed3410564.x.
- Rivera J, González T. The Fibromyalgia Impact Questionnaire: a validated Spanish version to assess the health status in women with fibromyalgia. Clin Exp Rheumatol 2004;22:554-60.
- Spielberger CD. State–Trait Anxiety Inventory. The Corsini Encyclopedia of Psychology. Hoboken: John Wiley & Sons, Inc.; 2010.
- McCracken LM, Vowles KE, Eccleston C. Acceptance of chronic pain: component analysis and a revised assessment method. Pain 2004;107:159-66. https://doi.org/10.1016/j.pain.2003.10.012.
- Scrimshaw SV, Maher C. Responsiveness of visual analogue and McGill pain scale measures. J Manipulative Physiol Ther 2001;24:501-4. https://doi.org/10.1067/mmt.2001.118208.
- Pengel LH, Refshauge KM, Maher CG. Responsiveness of pain, disability, and physical impairment outcomes in patients with low back pain. Spine 2004;29:879-83. https://doi.org/10.1097/00007632-200404150-00011.
- Roland M, Morris R. A study of the natural history of low-back pain. Part II: development of guidelines for trials of treatment in primary care. Spine 1983;8:145-50. https://doi.org/10.1097/00007632-198303000-00005.
- Sanson-Fisher RW, Perkins JJ. Adaptation and validation of the SF-36 Health Survey for use in Australia. J Clin Epidemiol 1998;51:961-7. https://doi.org/10.1016/S0895-4356(98)00087-0.
- Borkovec TD, Nau SD. Credibility of analogue therapy rationales. J Behav Ther Experimental Psychiatry 1972;3:257-60. https://doi.org/10.1016/0005-7916(72)90045-6.
- Pengel LH, Refshauge KM, Maher CG, Nicholas MK, Herbert RD, McNair P. Physiotherapist-directed exercise, advice, or both for subacute low back pain: a randomized trial. Ann Intern Med 2007;146:787-96. https://doi.org/10.7326/0003-4819-146-11-200706050-00007.
- Marques M, De Gucht V, Gouveia MJ, Cordeiro A, Leal I, Maes S. Psychometric Properties of the Portuguese Version of the Checklist of Individual Strength (CIS20-P) 2013. https://doi.org/10.5964/pch.v2i1.57.
- Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med Singapore 1994;23:129-38.
- Wendel-Vos GC, Schuit AJ, Saris WH, Kromhout D. Reproducibility and relative validity of the short questionnaire to assess health-enhancing physical activity. J Clin Epidemiol 2003;56:1163-9. https://doi.org/10.1016/S0895-4356(03)00220-8.
- Ware JE, Kosinski M, Turner-Bowker D, Gandek B. Version 2 of the SF-12 Health Survey. Boston, MA: Quality Metric; 2002.
- Derogatis L. Brief Symptom Inventory: Administration, Scoring and Procedures Manual. Minneapolis: National Computer System; 1993.
- Rief W, Hiller W, Geissner E, Fichter MM. Hypochondrie: erfassung und erste klinische ergebnisse. Zeitscrift Fur Klinische Psychologie 1994;23:34-42.
- Beck AT, Steer RA. Beck Depression Inventory Manual. San Antonio, TX: The Psychological Corporation; 1987.
- Franke G. BSI (Brief Symptoms Inventory). Weinheim: Beltz; 2000.
- Rief W, Hiller W. A new approach to the assessment of the treatment effects of somatoform disorders. Psychosomatics 2003;44:492-8. https://doi.org/10.1176/appi.psy.44.6.492.
- Lohaus A, Schmitt GM. Fragebogen zur Erhebung von Kontrolluberzeugungen zu Krankheit und Gesundheit. Gottingen: Hogrefe; 1990.
- Brown GK, Nicassio PM. Development of a questionnaire for the assessment of active and passive coping strategies in chronic pain patients. Pain 1987;31:53-64. https://doi.org/10.1016/0304-3959(87)90006-6.
- Jenkins CD, Stanton BA, Niemcryk SJ, Rose RM. A scale for the estimation of sleep problems in clinical research. J Clin Epidemiol 1988;41:313-21. https://doi.org/10.1016/0895-4356(88)90138-2.
- Roelofs J, Goubert L, Peters ML, Vlaeyen JW, Crombez G. The Tampa Scale for Kinesiophobia: further examination of psychometric properties in patients with chronic low back pain and fibromyalgia. Eur J Pain 2004;8:495-502. https://doi.org/10.1016/j.ejpain.2003.11.016.
- Morriss RK, Gask L. Treatment of patients with somatized mental disorder: effects of reattribution training on outcomes under the direct control of the family doctor. Psychosomatics 2002;43:394-9. https://doi.org/10.1176/appi.psy.43.5.394.
- Moss-Morris R, Weinman J, Petrie K, Horne R, Cameron L, Buick D. The revised illness perception questionnaire (IPQ-R). Psychol Health 2002;17:1-16. https://doi.org/10.1080/08870440290001494.
- Pilowsky I. Dimensions of hypochondriasis. Br J Psychiatry 1967;113:89-93. https://doi.org/10.1192/bjp.113.494.89.
- EuroQoL Group . EuroQoL: a new facility for the measurement of health related quality of life. Health Policy 1990;16:199-208. https://doi.org/10.1016/0168-8510(90)90421-9.
- Fulcher KY, White PD. Randomised controlled trial of graded exercise in patients with the chronic fatigue syndrome. BMJ 1997;314:1647-52. https://doi.org/10.1136/bmj.314.7095.1647.
- Ray C, Weir W, Stewart D, Miller P, Hyde G. Ways of coping with chronic fatigue syndrome: development of an illness management questionnaire. Soc Sci Med 1993;37:385-91. https://doi.org/10.1016/0277-9536(93)90268-9.
- Müller-Lissner S, Koch G, Talley NJ, Drossman D, Rueegg P, Dunger-Baldauf C, et al. Subject’s global assessment of relief: an appropriate method to assess the impact of treatment on irritable bowel syndrome-related symptoms in clinical trials. J Clin Epidemiol 2003;56:310-16. https://doi.org/10.1016/S0895-4356(03)00027-1.
- Mundt JC, Marks IM, Shear MK, Greist JH. The Work and Social Adjustment Scale: a simple measure of impairment in functioning. Br J Psychiatry 2002;180:461-4. https://doi.org/10.1192/bjp.180.5.461.
- Derogatis LR, Lipman RS, Rickels K, Uhlenhuth EH, Covi L. The Hopkins symptom checklist: a self-report symptom inventory. Behav Sci 1971;19:1-15. https://doi.org/10.1002/bs.3830190102.
- Main CJ. The modified somatic perception questionnaire (MSPQ). J Psychosom Res 1983;27:503-14. https://doi.org/10.1016/0022-3999(83)90040-5.
- Ware JE, Snow KK, Kosinski M, Gandek BSF. Health Survey Manual and Interpretation Guide. Boston, MA: The Health Institute, New England Medical Center; 1993.
- Caine TM, Foulds GA, Hope K. Manual of the Hostility and Direction of Hostility Questionnaire. London: University of London Press; 1967.
- Mooney V, Cairns D, Robertson J. A system for evaluating and treating chronic back disability. West J Med 1976;124:370-6.
- Sivik T. Alexithymia and hypersensitivity to touch and palpation. Integr Physiol Behav Sci 1993;28:130-6. https://doi.org/10.1007/BF02691215.
- Sifneos PE. The Schalling-Sifneos personality scale revised. Psychother Psychosom 1986;45:161-5. https://doi.org/10.1159/000287942.
- Weinryb RM, Gustavsson JP, Asberg M, Rössel RJ. The concept of alexithymia: an empirical study using psychodynamic ratings and self-reports. Acta Psychiatr Scand 1992;85:153-62. https://doi.org/10.1111/j.1600-0447.1992.tb01461.x.
- Powell R, Dolan R, Wessely S. Attributions and self-esteem in depression and chronic fatigue syndromes. J Psychosom Res 1990;34:665-73. https://doi.org/10.1016/0022-3999(90)90111-g.
- Dolan P, Gudex C, Kind P, Williams A. A Social Tariff for EuroQol: Results from a UK General Population Study. The University of York, York: Centre for Health Economics; 1995.
- Beck AT, Epstein N, Brown G, Steer RA. An inventory for measuring clinical anxiety: psychometric properties. J Consult Clin Psychol 1988;56. https://doi.org/10.1037/0022-006X.56.6.893.
- Bensoussan A, Talley NJ, Hing M, Menzies R, Guo A, Ngu M. Treatment of irritable bowel syndrome with Chinese herbal medicine: a randomized controlled trial. JAMA 1998;280:1585-9. https://doi.org/10.1001/jama.280.18.1585.
- Patrick DL, Drossman DA, Frederick IO, DiCesare J, Puder KL. Quality of life in persons with irritable bowel syndrome: development and validation of a new measure. Dig Dis Sci 1998;43:400-11. https://doi.org/10.1023/a:1018831127942.
- Ware JE, Kosinski M, Keller S. SF-36 Physical and Mental Health Summary Scales: A User’s Guide. Boston: The Health Institute, New England Medical Center; 1994.
- Bjorner JB, Damsgaard MT, Watt T, Bech P, Rasmussen NK, Kristensen TS, et al. Danish Manual for SF-36. Copenhagen: Lif; 1997.
- Fink P, Ewald H, Jensen J, Sørensen L, Engberg M, Holm M, et al. Screening for somatization and hypochondriasis in primary care and neurological in-patients: a seven-item scale for hypochondriasis and somatization. J Psychosom Res 1999;46:261-73. https://doi.org/10.1016/S0022-3999(98)00092-0.
- Zung WW. A self-rating depression scale. Arch Gen Psychiatr 1965;12:63-70. https://doi.org/10.1001/archpsyc.1965.01720310065008.
- Nyström C, Nyström O. Estimated stress verifies autonomic dysfunction. Lakartidningen 1996;93.
- Fairbank JC, Couper J, Davies JB, O’Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy 1980;66:271-3.
- Kaplan RM, Ganiats TG, Sieber WJ, Anderson JP. The Quality of Well-Being Scale: critical similarities and differences with SF-36. Int J Qual Health Care 1998;10:509-20. https://doi.org/10.1093/intqhc/10.6.509.
- Alonso J, Prieto L, Antó JM. The Spanish version of the SF-36 Health Survey (the SF-36 health questionnaire): an instrument for measuring clinical results. Med Clin 1995;104:771-6.
- Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatr 1961;4:561-71. https://doi.org/10.1001/archpsyc.1961.01710120031004.
- King S, Wessel J, Bhambhani Y, Maikala R, Sholter D, Maksymowych W. Validity and reliability of the 6 minute walk in persons with fibromyalgia. J Rheumatol 1999;26:2233-7.
- Whaley MH, Brubaker PH, Otto RM, Armstrong LE. Medicine ACoS. ACSM’s Guidelines for Exercise Testing and Prescription. Philadelphia, PA: Lippincott Williams & Wilkins; 2005.
- Bullinger M, Kirchberger I. SF-36. Fragebogen zum Gesundheitszustand. Handanweisung und Fragebogen (SF-36 und SF-12). Gottingen: Hogrefe; 1998.
- Kroenke K, Spitzer RL, Williams JB, Löwe B. The Patient Health Questionnaire Somatic, Anxiety, and Depressive Symptom Scales: a systematic review. Gen Hosp Psychiatry 2010;32:345-59. https://doi.org/10.1016/j.genhosppsych.2010.03.006.
- Löwe B, Gräfe K, Zipfel S, Spitzer RL, Herrmann-Lingen C, Witte S, et al. Detecting panic disorder in medical and psychosomatic outpatients: comparative validation of the Hospital Anxiety and Depression Scale, the Patient Health Questionnaire, a screening question, and physicians’ diagnosis. J Psychosom Res 2003;55:515-19. https://doi.org/10.1016/S0022-3999(03)00072-2.
- Schilte AF, Portegijs PJ, Blankenstein AH, Knottnerus JA. Somatisation in primary care: clinical judgement and standardised measurement compared. Soc Psychiatry Psychiatr Epidemiol 2000;35:276-82. https://doi.org/10.1007/s001270050239.
- Radloff LS. The CES-D scale: a self report depression scale for research in the general population. Appl Psychol Measurement 1977;1:385-401. https://doi.org/10.1177/014662167700100306.
- Smith RC, Lyles JS, Mettler J, Stoffelmayr BE, Van Egeren LF, Marshall AA, et al. The effectiveness of intensive training for residents in interviewing: a randomized, controlled study. Ann Int Med 1998;128:118-26. https://doi.org/10.7326/0003-4819-128-2-199801150-00008.
- Goldberg DP, Blackwell B. Psychiatric illness in general practice. A detailed study using a new method of case identification. Br Med J 1970;1:439-43. https://doi.org/10.1136/bmj.2.5707.439.
- de Silva N, Samarasinghe D. Acceptance of a psychiatric screening questionnaire by general practice attenders. Ceylon Med J 1990;35:105-8.
- Derogatis LR, Melisaratos N. The brief symptom inventory: an introductory report. Psychol Med 1983;13:595-60. https://doi.org/10.1017/S0033291700048017.
- Speckens AE, van Hemert AM, Spinhoven P, Hawton KE, Bolk JH, Rooijmans HG. Cognitive behavioural therapy for medically unexplained physical symptoms: a randomised controlled trial. BMJ 1995;311:1328-32. https://doi.org/10.1136/bmj.311.7016.1328.
- Bergner M, Bobbitt RA, Pollard WE, Martin DP, Gilson BS. The sickness impact profile: validation of a health status measure. Med Care 1976;14:57-6. https://doi.org/10.1097/00005650-197601000-00006.
- Meeuwesen L, Arrindell WA, Huyse FJ. Psychometric qualities of the SCL-90 in outpatients with abdominal or low back pain. Tijdsch Soc Gez 1992;70:123-31.
- Koopmans GT, Meeuwesen L, Huyse FJ, Meiland FJ, Donker JM. Effects of psychiatric consultation on medical consumption in medical outpatients with abdominal pain. Psychosomatics 1995;36:387-99. https://doi.org/10.1016/S0033-3182(95)71648-6.
- Kori SH, Miller RP, Todd DD. Kinesiophobia: a new view of chronic pain behaviour. Pain Management 1990;3:35-42.
- Kraaimat FW, Evers AW. Pain-coping strategies in chronic pain patients: psychometric characteristics of the pain-coping inventory (PCI). In J Behav Med 2003;10:343-63. https://doi.org/10.1207/S15327558IJBM1004_5.
- Nicholas MK. The pain self-efficacy questionnaire: taking pain into account. Eur J Pain 2007;11:153-63. https://doi.org/10.1016/j.ejpain.2005.12.008.
- Hakkaart-van Roijen L, Van Straten A, Donker M, Tiemens B. Manual Trimbos/iMTA Questionnaire for Costs Associated with Psychiatric Illness (TiC-P). Rotterdam: Institute for Medical Technology Assessment, Erasmus University; 2002.
- Baer RA, Smith GT, Lykins E, Button D, Krietemeyer J, Sauer S, et al. Construct validity of the five facet mindfulness questionnaire in meditating and nonmeditating samples. Assessment 2008;15:329-42. https://doi.org/10.1177/1073191107313003.
- Horn KK, Jennings S, Richardson G, Vliet DV, Hefford C, Abbott JH. The patient-specific functional scale: psychometrics, clinimetrics, and application as a clinical outcome measure. J Orthop Sports Phys Ther 2012;42:30-42. https://doi.org/10.2519/jospt.2012.3727.
- Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in chronic low back pain and disability. Pain 1993;52:157-68. https://doi.org/10.1016/0304-3959(93)90127-B.
- Catley MJ, Tabor A, Wand BM, Moseley GL. Assessing tactile acuity in rheumatology and musculoskeletal medicine – how reliable are two-point discrimination tests at the neck, hand, back and foot?. Rheumatology 2013;52:1454-61. https://doi.org/10.1093/rheumatology/ket140.
- Harwood RH, Rogers A, Dickinson E, Ebrahim S. Measuring handicap: the London Handicap Scale, a new outcome measure for chronic disease. Qual Health Care 1994;3:11-6. https://doi.org/10.1136/qshc.3.1.11.
- Arrindell WA, Ettema JHM. Symptom Checklist. Handleiding bij een multidimensionale psychopathologie indicator.[SCL-90. Handbook of a multidimensional psychopathology indicator]. Lisse: Swets & Zeitlinger; 2003.
- Higgins JPT, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ 2011;343. https://doi.org/10.1136/bmj.d5928.
- Netten A, Knight J, Dennett J, Cooley R, Slight A. A Ready Reckoner for Staff