Journals Library

Identification of gambling problems in primary care: A quantitative study

Project title

Identification of gambling problems in primary care: A quantitative study

Project reference

SPCR 244

Final report date

28 April 2017

Project start date

01 October 2015

Project end date

31 March 2017

Project duration

18 Months

Project keywords

gambling, general practice, cross-sectional study, England

Lead investigator(s)

Dr Sean Cowlishaw, School of Social and Community Medicine, University of Bristol

NIHR School Collaborators
  • Dr David Kessler, School of Social and Community Medicine, University of Bristol
  • Dr Alison Gregory, School of Social and Community Medicine, University of Bristol
  • Ms Lone Gale, School of Social and Community Medicine, University of Bristol (fixed term researcher)
  • Professor Jim McCambridge Chair in Addictive Behaviours and Public Health, University of York (external collaborator)



Project objectives 

The aims of this project were to improve awareness and recognition of gambling problems in general practices in England, and examine the need and scope for initiatives to identify and address these problems in primary care.

  1. Provide preliminary data on the prevalence of gambling problems among patients attending general practices in England;
  2. Examine variability in rates across practices providing services to different populations characterised by various levels of socio-economic deprivation and vulnerability; and
  3. Test associations with socio-demographic and clinical factors, thus indicating vulnerable groups and implications of gambling problems for patients and families.

Brief summary

Aims and objectives

It was originally intended that n=554 patients would be recruited from k=8 general practices. Data monitoring during the early stages of data collection indicated unexpected levels of missing data on the gambling problem items, and recruitment targets were increased to compensate. Our initial expectations regarding high refusal rates were found to be overly conservative, and these increased recruitment targets were achieved within allowances for researcher time that were originally specified.

The original proposals to conduct two interactive workshops with practitioners and policymakers was advised against by the Project Advisory Committee (PAC), who suggested that having sufficient numbers of practitioners / policymakers attend an event about a small project was unrealistic. These events were replaced with: (a) a presentation at the SW Public Health Conference in Bristol; (b) presentations of the findings to individual practices who took part in the study; and (c) one-on-one meetings with local policymakers (e.g., from Bristol City Council public health team).


Participants and procedure

The target population comprised patients attending general practices in the Bristol region of southwest England. Eleven practices were purposively sampled according to population deprivation and patient characteristics, as follows: (1) deprivation levels were quantified using data from the Office for National Statistics, which indicated four practices from deprived areas (top 30% for deprivation in England), two practices in areas of low deprivation (bottom 30%), and three practices in a moderate band (middle 40% for deprivation); (2) one practice provided care to young adults in a student health service, and one practice provided services to a homeless population. The latter were targeted to assess risk according to key population sub-groups.

Patients aged over 18 years and attending practices for any reason were eligible, but were excluded if they were unable to understand English, required immediate medical attention, or were unable to give consent. Patients were approached by a researcher in waiting rooms before appointments, and were provided with information about the study. Those who provided consent were given anonymous questionnaires. These were self-completed and returned in the waiting room or using pre-paid envelopes, and yielded n = 1,058 questionnaires. Across practices sampled according to deprivation, the patient numbers ranged from n = 58 to n = 122. There was n = 17 and n = 163 participants recruited from the practice for homeless patients and the student health service, respectively.


Brief measures identified mental health concerns and addictive behaviours. These included the 2-item Whooley scale for depression, and the GAD-2 scale for anxiety. Risky alcohol use was measured using the three consumption items from the Alcohol Use Disorders Identification Test (AUDIT-C). Non-prescription and recreational drug use was assessed using a Single-Item Screening Question (SISQ) for unhealthy drug use.

Gambling frequency was assessed using items derived from the British Gambling Prevalence Surveys, and participants asked about purchases of lottery or instant win / scratch tickets, play on bingo, casino table games, slot machines and other electronic gambling machines, games of skill against other individuals, or betting money on sporting events. These items used past year timeframes (0 = Never, 6 = 4+ times a week), along with an item about any other gambling. Patients reporting gambling were then asked to complete the Problem Gambling Severity Index (PGSI) [22], which consists of 9-items scored on 4-point response scales (0 = Never, 3 = Almost always) that relate to past year experiences. There was a single-item about whether family members or close relatives had ever had problems with gambling.

Data analyses

Data-file preparation was conducted using SPSS Version 21, while analyses were conducted using Program R. These comprised descriptive analyses of rates of gambling problems and other mental health concerns and addictive behaviours. Exploratory analyses of variability according to practice characteristics were conducted, followed by evaluations of associations with patient-level characteristics. These comprised Pearson c2-tests and logistic regression models that explored significant effects. The latter specified gambling problems as endogenous variables, and with patient characteristics treated as exogenous. These were evaluated in separate models, which thus estimated bivariate associations through Odds Ratios (ORs) and 95% Confidence Intervals (CIs).


Preliminary analyses indicated modest levels of missing data ranging from around 5% (depression) to 13% (alcohol) across most measures, and were managed through pairwise deletion. However, there were higher levels for the PGSI, with around 45% of eligible participants (i.e., reporting gambling in the past year) having missing data across items. Exploratory analyses indicated around 90% of these patients that reported gambling on lottery or with instant win tickets only, and suggested that missing data were attributable mainly to such patients failing to define these activities as gambling. Missing data were addressed using zero-fill techniques, and thus assumed no gambling problems.

Results indicated around 1% of patients demonstrating problem gambling (PGSI 5+), and 4% exhibiting problems that were low to moderate in severity (PGSI 1-4). Thus, a total of 5.2% of patients (95% CI = 4.0% to 6.8%) exhibited at least some gambling problems across a spectrum of severity. There were 7.2% of patients reporting gambling problems among family members, and this included eight patients reporting problems with their own gambling (PGSI 1+). Levels were lower than rates of other mental health problems and addictive behaviours, including depression (55.8%, 95% CI = 52.7% to 58.9%), anxiety (27.0%, 95% CI = 24.3% to 30.0%), risky alcohol use (32.4%, 95% CI = 29.4% to 35.5%), and unhealthy drug use (14.3%, 95% CI = 12.2% to 16.7%).

Subsequent analyses explored variability in gambling problems (PGSI 1+) according to practice characteristics. They yielded trends (p < .10) suggesting elevated rates in the student health service, when compared to practices characterised by low (OR = 2.57, 95% CI = 0.99 to 7.47) and moderate deprivation (OR = 2.12, 95% CI = 0.95 to 4.73). Modest elevations were observed for highly deprived practices but were not significantly different when compared to practices characterised by low (OR = 1.91, 95% CI = 0.81 to 5.25) or moderate deprivation (OR = 1.58, 95% CI = 0.95 to 4.73). These deprived practices included the clinic for homeless patients, which were too few for statistical comparison (n = 17), but exhibited extremely high rates of gambling problems (29.4%).

Bivariate associations involving any gambling problems (PGSI 1+) and patient characteristics indicated significant associations with gender, age and relationship status. Logistic regression illustrated higher rates among: males (compared to females: OR = 2.55, 95% CI = 1.44 to 4.55), patients aged 18-24 years (compared to 35-44 year olds: OR = 2.43, 95% CI = 1.21 to 5.06), and patients who were single / never married (compared to married or cohabitating: OR = 2.35, 95% CI = 1.32 to 4.29). Patients screening positive for depression demonstrated a 2-fold increase in rates of gambling problems (OR = 2.08, 95% CI = 1.15 to 3.94), while risky alcohol use was associated with a near 3-fold increase (OR = 2.78, 95% CI = 1.60 to 4.89). Drug use was associated with a 5-fold increase in gambling problems (OR = 5.03, 95% CI = 2.78 to 8.99).

Key Findings and Expected impact on the relevant field and conclusions

The study indicates around 1 in 20 patients that report some degree of gambling problem in routine primary care, and highlights need for improved acknowledgement and capacities to respond to these issues. It suggests that GPs and clinical staff should be vigilant for gambling problems, and particularly among young males and patients who are depressed or using alcohol and drugs. At a minimum, there should be training and support for clinical staff in identification and pathways to care. However, in the absence of visible signs of gambling problems that are low to moderate in severity, it seems unlikely that such strategies (which exclude questioning in the absence of visible risk factors) will identify many individuals who would benefit from early intervention. As such, it may also be that selective screening of high risk groups (e.g., depressed and/or young males), or within particular contexts (e.g., university clinics), are potentially appropriate.

Plain English summary

Studies suggest growing numbers of people are gambling in the UK, and also experiencing problems with gambling (e.g., debt, relationship breakdown). These problems tend to cluster with other lifestyle risk behaviours (e.g., alcohol misuse, smoking), and relate to harmful impacts on individuals (e.g., suicidality) and families (e.g., domestic violence). Studies suggest that gambling problems are over-represented in primary care, which may thus have a role to play in helping people control their behaviour and prevent problems; similar to ways GPs help reduce hazards from smoking or excessive drinking. However, in contrast with these other risk behaviours, there is limited awareness of gambling in primary care, such that most problems are unrecognised and unmanaged.

The aim of this project was to improve awareness of gambling problems in general practice, and examine the need for initiatives to identify and address these issues within primary care. It involved an exploratory quantitative study of gambling problems among patients attending general practices.

Eleven practices around Bristol took part in the study, and provided services to a range of populations; including those which were deprived and vulnerable to a range of lifestyle risk behaviours including gambling. Patients were approached in waiting rooms and were asked to completed questionnaires, which provided data on gambling problems and socio-demographic and clinical variables (e.g., drinking, depression), with analyses identifying vulnerable groups and impacts on patients. There were n = 1,058 patients who returned questionnaires and took part in the study.

The results indicated 0.9% of all patients that reported severe problem gambling, and another 4.3% reporting problems that were low to moderate in severity. Around 7% of patients reported gambling problems among family members. Further analyses indicated that rates of any gambling problems were higher among males and young adults, and potentially within a student healthcare setting. Levels were also high among patients exhibiting drug use, risky alcohol use and depression.

 These findings suggest around 1 in 20 patients attending their GP for any reason that report some degree of gambling problem, and highlight need for improved awareness and responses to these issues. It suggests that GPs and other practice staff should be alert for gambling problems, and particularly among young males and patients who are depressed or using alcohol and drugs. There should be training and support for GP staff about how to recognise gambling problems, and how to assist people to access relevant help services.


The primary paper on this project was published in the British Journal of General Practice (BJGP):

Cowlishaw S, Gale L, Gregory A, McCambridge J, Kessler D. (2017). Gambling problems among patients in primary care: A cross-sectional study of general practices. British Journal of General Practice, 67 (657), e275-79.

Preliminary findings were presented at the following conference:

Cowlishaw S. Identifying gambling problems in healthcare settings: Some evidence on general practice in England. Paper presented at the 2nd Welsh National Conference on Excessive Gambling. Cardiff, Wales: June 22, 2016.

 Findings from the BJGP paper were presented at the following conference:

 Cowlishaw S. Gambling problems in primary care: A study of general practices in England. Paper presented the South West Public Health Scientific Conference 2017. Bristol, England: March 14, 2017.

 In the context of publishing the BJGP paper, communications staff at the University of Bristol (Helen Bolton) and the SPCR (Kate Farrington) were successful in generating substantial media coverage in diverse outlets including:

  • Pulse Today
  • BMJ
  • The Times (twice - 14 March and 24 March)
  • The Guardian
  • The Conversation
  • Newsweek
  • Spectator
  • BBC Radio 5 Live
  • BBC World Service
  • Heart FM
  • International Business Times
  • Falkirk Herald
  • Today Breaking News
  • Yahoo News
  • Medical Xpress

Public involvement

A Project Advisory Committee (PAC) was assembled for the project, which included the research team in addition to:

  • Graham Box (patient/public involvement representative)
  • Gail Thornton (patient/public involvement representative)
  • Jody Clarke (Bristol City Council)
  • Graham England (Bristol Addiction Recovery Agency)


The PAC met twice across the duration of the study, and provided oversight and guidance about the research procedures and materials, as well as changes from the research protocol (for example, regarding revisions to the dissemination strategy). 


Academic impact has been achieved through publication of a journal article in the BJGP, which has a JCR Impact Factor of 2.74 and is ranked number 2 (out of 20) among journals in the primary health care field. Further impact is also reflected in the high levels of media coverage of the study (see above); particularly given the primary aim was to increase awareness and recognition of gambling problems in general practices.

Knowledge translation and dissemination activities are ongoing, and include continuing attempts to engage with policy makers at both local and national levels, as well as the generation of written summaries to communicate with diverse audiences. This includes an article about the study which has been written for The Conversation, and is scheduled for publication in May, 2017.

A major barrier to future impact and progress in this area is that almost all research and interventions for gambling in the UK are funded exclusively by the gambling industry, which does not have strong interests in reducing levels of problem gambling (given that problem gamblers contribute up to 60% of expenditure on some gambling activities). As such, an immediate objective (and indicator of research impact) will be increased visibility of the implications of gambling for health and health care services, and inclusion of these issues in the agendas for health services and public health agencies. In doing so, this will provide a platform for improved recognition and responses to gambling problems in mainstream health services, including primary care.  

This project was funded by the National Institute for Health Research School for Primary Care Research (project number SPCR 244)

Department of Health Disclaimer

The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the NIHR School for Primary Care Research, NIHR, NHS or the Department of Health.