Journals Library

Development and pre-­testing a tailored interactive voice response (IVR) intervention to support adherence to anti-­hypertensive medications

Project title

Development and pre-­testing a tailored interactive voice response (IVR) intervention to support adherence to anti-­hypertensive medications

Project reference


Final report date

14 December 2017

Project start date

01 September 2016

Project end date

31 August 2017

Project duration

12 months

Project keywords

Medication adherence, hypertension, Behaviour Change Techniques, digital intervention

Lead investigator(s)
  • Dr.Katerina Kassavou, Research Associate in Behavioural Science, University of Cambridge

NIHR School Collaborators


  • Professor Stephen Sutton, Professor of Behavioural Science, University of Cambridge (advised on the development of the intervention content & process of intervention tailoring)

  • Vikki Houghton, Research Assistant University of Cambridge, (Research assistant involved in recruitment process)

  • James Brimicombe, Data Manager, University of Cambridge is a collaborator, who developed the application to facilitate the delivery of the tailored intervention content. 

  • Simon Edwards is a collaborator, who developed the IVR application to deliver the intervention.


Project objectives

To  develop  and  pre-­test  an  IVR  intervention  to  support  adherence  to  anti-­hypertensive medications. Funding would enable us to solve the technical aspects of setting up a tailored IVR  intervention,  develop  some  intervention  content  and  pre-­test  it  in  a  small  sample  of patients from the target population.

Changes to project objectives

There were no changes to the project originally outlined in the proposal

Brief summary

This project had two main objectives: to (a) develop the content and the delivery mode of the intervention, and (b) test the intervention in a pre-­post pilot study.


To  develop  the  intervention  content,  a  systematic  literature  review  with  meta-­analysis  and meta-­regression   was   conducted   to   assess   whether   IVR  interventions   are   effective   at increasing medication adherence, and what is the active intervention content that associates with the intervention effectiveness, and a review  of empirical evidence about the theoretical determinants of medication adherence behaviour change. The evidence synthesis resulted in the   (a)   development  of   theoretical   framework   of   the   intervention,   (b)   the   tailoring questionnaire, and (c) the message file with Behaviour Change Techniques (BCTs).

To develop the intervention delivery mode, consultations with experts in the industry and academia were conducted, and an application to facilitate the delivery of the intervention was developed

A  prototype  intervention  was  developed  by  multiple  and  iterative  syntheses  of  the  data obtained  during  the  intervention  development  process.  The  intervention is  highly  tailored  to deliver   very   brief   (³³1   minute),   theory-­based   messages   at   participants’   preferred   time, telephone  to  receive  the  calls  (i.e.,  primary and/or  secondary  number),  and  at  frequency  of calls  (e.g.,  repeat  options).  Patients  use  the  voice  recognition  to  provide  feedback  during inbound or outbound calls. The intervention consisted of 29 messages, delivered daily for a duration of one prescription-­based month and at participants’ preferred time. Two messages were delivered  at  the  first  day  of  the  intervention,  with  feedback  tailored  to  participants’ responses  at  the  tailoring  questionnaire,  and  one  message  per  day  during the  following  27 days of the intervention.


The pilot study had three main objectives: to (a) assess uptake, retention, fidelity, and engagement with the intervention, (b) obtain participants’ views, understanding, and actions upon the intervention content, and (c) obtain their feedback and recommendations about the intervention content, delivery mode, and implementation procedures.

Patients with hypertension were recruited by three primary care practices located at different areas of deprivation within the Cambridgeshire and Peterborough CCG. One nurse/health care assistant (HCA) per practice was recruited. Nurses/HCA were eligible if they consulted patients about taking medications.

Patients  were  eligible  if  they:  (a)  had  a  primary  diagnosis  of  hypertension;;  (b)  had  been prescribed at least one anti-­hypertensive medication as confirmed by practice records for at least three months before recruitment;; (c) had gaps in collecting repeat prescriptions or had poorly controlled blood pressure as indicated by practice records;; and (d) were aged 40 years or older. Patients were excluded when they: (a) had hearing or speaking impairment, (b) had a  diagnosis  of  dementia,  aphasia  or  other  cognitive  difficulties  that  could  affect  the  study participation,  (c)  had  a  recent  severe  life-­threatening  event,  (d)  had  difficulty  in  speaking  or understanding  English,  or  (e)  participated  in  another  study.  A  practice  GP  double  screened the  list  of  selected  patients  against  the  inclusion  and  exclusion  criteria  to  confirm  eligibility. Eligible patients were invited (a) by post or telephone calls following up the postage invitation,

(b) face-­to-­face by practice nurses or health care assistants, or (c) by leaflets and posters in the  practice  waiting  room.  Practice  nurses  received  training  by  a  research  assistant  on  the face-­to-­face recruitment procedures. Participants completed measures at baseline (T1) and at the end of the intervention (T2).

Participants  completed  the  tailoring  questionnaire  (T1,  T2),  their  preferred  primary  and secondary telephone number to receive the calls, and the time intervals of repeated calls per telephone  number  (T1),  two  items  measuring  medication  adherence  (T1,  T2),  16  items measuring their experience and satisfaction with elements of the intervention (T2), and face-­ to-­face  interviews  with  a  member  of  the  research  team  (T2).  Medication  refill  data  for  the duration  of three  months  before  T1, and  at T2, were  collected  from  the  practice  dispensary records.  Telephone  log  files  assessed  uptake,  retention,  fidelity,  and  engagement  with  the intervention.  Participants’  inbound  calls  to  the  IVR  assessed  intervention  engagement  and intervention   impact.   Face-­to-­face   interviews   (T2)   assessed   users’   feedback   about   the intervention  content  and  delivery  mode,  and  obtained  recommendations  for  improvement. Face-­to-­face   interviews   with   practice   nurses   assessed   recruitment   and   data   collection procedures, and obtained recommendations for improvement.



Uptake   and   retention.   

Eighteen   participants   provided   written   informed   consent   and registered into the study, of whom 17 completed the one-­month intervention: one participant was excluded by the research team before registration because he/she was meeting one of the  exclusion  criteria,  one  participant  withdrew  before  registration  to  the  IVR, and  the  other participant  dropped  out  during  the  intervention.  All  seventeen  participants  completed  the measures at baseline and at the end of the intervention. All primary care practices provided completed   baseline   (e.g.,   refill   prescription   data   three   months   before   the   start   of   the intervention) and follow up data for all participants.

Fidelity  of,  and  engagement  with  the  intervention  delivery  and  intervention  content.

During the one-­month intervention on average 29 calls were scheduled, of which 22.52 calls received by participants. On average 37 calls were made, of which 2.8 failed to be made due to technical issues and 13 calls made but failed to go through because they were not picked up  by  the  participants.  On  average  three  calls  were repeated  and  received  by  participants. However, 10 out of the 17 participants had chosen the calls to be repeated, if not answered, and  five  out  of  the  17 participants  had  provided  a  secondary  number  for  the  calls  to  be repeated, if there was no answered at the primary number. Overall, participants were highly engaged  with  the  intervention  and  made  inbound  calls  to  change  the  schedule  of  the messages  or  report  about  intervention  content.  There  was  a  very  good fidelity  of  the intervention BCTs, with the majority of the BCTs having high fidelity scores (Table 1).

Table 1. Fidelity and engagement, intervention content (n=17)

Behaviour Change Techniques

IVR calls scheduled

IVR calls received

Tailored to baseline questionnaire






Information about health consequences



Information about emotional consequences



Action planning, implementation intentions



Report whether or not the behaviour was performed



Social reward



Habit formation



Social support (unspecified)



Note data from log files reported as average number

Participants’ feedback about the intervention content and delivery mode.

Participants reported that the intervention content increased their awareness about medication adherence and the risk and benefits of maintaining adherent to medication taking, reminded them to take medications as prescribed, and reinforced habit formation and the sense of social support (Table 2).

Table 2. Participants’ experiential feedback about the intervention content (n=17).

Messages content and tailoring: Understanding and acceptability of the theoretical determinants

Increase awareness of the importance to take medications as prescribed

So, that message came over quite clear, you know, that you must you mustn’t miss them.  You must take them, you know.  So, I thought it was quite good.” patient 01001X

Increase awareness of the benefits to keep taking medications as prescribed, and potential risks when not taking medications

  • Q. Can you remember a specific message?

  • R. “One message said how important it was to keep taking your pills to keep you fit and healthy” patient 02028K

  • R. “one [message] talked about the benefits of taking your medication regularly” patient 02076E, follow up interview

  • R. “If I’m honest, no. I’ve had a holiday in-­between. No, I don’t – I think probably, the one about – there was one about keeping – you’re taking it to keep well, yeah, probably that one, that’s the one that sticks in my mind” patient 01051L

Remind to take medication

For me, the short message was ideal, you know?  Who am I?  and yes, and, “Have you taken your medication?”” patients 01006W

I liked the idea of just a phone call to say, Is that (name)?  “Yeah.”  “Take your tablet.”

Straightforward, you know” patient 01001X

Oh, well, just the, you know, jogging the memory Yeah, it’s, you know, the telephone rings, somebody comes out with a message, are you so and so?  Don’t forget to take your tablets and so, you take them” patient 01025G

I will say I liked it because it reminded me, like helped me to remember to take my tablet” patient 01051L

I thought it was a good system to it just reminded me to take my tablets” patient 02028K

Habit formation 

"Well, the telephone call comes through at the same time every morning and so, you know you’re going to take the pills.” Patient 01025G, follow up interview

Well, I mean, yes, I mean, the phone call alone, without any message at all, would remind you” patient 01031Q

it helped the routine, I mean, that was just that reminder of doing things ... by the end I’d developed a bit of a routine that I’m sitting there with the kind of phone ready for 8 o’clock, right, and then I would go and take my tablets” patient 01057V


Social support (unsp)

it was interesting to do it and to think someone’s taking some notice of what I’m doing, trying to get rid of my high blood pressure” patient 01024X

Note: data from  patients' face-to-face follow up interviews

Participants suggested that the intervention was acceptable and easy to use (Table 3), and they provided their views about specific features of the intervention delivery mode  and content. Specifically, the tailored schedule of the messages, the personalisation, and the variation of the content was found to be particularly appealing, and were perceived to promote engagement with the intervention.

 Table 3. Participants’ experiential feedback on the intervention delivery mode (N=17)

Table 3

It was certainly easy to use.  There’s nothing complex about it really.  I mean, if you’ve got any questions, you can just ring the numbers.  I didn’t have any questions, personally, but if you’ve got any, you can ring the number and it’s quite easy and straightforward” patient 01006W

it was always dead on time too, and if I so, it didn’t recognise it, they phoned back, sort of, about five minutes later and then it was alright” patient 01024X

Well, it was very easy, just yeah, very easy.  I had clear numbers to call if there was a problem, which I did, on a couple of occasions.  No, it was very easy …  I think having the choice of time is best… Yes, yeah, no, I think that is important, the choice of time, definitely” patient 02028K

they were easy and as long as they were at a set time, which they were, and there’s and if there’s a second mobile phone, so if the first one misses, at least you can get hold of the person on the mobile if they’re out and about” patient 01025G

Participants reported being satisfied with the overall experience with the intervention, the availability of the intervention 24/7, and their ability to call in and leave a message. They also reported that they would recommend the intervention to other people who take prescribed medications (Table 4 and 5).Note:data from face -to-face follow up interviews

Satisfaction with

Mean (sd)

the experience with the intervention

4 (0.7)

the voice delivering the intervention messages

3.7 (1.1)

the content of the intervention messages

3.8 (0.8)

being able to call the intervention

4 (0.7)

the availability of the intervention 24/7

4.2 (0.8)

the response you receive from the intervention

3.4 (0.7)

the response you received to your particular queries/questions

3.4 (0.7)

Note: data from follow up questionnaires. Likert scale, 1 strongly agree to 5 strongly disagree.


Table 5. Participants’ perception of the intervention utility and future use


Mean (sd)

the IVR messages helped me to take my tablets as prescribed

3.2 (1.3)

the IVR reminded me to take my medications as prescribed

3.5 (1.3)

the IVR system was easy to use

4.3 (0.7)

the IVR addressed my personal needs

3.3 (1.1)

it would have been useful, if the IVR was part of the regular care

3.7 (0.9)

I  would  recommend  the  IVR  to  other  people  who  take  medications  for  long-­term conditions

4.2 (0.8)

Note: data from follow up questionnaires. Likert scale, 1 strongly agree to 5 strongly disagree.


A  highly  tailored  intervention  has  been  developed  to  support  adherence  to  medications prescribed  for  high  blood  pressure  and  co-­morbidities  (e.g.,  glucose  lowering  medications). The intervention is based on rigorous theory and evidence. It aims to increase adherence to daily prescribed medications and includes a combination of Behavioural Change Techniques (BCTs) tailored to patients’ beliefs (e.g., beliefs about medications, self-­efficacy, social norms, perceived regimens complexity) and prescription plan. The intervention has been found to be acceptable and feasible, with high retention, engagement and fidelity. To our knowledge, this is  the  first  medication  adherence  intervention  delivered  using  IVR  that  has  been  developed and piloted within the UK primary care setting. Future studies could usefully test the efficacy of such intervention to support medication adherence, and other health behaviours.

Plain English summary


Interactive  voice  response  (IVR)  interventions  are  effective  ways  to  increase adherence to medications, and can potentially be cost-­effective if applied into a larger scale. This  report  briefly  describes  the  systematic  development  and  piloting  of  a  tailored  IVR intervention to increase adherence to anti-­hypertensive medications. To our knowledge this is the  first  IVR  medication  adherence  intervention  that  has  been  developed  and  pre-­tested within the primary care setting.


The  intervention  content  was  based  on  rigorous  evidence  and  theory,  and  the development process was informed by a review of the literature andPPI input. Twenty people with  hypertension,  presumed  to  be  non-­adherent,  were  recruited  by  three  primary  care practices,  of  whom  17  tested  the  one-­month  intervention.  The  intervention  consisted  of  29 very  brief  messages  (³³1  minute),  delivered  daily  to  participants  phones  (landline  and/or mobiles).  Two  messages  were  delivered  at  the  first  day  of  the  intervention,  with  feedback tailored to participant’s responses at the tailoring questionnaire, and one tailored message per day  during  the  following  27  days  of  the  intervention.  Patients  used  the  voice  recognition  to provide feedback during inbound or outbound calls. A mixed methods study generated data from baseline and follow-­up questionnaires, call log files, inbound voice messages, and face-­ to-­face follow up interviews.


On average, 29 calls were scheduled and 23 messages were received. Log file data suggested that intervention fidelity and engagement was high. Participants’ experiential feedback suggested that the tailored schedule of the calls and the variation of the content were found to be particularly appealing, and were perceived to promote engagement with the intervention. Data suggested that the intervention messages increased awareness of the importance to take medications as prescribed, and facilitated habit formation and social support.


Participants  agreed  on  the  need  for,  and  the  potential  benefits  of,  an  IVR intervention  to  support adherence  to  anti-­hypertensive  medication,  as  an  adjunct  to  primary care  consultations.  Future  research  could  usefully  test  the  feasibility  of  a  tailored  IVR intervention to support medication adherence in the primary care.


Published articles

  • Kassavou, A., & Sutton S. Reasons for non-­ adherence to cardio-­metabolic medications, and acceptability of an interactive voice response intervention in patients with hypertension and type 2 diabetes in primary care: a qualitative study. BMJ Open 2017;;7:e015597. doi:10.1136/ bmjopen-­2016-­015597
  • Tsoli,   S.,   Sutton,   S.,   &   Kassavou,   A.   Interactive   Voice   Response   interventions targeting  behaviour  change:  A  systematic  literature  review  with  meta-­analysis  and meta-­regression.  BMJ Open 2017;;0:e018974. doi:10.1136/bmjopen-­2017-­018974

Public and participant involvement 

Two PPI members, recruited by the NIHR CBR PPI, pre-­tested the IVR intervention for one month,  and  provided  their  experiential  feedback  about  the  intervention  content  and  delivery mode during four, weekly telephone-­based interviews with a researcher, and completed the baseline  and  follow  up  questionnaire.  PPI  members  also  provided  their  input,  using  think aloud protocols, on the development of the tailoring questionnaire.

PPI input, with approximately 100 people during public events (e.g., Science Festival) or using emails, informed the content of the message file and the recruitment material.

The data generated from PPI refined the intervention, before testing it with patients recruited in primary care.

The PPI involvement also increased PPI awareness about the use of digital interventions to healthcare.


The  data  generated  from  this  study  has  impacted  in  the  development  of  a  highly  tailored intervention  to  support  adherence  to  anti-­hypertensive  medications. The  intervention  will  be tested in a feasibility trial. If it is effective, it could increase the number of patients who take their medications as prescribed in the long term and achieve associated health benefits, and it could  also  reduce  costs  to  the  NHS  from  health  complications  and  premature  deaths associated with these conditions and waste in prescribing. 

Project website:

This project was funded by the National Institute for Health Research School for Primary Care Research (project number 293)

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.