ann. behav. med.

(2013) 46:358–368
DOI 10.1007/s12160-013-9515-5

ORIGINAL ARTICLE

Improving Adherence to Medication in Stroke Survivors:

A Pilot Randomised Controlled Trial
Ronan E. O’Carroll, Ph.D. & Julie A. Chambers, Ph.D. &
Martin Dennis, M.D. & Cathie Sudlow, D.Phil. &
Marie Johnston, Ph.D.

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

Published online: 14 May 2013
# The Society of Behavioral Medicine 2013

Abstract effect of increased patient contact or mere measurement.

Background Adherence to preventive medication is often (http://controlled-trials.com, number ISRCTN38274953.)
poor, and current interventions have had limited success.
Purpose This study was conducted to pilot a randomised Keywords Stroke . Adherence . Medication beliefs .
controlled trial aimed at increasing adherence to preventive Implementation intentions . Antihypertensives
medication in stroke survivors using a brief, personalised
intervention.
Methods Sixty-two stroke survivors were randomly allocated Introduction
to either a two-session intervention aimed at increasing adher-
ence via (a) introducing a plan linked to environmental cues It is estimated that, in developed countries, only 50 % of
(implementation intentions) to help establish a better patients who suffer from chronic diseases adheres to treatment
medication-taking routine (habit) and (b) eliciting and modify- recommendations [1]. Adherence is lower in chronic condi-
ing any mistaken patient beliefs regarding medication/stroke or tions than in acute conditions and drops off dramatically after
a control group. Primary outcome was adherence to antihyper- the first 6 months of treatment [2]. A recent large-scale survey
tensive medication measured objectively over 3 months using of over 31,000 patients with hypertension in Italy found that
an electronic pill bottle. only 41 % had >80 % adherence (measured by dispensed
Results Fifty-eight people used the pill bottle and were medications) 2–3 years after they were first prescribed, and
analysed as allocated; 54 completed treatment. The inter- poor adherence was associated with higher incidence of myo-
vention resulted in 10 % more doses taken on schedule cardial infarction, stroke and all-cause death [3].
(intervention, 97 %; control, 87 %; 95 % CI for difference The current Cochrane review of interventions to improve
(0.2, 16.2); p=0.048). medication adherence concluded that ‘Current methods of im-
Conclusions A simple, brief intervention increased medica- proving adherence for chronic health problems are mostly
tion adherence in stroke survivors, over and above any complex and not very effective, so that the full benefits of
treatment cannot be realized. High priority should be given to
fundamental and applied research concerning innovations to
R. E. O’Carroll (*) : J. A. Chambers assist patients to follow medication prescriptions for long-term
Division of Psychology, School of Natural Sciences, University medical disorders’ [4, p. 2].
of Stirling, Stirling FK9 4LA Scotland, UK This pilot study aimed to evaluate a brief intervention to
e-mail: [email protected]
increase medication adherence in the secondary prevention of
M. Dennis : C. Sudlow stroke. Stroke is one of the most common causes of death in the
Division of Clinical Neurosciences, Western General Hospital USA and UK and is the most common cause of severe physical
and University of Edinburgh, Edinburgh, Scotland, UK disability amongst adults. The risk of a recurrent stroke is 30–
43 % within 5 years. Guidelines for secondary prevention after
M. Johnston
Aberdeen Health Psychology Group, University of Aberdeen, ischaemic stroke now recommend antiplatelet therapy and
Aberdeen, Scotland, UK reduction of both blood pressure (BP) and cholesterol level as
ann. behav. med. (2013) 46:358–368 359

key components in reducing the risk of future vascular events and only 30–50 % of patients regularly take their antihyper-
[5]. Despite this, adherence to prescribed medication in stroke tensive drugs as prescribed [15].
patients is often suboptimal. For instance, a recent US study of
2,888 stroke patients found that 25 % had discontinued one or
more of their medicines at just 3 months post-discharge [6]. Methods
Poor adherence may be both non-intentional and inten-
tional. Non-intentional non-adherence (e.g. forgetting) is Participants
often a consequence of cognitive impairment [7]. After a
stroke, the impact of cerebrovascular disease on cognitive In accordance with our published protocol [16], participants
function, particularly memory, may mitigate against adher- were recruited from consecutive discharges from the stroke
ence, particularly if the patient is elderly and the drug clinic and stroke unit at the Western General Hospital in

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

regime is complicated [8]. Edinburgh between January 2010 and October 2011. Inclu-
Brief and easy-to-complete implementation intentions in- sion criteria were first stroke or transient ischaemic attack
terventions have been shown to be effective at reducing (TIA), discharged to home and on any preventive stroke
forgetting and improving medication adherence [9]. These medication. All participants gave informed consent for the
involve patients writing down exactly when and where they study which was conducted in accordance with the ethical
will take their medication, using the format of an if–then standards of the Helsinki Declaration of 1975 (revised 2000)
plan (‘If it is time X in place Y and I am doing Z, then I will and had ethical approval from the South East Scotland
take my pill dose’, e.g. first cup of tea at breakfast in the Research Ethics Committee (REC ref. no. 09/S1102/36).
kitchen cues taking morning medication). If–then planning The study is registered with Current Controlled Trials,
makes people highly sensitive to written environmental http://controlled-trials.com, with the unique identifier num-
cues, establishes a habit and removes the burden of having ber ISRCTN38274953.
to think about and remember when to act by reducing the
load on prospective memory as habitual responses become Measures
established. Brown and colleagues showed that such an
approach was successful in improving adherence in a Medication Events Monitoring System (MEMS® Aardex
randomised controlled trial (RCT) in patients with epilepsy Ltd., Switzerland)
(e.g. doses taken on schedule, 78.8 % in the intervention
group versus 55.3 % in the control group, p<0.01) [10]. Our primary outcome measure was electronically recorded
Intentional non-adherence occurs when a patient deliber- openings using MEMS pill bottles for 3 months. MEMS pill
ately chooses not to take their medication, against medical bottles were used in both treatment arms as MEMS mea-
advice. This often depends on the patient’s beliefs concerning surement is not immune from the Hawthorne (or mere
their condition and/or their medication. Leventhal’s self- measurement) effect [17]. Following Brown and colleagues
regulation theory posits that patients have a common-sense [10], we calculated the percentage of (a) prescribed doses
model of their illness in terms of beliefs regarding how long it taken, (b) days on which the correct dose was taken and (c)
will last, whether it is acute or chronic, and so forth [11]. doses taken on schedule, i.e. within a 3-h window of the
Patients also have beliefs about treatment, particularly the median time taken. We considered the latter as particularly
perceived necessity of medication versus concerns about any important, as the implementation intentions intervention
possible harmful effects [12]. A recent study in stroke patients aims to establish regular behaviour patterns tied in to a
found that patients’ concerns about their medication (e.g. specific time. Median rather than mean time was used to
dependence, toxicity, too many tablets) were key determinants reduce the effect of any outliers.
of poor adherence, supporting the self-regulation theory [13].
We report the results of a pilot randomised trial of a brief The Medication Adherence Report Scale (MARS) [18]
intervention based on this theoretical framework. Our aim
was to increase medication adherence via (a) developing an The MARS consists of five items relating to taking medica-
implementation intentions plan to reduce non-intentional tion, each scored from 1 to 5 and totalled to give an overall
non-adherence and (b) eliciting and modifying erroneous MARS score. The MARS is worded in order to make
beliefs about medication and stroke to reduce intentional missing medication seem a normal behaviour, with the aim
non-adherence. Although we were interested in patients’ of reducing social desirability and eliciting honest re-
experiences regarding all of their stroke medication, antihy- sponses. As the MARS was initially used to detect less than
pertensive medication was targeted for measurement, as maximum adherence on patients’ antihypertensive medi-
poor adherence to antihypertensives has been associated cine, patients were asked to answer each question for how
with significantly increased risk of stroke and death [14], they took their ‘blood pressure medicines’. The MARS has
360 ann. behav. med. (2013) 46:358–368

been used extensively to measure adherence in patients with participants gave separate written informed consent for parts
chronic diseases and has shown good reliability (internal 1 and 2. Full details of the procedures can be obtained from
and test–retest) and validity (convergent and criterion). the study protocol [16].

Patient Beliefs About Medication and Illness Part 1: Screening for Intervention

The Beliefs about Medication Questionnaire (BMQ) [19] Patients consenting to part 1 were sent the MARS, BMQ,
was used to assess cognitive representations of medication. BIPQ and perception of benefits questionnaires around
We only used the specific subscales (BMQ-specific) which 3 months after discharge. They were also asked to list all
relate to a patient’s prescribed medication, as the BMQ- current medications taken and provide information on any
general scales (medication in general) were not associated help they currently received in remembering to take their

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

with adherence in our previous study [13]. The BMQ- pills.
specific has two five-item subscales representing (a) beliefs All patients who reported less than maximum adherence
about the necessity of medication (necessity) and (b) (i.e. MARS scores <25) were considered for the brief inter-
concerns/beliefs about the risks and/or negative effects of vention (part 2). Those indicating they were not responsible
taking medication (concerns). Patients were asked to give for their own medication, not on any antihypertensive med-
their personal views on the ‘medicines prescribed for your ication or already using pharmacy-supplied Dosette boxes
stroke or TIA (mini-stroke)’. The BMQ-specific subscales were excluded. Patients eligible for part 2 were sent an
have shown good reliability and validity amongst patients invitation letter plus information sheet, consent form and
with varied illnesses including heart conditions, asthma and stamped addressed envelope. A power calculation indicated
diabetes [19]. that a sample size of n=30 in each treatment arm would
Participants were also asked to indicate their perception detect a medium effect of the intervention on MEMS adher-
of the benefits (0–100 %) provided by their stroke medica- ence [16].
tion over the next 5 years following [20] and our previous
study [13]. Finally, the Brief Illness Perception Question- Part 2: Intervention Versus Control
naire (BIPQ) [21] was used to assess patients’ views of their
illness (i.e. stroke/TIA). The nine-item BIPQ provides an Two brief sessions, 2 weeks apart, were conducted by a
easy-to-complete, psychometrically robust measure of the trained research fellow for both the intervention and control
major components of illness perceptions. groups, either in the patient’s home or at the hospital-based
Wellcome Trust Clinical Research Facility. All interviews
Frenchay Aphasia Screening Test [22] were timed, digitally audio-recorded and transcribed for a
check on treatment fidelity.
This screening test, which is frequently used with stroke All patients were screened (at session 1) using the MMSE
patients, was developed as a quick and simple method to and the Frenchay Aphasia Screening Test; there were no
identify the presence of language disturbance. It has shown exclusions on either test. Patients in both treatment arms were
good test–retest and inter-rater reliability as well as good told that we were using the MEMS containers to collect
construct and criterion validity [22]. In order to ensure that information about how patients took their medication.
patients would be able to complete all self-report measures,
those scoring <13 were excluded from the intervention. Intervention Condition Session 1 focussed on helping each
patient establish a better medication-taking routine using an
Mini-Mental State Examination (MMSE) [23] implementation intentions approach to develop an individ-
ually tailored coping plan [24]. The plan was introduced at
The MMSE is a brief, valid and reliable assessment of session 1 so that patients could try it out over the 2-week
various components of cognitive function, which has been period prior to review in session 2.
widely used in stroke research. We excluded patients scor- Following Brown et al. [10], patients were advised that it
ing <23 from the intervention, as this could indicate cogni- was a good idea to make taking their medicines part of a
tive difficulties which could affect study participation. routine and were asked to make a plan to take their tablet(s)
at the same time as something else that they did every day.
Procedures They were then asked to write down the time, place and
what they would be doing at the time they would take the
The study was conducted in two stages: In part 1, partici- first dose of their antihypertensive medicine (i.e. that which
pants completed a questionnaire, which was used to screen would be put in the MEMS pill bottle) on an individualised
for eligibility for part 2, the pilot randomised trial. All worksheet (see Fig. 1). This was repeated for all daily doses
ann. behav. med. (2013) 46:358–368 361

they believed were very likely to result in adverse side

JOHN’S PLAN
effects (resulting from negative reports regarding statins in
If it is: 7am the UK media). In these instances, the research fellow aimed
And I am: in the bathroom to increase the patients’ belief in the necessity of their
And I: am going to clean my teeth medication by informing them of the current recommenda-
Then: I will take my blood pressure tablet tions for patients who have had a TIA or stroke (e.g. ‘A
statin should be prescribed to patients who have had an
ischaemic stroke, irrespective of cholesterol level’ [26])
MARY’S PLAN and explaining that having a cholesterol level of 4.0 or lower
If it is: first thing in the morning was likely to further reduce their risk of having another TIA
And I am: in the kitchen
or stroke. The research fellow also provided information on

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

the likelihood of experiencing any side effects mentioned
And I: am making my cup of tea
(e.g. less than one in ten people report experiencing this side
Then: I will take my tablets effect).

Note: ‘John’ took one anti-hypertensive a day; ‘Mary’ was on Control Condition The use of an objective outcome mea-
multiple medications, she took them all together in the morning sure of adherence in both groups, to control for any mea-
surement effect of using the MEMS pill bottle, meant we
Fig. 1 Examples of implementation intentions plans (following were not able to have a ‘usual care’ group. Hence, control
Brown et al. [10]) group participants received the same number of visits by the
research fellow, and all measures including BP readings
were collected at the same time points as the intervention
of this medication. Patients were then asked to repeat each group. During the first two sessions, the research fellow
plan up to three times, until they felt they were able to engaged the patient in non-medication-related conversation
remember it without looking at what they had written down. (e.g. what had happened when they had their stroke) to
The research fellow also took baseline BP readings using control for non-specific effects of attention/social contact.
an OMRON M10-IT BP monitor, according to a standard
protocol. A single reading was taken from each arm; then, Both Groups At the end of session 2, the research fellow
the arm with the highest systolic value was used to collect an filled each participant’s MEMS pill bottle with 1 month’s
average of three readings taken at 120-s intervals. supply of a single antihypertensive medication. For patients
Session 2 first reviewed the effectiveness of the imple- on more than one antihypertensive, the medication chosen
mentation intentions plan and any barriers/difficulties in was that taken most frequently (i.e. twice rather than once a
following it, and any required changes were developed day) or, if no difference, that which would most convenient-
collaboratively, following the methods outlined by Sniehotta ly fit into the MEMS pill bottle. Patients were instructed to
[24]. This approach helped ensure that the patient had de- only take out one dose of their medication whenever they
veloped a suitable implementation intentions plan before opened the pill bottle.
adherence measurement commenced. For each of the next 2 months, the research fellow made
The main focus of session 2 was to elicit and, if appro- another brief visit to refill the MEMS pill bottle and also take an
priate, challenge any mistaken beliefs about a patient’s electronic reading from the MEMS cap, downloading the data
illness (e.g. causes/effects of their stroke) and/or medication onto a laptop PC for later analysis. The participant also com-
(e.g. beliefs regarding toxicity, dependence, etc.), using re- pleted the BIPQ, BMQ and perceptions of medication benefits
sponses on the BMQ and BIPQ as a basis. This approach measures at the first of these visits. At 3 months, the research
was based on the model of Petrie et al. [25] who elicited and fellow made a final visit to take a last MEMS cap reading,
modified patients’ dysfunctional beliefs regarding their re- collect the MEMS pill bottle and final outcome measures and
cent myocardial infarction, resulting in faster return to work take the patient’s BP.
and lower angina symptoms at 3 months. Our aim was to
correct any misperceptions and provide evidence so that Randomisation
participants’ medication necessity beliefs regarding their
stroke medication came to outweigh their medication con- Patients were randomised by the web-based Edinburgh
cerns beliefs. As an example, many patients did not under- Clinical Trials Unit software to either the intervention or
stand why, as their cholesterol level was in the ‘normal’ control arm using a minimisation algorithm, together
range (i.e. 5.0 or less), they needed to take statins, which with a random element giving a one in ten chance of
362 ann. behav. med. (2013) 46:358–368

allocation to the opposite treatment from that deter- considered ‘missing at random’ [27] (i.e. 3 hospitalised for
mined by the algorithm. Based on our previous study, non-stroke reasons and 1 relocated). A further person chose
the minimisation variables (chosen to ensure that the to terminate a month early (for travel/time reasons) but
treatment arms did not differ on factors that might affect completed all outcome measures at session 4 (defined
adherence) were age, MARS scores and complexity of completer).
the medication regime [13]. There were no differences with regard to patterns of
As this was a pilot study, the recruitment, intervention, data missing data and treatment group (χ2(2)=0.5, p=0.766)
collection and analysis were all carried out by the same research nor with regard to any of the pretreatment outcome mea-
fellow who was not blind to the treatment allocation. Patient sures, gender, Scottish Index of Multiple Deprivation
contact time was controlled between treatment arms, and pa- (SIMD) scores, MMSE scores or overall MEMS scores
tients themselves were not informed which arm they were in. (data available from the authors). Dropouts had lower

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

pretreatment MARS scores, but there were no differences
Training of the Research Fellow by treatment group.

The research fellow was trained in the intervention by the Analysis

principal investigator, who had been previously trained in
the intervention used by Petrie et al. [25] in eliciting and Missing data were addressed by multiple imputation,
addressing mistaken beliefs. Both the intervention and con- which is currently the preferred method of imputing
trol group sessions (1 and 2) were piloted via role-play with missing data. Multiple imputation has been shown to
the principle investigator acting as a pseudo-patient. These perform well with both longitudinal data and small
sessions were video-recorded, and feedback was given to samples [28]. We used Imputation of Chained Equations
the research fellow, after which minor modifications were in the STATA software package to impute five datasets
made to the procedure. which were then analysed using SPSS version 19. T
tests and χ2 were used to test basic differences between
Fidelity of Intervention Check treatment groups, and repeated-measures analysis of var-
iance was used to compare changes in outcome mea-
The principle investigator checked the transcriptions of ses- sures over time. Where required, test statistics were
sions 1 and 2 for the first 11 participants in the intervention pooled using Rubin’s rules [29].
group (i.e. 38 % of all interviews) for adherence to the study
protocol on an ongoing basis. Minor suggestions and mod-
ifications were made to subsequent sessions, but there were Results
no violation issues. The fidelity check was terminated after
the 11th interview, as no issues arose from the last four Part 1 (Screening for Intervention)
interviews checked (i.e. 8–11).
Overall, 494 people completed consent forms for part 1, 35
Statistical Analysis of these opted not to take part and 52 proved ineligible (50
previous stroke/TIA and 2 no stroke diagnosis). Question-
Four people randomised to treatment did not use the MEMS naires were, therefore, sent to 407 patients at a mean of
pill bottle (three became ineligible between randomisation 132 days (SD, 76.5) after their stroke/TIA; 355 question-
and the intervention and one declined to use the MEMS pill naires (87.2 %) were returned.
bottle at session 2) (Fig. 2). An a priori decision had been Two hundred seventy people were excluded from part 2
made to exclude all participants with no data on the primary for reasons including maximum MARS adherence (53.5 %),
outcome measure (MEMS) from the analysis, in accordance not taking antihypertensives (20.6 %), receiving help taking
with our published protocol [16]. The remaining participants their medication (11.0 %) and using a pharmacy-supplied
were analysed as allocated to the treatment arm (n=29 in Dosette box (23.7 %) (Fig. 2).
each group).
Part 2 (Intervention Versus Control)
Missing Data
Part 2: Participation
Four (6.9 %) of the 58 people included in the analysis
discontinued participation in the study, all for reasons Eighty-five people were invited to take part in the interven-
unrelated to the study aim, meaning that the data can be tion; 62 consented and were randomised to either the
ann. behav. med. (2013) 46:358–368 363

Fig. 2 CONSORT diagram

Part 1: Completed consent forms and
Assessment assessed for eligibility for Part 1
(n=494) 33 declined to take part
for Intervention 2 incomplete forms
52 were not eligible for Part 1:
50 had had a previous stroke/TIA
2 had not had a stroke/TIA

Eligible for Part 1 and sent

baseline assessment
questionnaires (n=407) 19 declined to complete questionnaires
24 failed to return questionnaires
6 replied that hadn’t had stroke/TIA
3 ineligible for Part 2 (at phone call)
Part 1 assessment measures

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

returned and assessed for 270 excluded from Part 2 (main reason):
eligibility for Part 2 (n=355) 190 MARS=25
54 not on anti-hypertensives
10 had help taking medication
3 reported using Dosette box
13 other (7 missing data, 2 own box, 4 too late)
Part 2: Invited to take part in Part 2
Intervention intervention (n=85)
Not randomised:
21 declined to take part in Intervention
2 failed to respond to invitation

Randomised (n=62)

Allocated to brief intervention (n=31) Allocated to control group(n=31)

1st assessment (n=31) 1st assessment (n=31)

Received allocated intervention (n=30) Received allocated intervention (n=29)
Did not receive intervention (n=1): Did not receive intervention (n=2):
now ineligible: using Dosette box now ineligible: not taking anti-hypertensive

Lost to follow-up (n=2): Lost to follow-up (n=2)

2 hospitalised (non-stroke) 1 hospitalised (non-stroke); 1 relocated

Analysed (n=29) Analysed (n=29)

Excluded from analysis (n=2): Excluded from analysis (n=2):
1 did not receive intervention (ineligible), 2 did not receive intervention (ineligible)
1 declined to use MEMS

intervention or the control group (Fig. 2). Three people were Part 2 participants reported taking a mean of 5.5 (SD, 2.3;
excluded after randomisation and before starting the inter- range, 2–15) different regular oral medications, representing
vention: one (intervention) had started using a Dosette box a mean of 6.8 (SD, 4.1; range, 2–24) tablets each day, and a
before the research fellow’s first visit and two (both control) mean of 1.7 (SD, 1.0; range, 1–4) antihypertensive medica-
were no longer taking antihypertensive medication; there- tions per day.
fore, 59 people started the intervention. Baseline characteristics were similar in the intervention
There were no major differences with respect to age, and control groups, suggesting that the randomisation pro-
gender or any of the pretreatment beliefs measures between cedure was effective (Table 1). Total contact time was not
participants (n = 59) and non-participants (n = 296) (data significantly different between the intervention (mean,
available from the authors). However, part 2 participants 224 min (SD, 50)) and control (mean, 197 min (SD, 45))
(mean SIMD score=10.5 (SD, 9.6)) were from areas of groups (95 % CI for difference (−1, 55); p=0.056)). The
lower deprivation than non-participants (mean=14.3 (SD, slightly longer contact time in the intervention group largely
13.3), 95 % CI for difference (0.9, 6.7); p=0.011). occurred across sessions 1 and 2, when the intervention was
364 ann. behav. med. (2013) 46:358–368

Table 1 Demographics and baseline outcome measures by intervention group (n=58)

Intervention Control

Age 68.4 (11.3) 70.7 (10.5)

Range, 51–85 Range, 51–85
Gender (% male) 69 % 59 %
Ethnicity (% White British) 100 % 100 %
SIMD score (high=higher deprivation) 9.3 (10.8) 11.8 (8.4)
Work statusa 58 % retired 60 % retired
23 % working 30 % working
FAST total score 19.0 (1.1) 18.7 (1.1)

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

MMSE 28.6 (1.3) 28.3 (1.3)
Type of stroke (% ischaemic) 100 % 100 %
Stroke severityb
TIA 31.0 % 48.3 %
Minor/small stroke 20.7 % 10.3 %
Stroke 48.3 % 41.4 %
Days from stroke to randomisation 191 (84) 162 (77)
Total no. of different medications 4.8 (1.8) 6.0 (2.6)
No tablets per day×frequency 15.1 (9.6) 19.2 (15.7)
Total number of BP medications 1.5 (0.7) 1.9 (1.1)
Total contact time (min) 224 (50) 197 (45)
BIPQ timeline 3.7 (3.8) 3.1 (3.4)
BIPQ treatment control 7.5 (2.3) 7.7 (2.3)
BMQ-specific—necessity subscale 17.4 (3.1) 18.1 (3.0)
BMQ-specific—concerns subscale 13.2 (3.2) 12.9 (3.4)
BMQ-necessity minus concerns 4.2 (4.5) 5.3 (4.6)
Perception of benefits of medication (%) 81.9 (19.0) 79.1 (21.2)
MARS unintentional non-adherence (item 1) 3.8 (0.4) 3.8 (0.5)
MARS intentional non-adherence (items 2–5) 19.4 (1.3) 19.3 (1.4)
MARS total score 23.3 (1.2) 23.1 (1.6)

SIMD Scottish Index of Multiple Deprivation, with lower scores equated to lower deprivation, FAST Frenchay Aphasia Screening Test, MMSE
Mini-Mental State Examination, BIPQ Brief Illness Perception Questionnaire: timeline and treatment control subscales are reported as they have
shown associations with adherence in previous research [13], BMQ Beliefs about Medication Questionnaire, MARS Medication Adherence Report
Scale
a
Those not working or retired were housewives, unemployed or on incapacity benefit
b
As recorded by the consultant at the stroke unit in the letter to the patient’s general practitioner

delivered; we, therefore, estimate that the active part of the of total doses taken: mean difference, 5.1 %; 95 % CI (−1.6,
brief intervention lasted, on average, <30 min. 9.0); percentage of days correct dose taken: mean differ-
ence, 5.4 %; 95 % CI (−1.8, 9.4)). There were no effects of
Part 2: Primary Outcome Measure (MEMS) time or the group×time interaction in any of the MEMS
analyses.
MEMS data were recorded for a period of 3 months (mean
number of days=82.2 (SD, 17.4); range, 16–90). The inter- Part 2: Secondary Outcome Measures
vention group had higher adherence on all three MEMS
outcome measures than the control group (Table 2), al- MARS Scores Table 3 shows the changes in scores in self-
though this was only significant for doses taken on schedule reported adherence (MARS) from pre-intervention to
(Fig. 3) (i.e. percentage of doses taken on schedule: mean follow-up. There were significant time and interaction ef-
difference, 9.8 %; 95 % CI (0.2, 16.2); p=0.048; percentage fects of total MARS scores, with both groups reporting
ann. behav. med. (2013) 46:358–368 365

Table 2 Mean (SD) MEMS readings by time and treatment group (n=58)

Month 1 Month 2 Month 3 Pooled F

Group×time Group Time

% Doses taken Intervention 98.8 (2.1) 99.0 (2.2) 99.0 (2.1) 1.6, p=0.267 2.0, p=0.163 0.6, p=0.572
Control 94.1 (15.5) 92.8 (16.5) 94.7 (14.2)
% Days correct dose taken Intervention 98.6 (2.2) 99.0 (2.2) 98.7 (2.0) 1.6, p=0.229 1.9, p=0.178 0.5, p=0.662
Control 93.7 (15.4) 92.6 (16.5) 94.0 (14.4)
% Doses taken on schedule Intervention 95.9 (5.8) 97.4 (3.1) 96.8 (3.6) 0.9, p=0.456 4.3, p=0.048 0.6, p=0.555
Control 86.9 (22.2) 86.6 (22.5) 87.4 (23.3)

df for F are adjusted by inter-imputation variance to give pooled p values and, therefore, vary for each statistic

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

higher adherence at follow-up, but a significantly greater follow-up in the intervention versus the control group (mean
improvement in the intervention group (mean difference, difference, 1.3; 95 % CI (0.1, 2.5); p=0.047). There were no
0.61; 95 % CI (0.1, 1.2); p=0.027). significant effects of changes in beliefs about illness (BIPQ)
Self-reported adherence (i.e. MARS total scores) at or perceived benefit of medication as a result of the inter-
follow-up was highly correlated with all objective vention (Table 4).
MEMS measures (all p<0.001), i.e. percentage of doses
taken (r=0.71), percentage of days correct dose taken
(r =0.70) and percentage of doses taken on schedule Discussion
(r=0.66).
This simple, brief, two-session intervention resulted in a
Blood Pressure Both groups showed a reduction in BP 10 % increase in doses of antihypertensive medication taken
readings, with a significant effect of time on both systolic on schedule. This level of increase is likely to be clinically
and diastolic measures (Table 3), but there were no differ- important, as a large (n=47,479) retrospective cohort study,
ences between groups. using medical and pharmacy records to assess health out-
comes and adherence over 1–5 years, concluded that ‘in-
Beliefs About Illness and Medication Table 4 shows the pre, creasing adherence by one (anti-hypertensive) pill per week
post and follow-up scores on the BMQ-specific subscales. for a once-a-day regimen reduces the hazard of stroke by
There was a significant effect of time with BMQ-necessity 8–9 % and death by 7 %’ [14]. Furthermore, the main effect
minus concerns increasing and BMQ-concerns decreasing was observed on doses taken on schedule, precisely what
from pretreatment to follow-up in both groups. BMQ- the implementation intentions intervention aimed to change.
concerns also showed a significantly greater decrease by We also achieved our aim of a greater reduction in concerns

Fig. 3 Percentage of doses Intervention Control

taken on schedule by treatment 100
group for months 1–3
Percentage of doses taken on schedule

95

90

85

80
97.4 96.8
95.9

75
86.9 86.6 87.4

70

65

60
Month 1 Month 2 Month 3
Period of MEMS pill-bottle usage
366 ann. behav. med. (2013) 46:358–368

Table 3 Mean (SD) MARS adherence scores and BP readings by time and treatment group (n=58)

Pre Follow-up Pooled F

Group×time Group Time

MARS total Intervention 23.3 (1.2) 24.2 (0.5) 4.9, p=0.027 2.0, p=0.154 18.8, p<0.001
Control 23.1 (1.6) 23.6 (1.3)
BP systolic (mmHg) Intervention 133 (18) 131 (17) 3.5, p=0.110 1.2, p=0.319 8.5, p=0.018
Control 141 (22) 132 (20)
BP diastolic (mmHg) Intervention 81 (11) 80 (11) 2.7, p=0.112 0.1, p=0.710 7.0, p=0.049
Control 83 (13) 79 (13)

df for F for MARS scores are adjusted by inter-imputation variance to give pooled p values and, therefore, vary for each statistic; BP was measured

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

as an average of three readings taken from the arm which initially showed the highest single reading at the first visit. The same arm was used for the
average readings at both pre and follow-up

about medication in the intervention versus the control studies of adherence in stroke patients [6]. Patients’ adherence
group. did not appear to become worse over the 3 months of MEMS
Previous interventions to improve adherence have had pill bottle usage nor were there any differences in adherence
mixed results [4, 30], although adherence in clinical trials is over time between groups, contrary to our expectation of a
often higher than expected [31]. Implementation intentions Hawthorne (or mere measurement) effect [17]. It is important
interventions in other health conditions have been success- to emphasise that we did not compare our brief intervention
ful in improving adherence to medication in some (e.g. with treatment as usual; rather, our control condition involved
antiepileptic drugs [10]), but not all (e.g. antibiotics [31]) significant additional contact and electronic recording of pill-
studies. A recent RCT aimed at reducing concerns about taking for 3 months. It is highly likely that this resulted in
medication in 136 non-compliant hypertensive patients in improved adherence in the control group, e.g. when asked to
Jordan was strongly associated with higher adherence and rate the MEMS pill bottle during the follow-up interview,
reductions in BP [32]; however, this study did not control 46 % of control participants reported that they found it helpful
for patient contact time. In contrast, a controlled UK-based regarding adherence. This may also account for the fact that,
nurse-led support intervention which encouraged hyperten- although the intervention group were significantly more reg-
sive patients to discuss their medication concerns showed no ular in their pill-taking than the control group, the differences
effects on either adherence or BP [33]. in pills taken (i.e. 5 % more in the intervention group) was
The current study, using electronic pill monitoring, found non-significant. Although, due to cost, we were only able to
high levels of adherence overall, in contrast to some other use the MEMS pill bottle for one medication (a single

Table 4 Mean (SD) scores for medication and illness beliefs by time and treatment group (n=58)

Pre Post Follow-up Pooled F

Group×time Group Time

BMQ-specific necessity Intervention 17.4 (3.1) 18.1 (2.6) 18.5 (2.5) 0.8, p=0.460 0.1, p=0.743 2.7, p=0.066
Control 18.1 (3.0) 17.9 (3.0) 18.5 (2.7)
BMQ-specific concerns Intervention 13.2 (3.2) 11.9 (3.7) 11.0 (2.5) 3.1, p=0.047 0.01, p=0.924 9.2, p<0.001
Control 12.9 (3.4) 11.7 (3.7) 11.9 (3.0)
BMQ-necessity minus concerns Intervention 4.2 (4.5) 6.1 (4.9) 7.4 (3.4) 2.5, p=0.081 0.02, p=0.879 10.8, p<0.001
Control 5.3 (4.6) 6.2 (4.9) 6.6 (4.3)
BIPQ-timeline Intervention 3.7 (3.8) 4.4 (4.0) 4.1 (3.7) 0.9, p=0.418 2.3, p=0.132 0.1, p=0.952
Control 3.1 (3.4) 2.6 (3.2) 3.0 (2.9)
BIPQ-treatment control Intervention 7.5 (2.3) 8.2 (2.0) 8.7 (2.2) 1.8, p=0.160 0.7, p=0.417 1.4, p=0.256
Control 7.7 (2.3) 8.0 (2.3) 7.7 (1.9)
% Perceived benefit Intervention 81.9 (19.0) 81.1 (15.9) 84.8 (15.7) 1.2, p=0.328 0.3, p=0.622 0.3, p=0.758
Control 79.1 (21.2) 82.5 (19.1) 81.4 (17.4)

df for F are adjusted by inter-imputation variance to give pooled p values and, therefore, vary by each statistic. Timeline and treatment control
subscales are reported as they have shown associations with adherence in previous research [13]
BMQ Beliefs about Medication Questionnaire, BIPQ Brief Illness Perception Questionnaire
ann. behav. med. (2013) 46:358–368 367

antihypertensive), all patients took a number of medications to increases in self-reported adherence from pretreatment to
(range, 2–15), often at the same time. From the detailed follow-up, suggesting that our intervention may be effective
conversations the research fellow had with patients through- across social domains. We also acknowledge that other
out the study, we have no reason to believe that using the factors that were not measured in the current study, such
MEMS pill bottle for only one antihypertensive negatively as depression and social support, may also contribute to
impacted on other medication-taking nor that it resulted in non-adherence of medication.
patients being any more or less adherent to their remaining
medicines.
The intervention did not significantly increase patients’ Conclusions
beliefs in the necessity of their medication; rather, the effect
was a reduction in concerns. However, as many of the The current study has shown that a combined brief interven-

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

patients had been on multiple preventive medications such tion which addresses patients’ erroneous beliefs about medi-
as antihypertensives and statins for some years, many al- cation and stroke to reduce intentional non-adherence, in
ready viewed their treatment as ‘necessary’. conjunction with introducing an implementation intentions
Baseline data were not recorded for the MEMS and so the plan to reduce forgetting, can improve adherence to preven-
apparent differences between groups from month 1 may tive medication by 10 % in an older population of stroke
well reflect an immediate effect of the intervention on ad- patients, over and above any effects of measurement or high
herence. The significantly greater improvement in self- therapeutic contact. The effect found was equivalent to taking
report adherence (MARS) in the intervention group from one additional dose in ten within ±3 h of a regular time.
pretreatment to follow-up supports this observation. Using plans to establish routines may help older adults
Although the greater increase in MARS total scores from adhere to time-regulated tasks by making them automatic
pretreatment to follow-up in the intervention group did not habits, whilst addressing patients’ concerns may result in less
translate into greater reductions in BP, both groups did show reluctance to take medication. It has also been suggested that
a significant reduction over time, which may be due to the negative views about treatment may underlie unintentional, as
relatively high adherence observed in both groups. well as intentional, reasons for not taking medication [34];
thus, it is plausible that the observed reduction in medication
Limitations concerns led to decreases in both forgetting and choosing not
to take medication in the current study.
A limitation of the current study was that the same research This pilot was conducted in stroke survivors; however, it
fellow delivered the intervention and conducted the analysis may well be generalizable to other patients on multiple med-
and so was not blind to the treatment arm of the patients. ications. We estimate that the active part of the intervention
However, the main outcome measure was electronically took no more than 30 min, spread over sessions 1 and 2.
recorded MEMS readings, and the remaining (self-report) Although our approach may be more resource-intensive than
outcome measures were posted out to and completed by simpler interventions, such as telephone reminders or mobile
patients in advance of the meetings with the research fellow, phone alarms, our findings suggest that it is important to
so we do not believe that this would have greatly affected address not just non-intentional adherence but also intentional
the results. As we did not collect baseline MEMS readings, adherence by eliciting and addressing patients’ underlying
it is possible that the differences found may have been beliefs which are likely to affect their medication-taking.
observed at pretreatment and not as a result of the interven- Hence, there remains a need for personalised interventions
tion. However, there were no differences between groups in to increase adherence. We believe that this simple, relatively
baseline self-reported adherence (MARS) and we did ob- brief intervention could address this need, particularly if de-
serve significantly higher increases in MARS scores in the livered in a healthcare setting at the time when medication for
intervention group, suggesting that this was not the case. a chronic condition was first prescribed.
We planned to recruit patients 3 months after their stroke, We observed small to medium (0.17 to 0.59) effects on all
to allow time for the establishment of medication routines; the main outcome measures. Using G*Power [35] to estimate
however, many patients were prescribed antihypertensives the sample sizes required to detect these effects in a larger-scale
before their stroke and so the intervention may be more study, sample sizes varied from n=102 to n=240. Allowing for
effective if delivered earlier. All participants were White 20 % attrition, this would mean a total sample size of 288
British, had had an ischaemic stroke and tended towards patients would be large enough to detect the effects found in the
higher socio-economic status; thus, the intervention war- current study. The next step, therefore, is to confirm this effect
rants evaluation with more diverse populations. However, in an adequately powered RCT to demonstrate that the inter-
although those from more deprived areas reported lower vention can be successfully delivered by trained health pro-
adherence overall, socio-demographic status was not related fessionals (e.g. nurses) in different healthcare settings.
368 ann. behav. med. (2013) 46:358–368

Acknowledgments We would like to thank the doctors and nurses at 15. Stephenson J. Noncompliance may cause half of antihypertensive
the Western General Hospital stroke clinic and ward for their help in drug “failures”. JAMA. 1999;282:313-314.
patient recruitment and the patients for giving us their time to take part. 16. O’Carroll RE, Dennis M, Johnston M, Sudlow C. Improving Adher-
ence to Medication in Stroke Survivors (IAMSS): A randomised
Conflict of Interest The authors have no conflict of interest to controlled trial: Study protocol. BMC Neurol. 2010;10:15.
disclose. 17. Wetzels GE, Nelemans PJ, Schouten JS, van Wijk BL, Prins MH.
All that glisters is not gold: A comparison of electronic monitoring
versus filled prescriptions-an observational study. BMC Health
Sources of Funding This project was funded by a grant from the Serv Res. 2006;6:8.
Scottish Government, Department of Health, Chief Scientist Office; 18. Horne R. Measuring adherence: The case for self-report. Int J
reference number CZH/4/569. Behav Med. 2004;11(suppl):75.
19. Horne R, Weinman J, Hankins M. The Beliefs about Medicines
Questionnaire: The development and evaluation of a new method
for assessing the cognitive representation of medication. Psychol

Downloaded from https://academic.oup.com/abm/article-abstract/46/3/358/4564076 by guest on 16 October 2019

References Health. 1999;14:1-24.
20. Trewby PN, Reddy AV, Trewby CS, Ashton VJ, Brennan G, Inglis J.
Are preventive drugs preventive enough? A study of patients’ expec-
1. World Health Organisation. Adherence to long-term therapies: Evi- tation of benefit from preventive drugs. Clin Med. 2002;2:527-533.
dence for action. Geneva: WHO; 2003. 21. Broadbent E, Petrie KJ, Weinman J. The Brief Illness Perception
2. Osterberg L, Blaschke T. Adherence to medication. New Eng J Questionnaire (BIPQ): Validity and reliability. Int J Behav Med.
Med. 2005;353:487-497. 2004;11:278.
3. Esposti LD, Saragoni S, Benemei S, et al. Adherence to antihy- 22. Enderby PM, Wood VA, Wade DT, Hewer RL. The Frenchay
pertensive medications and health outcomes among newly-treated Aphasia Screening Test: A short, simple test for aphasia appropri-
hypertensive patients. Clinicoecon Outcomes Res. 2011;3:47-54. ate for non-specialists. Int Rehabil Med. 1987;8:166-170.
4. Haynes RB, Ackloo E, Sahota N, McDonald HP, Yao X. Interven- 23. Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A
tions for enhancing medication adherence. Cochrane Database practical method for grading the cognitive state of patients for
Syst Rev. 2008;(2):CD000011. the clinician. J Psychiatr Res. 1975;12:189-198.
5. Scottish Intercollegiate Guidelines Network. Management of pa- 24. Sniehotta FF, Scholz U, Schwarzer R. Action plans and coping
tients with stroke or TIA: Assessment, investigation, immediate plans for physical exercise: A longitudinal intervention study in
management and secondary prevention—A national guideline. cardiac rehabilitation. Brit J Health Psychol. 2006;11:23-37.
Guideline 108. Edinburgh: SIGN; 2008. 25. Petrie KJ, Cameron LD, Ellis CJ, Buick D, Weinman J. Changing
6. Bushnell CD, Zimmer LO, Pan W, et al. Persistence with stroke illness perceptions after myocardial infarction: An early intervention
prevention medications 3 months after hospitalization. Arch randomized controlled trial. Psychosom Med. 2002;64:580-586.
Neurol. 2010;67:1456-1463. 26. SIGN: Management of patients with stroke or TIA: assessment,
7. Hayes TL, Larimer N, Adami A, Kaye JA. Medication adherence investigation, immediate management and secondary prevention. A
in healthy elders: Small cognitive changes make a big difference. J national clinical guideline. Edinburgh: Scottish Intercollegiate
Aging Health. 2009;21:567-580. Guidelines Network; 2008.
8. Fischer B, Lehrl S, Weber E, Gundert-Remy U, Fischer U. Cere- 27. Lane P. Handling drop-out in longitudinal clinical trials: A compar-
brovascular insufficiency and compliance with drug therapy. Z ison of LOCF and MMRM approaches. Pharm Stat. 2008;7:93-106.
Gerontol. 1981;14:145-152. 28. Graham J. Missing data analysis: Making it work in the real world.
9. Webb TL, Sheeran P. Does changing behavioral intentions engen- Ann Rev Psychol. 2009;60:549-576.
der behavior change? A meta-analysis of the experimental evi- 29. Rubin DB. Multiple imputation for non-response in surveys. New
dence. Psychol Bull. 2006;132:249-268. York: Wiley; 1987.
10. Brown I, Sheeran P, Reuber M. Randomized controlled trial of 30. Hughes CM. Medication non-adherence in the elderly. How big is
an implementation intention intervention to enhance adherence the problem? Drugs Aging. 2004;21:793-811.
with antiepileptic drug treatment. Epilepsy Behav. 2009;16:634- 31. Jackson C, Lawton RJ, Raynor DK, et al. Promoting adherence to
639. antibiotics: A test of implementation intentions. Patient Educ
11. Leventhal H, Diefenbach M, Leventhal EA. Illness cognition: Couns. 2006;61:212-218.
Using common sense to understand treatment adherence and affect 32. Alhalaqia F, Nawafleh AH, Clark A, et al. Adherence therapy for
cognition interactions. Cognitive Ther Res. 1992;16:143-163. medication non-compliant patients with hypertension: A
12. Clifford S, Barber N, Horne R. Understanding different beliefs randomised controlled trial. J Hum Hypertens. 2012;26:117-126.
held by adherers, unintentional nonadherers, and intentional 33. Ellis G, Rodger J, McAlpine C, Langhorne P. The risk of stroke
nonadherers: Application of the Necessity-Concerns Framework. nurse specialist input on risk factor modification: A randomised
J Psychosom Res. 2008;64:41-46. controlled trial. Age Aging. 2005;34:389-392.
13. O’Carroll RE, Whittaker J, Hamilton B, Johnston M, Sudlow 34. Gadkari AS, McHorney CA. Unintentional non-adherence to
C, Dennis M. Predictors of adherence to secondary prevention chronic prescription medications: How unintentional is it really?
medication in stroke patients. Ann Behav Med. 2011;41:383- BMC Health Serv Res. 2012;12:98.
390. 35. Faul F, Erdfelder E, Lang A-G, Buchner A. G*Power 3: A flexible
14. Bailey JE, Wan JY, Tang J, Ghani MA, Cushman WC. J Gen statistical power analysis program for the social, behavioral and
Intern Med. 2010;25:495-503. biomedical sciences. Behav Res Methods. 2007;39:175-191.