© The Author (2005). Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oupjournals.

org
doi:10.1093/fampra/cmh717 Family Practice Advance Access originally published online on 17 January 2005

Nurse-led adherence support in hypertension:

a randomized controlled trial
Knut Schroedera, Tom Faheyb, Sandra Hollinghursta and
Tim J Petersa

Schroeder K, Fahey T, Hollinghurst S and Peters TJ. Nurse-led adherence support in

hypertension: a randomized controlled trial. Family Practice 2005; 22: 144–151.
Background. Lack of medication adherence is a common reason for poor control of blood
pressure in the community, increasing the risk of heart attacks and strokes.
Objective. To evaluate the effect of nurse-led adherence support for people with uncontrolled
high blood pressure compared with usual care.
Methods. We recruited 245 women and men with uncontrolled hypertension (150/90 mmHg)
from 21 general practices in Bristol, UK. Participants were randomized to receive nurse-led
adherence support or usual care alone. Main outcome measures were adherence to medication
(‘timing compliance’) and blood pressure.
Results. Mean baseline timing compliance (± SD) was high in both the intervention
(90.8 ± 15.6%) and the control group (94.5 ± 7.6%). There was no evidence of an effect of the
intervention on timing compliance at follow-up (adjusted difference in means 1.0%; 95%
confidence interval (CI) 5.1 to 3.1). There was also no difference at follow-up between the
groups with regard to systolic blood pressure (2.7 mmHg; 95% CI 7.2 to 1.8) or diastolic
blood pressure (0.2 mmHg; 95% CI 1.9 to 2.3). Projected costs for the primary care sector per
consultation were £6.60 for the intervention compared with £5.08 for usual care.
Conclusion. In this study, adherence to blood pressure medication was much higher than
previously reported. There was no evidence of an effect of nurse-led adherence support on
medication adherence or blood pressure compared with usual care. Nurse-led adherence
support was also more expensive from a primary care perspective.
Keywords. Clinical trial, hypertension, patient compliance, nursing care.

Introduction patients with hypertension.8 Motivational and complex

interventions demonstrated promising results; in some
Hypertension is a major risk factor for cardiovascular studies the interventions were delivered by nurses. The
disease.1 Although randomized trials (RCTs) have shown results of this review have to be interpreted with
that treating high blood pressure reduces the risk of heart caution, though, because the studies involved various
attacks and strokes,2 control of high blood pressure in the methodologies and were often of poor quality.
community is far from optimal,3 with lack of adherence Furthermore, it remains uncertain how applicable these
to medication being a common reason.4,5,6 Medication findings are to UK primary care, as the overwhelming
adherence in hypertension is estimated to be only around majority of trials were conducted in North America.
50% to 70%,5 and the World Health Organization has We therefore developed a nurse-led support inter-
recognized the importance of improving adherence.7 vention aimed at increasing adherence to medication and
A systematic review of RCTs investigated the effect- reduce blood pressure in uncontrolled hypertensive people,
iveness of interventions aimed at increasing adherence in evaluating its effectiveness and cost-effectiveness in a
pragmatic, primary care-based RCT.
Received 10 June 2004; Accepted 27 September 2004.
aPrimary Health Care, Department of Community Based

Medicine, University of Bristol and bTayside Centre for

General Practice, University of Dundee, UK. Correspondence Methods
to Dr Knut Schroeder, Primary Health Care, Department of
Community Based Medicine, University of Bristol, Cotham Setting and practice recruitment
House, Cotham Hill, Bristol BS6 6JL, UK; Email: We wrote to a random sample of 82 of the 171 general
[email protected] practices in Avon, UK. Of the 45 practices that replied,

144
 Nurse-led adherence support in hypertension 145
24 declined, giving time constraints and staff shortages as self-regulatory model of illness behaviour.12,13 This takes
the main reasons; 21 general practices in rural and urban into account the individual’s perception of symptoms,
settings took part in the study. Fourteen were teaching emotional responses to a health threat and coping
practices and 11 belonged to a Culyer-funded research strategies such as avoidance. The intervention was
consortium. All ran nurse-led clinics for hypertensive designed so that it could easily be introduced into
patients. routine practice. The only adherence-related training
provided for the practice nurses was an explanation of
Recruitment and eligibility criteria issues around medication adherence by KS, lasting
We recruited eligible participants from June–December between 20–30 min. We encouraged the nurses to find
2001, following them up until December 2002. individual solutions to patients’ problems, taking into
Participating practices performed a computerized account their experience and knowledge of their
search of their practice registers for patients coded as patients.
having hypertension and who had a latest blood pressure The control group received standard care delivered at
recording of 150 mmHg systolic and/or 90 mmHg their respective practices, apart from blood pressure
diastolic in the past six months.9 GPs and practice nurses checks at similar intervals as the participants in the
screened this list and excluded the following patients: intervention group. Wherever possible, these checks
individuals who did not control their medication intake were carried out by another practice nurse who was not
(such as some nursing home patients); secondary involved in delivering the intervention. Given the
hypertension; severe dementia; other reasons for not potential for contamination, all practice nurses were
approaching their patients such as recent bereavement. made aware of this risk and strongly and repeatedly
The practices then contacted an age–sex stratified encouraged not to change their ‘usual practice’ for the
random sample of the remaining patients (60 and control patients. Discussing non-adherence in detail was
60 years), using computer-generated random numbers not part of usual care.
supplied by the research team. Patients were asked to
return a reply slip to the research team, to receive Baseline data
detailed information and a consent form. We monitored baseline adherence in all participants for
30 days before randomization, using electronic medica-
Interventions tion monitors (MEMS®, Aardex Ltd, Zug, Switzerland),
Practice nurses invited participating individuals to which contain a microchip in their lids that registers the
attend, in addition to usual care, an adherence support time and date of each opening. For cost and feasibility
session lasting a maximum of 20 min, followed by a reasons, MEMS® could only be used for one antihyper-
shorter reinforcement session (10 min) two months tensive drug per patient. When participants were on
later. The aim of the intervention was to provide an more than one blood pressure lowering drug, the
opportunity for patients to talk about any problems with practice nurse chose the one to be monitored using a
their blood pressure lowering medication (Box 1) in a flow chart that favoured more commonly prescribed
safe, non-threatening atmosphere, largely led by the drugs (diuretics and beta-blockers) and medication with
patients themselves.10,11 During the consultation, fewer daily doses.
practice nurses investigated whether patients under-
stood their diagnosis and agreed with the treatment Outcomes
process. We encouraged practice nurses to address The primary outcome was adherence measured by
patient concerns with their medication and to agree MEMS® in the six months period following the
tailored strategies to resolve any medication problems.10 intervention, controlling for adherence in the baseline
We developed this pragmatic intervention using the period (Fig. 1). We defined adherence as ‘timing
compliance’, which is the number of doses taken at
24 ± 6 h intervals for a once daily regimen or 12 ± 3 h for
BOX 1 Common problem categories leading to difficulties twice daily doses, divided by the total number of days
adhering to a treatment regimen6,7,19,20,21 and multiplied by 100%. Secondary outcomes were two
less strict measures of adherence (‘correct dosing’, which
Side effects is the percentage of days on which the correct number of
Size or taste of tablets doses was taken and ‘taking compliance’, defined as the
Number of doses a day percentage of prescribed number of doses taken, which
is the equivalent to a ‘pill count’) as well as systolic
Non-acceptability of taking tablets
and diastolic blood pressure. MEMS® data were
Forgetfulness
downloaded in each centre and processed using
Non-comprehension PowerView software (Aardex Ltd, Zug, Switzerland),
Total number of different tablets but the results were not available to practice nurses or
study participants before the end of the study.
146 Family Practice—an international journal

patients stratified by age and sex to the intervention and

Start of the Baseline appointment
trial (20 minutes)
control groups using computer-generated random
numbers, which were assigned to an anonymized list of
Nurses give out electronic monitors
participants. The principal investigator (KS) passed the
Baseline questionnaire randomization schedule on to the practice nurses shortly
Blood pressure check before the appointment for delivering the intervention.

Blinding
In this open RCT both the study participants and the
1 month later Randomisation practice nurses were aware of the group assignment at
completion of the baseline period.
Intervention group Usual care group
(20 minutes) (10 minutes) Statistical analysis
Adherence support Blood pressure check Using Stata for all analyses,15 we first investigated
consultation
Downloading of baseline comparability of the groups using descriptive
Blood pressure check electronic monitors
statistics. The primary intention-to-treat analysis used
Downloading of multivariable regression models, adjusting for the values
electronic monitors
of the outcome variables at baseline as well as the
stratifying variables. The secondary analyses included
additional adjustment in these regression models in
terms of any potentially influential variables exhibiting
2 months after Re-inforcement Follow up
randomisation consultation appointment
imbalance at baseline. We also conducted pre-planned
(10 minutes) (5 minutes) subgroup analyses for age, sex, drug group and total
Blood pressure check Blood pressure check number of drugs prescribed, by introducing appropriate
Downloading of Downloading of interaction terms to investigate differential effects.
electronic monitors electronic monitors
Economic evaluation
We conducted a limited economic evaluation from the
6 months after Follow up Follow up viewpoint of the provider institutions in the primary care
randomisation (5 minutes) (5 minutes) sector to compare the direct cost of delivering the nurse-
Blood pressure check Blood pressure check led intervention with usual care, and estimate the pro-
Downloading of Downloading of jected cost in a typical general practice. Time spent on
electronic monitors electronic monitors consultations was recorded by the practice nurses and
valued using published sources.16 A simple one-way sens-
FIGURE 1 Overview of study appointment schedule itivity analysis was used to determine the effect of altering
the duration of the nurse appointments to a more realistic
level, likely to apply outside the constraints of a trial.
Sample size
From the literature, about 50–60% of patients would
be adherent to medication in the absence of the
Results
intervention.4,5 For a target difference in adherence of
15 percentage points (for example, 75% in the Recruitment and baseline comparability
intervention group and 60% in the control group), a Descriptive statistics for the participating practices and
2-sided 5% significance level and 80% power, we required practice nurses are available from Tables 1 and 2. Of 837
a total sample size of about 330 participants. With respect eligible patients invited to participate, 245 were random-
to blood pressure at follow-up, this sample size would be ized to receive nurse-led adherence support (n = 128)
sufficient (85% power, 2-sided 5% alpha) to detect and usual care (n = 117) (Fig. 2). A total of 204 particip-
differences between the two groups of 0.42 standard ants attended the six-month follow-up appointment.
deviations. For standard deviations in diastolic blood Table 3 shows key baseline descriptive statistics for both
pressure of about 9 mmHg, this corresponds to differences groups. Overall levels for ‘timing compliance’ were high,
between groups of about 3.2 mmHg. There is good evid- and mean blood pressures in both groups were under or
ence from RCTs that a difference of 5 to 6 mmHg in slightly above the British Hypertension Society audit
diastolic blood pressure leads to a reduction in stroke of standard,8 with 39% (94/245) of participants being
38% and in coronary heart disease of 16%.2,14 ‘uncontrolled’ according to this standard.

Randomization Adherence
One of the authors (TJP) who was not involved in Patients receiving nurse-led adherence support were no
practice and patient recruitment randomized eligible more adherent at six months than patients receiving usual
 Nurse-led adherence support in hypertension 147
TABLE 1 Characteristics of participating practices TABLE 2 Characteristics of participating practice nurses

Characteristic Study practices Characteristic Practice nurses in study

Number of nurses managing hypertension, 3 (1 to 7) Nursing grade, n (%)a

median (range) C 1 (4.6)
E 1 (4.6)
Total number of practice nurses, median (range) 3.5 (1 to 7)
F 12 (54.6)
Number of practices (%) receiving Culyer research 11 (50) G 5 (22.7)
funding H 3 (13.6)
Number of practices (%) involved in teaching Number of qualifications, median (range) 3 (1 to 4)
Undergraduate only 5 (22.7)
Length of time since qualification in years, 17 (6 to 42)
Postgraduate only 1 (4.6)
median (range)
Both 8 (36.4)
None 8 (36.4) Length of time working in the practice in 3.5 (2 to 18)
years, median (range)
Type of list, n (%)
Shareda 9 (40.9) Training in hypertension management, n (%)
Individualb 13 (59.1)
Practice based training only 2 (9.1)
Using a recall system for hypertensive patientsc 22 (100)
Dedicated hypertension clinic 8 (36.4) Both internal and external training 20 (90.9)
Using a standardised protocol, n (%) 22 (100)
a ‘Shared’ means that patients are free to choose any doctor for a
consultation, although they may be ‘registered’ with a particular GP. a Grades A, B and C are used for health care assistants, the D grade
b Patients are registered with a single GP whom they see for most if not
corresponds to a newly qualified staff nurse, and grade E is assigned to
all consultations.
c A system in which patients with high blood pressure are identified, more experienced nurses. An F grade is a senior staff nurse and
G grade is a charge nurse or sister. Any grade above the G grade
coded and, either by computer or other system, followed up
usually implies more management responsibilities.
systematically and according to an agreed practice protocol.

care on the primary outcome—‘timing compliance’ 87% group required on average an extra 13.4 min (9.3 min for
versus 90% as shown in Table 4. There was also no effect the adherence support group and 4.1 min for the
of the intervention on the less strict (secondary) reinforcement consultation). Projected costs for the
adherence measures of correct dosing and taking primary care sector per consultation were therefore
compliance. £9.50 for the intervention compared with £5.08 for usual
care. The sensitivity analysis, using a more realistic
Blood pressure duration of 20 min for the first consultation of the
There was no difference at six months between the intervention (compared with 24.7 min), produced an
groups with regard to systolic or diastolic blood pressure estimated cost of £6.60. These results suggest that for a
(Table 4). general practice of average list size, with a prevalence of
uncontrolled hypertensive patients equivalent to that
Secondary and subgroup analyses observed here, the additional cost of applying a one-off
Adjusting for variables that showed any imbalance at intervention would be £357.20.
baseline did not substantially alter the results. Pre-
specified subgroup analyses found no evidence that the
intervention effect on timing compliance differed by age
(interaction P = 0.36), gender (interaction P = 0.093), Discussion
the total number of drugs that were prescribed for a
Summary of main findings
patient (interaction P = 0.23) and the drug group used in
This RCT shows no evidence of an effect of nurse-led
the electronic monitor (interaction P = 0.12). Analysis
adherence support on timing compliance and systolic or
of process measures collected by the practice nurses
diastolic blood pressure compared with usual care. Costs
showed that amongst the 33 participants who agreed a
were also higher in the intervention group.
strategy to address barriers to adherence (out of 128
receiving the intervention), all stated that the strategy
used had been successful in that the problem had been The strengths and limitations of the study
addressed and ceased to exist. To our knowledge, this is the first RCT in the UK evalu-
ating an adherence improving strategy in hypertension
Economic evaluation that measured baseline levels of adherence and
The mean time allocated to appointments for usual care compared these with adherence levels at follow-up.
was similar across the two groups, at 15.4 min. In Moreover, this study used the current ‘gold standard’ to
addition to this, the practice nurses in the intervention measure timing compliance.
148 Family Practice—an international journal

837 eligible participants invited

(random sample from 22 practice registers)

466 did not respond

371 consented to receive more

information about the study

45 refused to participate

326 initially consented

77 dropped out prior to
first appointment
(5 left practice, 4 not on
medication anymore, 14 too
unwell, 3 died, 17 practice
nurse unwell, 34 reason not
known)

249 attended first appointment and

entered into run-in period 4 dropped out between
run-in and randomisation
(1 died, 1 too unwell, 1
moved away, 1 reason
unknown)

245 were randomised and allocated to treatment groups

128 randomised to 117 randomised to

receive nurse-led receive usual care alone
adherence support in
addition to usual care

8 did not receive re- 5 did not attend follow

inforcement appointment up at two months
(4 practice nurse had left, 4 (4 practice nurse had left,
reason not known) 1 reason unknown)

112 attended respective

120 completed follow up appointment at
intervention (i.e. 2 months
received re-inforcement
appointment at 2
months)
18 lost to follow up
10 lost to follow up (1 unwell, 17 no reason
(no reason given) given)

110 followed up at 6 94 followed up at 6

months months

FIGURE 2 CONSORT flow diagram

In terms of the precision attained by our study, the of the high levels of timing compliance at baseline, the
values in the confidence intervals for the primary chances of detecting a clinically important difference
outcome enable us to rule out a benefit or disbenefit in were low. The precision gained in this study highlights
excess of about five percentage points in timing the uncertainty that surrounded the original power
compliance. This is considerably lower than our target calculation in terms of the assumptions we made about
difference specified in advance and rules out that an the primary outcome. The findings from this study
important clinical difference is being missed, even should, therefore, be useful for future power calculations
though the target sample size was not reached. Because that use timing compliance as the main outcome.
 Nurse-led adherence support in hypertension 149
TABLE 3 Baseline characteristics of trial groups

Characteristic Adherence support group (n = 128a) Usual care group (n = 117a)

Age in years mean ± SD 67.9 ± 10.3 (n = 126) 68.2 ± 9.4 (n = 116)

Male: % 56 54
Timing compliance: mean % ± SDb 90.8 ± 15.6 (n = 101) 94.5 ± 7.6 (n = 88)
Mean blood pressure at baseline in mmHg ± SD
Systolic Diastolic 149.0 ± 15.2 (n = 127) 152.1 ± 17.5 (n = 114)
83.7 ± 9.3 (n = 127) 83.1 ± 9.9 (n = 114)
Uncontrolled blood pressure (BHS standard): % 40 37
Correct dosing: % of days correct number of
doses was taken ± SDb 95.6 ± 9.5 (n = 101) 97.6 ± 4.4 (n = 88)
Taking compliance: % of prescribed number of
doses taken ± SDb 98.6 ± 7.3 (n = 101) 99.4 ± 3.8 (n = 88)
Smokers: % 10 9
Cholesterol mmol/l: mean ± SDa 5.3 ± 1.3 (n = 99) 5.4 ± 1.2 (n = 82)
Diabetes mellitus: % 17.1 (n = 123) 16.2 (n = 117)
History of cardiovascular disease: % 36.6 (n = 123) 31.9 (n = 116)
Drug dispensed from electronic monitor (n = 127 and 115): %
Diuretic 47.2 53.9
Beta blocker 21.3 18.3
Calcium antagonist 11.0 17.4
ACE inhibitor 13.4 6.1
Other 7.1 4.4
Doses dispensed from electronic monitor per day: %
(n = 125 and 109)
Once daily 94.4 98.2
Twice daily 4.8 1.8
Three times daily 0.8 0
Duration of blood pressure treatment: %
less than 6 months 2.3 2.6
6 to 12 months 6.3 5.1
more than 1 year 91.4 92.3
Change in blood pressure medication in past 12 months: % 46.7 (57/122) 42.0 (47/112)

a These are the numbers included in the analyses unless stated otherwise.
b Numbers smaller due to downloading errors and lost monitors.

There is also a possibility that MEMS® may have on more than one antihypertensive drug, it is conceivable
altered patient behaviour, although it is unlikely that that timing compliance of all blood pressure lowering
this effect would have persisted throughout the whole drugs was not as good as for the one that was observed.
study period.17 Losses to follow-up were similar for both We do not know if the high adherence levels observed
groups, with no important differences between the in this RCT were due to a self-selected population, or
groups in terms of age, sex, or differential losses to whether the results reflect generally higher adherence
follow-up. There were differences in losses to follow-up levels in the UK. This raises the question as to whether
between the practices, because some practice nurses left doctors and researchers in the UK have made the wrong
the practice or were off sick without a replacement assumptions about their patients’ propensity for not
continuing to conduct the trial. taking their medicines as prescribed. Higher adherence
Baseline timing compliance, the strictest measure of levels in the UK would not be implausible, as the UK
adherence obtained through electronic monitoring, was National Health Service is different from health care
in excess of 90% in study participants, challenging the delivery in the US. Whereas patients in the UK pay a
widespread belief that adherence to blood pressure prescription charge, hypertensive people in the US
lowering medication is only about 50–60%. and Canada often pay for their medicine, which might
For reasons including cost and feasibility, the primary explain higher levels of non-adherence in North
outcome was timing compliance of a single antihyper- American settings. In addition, the UK is unique in its
tensive agent. In those instances where participants were organization of primary care, with GPs being in charge
150 Family Practice—an international journal

TABLE 4 Regression models comparing various outcomes between the intervention and control groups at six months, controlling for baseline
measurement of the outcome and the stratifying variables (general practice, age group and sex)

Nurse-led adherence Usual care Adjusted difference (AS-UC)

support (AS), mean (SD) (UC), mean (SD) between means (95% CI)a P-value

Timing compliance: % of 87.2 (20.1) 90.2 (16.2) 1.0 (5.1 to 3.1) 0.63
days correct number of
doses taken on time, n = 159
Correct dosing: % of days 90.8 (16.6) 92.4 (15.2) 0.5 (4.2 to 3.1) 0.77
correct number of doses
taken, n = 159
Taking compliance: % 95.6 (16.4) 95.6 (15.7) 0.6 (3.2 to 4.4) 0.76
prescribed number of
doses taken, n = 159
Systolic blood pressure: 142.9 (17.6) 147.7 (20.9) 2.7 (7.2 to 1.8) 0.24
mmHg, n = 200
Diastolic blood pressure: 80.4 (10.1) 79.9 (9.7) 0.2 (1.9 to 2.3) 0.85
mmHg, n = 200

a A positive difference indicates an increase in adherence for AS compared with UC; a negative difference indicates a reduction in blood pressure for

AS compared with UC.

of the overall care of their patients, which could medication is only about 50 to 60%. While the selec-
positively affect medication adherence. tion effects noted above may be influential, these
We chose patients with ‘uncontrolled hypertension’, previous studies often used relatively unreliable
using a widely agreed audit standard.9 This population of methods of measuring adherence. While the lack of an
patients was selected because we anticipated that low intervention effect in our study needs to be seen in the
adherence to therapy was a likely to be a significant con- context of already high adherence levels, it remains
tributory factor to their poorly controlled hypertension.18 the first randomised evaluation of such an intervention
Baseline blood pressures in both groups were close to our in the UK.
chosen cut-off point of 150/90, and only 94 out of 245
participants (39%) were ‘uncontrolled’. Thus there is Implications for future research
potential for bias, as patients with higher blood pressures, Further observational studies are required to investigate
who may also be less adherent to medication regimens, the epidemiology of adherence in treated hypertensive
were less likely to take part in this study. Due to data people in the UK at various levels of blood pressure
protection reasons, we were unable to obtain data on control. Previous studies often used measurements and
eligible participants who refused to take part, which would definitions of adherence that are varied and imprecise
have allowed us to investigate any systematic differences and had relatively short follow-up periods.5,8 Wherever
between these individuals and the study participants. possible, future studies should therefore use reliable
Although timing compliance was higher than measurements of medication adherence such as
expected,4,5,7 only 61% of participants had ‘controlled’ electronic monitoring, which can also provide valuable
blood pressure. This highlights that the relationship bet- data on the patterns of medication taking.
ween adherence and blood pressure is still poorly under- The World Health Organization and the Working
stood. It therefore appears that other factors may be Party on Concordance advocate a multidisciplinary
equally important when accounting for poor control, approach to make progress in the area of improving
including inaccurate measurement of blood pressure or medication adherence.7,10 Once more is known about
regression to the mean.18 the epidemiology of adherence, future research should
Lastly, with this study design there was potential for involve evaluation of complex interventions, which may
contamination. This is unlikely, however, since the median include nurse-led care. Interventions should target
number of nurses per practice was 3.5 (range 1 to 7), which those at greater levels of risk and always take patient
allowed the follow-up for both intervention groups to be preferences and concerns into account.10
conducted by different nurses within each practice.
Implications for clinical practice
How and why this study disagrees with the existing Nurse-led adherence support did not improve adherence
literature and was also more costly than usual care. The results of
Our study findings are not consistent with the widespread this study do not support the introduction of special
belief that adherence to blood pressure lowering nurse-led adherence support appointments in the
 Nurse-led adherence support in hypertension 151
management of hypertension. Since baseline timing References
compliance was high and many participants had
1 MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke and
relatively well controlled blood pressure, there may still
coronary heart disease. Part 1, prolonged differences in blood
be a need to identify and address adherence problems pressure: prospective observational studies corrected for the
in patients with higher blood pressure levels as one of regression dilution bias. Lancet 1990; 335: 765–774.
2
several potential reasons for inadequate control, Staessen J, Wang J-G, Thijs L. Cardiovascular protection and blood
including inaccurate measurement, white-coat hyperten- pressure reduction. Lancet 2001; 358: 1305–1315.
3 Colhoun HM, Dong W, Poulter NR. Blood pressure screening,
sion, sub-optimal treatment, antagonising substances, management and control in England: results from the health
co-existing conditions, and secondary hypertension.17 survey for England 1994. J Hypertension 1998; 16: 752.
4 Sackett DL, Haynes RB, Gibson ES et al. Randomized clinical trial
Although not based on evidence, there appear to be few
of strategies for improving medication compliance in primary
reasons why practice nurses, or indeed GPs, should not
hypertension. Lancet 1975; 1: 1205–1207.
routinely ask patients in more detail about their 5 Ebrahim S. Detection, adherence and control of hypertension for
medication taking, particularly if they show a less than the prevention of stroke: A systematic review. Health
expected response to their medical treatment. The Technology Assessment 1998; 11(2).
6 Benson J, Britten N. Patients’ decisions about whether or not to take
justification for this is that in the few study participants antihypertensive drugs: qualitative study. Br Med J 2002;
who stated that they did have medication problems, the 325: 873.
intervention appeared to be very successful in terms of 7 Sabate E. Adherence to long-term therapies: evidence for action.
addressing these barriers to adherence, although the Geneva: World Health Organization; 2003.
8 Schroeder K, Fahey T, Ebrahim S. How can we improve adherence
numbers involved were far too small to derive robust to blood pressure lowering medication? Systematic review of
conclusions in terms of the effects on timing compliance randomized controlled trials (RCTs). Archives of Internal
and blood pressure. However, asking patients about their Medicine 2004; 164: 722–723.
9 Ramsey LE, Williams B, Johnston G et al. Guidelines for the
medication taking and trying to identify individual
management of hypertension: report of the third working party
solutions could be performed opportunistically on of the British Hypertension Society. Journal of Human
selected patients with minimal extra costs and potential Hypertension 1999; 13: 569–592.
10 Marinker M, Blenkinsopp A, Bond C et al. From compliance to
benefits in terms of improved adherence, better blood
concordance—achieving shared goals in medicine taking.
pressure control—and a possible improvement in
London: Royal Pharmaceutical Society of Great Britain; 1997.
communication between patient and health professional. 11 Marinker M, Shaw J. Not to be taken as directed: putting
concordance for taking medicines into practice. Br Med J 2003;
Conclusion 326: 349–350.
12 Leventhal H, Diefenbach M, Leventhal E. Illness cognition: using
This pragmatic randomized trial has shown that nurse-
commonsense to understand treatment adherence and affect-
led adherence support was no better and more costly cognition interactions. Cognitive Research and Therapy 1992;
than usual care in improving adherence to blood 16: 143–163.
pressure lowering medication and reducing blood 13 Leventhal H, Cameron L. Behavioural theories and the problem of
pressure, particularly against a background of already compliance. Patient Education and Counselling 1987;
10: 117–138.
high adherence levels. 14 Collins OR, Peto R. Antihypertensive drug therapy: effects on
stroke and coronary heart disease. In Swales JD (ed.). Textbook
of hypertension. Oxford: Blackwell Scientific Publications;
Acknowledgements 1994, 1156–1164.
15 Stata Corporation. Stata Statistical Software: Release 8.0. College
We would like to thank the general practices, in Station, TX: Stata Corporation; 2003.
16 Netten A, Curtis L. Unit costs of health and social care 2002.
particular their practice nurses and patients, for taking
University of Kent: Personal Social Services Research Unit;
part in this study. Many thanks also to Alan Montgomery 2002.
for providing helpful comments on previous drafts of the 17 Urquhart J. The electronic medication event monitor—lessons
manuscript. for pharmacotherapy. Clinical Pharmacokinetics 1997; 32:
345–356.
18 Mulrow CD. Evidence-based hypertension. London: BMJ Books;
2001.
Declaration 19 Jokisalo E, Kumpusalo E, Enlund H, Takala J. Patients’ perceived
problems with hypertension and attitudes towards medical
Funding: this study was funded as part of a Medical treatment. J Human Hypertension 2001; 15: 755–761.
Research Council Training Fellowship in Health 20 Enlund H, Jokisalo E, Wallenius S, Korhonen M. Patient-perceived
Services Research (KS). problems, compliance, and the outcome of hypertension
treatment. Pharmacy World & Science 2001; 23: 60–64.
Ethical approval: the South West Multi-centre Research 21 Dowell J, Jones A, Snadden D. Exploring medication use to seek
Ethics Committee approved this study. concordance with ‘non-adherent’ patients: a qualitative study.
Conflicts of interest: none declared. Br J Gen Pract 2002; 52: 24–32.