The effects of education on patient adherence to medication

The effects of education on patient adherence to medication

ARTICLE IN PRESS Evidence-Based Healthcare & Public Health (2005) 9, 398–404 www.elsevier.com/locate/ebhph SYSTEMATIC REVIEW The effects of educati...

143KB Sizes 11 Downloads 34 Views

ARTICLE IN PRESS Evidence-Based Healthcare & Public Health (2005) 9, 398–404

www.elsevier.com/locate/ebhph

SYSTEMATIC REVIEW

The effects of education on patient adherence to medication Bazian Ltd London, UK

Key points

   





Educating patients about their condition and the medication prescribed to treat it is good medical practice. Many patients with chronic conditions need longterm therapy but do not fully adhere to their medication. There are few large studies of educational interventions aimed at increasing adherence to medication. Of the seven studies we found, two reported an increase in medication adherence, while in the other five, there was no significant difference between groups. In one of the two positive studies, the increased adherence was not matched by improvements in therapeutic response or quality of life. In the other, completion of treatment for tuberculosis reached 50%, still far short of the target of 90%. Providing educational information does not ensure patient adherence to medication.

50%.1 This limits the benefits that patients get from healthcare and, if the drug is dispensed but not taken, wastes money. The most cost-effective interventions to improve adherence are likely to be brief, yet will enable many patients to improve their adherence to medication. Such interventions are likely to include an educational component to help the patient understand their condition, the need for the medication and the importance of continuing the course of therapy according to the prescription. Providing some information is ethically mandatory, but we do not know if education improves adherence: more knowledge is often not enough to bring about changes in behaviour.

Review of the evidence In this review, we examine the effects of educational interventions on patient adherence to medication.

Background

Search strategy

Adherence to effective medication produces clinical benefits for patients, but low adherence is common.1 In chronic medical conditions, adherence to prescribed medication is often only around

In July 2005 we searched for systematic reviews and randomised controlled trials of measures to increase medication adherence. We searched Medline, EMBASE, Cinahl and the Cochrane Library.

1744-2249/$ - see front matter & 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ehbc.2005.09.004

ARTICLE IN PRESS The effects of education on patient adherence to medication

Inclusion/exclusion criteria We found no systematic review and 147 randomised trials, so we narrowed our search to studies of educational interventions in chronic conditions, with at least 250 participants. We included only studies in English.

Data extraction and synthesis We extracted data relating to the study design, the participants, the intervention and the main outcome measures reported. We assessed the methodological quality of the studies in terms of their randomisation methods, sample size, potential biases and confounding factors, use of correct statistical methods, and the validity and completeness of the outcome measures reported. The review is narrative.

Summary of studies We found seven studies that met the inclusion criteria (Table 1). Crilly and Esmail undertook a single-blind randomised controlled trial of providing an educational booklet about hypothyroidism compared with usual care, in an attempt to improve adherence to thyroxine therapy among patients with primary hypothyroidism.1 No statistically or clinically significant changes in adherence to medication were found between the groups or over the duration of the trial. Goujard and colleagues studied the impact on adherence to HIV medication of an individualised patient education programme, based on problems which patients might have with adherence.2 This intervention had no effect on knowledge about HIV and its treatment. Adherence to antiretroviral therapy apparently improved, but the outcome measures were not well described in the paper, so the clinical significance of an improvement of this size is unclear. The improved adherence to therapy was not matched by improvements in therapeutic response (measured by changes in CD4 cell counts and plasma viral load), and had no impact on quality of life. This suggests that the improved adherence was not enough to influence the course of the illness. In this trial, the control group crossed over to the educational intervention after the first year; at 18 months, when both groups had had the intervention, the former control group’s scores had improved in a similar way to the original intervention group. However, this increase in scores among the control group started prior to the time that

399

they received the intervention. This was thought to be due to contamination, as some people randomised to the control group attended the educational sessions, and the attitude of the staff was thought to have changed so that they did not want to withhold information that might help adherence in the control group. This contamination would have biased the study against showing an effect from the intervention. Grant and colleagues undertook a randomised controlled trial in which a pharmacist provided tailored education regarding medication, by phone, to patients with type 2 diabetes.3 Self-reported medication adherence did not change during the study; although this outcome measure is likely to be higher than true adherence, an unchanged result supports the trial’s negative findings. This study included a power calculation, but the calculated sample size was not reached, so the trial was underpowered. Herschorn and colleagues studied the impact of a brief health education intervention in patients with overactive bladder.4 The intervention consisted of a physician or nurse discussing with the participant an information sheet about overactive bladder and the medication (tolterodine) that was being prescribed. Three different information sheets appeared to be used; the duration of the intervention was less than 5 minutes. No difference was found between the groups in self-reported adherence to medication. The study was underpowered. The data were incorrectly analysed, using the wrong statistical test and not adjusting for a possible confounding factor (the use of more non-drug treatments for the bladder problem among the intervention than the control group). Hesselink and colleagues assessed in a randomised controlled trial the effectiveness of an education programme by a general practice assistant for patients with asthma or COPD.5 The intervention comprised 1–4 semi-structured consultations of 30 minutes each, focusing on the patient’s technical skills (such as inhaler technique) and how to cope with the disease. Inhaler technique was improved by the intervention, but adherence to medication, disease symptoms and quality of life were unchanged. Participants were also not representative of the underlying population: non-responders were significantly more likely to be male and were younger than those who participated, and over a third of participants were lost to follow up, so these results are not generalisable. Hovell and colleagues aimed to increase Latino adolescents’ adherence to treatment for latent tuberculosis infection in a controlled trial of

ARTICLE IN PRESS 400 Table 1

Bazian Ltd Studies of the effects of education on patient adherence to medication.

Key features of study

Main outcomes

Key quality issues

Crilly et al. Randomised controlled trial Participants: 332 adults with primary hypothyroidism out of 432 eligible in 3 GP practices in England; mean age around 62 yr; 92% female Intervention: Hypothyroid educational booklet (based on health action model) v usual care

Primary outcome measure: Adherence to thyroxine treatment measured by mean within-subject change in TSH: Booklet group: 0.11 mIU/L Usual care: 0.12 mIU/L Difference 0.01 mIU/L (95% CI –0.93 to +0.94 mIU/L): no significant difference Other outcome measures: SF-36 domains of vitality and general health; hypothyroid symptom index: changes minimal and no significant difference between groups

Single blinded intervention (TSH assessors blinded but participants not blinded) Power calculation done; power adequate to detect a difference of 1.5 mIU/L (pre-defined clinically relevant difference was 1.0 mIU/) Participants on average younger and more likely to have had recent TSH in reference range than non-participants, so may be more compliant sub-group of patients, although 22% admitted non-adherence

Goujard et al.2 Randomised controlled trial Participants: 326 outpatients with HIV on highly active antiretroviral therapy (at least 3 drugs for at least 3 months) in 3 hospital university-based centres in France; mean age around 40.5 yr; 80% male Intervention: Educational programme (individualised programme based on problems participants might have with adherence) v standard care; patients in control group crossed over to educational programme at 12 months

Knowledge about HIV and its treatment (11 questions; each answer right or wrong): number of incorrect answers stable in both groups Adherence to antiretroviral therapy (1 poor to 4 good, based on 2 questionnaires plus qualitative criteria): At 6 months: Intervention group: improved by +0.25 units from baseline Control group: deteriorated –0.19 units, p ¼ 0:025 At 12 months: Intervention group: improved by +0.22 units from baseline Control group: deteriorated –0.05 units, p ¼ 0:22 At 18 months (when control group had also received intervention): Intervention group: improved by +0.26 units from baseline Control group: improved 0.27 units, p ¼ 0:75 v intervention group Quality of life (HIV-46 questionnaire: 0 poor to 100 good): no significant difference Therapeutic response (change in CD4 cell count, plasma viral load): Intervention group: 56% improved at 6 months; 59% at 12 months Control group: 50% improved at 6 months; 59% at 12 months No significant difference

Increased adherence in control group prior to crossover to intervention may have been due to contamination (some control patients actually attended sessions) and possibly a change in the practice of the staff members towards the control group (e.g. staff feeling it was unethical to withhold information that might help increase adherence in control group). At least 20% of patients did not return questionnaires at each visit (possible source of bias although baseline adherence not associated with risk of dropout). High level of adherence at baseline (46% of patients scored maximum on this measure—possible ceiling effect, i.e. intervention could not increase adherence by much even if effective as there is less scope for improvement) QOL could also have a ceiling effect as these were mostly asymptomatic outpatients. Previous non-adherence can cause resistance to drug therapy, so then if adherence increased, still no effect on disease measures.

1

ARTICLE IN PRESS The effects of education on patient adherence to medication

401

Table 1 (continued ) Key features of study

Main outcomes

Key quality issues

Grant et al. Randomised controlled trial Participants: 232 patients with type 2 diabetes of 462 eligible (had to have had clinic visit in last 6 months) in one academically affiliated community health centre serving working class community in USA; around 51% female; mean age around 64 yr. Intervention: Pharmacist provided tailored education regarding medication use during a phone call with the patient and created summary of adherence barriers sent to primary care provider

Self-reported medication adherence and barriers (from 2 questionnaires): no change during intervention

Only patients with clinic visits in last 6 months eligible (excluded patients less engaged with medical system). Only 232 of 462 eligible patients randomised—230 ‘set aside’—reason not stated, so population may be biased and results not generalisable. The 232 included was more than the 110 in each arm required by the power calculation. However, it was not an intention-to-treat analysis; only 52% (120) of originally randomised patients assessed, i.e. underpowered. Baseline self-reported medication adherence very high (reported taking all prescribed medicine on 6.9 of previous 7 days; only 5% of patients in each arm reported less than perfect adherence) and barriers few (potential ceiling effect of intervention) Self-reported adherence less objective than pill counts or therapeutic drug monitoring and may be inaccurate

Herschorn et al.4 16-week open-label randomised trial Participants: 138 patients with overactive bladder in family medicine and urology clinics in Canada; mean age around 64 yr; 88% female Intervention: Tolterodine combined with an educational intervention (physician or nurse went through information sheets on overactive bladder and use of tolterodine—3 different information sheets used—intervention lasted 3–5 minutes) v tolterodine alone

Self-reported compliance with anticholinergic medication: Number of patients purchasing prescriptions: Intervention group: 39% Control group 31% No significant difference Severity of bladder symptoms (self-reported better, same or worse): Better: Intervention group: 45.4% v control group 20%, po0:05 Same: Intervention group: 42.4% v control group 66.7%, NS Worse: Intervention group: 12.1% v control group 13.3%, NS

No power calculation reported—study underpowered. 138 patients enrolled; only 84 evaluable (61%); may have biased the population. Participants also thought to be more compliant than ‘average’ patients as had to consent to a series of phone calls about their condition prior to participation. Self-reported adherence less objective than pill counts or therapeutic drug monitoring and may be inaccurate Wrong statistical tests used—should have used test for trend not compared proportion with ‘improved’ scores etc. Proportion of intervention group reporting improvement also confounded by more using non-drug treatments for overactive bladder in addition to the drug—not adjusted for in analysis.

Hesselink et al.5 Randomised controlled trial Participants: 276 patients with asthma or COPD using medication and experiencing pulmonary symptoms in

Primary outcome measures: Disease symptoms, healthrelated quality of life: No significant difference Secondary outcome measures: Inhalation technique correct: Intervention group: 66% at 1 yr and 75% at 2 yr

Sample size calculation based on improvement in health-related quality of life; required sample size reached but underpowered to detect a difference in adherence as scale only ‘compliant’ or ‘not compliant’. Non-responders more likely to be male and younger than participants, so participants not

3

ARTICLE IN PRESS 402

Bazian Ltd Table 1 (continued ) Key features of study

Main outcomes

Key quality issues

12 general practices in the Netherlands Intervention: Tailormade education (1–4 semi-structured consultations of 30 minutes regarding information about the disease, medication, compliance, inhaler technique, barriers to coping with the disease, advice about when to consult a doctor) by general practice assistant focusing on patient’s technical skills and coping with the disease

Usual care: 50% and 59% po0:05 for both time points Also compliance, smoking cessation, self-efficacy, coping: No significant difference

representative of underlying population 36% lost to follow up High quality of life scores at baseline so possible ceiling effect May not have been appropriate to have one intervention for both asthma and COPD patients

Hovell et al.6 Randomised controlled trial Participants: 302 out of 535 eligible Latino adolescents (aged 12–19 yr, mean 15.6 yr; 55.6% male) with latent tuberculosis infection Intervention: Adherence coaching (interviewing, setting adherence goals, contingency contracting and ‘shaping’ procedures; the latter involving gradual steps towards solving adherence problems), attention (self-esteem counselling) control or usual care: 12 intervention sessions over 6 months (5 face to face+7 by telephone)

Primary outcome measure: Adherence to isoniazid at 9 months (total number of pills taken): monthly interviews validated with random urine assays: Adherence coaching: Mean number of pills taken 179.93 (57.01), po0:05 v each of the control groups Attention control: 155.37 (69.91) Usual care: 150.98 (73.75)

No power calculation, but appears a reasonable sample size. Best group only achieved 51.1% completion of treatment (i.e. took 180 pills within 9 months), whereas target rate (from Healthy People 2010) is 90%—so although beneficial, the intervention was not strong enough to achieve the 90% target.

Morisky et al.7 Randomised experimental 4-group design Participants: 794 adolescents (aged 11–19 yr, mean 15.2 yr; 51% female) with latent tuberculosis infection at 2 clinics in the USA Interventions: (1) peer counseling

Completion of care (typically isoniazid for at least 6 months): No significant difference across groups Peer counselling appeared to improve ‘self-efficacy for medication-taking behaviour’ (e.g. items assessing the participant’s feelings about their ability to remember to take medication), ‘mastery’ (a measure of perception of control

All participants, including the ‘usual care’ group were interviewed 3 times during the study; may have increased adherence due to the ‘Hawthorne’ effect. No proof was required that the incentives agreed between the participants and parents were actually provided. The questionnaires were translated into Spanish for some participants—some had difficulty understanding the scale of responses (strongly agree to strongly disagree) as there are no exact equivalents to this in Spanish, so

ARTICLE IN PRESS The effects of education on patient adherence to medication

403

Table 1 (continued ) Key features of study

Main outcomes

Key quality issues

(adolescents who had completed therapy for latent TB acted as counsellors by phone at least every 2 weeks, stressing the importance of taking medication), (2) participant-parent contingency contract (negotiation of an incentive provided by the parent if the adolescent adhered to treatment), (3) both (4) usual care

over situations) and self esteem; all 3 of which were stated to be related to medication-taking behaviour. However, this was not translated into a benefit in terms of participants actually completing the course of medication (see above).

questionnaires may not be valid in this population. The data relating each group individually with self-efficacy, mastery and self esteem were not presented; it is not clear whether the results for the peer counselling group presented apply to the participants who had this intervention only (i.e. group 1) or all those who had this intervention (i.e. group 1 plus group 3)—this seems to be selective reporting. These variables measured at baseline were also related to medication taking (i.e. before the participants had had the intervention)—the outcome analyses suggesting that peer counselling improved the variables during the course of the study did not appear to control for these at the start of the study, i.e. the ‘effect’ could be a bias between the groups at entry rather than a true effect of the intervention.

adherence coaching compared with an attention control (an intervention comprising self-esteem counselling) or usual care.6 Adherence coaching increased the mean number of pills taken (assessed using monthly questionnaires validated by random urine assays) from around 150 to around 180 over 9 months. However, completion of treatment (the number of pills over 9 months needed to eradicate tuberculosis) is defined as 180. So even after the intervention, only half the participants completed treatment, compared with a United States Government target of 90%. In another trial in adolescents with latent tuberculosis, Morisky and colleagues randomised participants to peer counselling, a contingency contract where parents offered participants incentives to complete treatment, both of these interventions, or neither.7 There was no significant difference in the proportion of participants completing the course of treatment between the groups. Peer counselling appeared to improve ‘self-efficacy for medication-taking behaviour’ (e.g. items assessing the participant’s feelings about their ability to remember to take medication), ‘mastery’ (a measure of perception of control over situations) and self esteem, all three of which the authors thought were related to medication-taking behaviour. However, these changes were not translated into a benefit in terms of participants actually completing the course of medication.

This study had a number of limitations. Firstly, all participants, including the usual care group, were interviewed three times during the study; this may have increased adherence due to the Hawthorne effect. Secondly, no proof was required that the incentives agreed between the participants and their parents were actually provided. Thirdly, the questionnaires were translated into Spanish, but some participants had difficulty understanding the scale of responses for which there were no exact equivalents in Spanish. The questionnaires may not have been valid in Spanish speakers. Fourthly, the data were selectively reported, with those relating each group individually with self-efficacy, mastery and self esteem not presented. It is not clear whether the results presented for the peer counselling group apply to the participants who only had this intervention or included those who also had the contingency contract. Finally, there may be important confounding because of uneven randomisation: the groups differed at baseline with respect to self-efficacy, mastery and self esteem but there was no correction for this in the analysis.

Review findings We found few trials that studied the effects of educational interventions on patient adherence to medication. The studies we found were in general

ARTICLE IN PRESS 404 small and had methodological flaws that made their findings unreliable. In the one study where an educational intervention was associated with an improved inhaler technique in patients with asthma or COPD, this was not matched by improved adherence to medication, disease symptoms or quality of life.5 Similarly, in a study where an intervention increased adherence to antiretroviral therapy, this was not matched by improvements in therapeutic response or quality of life.2 Adherence coaching comprising 12 intervention sessions over 6 months increased adherence to the treatment of latent tuberculosis among Latino adolescents, although it failed to achieve the target of ensuring 90% of participants completed the required treatment. Providing information may be necessary but it is not sufficient to ensure patient adherence to medication. Other factors such as the doctorpatient relationship, the patient’s attitude to their illness and involvement in their care and their beliefs about the convenience, efficacy, safety and side-effects of the medication may be more important than information. Also, the interventions studied were mostly brief, and may not have been sufficient to provide motivation or maintain it in the long term. Participants in these studies may not be widely representative, due to greater engage-

Bazian Ltd ment with the medical system in the first place. Opportunities for improving adherence may be greater in populations where initial adherence is lower.

References 1. Crilly M, Esmail A. Randomised controlled trial of a hypothyroid educational booklet to improve thyroxine adherence. Br J Gen Pract 2005;55:362–8. 2. Goujard C, Bernard N, Sohier N, et al. Impact of a patient education program on adherence to HIV medication. J Acquir Immune Defic Syndr 2003;34:191–4. 3. Grant R, Devita N, Singer D, Meigs J. Improving adherence and reducing medication discrepancies in patients with diabetes. Ann Pharmacother 2003;37:962–9. 4. Herschorn S, Becker D, Miller E, Thompson M, Forte L. Impact of a health education intervention in overactive bladder patients. Can J Urol 2004;11:2430–7. 5. Hesselink AE, Penninx BW, van der Windt DA, et al. Effectiveness of an education programme by a general practice assistant for asthma and COPD patients: results from a randomised controlled trial. Patient Educ Couns 2004;55:121–8. 6. Hovell MF, Sipan CL, Blumberg EJ, et al. Increasing Latino adolescents’ adherence to treatment for latent tuberculosis infection: a controlled trial. Am J Public Health 2003;93:1871–7. 7. Morisky DE, Malotte CK, Ebin V, et al. Behavioral interventions for the control of tuberculosis among adolescents. Public Health Rep 2001;116:568–74.