Strategies to improve adherence to medications for cardiovascular diseases in socioeconomically disadvantaged populations: A systematic review

Strategies to improve adherence to medications for cardiovascular diseases in socioeconomically disadvantaged populations: A systematic review

International Journal of Cardiology 167 (2013) 2430–2440 Contents lists available at ScienceDirect International Journal of Cardiology journal homep...

915KB Sizes 0 Downloads 47 Views

International Journal of Cardiology 167 (2013) 2430–2440

Contents lists available at ScienceDirect

International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard

Review

Strategies to improve adherence to medications for cardiovascular diseases in socioeconomically disadvantaged populations: A systematic review Tracey-Lea Laba a, b,⁎, Jonathan Bleasel a, c, Jo-anne Brien b, c, d, Alan Cass a, Kirsten Howard f, David Peiris a, Julie Redfern a, e, Abdul Salam g, Tim Usherwood e, Stephen Jan a, e a

The George Institute for Global Health, Sydney, Australia Faculty of Pharmacy, University of Sydney, Australia Faculty of Medicine, University of New South Wales, Sydney, Australia d Department of Pharmacy, St Vincent's Hospital, Sydney, Australia e Sydney Medical School, University of Sydney, Sydney, Australia f Sydney School of Public Health, University of Sydney, Australia g The George Institute for Global Health, Hyderabad, India b c

a r t i c l e

i n f o

Article history: Received 19 November 2012 Accepted 18 January 2013 Available online 13 February 2013 Keywords: Medication adherence Socioeconomic disadvantage Cardiovascular disease

a b s t r a c t Medication non-adherence poses a major barrier to reducing cardiovascular disease (CVD) burden globally, and is increasingly recognised as a socioeconomically determined problem. Strategies promoting CVD medication adherence appear of moderate effectiveness and cost-effectiveness. Potentially, ‘one-size-fits-all’ measures are ill-equipped to address heterogeneous adherence behaviour between social groups. This review aims to determine the effects of strategies to improve adherence to CVD-related medications in socioeconomically disadvantaged groups. Randomised/quasi-randomised controlled trials (1996-June 2012, English), testing strategies to increase adherence to CVD-related medications prescribed to adult patients who may experience health inequity (place of residence, occupation, education, or socioeconomic position) were reviewed. 772 abstracts were screened, 111 full-text articles retrieved, and 16 full-text articles reporting on 14 studies, involving 7739 patients (age range 41–66 years), were included. Methodological and clinical heterogeneity precluded quantitative data synthesis. Studies were thematically grouped by targeted outcomes; underlying interventions and policies were classified using Michie et al.'s Behaviour Change Wheel. Contrasting with patient or physician/practice strategies, those simultaneously directed at patients and physicians/practices resulted in statistically significant improvements in relative adherence (16–169%). Comparative cost and cost-effectiveness analyses from three studies did not find cost-saving or cost-effective strategies. Unlike much current evidence in general populations, promising evidence exists about what strategies improve adherence in disadvantaged groups. These strategies were generally complex: simultaneously targeting patients and physicians; addressing social, financial, and treatment-related adherence barriers; and supported by broader guidelines, regulatory and communication-based policies. Given their complexity and potential resource implications, comprehensive process evaluations and cost and cost-effectiveness evidence are urgently needed. © 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction In high-income countries, it is estimated that up to 50% of patients who are prescribed cardiovascular medications do not, at some point, adhere to therapy [1,2]; with rates of non-adherence greater still among marginalised groups and in low and middle income countries [2,3]. Although a well-established evidence-base for efficacious cardiovascular ⁎ Corresponding author at: The George Institute for Global Health, Level 10 King George V Building, Missenden Road, Camperdown NSW 2050, Australia. Tel.: +61 403 987 464. E-mail address: [email protected] (T.-L. Laba). 0167-5273/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijcard.2013.01.049

disease (CVD) medicines exists, low adherence rates undermine the translation of this evidence into practice, thereby limiting the contribution of these proven treatments to reducing the chronic disease burden globally [2,4,5]. Observational research evidence has to date identified several patient, medication, disease and environmental factors that may explain variations in adherence across populations [2,6,7]. These factors are commonly dichotomised into ‘intentional’ versus ‘unintentional’ non-adherence [4,6,8]. Intentional non-adherence arises from the beliefs, attitudes and expectations that influence patients' motivation to begin and persist with prescribed therapy. By contrast, unintentional

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

2431

non-adherence arises from resource and other limitations (such as a lack of information) faced by individuals and groups [6,8]. Medication non-adherence is increasingly recognised to be associated with socioeconomic hardship. In other chronic conditions such as HIV/AIDS, factors such as poverty and in particular food insufficiency and hunger [9], and unstable housing [10] have been associated with medication non-adherence. In relation to CVD, low socioeconomic status (SES) has been found to be associated with low adherence in a number of different contexts [11–13]. Although there is evidence that the elimination of co-payments for medications prescribed after myocardial infarction improves medication adherence rates, sub-optimal adherence has also been found to exist despite low out-of-pocket prescription costs and free access to health services [14]. This suggests that intentional as well as unintentional non-adherence have a role to play in nonadherence and therefore multi-faceted strategies need to be considered. Previous systematic reviews indicate that intervention studies within CVD have been of low quality and use inconsistent methodology, particularly pertaining to the measurement and conceptualisation of adherence [15,16]. Furthermore, complex, multi-factorial strategies, primarily addressing unintentional or intentional adherence separately have been mostly investigated without assessment of their individual components. Interventions such as simplifying dosage regimens and fixed combination pills appear to be the most effective [15,16]. In general, despite the breadth of adherence research to date, strategies to promote adherence to medication prescribed for chronic conditions [4] including cardiovascular diseases [15,16] have been shown, at best, to be of moderate effectiveness and cost-effectiveness. One potential explanation is that ‘one-size-fits-all’ measures are ill-equipped to address the heterogeneity in adherence behaviour that exists between social groups. Strategies may therefore be more effective if: 1) designed for, and targeted to, specific groups such as those of socioeconomic disadvantage; and 2) they take into account multiple pathways though which social disadvantage may influence patients' medication-taking behaviour. This systematic review aims to determine the effects of strategies to improve adherence to medications prescribed for the primary and secondary prevention of cardiovascular diseases in socioeconomically disadvantaged groups and to identify the characteristics of those strategies that appear to be effective.

developed from previous Cochrane systematic reviews [4,15,16,18,19] incorporating the Cochrane Highly Sensitive search strategy for identifying randomised controlled trials [20]. The references of retrieved articles were screened to identify additional references. One author (TL) carried out the search and screened all of the titles, abstracts and keywords of publications retrieved to identify publications that appeared to meet the inclusion criteria. Two authors (TL, JBl) examined the full-text versions of all studies that appeared to meet the inclusion criteria, using a standardised study eligibility assessment form developed for this review (available upon request). The final decision on which studies to review was confirmed by consensus between the two reviewing authors.

2. Methods

Descriptions of the health care setting including the medication reimbursement environment were described in 7 articles [24,26,27,30,32, 37,37]. For trials conducted in the USA [24–27,30,31,33,36,37], Australia [28] and Spain [29], inclusion was assessed by reported socioeconomic baseline characteristics rather than being based on specific inclusion criteria of socioeconomic disadvantage. Haskell et al. (2006) was one exception to this where low socioeconomic status was an explicit inclusion criterion. The dimensions of socioeconomic disadvantage assessed at baseline included income (n=7) [24,25,27,30,33,36,37] and/or education (n=2) [29,33] or residence (n=2) [28,37].

A pre-defined plan was used (available on request). The authors of this manuscript have certified that they comply with the Principles of Ethical Publishing in the International Journal of Cardiology. 2.1. Criteria for considering studies in this review 2.1.1. Study type, participants, setting, and strategies Randomised or quasi-randomised controlled trials comparing a strategy of any type intended to increase patient adherence to cardiovascular medications to no strategy or usual care were considered. Studies involving adult patients prescribed medications for the prevention or treatment of cardiovascular disease in a primary care, outpatient, or other community setting who may be at risk to some form of health inequity according to place of residence, occupation, education and socioeconomic position (as outlined in the PROGRESS-plus framework) [17] were eligible for inclusion. Studies of targeted-ethnic groups were excluded unless specifically attributed to socio-economic disadvantage. Studies involving psychiatric, military or institutionalised patients were excluded to avoid the potential influence of psychosocial or institutional controls over adherence. 2.1.2. Outcome measures Indirect (e.g. pill count, prescription refill rate, electronic monitoring), direct (e.g. tracer substances in blood), and subjective measures (e.g. self-report) of patient adherence to cardiovascular medications. Proxy measures for adherence such as physiological indicators or health outcomes were only included if these were reported in association with specific adherence outcomes. 2.2. Search strategy The following electronic databases were searched for English language articles from 1996 to end June wk2, 2012: Cochrane Register of Controlled Trials, Medline, Embase, PsychInfo and Cinahl. The search was arbitrarily limited to studies published after 1996 to include only contemporary evidence. Search strategies (Table S1) were

2.3. Data abstraction and analysis Two authors (TL, JB) independently extracted data from the full papers using predefined data collection forms in a spreadsheet developed and piloted for this review [21,22]. Extracted information included study characteristics, strategy/control details including the underlying behavioural change techniques described, and outcomes and results (adherence, physiological, lifestyle, health, adverse effects and economic). Disagreements were handled in the same way as for study selection. Studies were grouped based on the target/s of the strategy and the outcomes compared independently. Relative risk (of improved adherence) was used as a measure of effect for dichotomous and continuous data [23]. Where possible, outcomes were recalculated, including 95% confidence intervals. Trial authors were contacted for missing information and data. 2.4. Assessment of risk of bias in included studies The methodological quality of the included studies was assessed using the Cochrane “Risk of bias” guidelines [21]. Each of the following attributes was assessed as being either present, absent or unclear: adequate sequence generation, allocation concealment, blinding of participants, personnel, or outcome assessors, report of losses to follow-up and intention-to-treat analysis, and selective outcome reporting. The method used to measure adherence was also taken into consideration, as some methods are more prone to bias than others.

3. Results The search retrieved 772 citations from all sources and 111 full manuscripts were reviewed (see Fig. 1). In total, 16 full-text articles reporting on 14 studies involving a total of 7739 patients were included [24–37]. Table S2 summarises the characteristics of included studies. The characteristics of excluded studies are available on request. 3.1. Participants and setting

3.2. Strategies The strategies tested were grouped first according to whether they were directed at patients, providers or both. The strategies were then categorised according to Michie's ‘Behaviour Change Wheel (BCW)’ [38], a framework for categorising behaviour change interventions. The BCW comprises a behaviour system at the hub involving three essential conditions: capability, opportunity, and motivation, that is encircled by nine intervention functions, aimed at addressing deficits in one or more of these conditions (which Michie labels simply as ‘interventions’) and comprised of one or more behaviour change techniques; around which are positioned seven policy categories, which are broader based population-level strategies that could enable those interventions to occur (which Michie labels simply as ‘policies’) (see Fig. 2). The intervention and policy codes within this taxonomy are provided in Table 1.

2432

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

Fig. 1. PRISMA flowchart of included/excluded studies.

Table 2 displays the intervention and policy coding for the included studies. The more fine-grained behaviour change techniques underlying the interventions are given in Table S3 [39]. In general, the strategies tested were multi-faceted, testing a broad range of interventions and policies in a variety of combinations and using a number of different behavioural change techniques. Education and enablement were the most commonly tested intervention and, in addition to service provision, were frequently supported by policies that incorporated guidelines

to either mandate or recommend practice, as well as using print, telephone or electronic media for communication. Patient incentivisation or stronger coercive measures were not tested. Additionally, no strategy tested legislative or fiscal policies, or those that were designed to change the broader physical or social environment as a means to support the interventions tested. With respect to behaviour change techniques, the use of follow-up prompts, prompting patients and/or physicians to identify potential

Fig. 2. The Behaviour Change Wheel [38].

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440 Table 1 Intervention and policy coding [38]. Category (code) Interventions Education Persuasion Incentivisation Enablement

Training Coercion Restriction

Environmental restructuring Modelling Policies Communication/ marketing Guidelines Fiscal Regulation Legislation Environmental/ social planning Service provision

Definition Increasing knowledge or understanding Using communication to induce positive or negative feelings or stimulate action Creating expectation of reward Increasing means/reducing barriers to increase capability or opportunity beyond education and training and environmental restructuring respectively Imparting skills Creating expectation of punishment or cost Using rules to reduce the opportunity to engage in the target behaviour (or increase the opportunity to engage in competing behaviours) Changing the physical and social context Providing an example for people to aspire to or imitate

Using print, electronic, telephonic or broadcast media Creating documents that recommend or mandate practice Using the tax system to reduce or increase the financial cost Establishing rules or principles of behaviour or practice Making or changing laws Designing and/or controlling the physical or social environment

Delivering a service

barriers to adhering to medications, and providing instructions to patients and/or information on the health link were most commonly tested [see Table S3]. Other behaviour change techniques that are listed in the taxonomy used but have not been included in the strategies tested were providing information about others approval regarding adherent behaviour, providing rewards contingent on adhering to medication, prompting patients to form intentions to

2433

adhere to therapy, prompting patients to use self-talk, or using strategies to prevent relapses of non-adherent behaviour. Strategies specific to the socioeconomic disadvantage of the target populations were included in only five studies but never as standalone strategies. These included medication subsidies (n=5) [24,26,30,32,33], and referral to social services involving employment, housing and transportation (n=2) [24,33]. Only 5 studies explicitly cited any underlying theoretical basis for the strategy designed [25,26,31,36,37]. Only one study [34] tested patient decision support systems. There were no studies in which single (patient level) interventions were tested without being reinforced by at least one additional strategy at a patient or population level. 3.2.1. Patient-directed strategies [25,29,31,35,37] Five studies exclusively directed strategies towards patients. These were patient education sessions about the disease, treatment and healthy lifestyles [25,29,35,37], persuasion through goal setting [35], the use and revision of behavioural contracts [31] or ambulatory blood pressure monitoring reports [25], modelling behaviour via a computer interface [37], and increasing the possibility of patients adhering with prescribed medication regimens via addressing adherence barriers such as lack of motivation or general concerns about taking medications. The number of intervention sessions delivered varied from 5 [31,35] to over 18 [25]. To support these interventions, guideline-based policies were used to direct advice and interviewing [31,35]. Increased patient follow-up at the patient's home or in a dedicated clinic was common. Telephonic [25,29,35,37] and/or print media [29,37] were also used to enhance communication. Additionally, Ogedegbe et al. [31] sought to influence the behaviour of patients via medication-taking contracts. 3.2.2. Physician/practice-directed strategies [28,32] Two studies specifically targeted physician and/or practices to improve patient medication-taking behaviour. Alongside training, the interventions enabled General Practitioners (GPs) to influence patient behaviour by providing treatment algorithms [32], and by instilling competency assessments of participants [28]. Accreditation was used

Table 2 Intervention and policy classification of included studies. Study citation

Interventions Education Persuasion Incentivisation Enablement Training Coercion Restriction Environmental Restructuring Modelling Policies Communication/marketing Guidelines Fiscal Regulation Legislation Environmental/social planning Service Provision Unclassified4 1

Patient

Physician/ practice

[25]

[29]

✓ ✓



✓ ✓



[31]

[35]

[37]



✓ ✓

✓ ✓



[28]

✓ ✓

✓ ✓

✓ ✓



Patient and physician/practice [32]

[24]



✓1 ✓1

✓ ✓ ✓

✓1

[26]











✓ ✓









[35]

[34]

[36]

✓1

✓1 ✓1

✓1

✓1

✓1 ✓3

✓1 ✓3

✓1 ✓1

✓2 ✓1 ✓3 ✓1 ✓2

✓3



✓ ✓

✓1 ✓3

✓1 ✓3

✓3

✓2

✓2 ✓3

✓2 ✓

✓1 ✓2

✓2 ✓3

✓1 ✓b

✓1 ✓c

✓1

✓1

✓3





✓1 ✓3 ✓1







[30]

✓1

✓3 ✓

[27]



✓1

✓1 ✓a

✓1

Directed at the patient. Directed at the patient and physician/practice. Directed at the physician/practice. 4 Interventions and policies that were statements of general exhortation were categorised as Unclassifiable. a. “Patient Counselling”; b. “provide motivation to take medications”; c. “Intensive counselling sessions with clinical pharmacists”; “medication management with patients”; “Diabetes care co-ordinator addressed “issues” relating to health behaviour and health education”. 2 3

2434

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

as an incentive. For the GP hypertension training intervention [32], in addition to education about non-pharmacological and pharmacological treatments (i.e. appropriately titrated, low-cost, single dose drugs), case scenarios were used to model the clinical encounter. For each of the services delivered, guideline-based policies were provided to guide [32] or mandate [28] clinical practice. Rules and principles of Point of Care Testing (PoCT) were regulated through quality assurance and control testing. Doctor-related characteristics associated with the clinical encounter were collected from the patient in the Qureshi study [32]. In this study, the act of explaining the purpose of the drug to the patient was found to be an independent predictor of patient medication adherence in the final multivariable model. However, complexities and regional diversity of general practice and the heterogeneity in usual care (e.g. use of multiple pathology providers) meant that concerns could be raised about generalisability of such strategies [28]. 3.2.3. Patient and physician/practice-directed strategies [24,26,27,30,33,34,36] Seven studies tested strategies that were directed to both physicians/practices and their patients. Multidisciplinary disease or risk management services, providing disease, treatment and lifestyle patient education alongside the use of treatment algorithms and protocols for physicians/prescribers were common. One study focussed on communication techniques for physicians and patients by providing continuing medical education, incentivised via professional accreditation, and offering patients the opportunity to role-play the clinical interaction [36]. Medication adherence was enabled via interventions to reduce patient barriers to adherence that could be classed as either: social (i.e. transportation, housing, employment) [24,33]; financial (i.e. medication costs) [24,26,30,33]; medical (i.e. regimen simplification and visual reminders) [30], and others (e.g. engage, inform, and support in clinical matters such as disease knowledge) [36]. In some cases these were reinforced by patient-directed interventions aimed at empowerment [26,36], stress management [34,36], and self-efficacy [30,33,36]. Other interventions used across the studies included the use of encouragement cards, measured adherence records and telephone follow-up to persuade patients [24,30,36], modelling behaviour through demonstration [30] or through photonovels [36], and training in the use of ambulatory blood pressure monitoring for patients [34] or in communication skills [36]. Finally, restrictions were placed on patients [34] (i.e. type/dose/timing of medicine according to chronotherapeutics) and physicians behaviour [27] (i.e. compulsory in-patient pre-discharge carvedilol dosing) to discourage non-adherence with guideline-recommended therapy. In addition to the provision of practice guidelines, communication to patients and/or physicians via telephone, written or electronic media, and regulation of prescriber behaviour via the implementation of indigent drug programmes were the only other policies described. Process evaluations were reported in three studies [26,30,33] in this category. Specifically, the frequency and duration of disease management visits were recorded in two studies [26,33]. By contrast, Murray et al. reported a log of the intervention pharmacist activities, which revealed that patient education, resolving medication related problems, reminding patients about the importance of medication adherence and obtaining refills, reinforcing physician instructions, communicating with physicians, and encouraging patient lifestyle changes were most frequently performed. 3.3. Outcome measures 3.3.1. Adherence In all but two studies [24,29], adherence was the primary outcome (Table 3). Three studies used more than 1 method to measure adherence, including in each case some measure based on self-report

[24,30,31]. The conceptualisation of adherence included: headcount (i.e. percentage of participants) based on medication utilisation (n = 5) [24–27,33], pre-specified percentage of doses needed to be taken to be classified as adherent (n = 2) [29,37], or the numbers of perfect adherence scores achieved in a survey (n = 3) [28,35,36]. For the Gialamas study [28], as some participants returned more than one survey, the percentage of returned surveys with a perfect adherence score as opposed to the percentage of participants was used. Three studies reported the percentage of doses taken or taken on time averaged across all patients [30–32] and one reported the percentage of prescriptions collected relative to supplied, also as an average across all patients [30]. Finally, two studies reported the average adherence scores from self-report surveys across the populations [30,34], however neither study described the scale nor boundaries of the scores. 3.3.2. Other outcomes The other outcomes collected across the studies are reported in Table 4. 3.3.2.1. Risk of bias in included studies. When adequately reported, the risk of bias of the included studies (Fig. 3), based on the Cochrane “Risk of bias” guidelines [21] was generally low. However, frequent use of self-report measures could have led to overestimates of adherence levels [40]. The majority of studies (n = 11) [26–33,35–37] randomly allocated participants using a computer-generated procedure, however adequate concealment of the allocation was only detailed in 6 studies [27,28,30,31,33,36]. Blinding of participants and personnel was not well reported, though considering the type of strategies reviewed, the ability to do this would have been limited. Blinding of the outcome assessor did not appear to be performed in two studies [25,29], placing these studies at risk of detection bias. Losses to follow-up were reported and taken into account in the analysis in the majority of studies (n = 10) [25,26,28,30–32,34–37]. Losses to follow-up ranged from 3% to 47%. Based on the information obtained from published reports, including protocols where available, and correspondence with authors, a high risk of reporting bias could not be excluded in two studies [30,31]. These did not report all the reported adherence measures that were declared in the protocol. 3.4. Effects of strategies on adherence The effects of the strategies on adherence are summarised in Table 3 and Figs. 4–6. For the ten studies that effect estimates could be calculated, five significantly improved adherence across a large range of values [RR: 1.16–2.69] [26,27,30,33,35]. Four of those studies tested strategies directed towards patients and physicians; no study tested strategies directed exclusively at physicians/practices. The largest increase in relative adherence (RR 2.69, CI 95% 1.59, 4.56) was for a disease management approach to multifactor cardiovascular risk reduction specifically targeted to patients at high risk of cardiovascular disease with limited or no health insurance and low family income [26]. However, as with three other studies examining adherence with multiple medications, such improvements were not generally achieved for all cardiovascular medications investigated. Although it is not possible to determine the degree to which individual interventions and policies are impacting adherence, comparison between the studies suggests that targeting physician/practice prescribing behaviour is an effective intervention. Furthermore, the use of policies that regulate physician/practice behaviour is effective when the strategy is targeted to patients and physicians/practices but not when targeting physicians/practices alone. By contrast, incentivisation of prescribers/practices, environmental restructuring of the patient's and/or physician/practice's social or physical context, modelling physician behaviour, and training patients have not been shown to be effective. Additionally, reducing barriers of or increasing

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

2435

Table 3 Effects of strategies on adherence outcome. Author

Condition

Patient directed strategies Han [25] Hypertension

Measure

Adherence conceptualisation

Effect on adherence (strategy, control)

Self-reporta

Headcount-Medication Utilisationb Headcount-thresholdb (95–100% of prescribed doses)

Between: no statistical difference (71.5%, 70.7%); Within: statistically higher (65% to 71.5%, P b 0.05); Between: no statistical difference (85%, 73.9%)

Headcount-thresholdb (≥80% prescribed doses) Doses-takingd

Between: no statistical difference (51%, 49%)

Lopez-Cabezas Heart failure [29]

Pill count

Martin [37]

Pill count

Hypertension

Ogedegbe [31] Hypertension

Zhao-2009 [35]

Angina/ myocardial infarction

MEMSc

Self-reporta Self reporta

Physician/practice directed strategies Self-reporta Gialamas [28] Dyslipidemia or anticoagulant/ anti-lipidemic therapy Qureshi [32] Hypertension MEMSc Patient & physician/practice directed strategies Cooper [36] Hypertension Self-reporta

Dennison [24]

Hypertension

2 × Self-reporta

Haskell [26]

High CVDc risk Heart failure

Self-report

Krantz [27] Murray [30]

Heart failure

Pill count MEMS

d

Prescriptions

Rothman [33] Zang [34]

High CVDc risk Hypertension

Self-report Self-report: Self reporta

Comments

Between: no statistical difference (57% 43%)

Intervention halts declining adherence

Not reported Headcount-thresholdb (100% adherence score)

Not reported Between: Statistically higher (86%, 51%, P b 0.001)

Headcount-thresholdb (100% adherence score)

Between: no statistical difference (39.3%, 37%, 90%CI −0.1%–4.6%); non-inferior to control (Pb0.001)

Doses-takingd

Between: statistically higher (48.1%, 35.8%, Pb0.05)

Headcount-thresholdb (100% ‘adherence score’) Headcount-medication utilisationb,

Between: no statistical difference compared to minimal intervention

Headcount-medication utilisationb Headcount-medication utilisationb Doses-takingd schedulingd

Between: statistically higher for aspirin (69%, 26% P b 0.0001), statins (71%, 33% P b 0.0001), ACEI/ARBsc (65%,48% P b 0.05), BBc (33%, 29% P b 0.05) only Between: BBc utilisation statistically higher (96.2%, 47.8%, P b 0.001) only

Refill: supplies received relative to prescribed Overall usef Headcount-medication utilisationb Taking compliancef

Secondary endpoint; dissipation of effects

Non-inferiority trial

Between: statistically higher (P b 0.05)e.

Secondary end-point; dissipation of effects

c

Between: Taking: statistically higher except spironolactone and ARBs . Scheduling: statistically higher except loop diuretics, digoxin, spironolactone and ARBsc. Between: Refill: statistically higher overall (109.4%,105.2% P = 0.007); Statistically lower for ACEIsd (96.9%,98.3% P = 0.018)

Small sample size Dissipation of effects

No statistical difference Between: statistically higher for aspirin (91%, 58% P b 0.0001) Between: statistically higher (14.6, 13.1, P = 0.001)

a

Dennison [24]: Hill Bone Medication Compliance Subscale (validated) and Hypertension Medication Utilisation; Han [25]: (Medication-taking); Zang [34]: Chronotherapeutic Compliance Questionnaire for hyperpietic (seven day recall); Zhao [35]: 7-day recall; Haskell [26]: “Compliance to therapies”; Ogedegbe [31], Cooper [36]: Morisky-4 (validated); Gialamas [28]: MARS-5 (validated). b Headcount-Medication utilisation is the percentage of participants who have reported use of the medication; Headcount-threshold is the percentage of participants who have been classified as adherent based on a pre-specified threshold. For Gialamas [28], some respondents returned >1 survey, therefore reported as % surveys returned. c BB: beta blockers; ARBs: angiotensin 2 receptor blockers; ACEIs: angiotensin converting enzyme inhibitors; CVD: cardiovascular disease; MEMS: Medication Event Monitoring System. d Doses-taking: the average % doses taken as prescribed across participants; Scheduling: the average % doses taken on time across participants. e Adherence rates not available. f Figures based on an average “adherence score” across respondents; limits/scale of score not provided.

the means for patients to adhere to medications does not appear to be effective unless used in conjunction with other physician/practice targeted interventions and policies. Finally, despite the frequency with which it has been trialled, the effect of patient education remains unclear. 3.4.1. Time course of effect of strategies on adherence Two studies [24,29], conducted over 12 months and 5 years respectively, reported a marked dissipation of the effects on adherence by the end of follow-up. In both studies, a loss in statistical power due to death and/or incarceration was identified as a potential cause. Other possible explanations included a lack of ongoing resource capacity, reduced

participant acceptance of ongoing visits, and a decreased uniqueness of the intervention over time. In both studies, a reinforced strategy was suggested throughout the time-course of therapy. In contrast, Ogedegbe et al. [31] analysed the observed differences in adherence patterns across time. Although there was a significant drop between baseline and post-intervention in the usual care group, a slight but non-significant increase in the intervention group suggests that there may have been an effect of the strategy to slow the decline in medication adherence. Consistently increasing adherence rates over the course of the strategy were reported in three studies [27,28,35]. One study [30] measured adherence 3 months after the strategy had ceased and in

2436

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

Table 4 Other outcomes collected: intervention effects. Study citation

Patient

Physician/practice

Patient & physician/practice

[25]

[29]

[31]

[35]

[37]

[32]

[28]a

[24]

[26]

[27]

[30]

[33]

[34]

[36]

Physiological BPb control Mean SBPb Mean DBPb Lipids Glucose/HbA1c INR Echocardiogram

– – – – – – –

– – – – – – –

– ✘ ✘ – – – –

– – – – – – –

– – – – – – –

– – – – – – –

– – – ✓ not HDLb ✓ ✘ –

✓c ✓c ✓c – – – ✓

– ✓ ✓ ✓ ✓ – –

– ✘ ✘ – ✓ – ✓

– – – – – – –

– ✓ ✓ ✘ ✓ – –

✓ ✓ ✓ – – – –

✘ ✘ ✘ – – – –

Health outcomes Health service utilisation Mortality QoLb

– – –

✘ ✓ ✘

– – –

✘/✓d – –

– – –

– – –

– – –

– – –

– – ✘

✓ – ✘

✓ – ✘

✘ – –

– – –

– – –

Other lifestyle Smoking/alcohol Nutrition Physical activity Stress

✘ – ✘ –

– – – –

– – – –

✓ ✓ ✓ –

– – – –

– – – –

– ✘ ✘ –

✘ ✘ – –

✘ ✓ ✓ ✘

– – – –

– – – –

– – – –

✘ ✓ ✓ –

– – – –

Adverse effect Adverse event





























Economic Cost comparison Cost effectiveness

– –

– –

– –

– –



– ✘

✘/✓e ✓ ACRb only

– –

– –

– –

✘ –

– –

– –

– –

✓ Statistically significant improvement in favour of the intervention; ✘ No statistical difference between groups; – not measured. a Non-inferiority trial design. b BP (blood pressure); SBP (systolic blood pressure); DBP (diastolic blood pressure); QoL (quality of life); HDL (high density lipoprotein); ACR (albumin creatinine ratio). c Effect not sustained to 5 years. d Not statistically significant for hospital re-admissions; statistically lower clinic visits for control arm. e Dependent on test and costs compared.

Fig. 3. Risk of bias: summary representing the proportion of studies with each of the authors' judgement about each item risk. High/Low/Unsure categories determined according to recommendations from the Cochrane “Risk of bias” guidelines [21].

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

Fig. 4. Effect estimates (relative risk): patient-targeted strategies.

which case the initial effects of the strategy on adherence had dissipated. 3.5. Effect of strategies on other outcomes When reported, significant improvements in physiological and other lifestyle outcomes were generally seen across the studies where there were significant improvements in directly measured adherence [26,27,30,33–35]. In contrast no effects were generally reported in relation to health outcomes, particularly quality of life (disease and/or non-specific) and adverse events. In the Zhao study, the number of clinic visits by intervention patients was statistically higher potentially due to patient sensitisation to health care needs and medical-help seeking behaviour. Three studies addressed the economic implications of the strategies tested [28,30,32]. In the study by Murray et al., comparison of the direct and variable costs found an overall cost savings in the intervention group of $2960 (CI, -$7603 to 1338) per patient [30]. However, due to the large variability in costs, no statistical difference between the groups was found. The comparative cost analysis of PoCT and pathology laboratory testing, using a societal perspective, in the Gialamas study varied by the type of testing investigated [28]. Generally, PoCT led to small savings to patients and families in terms of indirect costs and patient travel. Regarding direct health care sector costs, PoCT testing was associated with significantly higher costs for GP consultations and/or pharmaceuticals for INR and lipid testing. 4. Discussion This review assessed the effects of strategies to improve adherence to medications prescribed for the prevention of CVD in socioeconomically disadvantaged groups. In the reviewed studies, disadvantage was defined by place of residence, education and income. In fourteen studies that fulfilled our inclusion criteria, the strategies tested were mostly multi-factorial varying in composition, intensity and follow-up, and generally lacked any specified underlying theory [39]. Strategies

2437

simultaneously directed at patients and physicians/practices, targeting physician prescribing behaviour as well as interventions to reduce patient social (i.e. transportation, housing, employment), financial (i.e. medication costs) and treatment-related barriers to enable patients to adhere to prescribed therapy have been found to be most effective, resulting in significant improvements in the risk of adherence of 16% to 169%. Importantly these interventions were generally supported by policies that incorporated practice guidelines and/or regulated physician/practice behaviour, as well as using telephone follow-up or print and electronic media. Furthermore, the most successful strategy was targeted to and specifically tested among a socioeconomically disadvantaged population. The apparently synergistic effects of physician- and patient-targeted strategies on medication adherence in this review suggest important overlapping roles of physicians and patients upon medication-taking behaviour within socioeconomically disadvantaged groups. This finding is in contrast to a recent systematic review evaluating the role of physicians in improving adherence to cardiovascular and diabetic medications among general populations [41], which found physician noninvolved studies more likely to show larger improvements in adherence compared to physician involved studies. Such disparity potentially reflects the differences that exist in adherence behaviour between social groups, and supports the notion of designing strategies specific for socioeconomically disadvantaged populations. Despite testing strategies within socioeconomically disadvantaged groups, relatively few interventions or supporting policies addressed directly the dimensions for socioeconomic disadvantage potentially causing non-adherence such as affordability and other costs associated with accessing treatment. For the five studies that did, three demonstrated the largest relative improvements in adherence [26,30,33]. Specifically, these interventions included reducing medication costs, and improving access to services such as insurance and transportation. The extent to which these specific socioeconomic measures contributed towards the measured effect on adherence, relative to the other intervention components, is not clear and could in future studies be addressed by mixed-methods process evaluations [42]. In this review, the generally positive effect of the more successful strategies on physiological and lifestyle outcomes suggests that improvements in adherence may result in significant clinical benefits. By extension, and despite the lack of evidence on effect on health outcomes, addressing inequalities in medication adherence should be expected to reduce some of the inequality in health status. Of particular concern for resource-poor settings is the scarcity of cost-effectiveness data to guide implementation decisions. In this review, cost-effectiveness was not demonstrated in the two studies reporting this outcome [28,32]. In one of those studies [28], the costs were particularly sensitive to hospital admission data. In this respect, the effects of the strategies on increasing costs due to increasing health-seeking behaviour, as suggested in the Zang study [34], warrant further investigation. Additionally, considering the lack of intervention effects on quality of life, the impact of strategies on effectiveness measures such as Quality Adjusted Life Years Gained (QALYs) needs clarification. Furthermore, in light of the chronic nature of CVD and associated therapies, the effects on cost-effectiveness resulting from the dissipation of adherence effects post-intervention seen in some of the studies reviewed require further investigation to inform future implementation decisions. The results of this review have identified gaps in research on the effectiveness of adherence-enhancing strategies in socioeconomically disadvantaged groups. First, although the strategies were all aimed at socioeconomically disadvantaged groups, 13 of the 14 studies did not recruit based on such measures. As a result, the populations studied were largely though not exclusively disadvantaged. Despite this, only one study conducted a sub-group analysis to ascertain any gradient or gap between populations in terms of adherence. Combined with the low number of adequately designed randomised controlled

2438

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

Fig. 5. Effect estimates (relative risk): patient and physician-targeted strategies.

trials relative to those within more advantaged populations, a need exists to either implement more research in targeted populations or utilise gap methods, such as adequately designed sub-group analysis, to evaluate the relevance and applicability of strategies in different socioeconomic groups.

With respect to strategies, priority needs to be given to evaluating less resource intensive interventions, particularly those that have proven to be effective in general populations, such as simplifying dosage regimens, and using behaviour change techniques not presently tested. Despite its popularity, the role of patient education

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

2439

Abbreviations CVD cardiovascular disease GP general practitioner ICER Incremental cost effectiveness ratio HIV/AIDS human immunodeficiency virus/acquired immunodeficiency syndrome INR international normalised ratio MEMS medication events monitoring system SES socioeconomic status PoCT point of care testing QALY Quality adjusted fife years SD standard deviation

Author contributions All authors were involved in initial conception of the paper and in the design of the systematic review protocol. All authors contributed to the preparation of the final manuscript. TL and SJ take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. Financial disclosure

Fig. 6. Effect estimates (relative risk): physician-targeted strategies.

also requires clarification. Furthermore, the effect of policies that were not uncovered in this review, such as fiscal, legislative, and environmental restructuring or social planning policies, requires investigation. As the effects of some strategies appeared to dissipate over time, long-term studies are needed to assess the extent to which interventions need to be reinforced over time. This review must be viewed in light of its limitations. First, outcome-reporting bias cannot be excluded; authors of excluded trials were not contacted to see if adherence was measured but not reported. Second, the search was restricted to English language articles published from 1996; the study results may not therefore represent less contemporary non-English publications. Finally, although authors were contacted, the relative effect on adherence of four studies could not be calculated [24,31,34,36].

5. Conclusion Medication adherence is a complex health behaviour influenced by patient beliefs, attitudes and expectations and which is subject to the constraints on resources faced by individuals and groups. Whilst the evidence about strategies to improve adherence in general populations have shown, at best, to have had moderate effectiveness and cost-effectiveness, there is emerging evidence that among socioeconomically disadvantaged groups, strategies that simultaneously target patients and physicians are highly effective. These strategies are typically complex: incorporating interventions targeting physician prescribing behaviour, addressing social, financial, and treatment-related barriers to adherence, and are supported by broad-based guidelines, communication and regulating policies. Given such potential complexity and consequent resource implications, comprehensive process evaluations as well as evidence regarding costs and cost-effectiveness of these strategies are urgently needed. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.ijcard.2013.01.049.

National Health and Medical Research Council (NHMRC) Project grant number: 1004623. TL is the recipient of an NHMRC Post-graduate scholarship and Vice-chancellor's research scholarship (University of Sydney) and NHMRC Capacity Building Grant (571372). AC and SJ are funded by Senior Research Fellowships from the NHMRC. JR is funded by a Postdoctoral Fellowship co-funded by the NHMRC and the National Heart Foundation (632933). No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Acknowledgements PEAK study team members (Anushka Patel, John Rose, Noel Hayman, Deborah Blair, Ashley Page, Anne-Marie Eades, Frances Stewart, Barry Fewquandie, Chris Lawrence), Kanyini Vascular Collaboration. References [1] DiMatteo MR. Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med Care 2004;42:200–9. [2] Sabate E. Adherence to long term therapies: evidence for action. (available at http:// www.who.int/chronic_conditions/adherencereport/en/)Geneva, Switzerland: World Health Organisation; 2003. [3] Humphrey K, Weeramanthri T, Fitz J. Forgetting compliance. Northern Territory University Press; 2001. [4] Haynes RB, Ackloo E, Sahota N, McDonald HP, Yao X. Interventions for enhancing medication adherence. [update of Cochrane Database Syst Rev. 2005;(4):CD000011; PMID: 16235271]Cochrane Database Syst Rev 2008:CD000011. [5] Suhrcke M, Boluarte TA, Niessen L. A systematic review of economic evaluations of interventions to tackle cardiovascular disease in low- and middle-income countries. BMC Public Health 2012;12:2. [6] Horne R, Weinman J, Barber N, Elliott R, Morgan M. Concordance, adherence and compliance in medicine taking. Report for the National Co-ordinating Centre for NHS Delivery and Organisation R & D (NCCSDO); 2005. [7] Pound P, Britten N, Morgan M, et al. Resisting medicines: a synthesis of qualitative studies of medicine taking. Soc Sci Med 2005;61:133–55. [8] Lehane E, McCarthy G. Intentional and unintentional medication non-adherence: a comprehensive framework for clinical research and practice? A discussion paper. Int J Nurs Stud 2007;44:1468–77. [9] Kalichman SC, Grebler T. Stress and poverty predictors of treatment adherence among people with low-literacy living with HIV/AIDS. Psychosom Med 2010;72: 810–6. [10] Parashar S, Palmer AK, O'Brien N, et al. Sticking to it: the effect of maximally assisted therapy on antiretroviral treatment adherence among individuals living with HIV who are unstably housed. AIDS Behav 2011;15:1612–22. [11] Wong MC, Jiang JY, Griffiths SM. Factors associated with compliance to thiazide diuretics among 8551 Chinese patients. J Clin Pharm Ther 2011;36:179–86.

2440

T.-L. Laba et al. / International Journal of Cardiology 167 (2013) 2430–2440

[12] Shimony A, Zahger D, Ilia R, Shalev A, Cafri C. Impact of the community's socioeconomic status on characteristics and outcomes of patients undergoing percutaneous coronary intervention. Int J Cardiol 2010;144:379–82. [13] Wu JR, Moser DK, Chung ML, Lennie TA. Predictors of medication adherence using a multidimensional adherence model in patients with heart failure. J Card Fail 2008;14:603–14. [14] Choudhry NK, Avorn J, Glynn RJ, et al. Full coverage for preventive medications after myocardial infarction. N Engl J Med 2011;365:2088–97. [15] Schedlbauer A, Davies P, Fahey T. Interventions to improve adherence to lipid lowering medication. Cochrane Database Syst Rev 2010:CD004371. [16] Schroeder K, Fahey T, Ebrahim S. Interventions for improving adherence to treatment in patients with high blood pressure in ambulatory settings. Cochrane Database Syst Rev 2004:CD004804. [17] Tugwell P, Petticrew M, Kristjansson E, et al. Assessing equity in systematic reviews: realising the recommendations of the Commission on Social Determinants of Health. BMJ 2010;341:c4739. [18] Al-Aqeel S, Al-Sabhan J. Strategies for improving adherence to antiepileptic drug treatment in patients with epilepsy. Cochrane Database Syst Rev 2011:CD008312. [19] Welch V, Tugwell P, Petticrew M, et al. How effects on health equity are assessed in systematic reviews of interventions. Cochrane Database Syst Rev 2010:MR000028. [20] Lefebvre C, Manheimer E, Glanville J. In: Higgins J, Green S, editors. Cochrane handbook for systematic reviews of interventions version 502 (updated September 2009). The Cochrane Collaboration; 2009 [Available from www.cochrane-handbook.org The Cochrane Collaboration]. [21] Higgins J, Altman D. Chapter 8: assessing risk of bias in included studies. In: Higgins J, Green S, editors. Cochrane handbook for systematic reviews of interventions version 501 (updated September 2008). Oxford, UK: The Cochrane Collaboration; 2008 Available from www.cochrane-handbook.org; 2008. [22] Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med 2009;6:e1000100. [23] Cochrane handbook for systematic reviews of interventions. Available from www.cochrane-handbook.orgThe Cochrane Collaboration; 2009. [24] Dennison CR, Post WS, Kim MT, et al. Underserved urban African American men: hypertension trial outcomes and mortality during 5 years. Am J Hypertens 2007;20: 164–71. [25] Han H-R, Kim J, Kim KB, et al. Implementation and success of nurse telephone counseling in linguistically isolated Korean American patients with high blood pressure. Patient Educ Couns 2010;80:130–4. [26] Haskell WL, Berra K, Arias E, et al. Multifactor cardiovascular disease risk reduction in medically underserved, high-risk patients. Am J Cardiol 2006;98:1472–9. [27] Krantz MJ, Havranek EP, Haynes DK, Smith I, Bucher-Bartelson B, Long CS. Inpatient initiation of beta-blockade plus nurse management in vulnerable heart failure patients: a randomized study. J Card Fail 2008;14:303–9.

[28] Gialamas A, Yelland LN, Ryan P, et al. Does point-of-care testing lead to the same or better adherence to medication? A randomised controlled trial: the PoCT in General Practice Trial. Med J Aust 2009;191:487–91. [29] Lopez Cabezas C, Falces Salvador C, Cubi Quadrada D, et al. Randomized clinical trial of a postdischarge pharmaceutical care program vs regular follow-up in patients with heart failure. Farm 2006;30:328–42. [30] Murray MD, Young J, Hoke S, et al. Pharmacist intervention to improve medication adherence in heart failure: a randomized trial. Ann Intern Med 2007;146:714–25. [31] Ogedegbe G, Chaplin W, Schoenthaler A, et al. A practice-based trial of motivational interviewing and adherence in hypertensive African Americans. Am J Hypertens 2008;21:1137–43. [32] Qureshi NN, Hatcher J, Chaturvedi N, Jafar TH. Effect of general practitioner education on adherence to antihypertensive drugs: cluster randomised controlled trial. BMJ 2007;335:1030. [33] Rothman RL, Malone R, Bryant B, Sequist TD. A pharmacist-driven intervention improves diabetes care. J Clin Outcomes Manage 2005;12:188–9. [34] Zang X, Liu J, Chai Y, Wong FKY, Zhao Y. Effect on blood pressure of a continued nursing intervention using chronotherapeutics for adult Chinese hypertensive patients. J Clin Nurs 2010;19:1149–56. [35] Zhao Y, Wong FK. Effects of a postdischarge transitional care programme for patients with coronary heart disease in China: a randomised controlled trial. J Clin Nurs 2009;18:2444–55. [36] Cooper LA, Roter DL, Carson KA, et al. A randomized trial to improve patient-centered care and hypertension control in underserved primary care patients. J Gen Intern Med 2011;26:1297–304. [37] Martin MY, Kim YI, Kratt P, et al. Medication adherence among rural, low-income hypertensive adults: a randomized trial of a multimedia community-based intervention. Am J Health Promot 2011;25:372–8. [38] Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci 2011;6:42. [39] Abraham C, Michie S. A taxonomy of behavior change techniques used in interventions. Health Psychol 2008;27:379–87. [40] Osterberg L, Blaschke T. Adherence to medication. [see comment]. N Engl J Med 2005;353:487–97. [41] Cutrona SL, Choudhry NK, Stedman M, et al. Physician effectiveness in interventions to improve cardiovascular medication adherence: a systematic review. J Gen Intern Med 2010;25:1090–6. [42] Jan S, Usherwood T, Brien JA, et al. What determines adherence to treatment in cardiovascular disease prevention? Protocol for a mixed methods preference study. BMJ Open 2011;1:e000372.