Article Text

Download PDFPDF

Improving medication adherence in patients with cardiovascular disease: a systematic review
  1. Rosemary Hines Fuller1,2,
  2. Pablo Perel3,
  3. Tamara Navarro-Ruan4,
  4. Robby Nieuwlaat4,
  5. Robert Brian Haynes4,
  6. Mark D Huffman1,2
  1. 1 Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA
  2. 2 Division of Cardiology, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA
  3. 3 Non-communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK
  4. 4 Department of Health Research Methods, Evidence and Impact, McMaster University, Hamilton, Canada
  1. Correspondence to Dr Mark D Huffman, Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USA; m-huffman{at}


Objective To evaluate and compare the effect of interventions for improving adherence to medications for atherosclerotic cardiovascular disease (ASCVD) secondary prevention.

Methods We extracted eligible trials from a 2014 Cochrane systematic review on adherence for any condition. We updated the search from CENTRAL, Medline, Embase, PsycINFO, CINAHL, Sociological Abstracts and trial registers through November 2016. Study reports needed to be from a randomised controlled trial, incorporate participants identified as having ASCVD and interventions aimed at improving adherence to medicines for secondary prevention of ASCVD and measure both adherence and a clinical outcome. Two reviewers independently determined the eligibility of studies, extracted data and conducted a narrative synthesis.

Results We identified 17 trials (n=17 448 participants). Most trials had high risk of bias in at least one domain. The intervention group adherence rates ranged from 44%to99% and the comparator group adherence rates ranged from 13% to 96%. Three distinct interventions reported improvements in both adherence and clinical outcomes: short message service (65% vs 13% of participants with high adherence in the intervention vs control group), a fixed-dose combination pill (86% vs 65% adherence, risk ratio of being adherent, 1.33; 95% CI 1.26 to 1.41) and a community health worker-based intervention (97% in the intervention group compared with 92% in the control group; OR=2.62, 95% CI 1.32 to 5.19).

Conclusions We identified three interventions that demonstrated improvements in adherence and clinical outcomes. Ongoing, longer-term trials will help determine whether short-term changes in adherence can be maintained and lead to differences in clinical events.

  • medication adherence
  • coronary artery disease
  • systemic review

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Cardiovascular diseases (CVDs) are the leading cause of death worldwide. In 2015, an estimated 17.9 million people died from CVDs.1 2 At least three quarters of those deaths occurred in countries with limited resources and one-quarter to one-third of CVD events occurred in individuals with prevalent CVD.1 3 Aspirin, statins and blood pressure lowering drugs can substantially reduce CVD morbidity and mortality among individuals with prevalent atherosclerotic CVD (ASCVD).4

However, those secondary prevention medications fail to achieve their therapeutic potential when non-adherence or discontinuation rates are high.5 Medication non-adherence is often defined as taking less than 80% of prescribed doses, but it can also include taking too many doses and is associated with an increased risk for poor health, adverse clinical events and mortality.6 Reasons for non-adherence to ASCVD medications are varied and can include costs, lack of availability, side effects, misunderstandings of potential benefits and harms, among others.7 8 While medication non-adherence is a global problem, patients in resource-limited settings are affected to a greater degree. Adherence rates to long-term pharmacotherapy in those low-income countries are as low as 10% in patients with previous CVD.6 As the number of individuals with ASCVD rises globally, non-adherence will likely prove a salient cause of poor health.2

The 2014 Cochrane systematic review, ‘Interventions for enhancing medication adherence,’ assessed the effects and outcomes of interventions intended to improve patient adherence to prescribed medications for a variety of medical conditions.9 While several interventions to improve medication adherence have been tested in high-income countries, many are marginally effective at best, resource-intensive and not readily scalable to have a global reach. Further, strategies to improve adherence to ASCVD pharmacotherapy in resource-limited areas remain understudied and undeveloped and may be relevant to low-resource settings in high-income countries.

The objective of this systematic review is to evaluate and compare the effects of strategies to improve medication adherence in conjunction with clinical outcomes for the secondary prevention of ASCVD. Ultimately, our aim is to draw conclusions about the effectiveness of these interventions for adaptation, implementation, scalability and sustainability globally.


Data sources

Our systematic review builds on the broader 2014 Cochrane review on strategies to improve medication adherence and serves as a focused update.9 We searched the included studies from the systematic review by Nieuwlaat et al for trials evaluating strategies for improving adherence in participants with prevalent ASCVD. To supplement these data, we searched the following sources from 2013 through November 2016 and placed no restrictions on language of publication.

  • Cochrane Central Register of Controlled Trials (CENTRAL)

  • MEDLINE (Ovid)

  • EMBASE (

  • CINAHL (via EBSCO)

  • PsycINFO (via Ovid)


  • Sociological Abstracts (via ProQuest)

A detailed search strategy can be found in online supplementary appendix 1. We identified other potentially eligible trials or ancillary publications by searching the reference lists of included trials, systematic reviews, meta-analyses and health technology assessment reports. We also contacted study authors of included or registered trials to identify any further studies we may have missed. We supplemented this search with hand searching, including major trials published in 2017.

Supplementary file 1

Study selection

We (RHF, PP, MDH) reviewed all titles retrieved by the literature search in duplicate. Titles needed to appear potentially relevant to the study area. We independently assessed abstracts against four criteria to determine: first, that the study was a randomised controlled trial; second, that a defined group of patients were identified as having ASCVD; third, that the intervention aimed at improving adherence to medicines specifically for secondary prevention of ASCVD and fourth, that the trial measured both adherence and a clinical outcome or risk factor (eg, blood pressure). We then independently assessed and retrieved studies for their suitability for inclusion in the narrative synthesis. Differences were resolved by discussion with reference to a third author (TN-R) if necessary. We present a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram showing the process of study selection.

Data extraction

We extracted data (RHF, PP, MDH) in duplicate on to an electronic database based on a previously piloted data extraction form. Extracted data included methodology, participants, interventions, outcomes, risk of bias, intervention details, comparator group, adherence outcomes and clinical outcomes. We imported data from the 2014 Cochrane review by Nieuwlaat et al into the updated database.9


The predefined primary outcome was adherence to ASCVD-related medication (antiplatelet, blood pressure lowering and lipid lowering drugs) measured by each study’s authors. Secondary outcomes included clinical measures and risk factors including systolic blood pressure, low-density lipoprotein (LDL) cholesterol, investigator-defined adverse events and fatal and non-fatal ASCVD events.

Quality assessment

We judged risk of bias as ‘low’, ‘high’ or ‘unclear’ across individual bias domains as described in the Cochrane Handbook for Systematic Reviews of Interventions using the Cochrane Risk of Bias Tool.10 Differences in risk of bias assessment were resolved by discussion and contribution of a third author (TN-R). We present the overall quality of the evidence for adherence outcomes as well as ASCVD events, change in blood pressure and change in cholesterol according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, which takes into account issues related to internal validity (risk of bias, inconsistency, imprecision, publication bias) and to external validity, such as directness of results.11

Data synthesis

Due to the substantial heterogeneity in participant selection criteria, medication regimens, interventions, adherence measures and clinical outcome measures, we did not pursue quantitative pooling. Instead, we conducted a narrative synthesis of the included studies.10


Search results

Figure 1 includes the PRISMA flowchart of included studies. After deduplication, we screened 1096 titles and abstracts, which yielded 90 potentially relevant studies. Twelve met the inclusion criteria. We included 5 from the 2014 Cochrane review for a total of 17 trials randomising 17 448 participants.12–27 Online supplementary appendix 2 includes the list of excluded studies and reasons for exclusion.

Supplementary file 2

Figure 1

PRISMA flowchart of included studies. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Study characteristics

Online supplementary table 1 includes detailed characteristics of included studies. The median sample size was 623  participants (range: 62–5855 participants). Four trials were small (<200 participants) and were performed in a single centre or country.15 16 18 19 Eleven trials were conducted in high-income countries,12 14 16 17 20–22 24 25 27 two trials in upper middle-income countries (Brazil and Malaysia) 15 ,19 ,28 and two in lower middle countries (India and Pakistan).18 23 Two trials were conducted in a mix of high-income, upper middle-income and lower middle-income countries.13 26 The mean age across 14 trials ranged from 54 to 68.8 years and the proportion of women in each study ranged from 2% to 37%. Three trials did not report gender or age.13 16 20

Supplementary file 3

Ten trials included participants following acute coronary syndrome and myocardial infarction.12–15 17 19 21 23 27 Two trials included participants following stroke only.18 20 Two trials included participants following percutaneous coronary intervention.16 22 Three trials included a combination of participants with either established disease (coronary, cerebrovascular or peripheral vascular) or without established disease but at high predicted risk for an incident ASCVD event.24–26


The intervention types varied among trials. Four trials used a multifaceted intervention, comprised of discharge educational materials, structured and early follow-up clinical evaluations and refill reminders.12 15 17 23 Four trials used a fixed-dose combination or polypill intervention.13 24–26 Two trials used discharge instructions alone,16 21 and three  trials studied short message service reminders.18 19 One trial reported on a telephone follow-up intervention,22 and one trial studied a motivational interviewing intervention22 28 One trial used a full-prescription coverage intervention14 and another used a combination of incentives and social support.27 Adherence was assessed through self-report,12 13 15 16 18–21 23–26 pharmacy data,14 17 22 25 pill count13 and electronic pill bottle opening.27

Risk of bias

Figure 2 and online supplementary figure 1 demonstrate trial-specific risk of bias assessment.  Fifteen trials had a low risk of selection bias based on reported methods of sequence generation and ten trials had a low risk of selection bias based on reported methods of allocation concealment. All others had an unclear risk of selection bias. No trials specifically blinded the study personnel and participants for a high risk of performance bias. Ten trials had a low risk of detection bias using blinded outcome assessors for both adherence and clinical outcome measures. Seven  trials had a high risk of detection bias. Eleven trials had low risk of reporting bias based on previously published protocols and adherence to those protocols.

Figure 2

Risk of bias summary.

Effects on adherence and clinical outcome

Table 1 describes the summary of findings using the GRADE framework and online supplementary table 2 shows the intervention effects on both adherence and clinical outcomes for each trial. None of the nine trials (n=9051 participants) that evaluated ASCVD events demonstrated an effect of the intervention on these outcomes. On the other hand, none of the trials were designed or sufficiently powered to detect a difference in ASCVD events. Using the GRADE framework, the quality of evidence for this outcome is very low, suggesting that the ‘true effect is likely to be substantially different from the estimate of effect’.11 The intervention group adherence rates ranged from 44% to 99% and the comparator group adherence rates ranged from 13% to 96%. However, the definitions and presentation of summary effects differed across trials, which precludes pooling for and reporting of summary relative effect estimates. Based on the GRADE framework, there was moderate quality evidence that adherence interventions led to improved medication adherence (12 out of 17 randomised controlled trials (n=17 448  participants) with a significant effect on adherence), which suggests that ‘true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different’.11

Table 1

Summary of findings

Data demonstrating changes in blood pressure and cholesterol were more variable and less robust. The mean change in systolic blood pressure ranged across control groups from −13.2 mm Hg to 7.3 mm Hg and from −22.0 mm Hg to 0.3 mm Hg across intervention groups. The mean change in LDL cholesterol ranged across control groups from −25.1 mg/dL to 2.2 mg/dL and from −33.2 mg/dL to 5.3 mg/dL across intervention groups. The quality of evidence was low for the effect on systolic blood pressure due to study limitations and heterogeneity and very low for the effect on LDL cholesterol due to study limitations, heterogeneity and imprecision.

Three trials with three distinct interventions demonstrated an effect on both adherence and one or more clinical outcomes.19 23 26 The interventions in each of these trials included a short message service reminder, a community health worker-based intervention and a fixed-dose combination intervention. In Khonsari et al 2015 study, participants in the intervention group received a short message service reminder before every intake of cardiovascular medication for 8 weeks following discharge after acute myocardial infarction.19 Two of every three (65%) participants in the intervention group who received SMS reminders had a high adherence level compared with approximately one of every seven (13%) participants in the usual care group. SPREAD 2016 demonstrated improved adherence among the community health worker-based intervention group compared with usual care (97% in the intervention group compared with 92% in the control group; OR=2.62, 95% CI 1.32 to 5.19) as well as lower mean (SD) systolic blood pressure (124.4 (13.5) mm Hg versus 128.0 (15.9) mm Hg; P=0.002) and body mass index (24.4 (3.7) kg/m2 versus 25.0 (3.8) kg/m2, P<0.0001) in the intervention group.23 In the UMPIRE trial, the fixed-dose combination group had improved adherence versus usual care (86% vs 65%; risk ratio of being adherent, 1.33; 95% CI 1.26 to 1.41; P<0.001) with concurrent reductions in SBP (−2.6 mm Hg; 95% CI −4.0 to –1.1 mm Hg) and LDL-C (−4.2 mg/dL; 95% CI −6.6 to –1.9 mg/dL).26 All three trials had high risk of bias for at least two domains.


In this systematic review, we identified 17 trials that evaluated the effect of interventions to improve adherence to medications for the secondary prevention of ASCVD. Other reviews thus far have focused on a range of disease processes or included trials that did not necessarily report clinical outcomes in addition to adherence outcomes.29 Our review builds on existing literature by demonstrating improvements in medication adherence in some trials in conjunction with concurrent benefits in clinical and health outcomes.

This review highlights several key findings, especially in thinking about scalability of adherence interventions in low-income and middle-income countries. The 2014 Cochrane adherence review, for which this review is a focused update, found that interventions to improve adherence to medications have been mostly complex in nature, which makes adaptation, implementation, scalability and sustainability difficult in practice.9 The review found that of the 182 trials, only five had the lowest risk of bias and reported improvements in both adherence and clinical outcomes. Our present review demonstrated differences in the effects of interventions to improve adherence in this study population. This may be attributed to our focus on ASCVD, compared with the inclusion of any medical condition in the previous review. For instance, interventions aimed at improving adherence to psychiatric drugs may be more complex and less likely to have immediate success. Additionally, the trials in our review reflect the evolution towards simpler and technology-enabled interventions, elements that may explain our different findings. For example, polypills and SMS reminders, two of the interventions with improvements in both adherence and clinical outcome, were in earlier stages of development at the time of the 2014 review. Low-cost, relatively simple and—most importantly—effective interventions could bring scalable adherence solutions to low-income and middle-income countries and other low-resource settings, which are disproportionately affected by both CVD and health system resource limitations.

There are important limitations in this review. We found the interventions to be diverse, ranging in format and complexity. Given the lack of uniformity in intervention type and in measuring adherence, our results and conclusions were based on a narrative analysis. We also acknowledge that few studies reported adherence at multiple time points. Thus, it is difficult to distinguish the influence of the interventions on initiation, persistence or both. Additionally, we acknowledge that our results and conclusions may have been limited by our stringent inclusion criteria and study limitations of more than half of the trials There have been recent studies that only reported either adherence or clinical outcomes and not both and thus were excluded from this review.Similar to the 2014 Cochrane review, this review assessed interventions to improve medication adherence based on randomised controlled trials that assessed both adherence and clinical outcomes. While our criterion decreased the number of included studies, improving adherence is not a goal unto itself but rather a strategy to improve outcomes. Thus, evaluating trials with both outcomes may be more informative. Unfortunately, there was no evidence that ASCVD outcomes were improved, which may be due to limited power to detect this difference as well as short-term follow-up of these largely phase II trials for long-term chronic disease problems.

In conclusion, the trials were distinct in design, participants, intervention and settings. Few trials demonstrate improvements in both adherence and clinical outcomes. We identified three interventions that demonstrated significant impact in both domains. Two of the three trials involved simple, low-cost interventions with potential for scaling in a low-resource setting. Ongoing and new well-powered, longer-term trials are needed to determine whether short-term changes in adherence can be maintained and lead to differences in clinical events.



  • Contributors All authors designed the review. RHF, MDH and PP screened and extracted data. TN-R adjudicated data. RHF analysed data and wrote the first draft of the manuscript. All authors contributed important intellectual content to the drafted manuscript and approved it for publication.

  • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests MDH receives grant support from the World Heart Federation to serve as its senior program advisor for the Emerging Leaders program, which is sponsored by Boehringer Ingelheim and Novartis with prior support from BUPA and AstraZeneca. MDH has received grant support from the Cochrane Collaboration to complete a systematic review update on fixed-dose combination, or polypills, and travel support from the World Heart Federation to attend a meeting on polypills. PP led several studies on improving adherence for cardiovascular prevention for which his institution has received grants. He is Senior Science Advisor at the World Heart Federation, which is supported, among other funders, by pharmaceutical companies.

  • Patient consent None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement There are no unpublished data from the study.