Medication Adherence After Acute Coronary Syndrome in Women Compared With Men: A Systematic Review and Meta-Analysis

Introduction: Pharmacological treatment is an important component of secondary prevention in acute coronary syndrome (ACS) survivors. However, adherence to medication regimens is often suboptimal, reducing the effectiveness of treatment. It has been suggested that sex influences adherence to cardiovascular medication, but results differ across studies, and a systematic overview is lacking. Methods: We performed a systematic search of PubMed and EMBASE on 16 October 2019. Studies that reported sex-specific adherence for one or more specific medication classes for ACS patients were included. Odds ratios, or equivalent, were extracted per medication class and combined using a random effects model. Results: In total, we included 28 studies of which some had adherence data for more than one medication group. There were 7 studies for angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin II receptor blockers (ARBs) (n = 100,909, 37% women), 8 studies for antiplatelet medication (n = 37,804, 27% women), 11 studies for beta-blockers (n = 191,339, 38% women), and 17 studies for lipid-lowering medication (n = 318,837, 35% women). Women were less adherent to lipid-lowering medication than men (OR = 0.87, 95% CI 0.82–0.92), but this sex difference was not observed for antiplatelet medication (OR = 0.95, 95% CI 0.83–1.09), ACEIs/ARBs (OR = 0.95, 95% CI 0.78–1.17), or beta-blockers (OR = 0.97, 95% CI 0.86–1.11). Conclusion: Women with ACS have poorer adherence to lipid-lowering medication than men with the same condition. There are no differences in adherence to antiplatelet medication, ACEIs/ARBs, and beta-blockers between women and men with ACS.


INTRODUCTION
Patients who survive an acute coronary syndrome (ACS) are at high risk of recurrent events (1)(2)(3). Secondary prevention through pharmacological therapy reduces the risk of recurrent events and mortality in this population (1,2), but its effectiveness is attenuated by suboptimal patient adherence (4,5). Poor adherence to medication regimens is an important obstacle in improving outcomes for ACS patients and has proven difficult to solve (6). Two large meta-analyses evaluating adherence to cardiovascular medication found that patient sex was an important factor in predicting adherence (7,8). However, these meta-analyses did not investigate which sex was at higher risk of non-adherence.
There is some evidence on adherence to, for example, statins that suggests women have poorer adherence because they experience more adverse drug reactions (9), but a structured overview of the literature is still lacking. We performed a systematic review with meta-analyses on sex differences in adherence to cardiovascular medication in patients with ACS. We hypothesize that women, in general, have poorer adherence than men.

Terminology
It is important to recognize that sex and gender describe two different concepts. Sex refers to the biological differences between females and males, whereas gender refers to social differences between women and men. Both play an important role in health and disease, although through different mechanisms (10). This manuscript evaluates sex differences, meaning the linguistically correct terms to use would be "female" and "male." However, all studies included in our review used the terms "women" and "men" to refer to patient sex, as is common in medical literature. We therefore also use the terms "women" and "men" to refer to patients of the female and male sex, respectively.

Search Strategy and Selection Criteria
We searched both PubMed and EMBASE on 16 October 2019 using a pre-defined search term consisting of both text words and MeSH headings (Supplementary Files). The text words were limited to title and abstract only. The retrieved articles were screened by two independent reviewers who also resolved any conflicts that arose with help of a third reviewer, if necessary. The reference lists of relevant articles were screened for any additional articles. A modified version of the Newcastle-Ottawa Scale was used to assess the quality of included studies (Supplementary Files).
Only original research articles written in English that evaluated adherence at the individual patient-level were eligible for inclusion. Articles were included if they reported sexspecific data on medication adherence in patients with ACS, defined as either myocardial infarction or unstable angina (11). We excluded studies with too few participants to evaluate sex differences (n < 100), studies where ACS was included alongside other cardiovascular diseases and results could not be separated based on disease subgroup, and studies that included only men or only women. We also excluded studies that evaluated adherence to a combination of medications instead of per specific medication group. Finally, we excluded all studies for which the full text could not be retrieved.
We extracted population size, the percentage of women, mean age, total duration of follow-up, and measure of adherence used from all included studies. In addition, we extracted the number of adherent and non-adherent women and men or, if unavailable, unadjusted relative risk estimates (or equivalent). We also extracted adjusted relative risk estimates when available.

Statistical Analysis
The meta-analysis was conducted conforming with the Meta-Analyses and Systematic Reviews of Observational studies (MOOSE) guidelines (12). We chose "good adherence, " as defined by each study, as our outcome and men as the reference category to facilitate interpretation of the results. We pooled the sex-specific odds ratios (ORs) using random effects meta-analysis because the included studies applied varying definitions of adherence and thus the estimated effect of sex on adherence can vary across these studies. In these situations, it is recommended  to apply random effects meta-analysis instead of fixed effects meta-analyses (13).
We calculated the average sex-specific adherence across studies weighted by study size. We calculated unadjusted ORs for studies that presented number of adherent and non-adherent women and men. We converted the risk estimates from studies that used either a different outcome (poor adherence) or reference category (women) to fit our analysis. When studies stratified their analysis by subgroups, we pooled reported risk estimates using fixed effects meta-analysis and included the pooled risk estimate in our overall meta-analysis. We performed an additional analysis using only adjusted ORs to see whether adjustment would affect our crude estimates. We created funnel plots to check for publication bias.
All analyses were performed in R (R Core Team, Vienna, Austria). A p < 0.05 was considered statistically significant.

Study Characteristics
In total, we included 28 studies of which some had adherence data for more than one medication group (Figure 1). The medication groups included were angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers (ACEIs/ARBs), antiplatelet therapy, beta-blockers, and lipid-lowering medication. Nine studies were of high quality (4 stars), 17 studies were of moderate quality (2 or 3 stars), and two studies were of poor quality (0 or 1 star). A complete overview of the included studies can be found in Table 1.
In the adjusted analyses, we included three studies for ACEIs/ARBs and beta-blockers and six studies for lipid-lowering medication. There was only one study with adjusted risk estimates for antiplatelet medication. These analyses showed a significantly poorer adherence in women for ACEIs/ARBs (OR = 0.88, 95% CI 0.80-0.96) and beta-blockers (OR = 0.91, 95% CI 0.88-0.93) but not for lipid-lowering medication (OR = 0.97, 95% CI 0.89-1.05) (Figure 3).
The funnel plots for ACEIs/ARBs and lipid-lowering medication were relatively balanced, suggesting little publication bias. For antiplatelet medication and beta-blockers, however, smaller studies showing poorer adherence in women seemed to be lacking compared with the number of such studies showing poorer adherence in men (Figure 4).

DISCUSSION
Adherence to ACEIs/ARBs, antiplatelet medication, betablockers, and lipid-lowering medication lies between 60 and 70% in both women and men surviving an ACS. Women had poorer adherence than men for lipid-lowering medication but not for the other medication groups, where adherence was similar between the sexes.
A previous systematic review on adherence to cardiovascular medication in coronary heart disease patients also found adherence to be 60-70% (42), suggesting that adherence is reasonable in secondary prevention of ACS. However, they did not find any sex differences (42), whereas our results suggest that those may be present at least for lipid-lowering drugs. This finding is supported by previous work showing that women have poorer adherence to statins in both primary and secondary prevention (43). This may be due to biological or social reasons, or a combination of both. There are known biological differences in drug metabolism between women and men (44), which may increase the risk of statin-related adverse drug reactions in women (45). This may also be true for the other medication groups included in our review, but the lack of sex-specific data on medication efficacy, safety, and metabolism prevents researchers from drawing sound conclusions on this topic (46)(47)(48). Gender differences may also play a role in adherence, with women for example more often refusing or discontinuing statins because they do not believe the medication is safe (49). Given that women derive equal benefit from statin therapy as men (50), it is important to improve statin adherence in women through both collecting more high-quality sex-specific data on this topic and adapting treatment to individual patients by for example using lower dosages to reduce the risk of side effects (45).
The main strength of this review is that it combines data from 28 studies. However, it is limited by the quality of the available data. The majority of studies included in this review were of moderate quality, and the data were heterogeneous on several important points. Both the chosen measure of adherence and the definition of "good adherence" varied greatly across studies. Approximately half of the included studies used a standardized measure of adherence, such as the medication possession ratio (18% of studies) or proportion of days covered (36%), but others used either self-report (11%) or another, sometimes selfdevised, measure (35%). This makes meta-analyzing such data and interpreting the results difficult. To alleviate this issue, it is important that future studies use both standardized measures of adherence and standardized cut-off values to denote good and poor adherence. We also saw that smaller studies showing poorer adherence in women were less likely to be published, and that studies showing poorer adherence in women more often provided adjusted risk estimates. This differential approach may introduce bias in meta-analyses such as ours and complicate the interpretation of our findings.
In conclusion, we show that adherence to cardiovascular medication is reasonable in women and men surviving an ACS. Women had poorer adherence to lipid-lowering medication than men, but this difference was not observed for the other cardiovascular medication groups. However, a standardized approach to the measurement and evaluation of adherence is needed to improve the quality of research performed in this field.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
SB and JI performed the systematic search. SB analyzed the data and wrote the manuscript. SP conceived the project, supported data analyses, and critically reviewed the manuscript. All authors contributed to the article and approved the submitted version.