Adherence with direct oral anticoagulants in patients with atrial fibrillation: Trends, risk factors, and outcomes

Abstract Background Adherence to direct oral anticoagulants (DOACs) remains a concern among non‐valvular atrial fibrillation (AF) patients. We aimed to assess patterns of adherence with DOACs and examine their association with ischemic stroke and systemic embolism (SE). Methods This retrospective cohort study includes all adult members of Clalit Health Services, the largest healthcare provider in Israel, with newly diagnosed non‐valvular AF between January 2014 and March 2019, who initiated DOACs within 90 days of AF diagnosis and used DOACs exclusively. Adherence was assessed using the proportion of days covered (PDC) over the first year of treatment, and high adherence was defined as PDC ≥80%. Regression models were used to identify predictors of high adherence to DOACs and to examine the association between adherence and stroke or SE. Results Overall 15,255 patients were included in this study. The proportion of highly adherent (PDC ≥80%) DOACs users was around 75% and decreased slightly over the years. On multivariable analyses, the likelihood of high adherence to DOACs increased with age and across higher socioeconomic classes, and was more likely among females, Jews, statins users, and patients with CHA2DS2‐VASc score ≥2. Risk of stroke and SE was lower among highly adherent DOACs users; adjusted HR 0.56 (95% CI, 0.45–0.71), compared to users with PDC <80%. Conclusions Adherence with DOACs is still sub‐optimal among non‐valvular AF patients, resulting in a higher risk of stroke and SE.


| INTRODUC TI ON
Atrial fibrillation (AF) is a common cardiac rhythm disorder which poses a significant risk for cerebrovascular morbidity and mortality. 1 When used appropriately, oral anticoagulants (OACs) have shown to reduce the incidence of embolic stroke in AF by more than 50%. 2 Vitamin K antagonists (VKAs) have long been considered the main OAC used worldwide. VKAs have several limitations such as high bleeding risk, slow onset of action, the need for frequent monitoring, and numerous drug and diet interactions. Direct oral anticoagulants (DOACs) have been recently introduced and have been gradually replacing VKAs owing to their distinct benefits in terms of efficacy, safety, convenience, more predictable effect and fewer drug and diet interactions. 3 However, their short half-life time and their rapid onset of action require a strict treatment compliance and adherence in order to maintain the desirable antithrombotic effect. [4][5][6] It has been shown that less than half of the patients adhere to OAC treatment over time. 7 There was an expectation that DOACs introduction, given their advantages, would be translated into improved adherence. Unfortunately, adherence to DOACs remains poor in patients with AF, 7-12 and appears to be associated with increased risk of stroke, [11][12] yet real-life data is scant. [7][8][9][10][11][12] Adherence patterns are an integral part of clinical decision making and depend on various factors, either social or clinical.
Understanding those patterns and factors is clinically relevant and may be helpful in better resources utilization and health planning.
In this study, we use a population-based real-life data to assess patterns and trends of patients' adherence with DOACs treatment.
This study also aims to provide specific insight as for the implication of non-adherence to DOACs on the risk of ischemic stroke and systemic embolism (SE).

| Study outcomes and definition of terms
Adherence to DOACs was assessed using the proportion of days covered (PDC), as recommended by the Pharmacy Quality Alliance and SE (ICD-9; 444.X, 445.X).

| Covariates
Demographic characteristics, and data on comorbidities were retrieved from the CHS-computerized database for the calculation of CHA 2 DS 2 -VASc score, a widely used risk stratification score for stroke prediction in patients with AF. 16

| Predictors of high adherence to DOACs
Multivariable logistic regression analysis revealed that the likelihood of high adherence to DOACs (PDC ≥80%), during the first year of treatment, increased with age and with increasing socioeconomic classes, and was more likely among females compared to males, among Jews compared to Arabs, among patients treated with statins and among patients with CHA 2 DS 2 -VASc score ≥2 compared to those with CHA 2 DS 2 -VASc score <2. The likelihood of high adherence was lower in smokers compared to nonsmokers. Compared to rivaroxaban use, the likelihood of high adherence was significantly lower with dabigatran use, whereas no statistically significant difference was observed with apixaban use (Table 3).

| Association between adherence and stroke or SE
The risk of ischemic stroke and SE, after adjustments to CHA 2 DS 2 -VASc score, was found to be 44% lower (adjusted HR 0.56, 95% CI 0.45-0.71) among highly adherent DOACs users (PDC ≥80%) compared to DOACs users with PDC <80% (Table 4). Using PDC as a ordinal variable with the lowest category (PDC <40%) serving as reference category, it was shown that increasing PDC up to 80% was not associated with stroke and SE risk reduction, and that only high adherence (PDC ≥80%) was associated with a statistically significant decrease of stroke and SE; adjusted HR 0.59 (95% CI, 0.42-0.83) ( Table 4 and Figure 4). A protective effect was also demonstrated when using PDC as a continuous variable as well, with a 9% decrease (adjusted HR 0.91, 95% CI 0.88-0.95) in the risk of stroke and SE for each 10% increase in PDC (Table 4).

| DISCUSS ION
This is one of the few large-scale population-based studies to assess adherence to DOACs. In this study, we found that approximately 75% of incident AF patients treated with DOACs were highly adherent to treatment, namely, were covered at least 80% of the days during the first year of treatment. The proportion of high adherence to DOACs in this study is consistent with the rates described in two recent large cohort studies conducted in France and Canada. 17,18 TA B L E 4 Multivariable a hazard ratios (HRs) for the association between DOACs use adherence, as estimated by PDC in the first years of treatment, and the risk of ischemic stroke and systemic thromboembolism Nevertheless, as much as these results seem encouraging compared to previous studies, [7][8][9] and despite the growing use of DOACs, non-adherence to DOACs is still a concern, with the rate reaching up to 25% over a year period. Moreover, there is a worrisome pro- However, direct comparisons to those studies are difficult because of methodological differences.
A notable strength of this study is being a population-based study with a relatively large number of AF patients. The fact that healthcare services in Israel are public and medication copayment is low makes the documentation of medications prescribing and purchases, reliable. In addition, relying on these documentations rather than patients' self-reports avoids the potential recall bias.
Nevertheless, our study is subject to some limitations. First, as an observational study, in which data is extracted from electronic health records, it is prone to misclassification bias or lacks data. Second, we relied on prescriptions and purchases data rather than actual taking the drugs by the patients. Thus, there might be an overestimation of adherence rates in cases in which patients stopped taking DOACs even though they had purchased the drugs. In addition, patients could have taken lower doses than purchased, hence being exposed to higher embolic risk. Finally, our study is prone to the healthy adherer effect. 24 In other words, patients who are adherent to DOACs, may also adhere to other therapies as well as to healthier lifestyle and to medical preventive services, providing another explanation for the lower risk of embolic complications.

| CON CLUS IONS
Adherence with DOACs treatment is still far from optimal, resulting in a substantial higher risk of embolic complications. More efforts should be done in order to increase physicians' and AF patients' awareness of the importance of compliance to DOACs.

ACK N OWLED G M ENTS
Walid Saliba and Zomoroda Abu-Ful conceived and designed the study, did the analysis, and took responsibility for the integrity of the data and the accuracy of the data analysis. All authors had full access to all of the data in the study. Anat Arbel and Walid Saliba drafted the manuscript.
All authors critically revised the manuscript for important intellectual content and gave final approval for the version to be published.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest for this article.

E TH I C S A PPROVA L
The study was approved by the Review Board of the Lady Davis Medical Centre and conducted in accordance with the Declaration of Helsinki.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.