Background: Stroke reduction with direct oral anticoagulants (DOACs) in atrial fibrillation (AF) is dependent on adherence and persistence in the real-world setting. Individual study estimates of DOAC adherence/persistence rates have been discordant. Our aims were to characterize real-world observational evidence for DOAC adherence/persistence and evaluate associated clinical outcomes in patients with AF. Methods and Results: PubMed, EMBASE, and CINAHL were searched from inception to June 2018. Observational studies that reported real-world DOAC adherence/persistence in patients with AF were included. Study quality was assessed using the Newcastle-Ottawa Scale. Meta-analyses for pooled estimates were performed using DerSimonian and Laird random-effects models. Outcomes included DOAC mean proportion of days covered or medication possession ratio, proportion of good adherence (proportion of days covered/medication possession ratio ≥80%), persistence, DOAC versus vitamin K antagonists persistence, and clinical outcomes associated with nonadherence/nonpersistence. Forty-eight observational studies with 594 784 unique patients with AF (59% male; mean age 71 years) were included. The overall pooled mean proportion of days covered/medication possession ratio was 77% (95% CI, 75%–80%), proportion of patients with good adherence was 66% (95% CI, 63%–70%), and proportion persistent was 69% (95% CI, 65%–72%). The pooled proportion of patients with good adherence was 71% (95% CI, 64%–78%) for apixaban, 60% (95% CI, 52%–68%) for dabigatran, and 70% (95% CI, 64%–75%) for rivaroxaban. Similar patterns were found for pooled persistence by agent. The pooled persistence was higher with DOACs than vitamin K antagonists (odds ratio, 1.44 [95% CI, 1.12–.86]). DOAC nonadherence was associated with an increased risk of stroke (hazard ratio, 1.39 [95% CI, 1.06–1.81]). Conclusions: Suboptimal adherence and persistence to DOACs was common in patients with AF, with 1 in 3 patients adhering to their DOAC <80% of the time, which was associated with poor clinical outcomes in nonadherent patients. Although it is convenient that DOACs do not require laboratory monitoring, greater effort in monitoring for and interventions to prevent nonadherence may be necessary to optimize stroke prevention. Increased clinician awareness of DOAC nonadherence may help identify at-risk patients.

No abstract

No abstract

Background: Direct oral anticoagulants (DOACs) are increasing in popularity for stroke prevention in atrial fibrillation (AF) due to ease of use. However, varying levels of DOAC adherence and persistence has been reported, which may result in poor clinical outcomes. We undertook a systematic review to characterize the real-world observational evidence for DOAC adherence and persistence. Methods: We searched MEDLINE, EMBASE, and CINAHL from inception to June 2018. We included studies that reported DOAC adherence and/or persistence rates in patients with AF. Two investigators independently screened abstracts and full text articles for inclusion, abstracted data from eligible studies, and assessed risk of bias using the Newcastle-Ottawa Scale. Prevalence rates of nonadherence were pooled using a DerSimonian and Laird random-effects model. Heterogeneity was assessed using I 2 . Results: We identified 48 studies including 576,360 patients in 40 countries. Seventeen studies (35%) reported proportion of days covered (PDC) as the adherence measure, 5 (10%) reported medication possession ratio (MPR), 36 (75%) reported persistence, and 6 (13%) evaluated clinical outcomes associated with DOAC adherence/persistence. Seventeen studies (35%) evaluated DOACs as a class, 11 (23%) evaluated apixaban, 35 (73%) evaluated dabigatran, and 28 (58%) evaluated rivaroxaban. The overall pooled proportion of patients who had good adherence (PDC/MPR≥80%) at all times was 66% (CI: 63%-70%) and at 1 year was 69% (CI: 63%-74%). There was wide variation in persistence definitions used in individual studies, with permissible gaps varying from 14 to 365 days. The overall pooled persistence rate was 71% (CI: 69%-74%) at all times and was 63% (CI: 58%-68%) at 1 year. Heterogeneity was high for all analyses. In subgroup analyses by agent, the pooled 1-year persistence rate was 72% (CI: 60%-83%) for apixaban, 54% (CI: 46%-62%) for dabigatran, and 67% (CI: 59%-75%) for rivaroxaban. Similar patterns were found for pooled adherence rates by agent. Conclusions: We found that DOAC adherence and persistence is low among patients with AF, with 1 in 3 patients not taking a DOAC after initiation. Clinicians need to be aware of the low real-world adherence and persistence rates when using DOACs in AF. High heterogeneity warrants further subgroup analyses.