Objective: To evaluate the incidence of key adverse outcomes among patients with nonvalvular atrial fibrillation receiving oral anticoagulants by CHA2DS2-VASc risk score range, thromboembolic event history, and age group.

Design, setting, and participants: This cohort study was a retrospective claims data analysis using combined data sets from 5 large health claims databases. Eligible participants were adult patients with nonvalvular atrial fibrillation who initiated oral anticoagulants. Data were analyzed between January 2012 and June 2019.

Exposure: Initiation of oral anticoagulants.

Main outcomes and measures: We observed clinical outcomes (including stroke or systemic embolism, major bleeding, and a composite outcome) on treatment through study end, censoring for discontinuation of oral anticoagulants, death, and insurance disenrollment. The population was stratified by CHA2DS2-VASc risk score; history of stroke, systemic embolism, or transient ischemic attack; and age groups. We calculated time to event, incidence rates, and cumulative incidence for outcomes.

Results: We identified 1 141 097 patients with nonvalvular atrial fibrillation; the mean (SD) age was 75.0 (10.5) years, 608 127 patients (53.3%) were men, and over 1 million were placed in the 2 highest risk categories (high risk 1, 327 766 participants; high risk 2, 688 449 participants). Deyo-Charlson Comorbidity Index scores ranged progressively alongside CHA2DS2-VASc risk score strata (mean [SD] scores: low risk, 0.4 [1.0]; high risk 2, 4.1 [2.9]). The crude incidence of stroke and systemic embolism generally progressed alongside risk score strata (low risk, 0.25 events per 100 person-years [95% CI, 0.18-0.34 events]; high risk 2, 3.43 events per 100 person-years [95% CI, 3.06-4.20 events]); patients at the second-highest risk strata with thromboembolic event history had higher stroke incidence vs patients at the highest risk score strata without event history (2.06 events per 100 person-years [95% CI, 2.00-3.12 events] vs 1.18 events per 100 person-years [95% CI, 1.14-1.30 events]). Major bleeding and composite incidence also increased progressively alongside risk score strata (major bleeding: low risk, 0.68 events per 100 person-years [95% CI, 0.56-0.82 events]; high risk 2, 6.29 events per 100 person-years [95% CI, 6.21-6.62 events]; composite incidence: 1.22 events per 100 person-years [95% CI, 1.06-1.41 events]; high risk 2, 10.67 events per 100 person-years [95% CI, 10.26-11.48 events]). The 12-month cumulative incidence proportions for stroke and systemic embolism, major bleeding, and composite outcomes progressed alongside risk score strata (stroke or systemic embolism, 0.30%-1.85%; major bleeding, 0.55%-5.55%; composite, 1.05%-8.23%). Age subgroup analysis followed similar trends.

Conclusions and relevance: The observed incidence of stroke or systemic embolism and major bleeding events generally conformed to an expected increasing incidence by risk score, adding insight into the importance of specific risk score range, thromboembolic event history, and age group strata. These results can help inform clinical decision-making, research, and policy.