Background and Purpose- Prevalence of atrial fibrillation (AF) is increasing, but the impact on overall burden of stroke is uncertain, as is the proportion that could be attributed to under anticoagulation. We did a population-based study of AF-associated stroke and a systematic review of time trends in other stroke incidence studies and of rates of premorbid anticoagulation. Methods- The proportion of incident strokes with associated AF was determined in the OXVASC (Oxford Vascular Study; 2002-2017) and in other prospective, population-based stroke incidence studies published before December 2017. Proportions were pooled by Mantel Haenszel methods, and the pooled percentage of cases with premorbid anticoagulation was determined. Analyses were stratified by the age of study population, mid-study year, country, and ethnicity. Results- Of 1928 patients with incident ischemic stroke in OXVASC, 629 (32.6%; 95% CI, 30.5-34.7) were AF associated, consistent with the pooled estimate from 4 smaller studies over the same study period (608/1948; 31.2%, 30.0-32.4; P_het=0.80). The pooled estimate from all studies reporting premorbid AF over 25 million person-years of observation (1960 onwards; 33 reports) was lower (18.6%, 16.8-20.3) and more heterogeneous (P_het<0.0001), but 62% of heterogeneity was explained by the age of study population, study period, country, and ethnicity. The proportion of incident strokes on premorbid anticoagulation increased over time, both for ischemic stroke in OXVASC (2002-2007: 15.1%, 2008-2012: 19.6%, and 2013-2017: 35.9%; P_trend<0.0001), and across all studies (P=0.002), but the pooled estimates suggested substantial undertreatment even in the most recent periods (2001-2015: 25.7%, 21.1-30.3 and ≥2010: 31.6%, 18.2-44.9). Conclusions- About 1 in 3 incident ischemic strokes are still AF associated, due partly to low rates of anticoagulation for known prior AF, which therefore represents a major public health opportunity to reduce the burden of stroke.

	Year	Citations