The cumulative incidence of atrial fibrillation in the hospitalized medical patient: A systematic review and Meta-analysis

Abstract Introduction New-onset atrial fibrillation (NOAF) is the most common cardiac arrhythmia occurring in hospitalised patients. NOAF confers a significant risk of cardiovascular morbidity and mortality in the form of symptomatic tachyarrhythmia, heart failure, stroke and sudden cardiac death. It is unclear what the cumulative incidence of NOAF is in the hospitalised medical (non-surgical) patient. Methods MEDLINE and Embase were searched for studies published between 01/01/2000 and 10/03/2026. All studies reporting the cumulative incidence of NOAF in hospitalised medical (non-surgical) patients were included. The pooled cumulative incidence of NOAF, 95% confidence intervals (CI), and 95% prediction intervals (PI) were computed using a random-effects model. The inconsistency index (I2) was calculated to measure heterogeneity. Subgroup analyses were also performed. A study protocol was registered with the PROSPERO database of systematic reviews (CRD42024626333). Results A total of 10,323 articles were identified from our searches, and of these 62 met the inclusion criteria. The cumulative incidence of NOAF across all studies was 99,474 (crude incidence 2.4%) out of 3,608,663 patients. The pooled cumulative incidence of NOAF was 9% (95% CI 7–11%), with substantial heterogeneity (I2 = 99.9%) and a wide prediction interval (1–42%). In meta-regression, ICU setting (OR 2.46, 95% CI 1.22–5.00; p = 0.013) and prospective study design (OR 1.73, 95% CI 1.01–2.95; p = 0.044) were independently associated with higher NOAF incidence, while year of publication and use of administrative data were not significant. Subgroup analyses demonstrated consistently higher incidence in prospective and critically ill populations. Conclusion The cumulative incidence of NOAF in patients hospitalised for acute medical illness varies significantly dependent on disease severity and methods to detect it. Attempts to pool data across studies are limited by differences in study design, patient populations and disease severity.