Incidence and predictors of recurrent stroke in patients with cryptogenic stroke later diagnosed with atrial fibrillation: a 5-year cohort study from a tertiary care hospital

Background: Cryptogenic stroke (CS) accounts for a substantial proportion of ischemic strokes, with atrial fibrillation (AF) often identified as the underlying cause. In countries with low-resource settings like Pakistan, limited access to extended cardiac monitoring delays AF detection, increasing the risk of recurrent stroke. Objective: To determine the incidence and predictors of recurrent ischemic stroke in patients initially diagnosed with CS and later diagnosed with AF at a tertiary care hospital in Pakistan. Methods: A retrospective cohort study was conducted at tertiary care Hospital, from January 2018 to December 2023. Adult patients with CS and a subsequent diagnosis of AF within 12 months were included. Data on demographics, diagnostic modality, echocardiography, anticoagulation adherence, and outcomes were collected. Cox regression was used to identify independent predictors of recurrent stroke. Results: Out of 406 patients with CS, 109 (26.8%) were later diagnosed with AF. Recurrent ischemic stroke occurred in 32 patients (29.4%) over a median follow-up of 13.8 months. Only 11 of the recurrent cases (34.4%) were on anticoagulation therapy at the time of recurrence. Multivariate analysis identified age >65 years (HR: 2.21; 95% CI: 1.18-4.14), delayed AF detection beyond 6 months (HR: 2.58; 95% CI: 1.36-4.91), non-adherence to anticoagulation (HR: 3.44; 95% CI: 1.88-6.27), and left atrial enlargement (HR: 2.03; 95% CI: 1.15-3.57) as independent predictors of recurrence. Mortality (18.8%) and functional disability (mRS >3 in 59.3%) were significantly higher in the recurrence group. Conclusion: Delayed AF detection and poor anticoagulation adherence significantly increase the risk of recurrent stroke in CS-AF patients. Prolonged cardiac monitoring and improved access to anticoagulants, particularly NOACs, are essential to reduce recurrent stroke burden in low-resource settings.