Abstract / Summary
Purpose: Atrial fibrillation (AF) is the best-established thromboembolic risk factor in rheumatic heart disease (RHD), but stroke and thromboembolism also occur without documented AF. We systematically synthesised associations between non-AF clinical/echocardiographic predictors and thromboembolism in RHD, comparing AF-adjusted with unadjusted estimates.
Methods: PubMed/MEDLINE, Embase, Scopus, and Web of Science were searched, supplemented by Google Scholar and citation screening. Eligible studies evaluated RHD patients with ≥1 non-AF predictor of stroke or thromboembolism. Two reviewers performed selection, extraction, and QUIPS risk-of-bias assessment; certainty was graded using GRADE. Unadjusted, adjusted, and AF-adjusted estimates were extracted separately, and random-effects meta-analysis was performed when feasible.
Results: Nineteen studies (∼8,206 patients) met eligibility criteria. SEC, LA size/volume, LA/appendage thrombus, LA appendage flow, LA mechanical function, mitral valve severity, age, and previous thromboembolism were most frequently examined. Four studies provided comparable crude ORs for SEC (pooled OR 8.27, 95% CI 3.21-21.29; I2 = 66%), substantially attenuated after AF adjustment (as low as OR 1.7, 95% CI 1.1-3.0). Older age and previous stroke remained associated with thromboembolism in AF-inclusive models, and LA mechanical/reservoir function showed consistent inverse associations, including in sinus rhythm. Certainty was low for SEC, age, and LA function; moderate for previous stroke/thromboembolism; very low for anticoagulation non-use.
Conclusion: In RHD, thromboembolic risk extends beyond documented AF. SEC showed the strongest unadjusted association but was substantially attenuated after AF adjustment, likely partly reflecting subclinical AF; it is TEE-dependent and inconsistently graded. Older age, previous thromboembolism, and impaired LA mechanical/reservoir function showed more persistent associations after AF-inclusive adjustment; prospective validation is needed.