Clinical characteristics and predictors of adverse outcomes in children with acute rheumatic fever from a high-endemic region of Brazil

Introduction Acute rheumatic fever (ARF) remains a leading cause of pediatric cardiovascular morbidity and a major driver of rheumatic heart disease (RHD) in low- and middle-income countries. Early identification of children at highest risk for adverse outcomes is essential to guide timely management and referral, but remains challenging in resource-limited settings. This study aimed to characterize the clinical profile of pediatric ARF and identify readily available predictors of adverse outcomes in a high-endemic region of Brazil. Methods We conducted a cohort study of children treated for ARF at a tertiary pediatric center in Feira de Santana, Brazil, between 2010 and 2023, including both retrospectively identified cases and a prospectively enrolled subset through the Acute Rheumatic Fever Diagnosis Collaborative (ARC) Network. Clinical, laboratory, treatment, and outcome data were obtained from hospital and outpatient records. The primary outcome was a composite of cardiac surgery or death. Independent predictors were identified using multivariable logistic regression with covariate selection based on clinical relevance and model parsimony. Model performance was assessed using discrimination (C-statistic) and risk reclassification metrics. Results A total of 151 children were included (median age, 9 years; 46% female). Carditis was present in 76%, with moderate-to-severe involvement in 49%, predominantly affecting the mitral valve. During a median follow-up of 17.5 months, 20 children (13%) experienced the primary outcome. In multivariable analysis, absence of fever (OR 0.14, 95% CI 0.31–0.65), higher ESR (OR 1.04 per mm/h, 95% CI 1.01–1.07), and lower hematocrit (OR 0.85, 95% CI 0.74–0.98) were independently associated with adverse outcomes. Incorporating erythrocyte sedimentation rate (ESR) into the model improved risk prediction, with a net reclassification improvement (NRI) of 0.79 ( P = 0.005) and an integrated discrimination improvement (IDI) of 0.11 ( P = 0.032). The final model demonstrated good discrimination (C-statistic 0.82). Results were consistent after multiple imputation. Conclusions In children with ARF, simple inflammatory and hematologic markers are independently associated with progression to severe outcomes. ESR, an inexpensive and widely accessible test, provides incremental prognostic information and may support scalable risk stratification strategies aimed at reducing progression to advanced RHD in resource-limited endemic settings.