Cardiovascular adverse events of tocilizumab and baricitinib in COVID-19 and rheumatoid arthritis

Abstract

Tocilizumab and baricitinib are used in rheumatoid arthritis (RA) and COVID-19, where acute hyperinflammation and systemic stress may amplify cardiovascular risk. Comparative evidence on indication-specific cardiovascular signals remains scarce; we therefore assessed these signals by indication. We analyzed the FDA Adverse Event Reporting System data from January 2010 to June 2025 to compare cardiovascular adverse events (CVAEs) associated with baricitinib and tocilizumab. Primary outcomes included 6 CVAE categories: cardiac arrhythmias, central nervous system vascular disorders, embolic and thrombotic events, ischemic heart disease, noninfectious myocarditis/pericarditis, and other cardiac disorders. Disproportionality was assessed using the information component (IC) and reporting odds ratio (ROR), stratified by pre- and post-COVID-19 Emergency Use Authorization (EUA) periods and indications (COVID-19 vs RA). Machine learning models were additionally trained to characterize risk patterns for each drug, with feature contributions interpreted using SHapley Additive exPlanations. Among 17,750,053 reports, 5994 CVAEs were identified (baricitinib, n = 965; tocilizumab, n = 5029). Disproportionality was higher in the post-EUA than the pre-EUA period for both baricitinib (IC, 0.84 [IC 025 , 0.59] to 1.51 [1.39]; ROR, 1.89 [95% confidence interval, 1.62–2.21] to 3.22 [2.98–3.48]) and tocilizumab (0.22 [0.16] to 0.54 [0.47]; 1.18 [1.14–1.23] to 1.50 [1.43–1.57]). Baricitinib for COVID-19 showed the highest disproportionality (IC, 2.90 [IC 025 , 2.70]; ROR, 11.95 [95% confidence interval, 10.33–13.83]). Gradient boosting machine achieved the highest AUROC (baricitinib, 0.728; tocilizumab, 0.702), and SHapley Additive exPlanations highlighted COVID-19 indication for baricitinib and RA for tocilizumab. Our findings suggest indication-dependent cardiovascular signals when immunomodulators are repurposed for acute viral infections, underscoring the need for context-aware evaluation in drug repurposing.

The authors' abstract, as published at the source. Medicine, 2026 · DOI ↗