Autoimmune Reporting Signals in FAERS Reports Listing COVID-19 Vaccines as Suspect Products: An Active-Comparator Disproportionality Analysis.

BACKGROUND/OBJECTIVES: Autoimmune and immune-mediated adverse events coded in reports listing COVID-19 vaccines as suspect products are rare, heterogeneous, and difficult to evaluate in spontaneous reporting systems. We characterized autoimmune reporting signals using strict case definitions and an active-comparator disproportionality design. METHODS: FAERS reports from 2021Q1 to 2023Q4 were deduplicated and harmonized by vaccine product, platform, sex, age group, and MedDRA Preferred Term (PT). COVID-19 vaccine reports recorded as primary or secondary suspect products were compared with traditional non-COVID vaccine reports. Strict autoimmune events were grouped into neurological, endocrine, rheumatological, hematological, dermatological, and cardiac categories. Reporting odds ratios (RORs), proportional reporting ratios (PRRs), PT-level rankings, masking audits, and sensitivity analyses were evaluated. RESULTS: The primary analysis included 4191 COVID-19 vaccine reports, 2817 comparator reports, and 311 strict-autoimmune reports contributing 316 report-category observations. Positive category-level signals were identified for dermatological (ROR 10.49; 95% CI: 2.51-43.86), endocrine (ROR 5.39; 95% CI: 1.24-23.48), hematological (ROR 4.68; 95% CI: 2.32-9.44), and neurological events (ROR 1.63; 95% CI: 1.16-2.29). Rheumatological reporting was lower, strict cardiac events were absent, and hematological signals were most consistent across sensitivity analyses. Among five adjusted-positive finite PT signals, immune thrombocytopenia was more stable, multiple sclerosis intermediate, and acquired hemophilia, myasthenia gravis, and optic neuritis sparse/imprecise. A post hoc audit identified four potential clusters among 20 acquired-hemophilia reports; cluster collapse reduced the ROR to 2.70 (95% CI 0.30-24.18; p = 0.654). CONCLUSIONS: These findings are hypothesis-generating and do not establish incidence, absolute risk, or causality.