Exploring vaccine safety and adverse events in major autoimmune diseases

Abstract

This study evaluates post-vaccination adverse events by analyzing a large dataset of patients with major autoimmune diseases, including Hashimoto’s $n=26,330$; Rheumatoid Arthritis $n=9,251$; psoriasis $n=5,589$; Systemic Lupus Erythematosus $n=4,208$; Inflammatory Bowel Disease $n=5,831$; type 1 diabetes $n=2,235$; vasculitis $n=466$; Guillain-Barré Syndrome $n=185$; Immune Thrombocytopenic Purpura $n=623$; ankylosing spondylitis $n=926$; Sjögren’s syndrome $n=269$; psoriatic arthritis $n=2,355$; polymyositis $n=169$; dermatomyositis $n=130$. Our objective is not to refute the importance of vaccines, but to raise awareness about potential risks observed in autoimmune patients by analyzing CDC (Centers for Disease Control and Prevention) data. The sex distribution analysis in vaccine adverse events highlights a consistent female predominance across most autoimmune conditions. We designed machine learning predictive classification models by identifying key predictors to predict severe adverse events (hospitalization or death) following vaccination based on clinical and demographic predictors including age, sex, vaccine type, dose series, and vaccine route. Our models identified distinct risk profiles for severe events across diseases. Example AUC values ranged from 0.90 for dermatomyositis and GBS to 0.98 for Psoriatic Arthritis with accuracy 96% observed for ankylosing spondylitis. Vasculitis and Sjögren’s showed peak precision scores, while polymyositis showed peak recall (97%). Moreover, the reported adverse events in the first week and after the 6th week of vaccine administration are one order of magnitude larger than reported incidents in other time intervals for all diseases. Understanding these differences can inform safer vaccination strategies. We recognize the essential public health role of vaccines and underscore the importance of vigilant post-vaccination monitoring in autoimmune populations.