Uncovering Adverse reactions following COVID-19 Monovalent XBB.1.5 Vaccination from Active Surveillance: A Text Mining Approach

Abstract

Background Unstructured text data collected through a surveillance system for vaccine safety monitoring can identify previously unreported adverse reactions and provide the information necessary to improve the surveillance system. Therefore, this study explored adverse reactions using text data gathered through an active surveillance system following monovalent XBB.1.5 COVID-19 vaccination. Methods A text mining analysis was conducted on 2,608 records from 1,864 individuals who reported any health conditions experienced within 7 days after vaccination in text format. Frequency analysis of key terms was performed, with subsequent analysis by sex, age, and concurrent influenza vaccination. Furthermore, semantic network analysis was conducted on terms reported simultaneously. Results The analysis identified various common (≥ 1%) adverse events, such as sleep disturbances, lumbago, and indigestion, which had not been frequently reported in prior literature. Moreover, although not common (≥ 0.1% to < 1%), adverse reactions affecting the eyes, ears, and oral cavity were also noted. These adverse reactions showed no significant differences in occurrence with or without simultaneous influenza vaccination. Through cooccurrence analysis and correlation coefficient assessments, associations were found between diarrhea and abdominal pain, as well as between musculoskeletal symptoms and cold-related symptoms. Conclusion This study used text mining to reveal previously unrecognized adverse reactions related to COVID-19 vaccination, thus expanding our understanding of the vaccine’s safety profile. The insights gained could further the scope of future investigations into adverse reactions to vaccines and improve the processing of text data in surveillance systems.