Text Mining of Symptom Descriptions in the Vaccine Adverse Event Reporting System for Human Papillomavirus Vaccination.

introductionThe World Health Organizationhas reaffirmed the efficacy and safety of human papillomavirus (HPV) vaccines in a statement. While vaccine safety has been extensively studied, little is known about the descriptive language used in reports of adverse events. The Vaccine Adverse Event Reporting System (VAERS), managed by the US Food and Drug Administration and the Centers for Disease Control and Prevention, collects free-text narratives on adverse events following vaccination. This study aimed to examine these narratives to describe vocabulary patterns associated with HPV vaccination.

methodsWe conducted a retrospective, cross-sectional observational study using quantitative text mining techniques. Symptom descriptions related to HPV vaccination were extracted from the Vaccine Adverse Event Reporting System (VAERS, 2009-2023). Sentiment analysis was performed with the AFINN lexicon in R ("tidytext" package), which assigns numerical sentiment scores to words. This quantitative scoring approach enabled us to describe vocabulary patterns without qualitative inference of context or emotions.

resultsSentiment analysis was performed using the R "tidytext" package with the AFINN lexicon. Reports on suspected adverse events were obtained from the VAERS reports spanning 2009 to 2023, comprising 55,919 suspected adverse events. Approximately 6% of these reports involved product handling issues. The analysis showed that sentiments toward vaccination were more negative among females than males.

conclusionHealthcare providers supporting HPV vaccination should provide patients with accurate and comprehensive information on vaccine safety and potential adverse reactions.