Human Papillomavirus Vaccine Discourse and Sentiment on Reddit Before and After COVID-19: Mixed Methods Retrospective Cross-Sectional Study.

Résumé

BACKGROUND: Human papillomavirus (HPV) is a sexually transmitted virus that causes various oropharyngeal and anogenital cancers. The HPV vaccine provides protection against several strains of HPV and is a key preventative tool against HPV-related cancers; however, vaccination rates remain suboptimal in the United States due to variable state mandates and misperceptions of vaccine efficacy and risks. As social media becomes an increasingly popular avenue for health discussions, platforms such as Reddit offer opportunities to understand public vaccine discourse, particularly among underrepresented groups. Furthermore, vaccine hesitancy and mistrust generally increased during the COVID-19 pandemic, potentially impacting HPV vaccination rates.OBJECTIVE: This study aimed to characterize HPV vaccine-related discussions on Reddit by (1) identifying dominant themes, (2) assessing sentiment, and (3) examining shifts in themes and sentiment before and after the start of the COVID-19 pandemic.METHODS: This convergent mixed methods analysis used a Python-based script to extract Reddit posts and comments referencing the HPV vaccine from multiple subreddits from September 13, 2016, to February 13, 2025. Entries were manually labeled as intentional or incidental, with intentional entries further categorized into 10 thematic categories. Sentiment analysis was conducted using the VADER (Valence Aware Dictionary for Sentiment Reasoning) algorithm. Pre- vs post-COVID-19 comparisons used March 11, 2020, as the cutoff date. Temporal trends were assessed using pre- and post-COVID-19 stratification. Chi-square tests, Mann-Whitney U tests, and linear regression were used for statistical analysis.RESULTS: Of 4235 collected entries, 2801 (66.1%) intentional posts and comments were analyzed. The most common themes were factual claims (703/2801, 25.1%), personal experiences (571/2801, 20.4%), and offering advice (465/2801, 16.6%). Overall sentiment was 51.7% (1448/2801) positive (95% CI 49.8%-53.5%), 37.6% (1053/2801) negative, and 10.8% (300/2801) neutral, with a median sentiment score of 0.13 (IQR -0.49 to 0.71). Pre-COVID-19 entries (n=278) were 124 (44.6%) positive (95% CI 38.7%-50.7%), with a median sentiment of 0.00 (-0.66 to 0.58). Post-COVID-19 entries (n=2523) were 1324 (52.4%) positive (95% CI 50.5%-54.4%), with a median sentiment of 0.17 (-0.46 to 0.72). Sentiment increased significantly post-COVID-19 (P=.01). Theme distribution differed before vs after COVID-19 (χ²9=114.47, P