The effectiveness of public health measures (PHMs) depends on their adoption by the population. Social media posts could be used as a proxy for adherence, but it is unclear if the results would be representative of the general population. The objective of this repeated cross-sectional study was to compare sociodemographic characteristics and self-reported adherence to PHMs among people with different Twitter use.

Seventeen cross-sectional surveys were conducted between September 2020 and August 2021. Using Bayesian logistic regression models, we investigated the association between Twitter use and 1) poor adherence to mask-wearing, defined as mask-wearing some of the time or less in public places, or 2) poor adherence to vaccination recommendations, defined as not willing to be vaccinated.

Of the 26,663 survey responses, 9% reported poor adherence to both mask-wearing and vaccination and 22% had used Twitter, among whom 30% tweeted about COVID-19. The proportion of Twitter users who tweeted or not about COVID-19 remained stable throughout the study period. A larger proportion of Twitter users lived in metropolitan areas, were younger, male, with a university education, or with an annual income greater than $100,000. The odds ratios (ORs) of poor adherence to mask-wearing and vaccination for those who tweeted about COVID-19 ranged from 1.6 to 5.1 compared to those not using Twitter; the opposite was true for those who tweeted about something other than COVID-19 with ORs ranging from 0.6 to 0.8. The association with mask-wearing was modified by geographical region and age.

The sociodemographic characteristics and adherence of Canadians who tweeted about COVID-19 differed considerably from that of non-Twitter users. This disparity between Twitter users and non-users must be considered if social media posts are to be used as a proxy to assess behaviours and opinions in the general population.