Background Boxing headgear has previously been mandatory at the Olympic Games but was removed from men's competition in 2016, citing evidence that it increases concussion risk.

Objective To describe the dissemination of concussion information on social media in response to headgear removal during Olympic competition.

Design Cross-sectional media analysis.

Setting User-generated content posted on the social networking site Twitter during the 2016 Olympic Summer Games.

Participants Twitter users posting relevant content between 5th-21st August, 2016.

Assessment of Risk Factors Twitter data (“tweets”) were extracted and coded as either for (supporting) or against (opposed to) headgear removal, and to indicate the user's belief that headgear prevents or increases the risk of concussion.

Main Outcome Measurements The total number of tweets and the number of users reached were extracted. Pearson chi² was used to assess the effect of concussion risk beliefs on the type of disseminated message (e.g., personal opinion versus evidence-based).

Results 646 tweets were captured, reaching ∼13.2 million Twitter users. Concussion was mentioned in 91 (14.1%) cases; increased laceration risk after headgear removal was referenced in 100 (15.5%). External links were included in 219 tweets (33.9%), comprising scientific publications/resources (n=27, 12.3%), news articles (n=141, 64.4%), and blog/opinion sites (n=51, 23.3%). Only 277 tweets were framed as either for (n=121, 43.7%) or against (n=156, 56.3%) headgear removal. Of these, 64 (45.4%) indicated that headgear increases concussion risk, whereas 77 (54.5%) suggested that it prevents concussions. Users who believed that headgear prevents concussions were more likely to base this on personal opinion (n=66, 85.7%) than those believing it increases risk (n=41, 64.1%) (x²=8.95, p=0.003).

Conclusions Personal opinion and news articles were the most commonly disseminated concussion information sources, leading to conflicting positions on the counterintuitive removal of boxing headgear. These findings have implications for real-time knowledge translation strategies during intervention implementation across contexts.