The institutional review boards of the University of Pennsylvania (828972) and Cedars-Sinai Medical Center (STUDY00002429) deemed this study exempt. The data used in this study were publicly available at the time of collection and were collected and analyzed in accordance with the Twitter terms of service. We have slightly paraphrased the sample tweets to deidentify the users.

We leveraged a dataset from our prior work containing 70 million tweets that were posted by 23,276 users in the United States who self-reported on Twitter that they are sexual minority men [9]. We searched these tweets for parental rights in education or don’t say gay, including in hashtags, or keywords related to both legislation (legislation, bill, bills, law, laws, policy, policies, ban, bans, banning, proposal, proposals) and LGBTQ+ (lgbt, lgbtq, gay, bisexual, trans, transgender, queer, sex, gender, saygay). Our query returned 8486 tweets, excluding retweets, posted by 2857 users between September 2009 and September 2022.

We used 200 of the 8486 tweets—1 random tweet per user—to develop a codebook for identifying those that expressed health-related concerns of anti-LGBTQ+ legislation (Multimedia Appendix 1). We adopted the World Health Organization’s (WHO’s) broad definition of health as “a state of complete physical, mental and social well-being” [10]. We included nonpersonal experiences but excluded references to federal, international, or equitable legislation. To assess interannotator agreement, 2 authors (AZK and KS) coded an additional 200 tweets. After resolving disagreements, one of the authors (KS) labeled the remaining 8086 tweets.

We used 50 tweets to develop a coding scheme for categorizing the health-related concerns into one or more themes. In contrast to a deductive (theory-driven) approach, we used an inductive (data-driven) approach to thematic analysis and focused more on themes that captured explicit, rather than latent, patterns of meaning [11]. To assess interannotator agreement, 2 authors (AZK and KS) coded an additional 50 tweets. After resolving disagreements, one author (KS) coded the remaining tweets. For subgroup analysis, we used automated tools to identify the users’ geographic location based on the metadata of their tweets and user profiles (Carmen) [12] and age based on explicit self-reports in their tweets (ReportAGE) [13].

In this thematic analysis, health care was the top concern overall and across all regions and age groups. Our finding that safety was more of a concern in the Midwest and South than in the Northeast and West may reflect that many bills are being proposed in the Midwest and South, including 411 (77.5%) in 2024 [1]. Thus, concerns in the regions that would be more directly impacted by anti-LGBTQ+ legislation have focused more on specific and personal harms such as physical and life-threatening ones. Likewise, mental health was more prevalent than general harm in the Midwest and South, and support was more prevalent than general harm and human rights in the Midwest. In contrast, human rights was more prevalent than mental health and support in the West.

The variation of mental health across age groups may also reflect those who would be more directly impacted by anti-LGBTQ+ legislation—in particular, youth. Our finding that mental health was more prevalent among younger age groups supports findings that, among the general population, younger generations are more likely to report mental health concerns [16]. Although it may appear surprising that support was more prevalent in most of the older age groups, their tweets reveal many personal accounts reflecting on past experiences that raise concerns about the current generation of youth: “Had a friend text me about ‘Don’t say Gay’ bill and wanted my thoughts on it. Told him my experience as a kid and that I wish I had someone to talk to or resources about how I was feeling.”

Our findings demonstrate that social media can be used to capture health-related concerns expressed by those who are directly impacted by anti-LGBTQ+ legislation. Based on our exploratory approach using thematic analysis, we found that safety was as much of a concern as mental health [2-5] overall and across the United States and most age groups. Our sample (n=288) was substantially larger than that in prior thematic analyses [6,7], addressed their limited distribution of age [6] and geographic location [7], and given that over 40% of the users were in the South, captured the voices of individuals who are least likely to discuss their sexual orientation in traditional settings [17]. Despite these strengths, our sample primarily included cisgender men and was collected solely from Twitter, which limits the generalizability of our results to the broader LGBTQ+ community and users of other social media platforms.

We used social media for a thematic analysis of health-related concerns of anti-LGBTQ+ legislation among sexual minority men in the United States. Our findings may inform a broader range of health interventions and approaches for targeting them at specific populations of sexual minority men. By demonstrating that these concerns are expressed on social media, our findings can be leveraged by advocacy groups to amplify voices and rally public support for countering anti-LGBTQ+ bills. Directions for future work include incorporating data from more diverse LGBTQ+ identities and social media platforms.