Over time, Facebook has evolved into one of the most globally used platforms, attracting over 3 billion annual users [17]. Approximately 71% of US adults report using the platform with disproportionately higher usage among older adults and women compared to other platforms [18]. This pattern aligns with GLP-1 use among women who are more likely to be currently taking a GLP-1 RA [19]. Facebook posts were collected using CrowdTangle, a publicly available tool that enables systematic exploration of content across social media platforms [20]. CrowdTangle provides data from public accounts (7 million Facebook pages, groups, and verified profiles) and allows users to refine results using parameters such as keywords, timeframes, and geolocation [20]. Currently available tools for researchers include Meta Content Library and Content Library API [21]. For this study, data were collected using keyword-based filters, yielding information such as post content, image URLs, captions, dates of posts, and engagement metrics.

Data from English-language posts originating from US-based users were collected between January 1, 2022, and May 31, 2024. GLP-1 RA use has surged since 2022. Tirzepatide was approved by the US Food and Drug Administration (FDA) for diabetes management in May 2022 and was quickly used for off-label obesity management before receiving FDA approval in June 2023 [22-24]. To examine the rise in GLP-1 RA discussions online, we used data collected through CrowdTangle from January 2022 to May 2024; CrowdTangle was discontinued in August 2024 [21]. The keyword list (Table 1) was developed based on a combination of prior literature and clinical expertise. A total of 64,202 posts were retrieved. Text preprocessing included the removal of symbols, hyperlinks, HTML tags, and other special characters. Key terms were identified using regular expression methods.

To identify key topics across Facebook posts, we applied BERTopic, a modern topic modeling approach that uses advanced language models to group similar content based on meaning [25]. Unlike traditional techniques such as latent Dirichlet allocation, which rely on word frequency patterns, BERTopic captures deeper contextual relationships between words, allowing for more meaningful and nuanced topic discovery [26]. The process began by converting each post into a numerical representation of its meaning using a pretrained language model. These representations were then simplified using uniform manifold approximation and projection (UMAP), which helps organize complex data. Next, a clustering algorithm known as hierarchical density-based spatial clustering of applications with noise grouped the posts into clusters based on similarity [27,28]. To interpret each topic, we used a modified version of term frequency–inverse document frequency (TF-IDF) called class-based TF-IDF, which highlights the most distinctive words within each group [25]. This technique improves the clarity of each topic and sets BERTopic apart from other topic modeling tools.

BERTopic was implemented in Python (Python Software Foundation) using the BERTopic library, with default settings for dimensionality reduction and clustering [25]. Several exploratory runs were conducted by adjusting parameters to evaluate the stability and interpretability of the results. Four parameters in total were adjusted: UMAP dimensionality reduction parameters (n_neighbors, n_components) and BERTopic parameters (nr_topics, min_topic_size). In the final model, we used n_neighbors=13, n_components=9, nr_topics=5, and min_topic_size=100. The final model was selected based on semantic coherence and the clarity of topic keywords. Following deduplication, the dataset consisted of 50,013 unique Facebook posts. These topics provided an initial thematic structure that helped organize the large Facebook dataset and informed the subsequent qualitative content analysis.

A thematic analysis was conducted on a random sample of 500 Facebook posts for each of the 5 topics identified using BERTopic, for a total of 2500 posts. Thematic analysis is a widely used qualitative method that allows for the analysis of free-response data using an established codebook [29]. First, the authors implemented an inductive approach by reviewing 20 posts from each subcategory and discussing emerging themes to establish an understanding of the dataset through familiarization [29]. Then, the authors constructed themes through pattern formation and identification, assessed thematic diversity, and tested the codebook’s operational abilities. The final codebook comprised 13 overarching themes, each containing unique subthemes (Table 2). Once the codebook was finalized, 2 authors independently conducted postscreening of the 2500 posts. Conflicts were resolved through discussion between the two reviewers until consensus was reached based on detailed justifications. The finalized themes and subthemes were then applied to guide a comprehensive thematic analysis of the sample, offering insight into the range of topics discussed in public discourse surrounding GLP-1 RA medications on Facebook.

This study did not involve human participants and was conducted in accordance with ethical standards determined by the University of Maryland College Park Institutional Review Board (2072551‐1). Given the nature of the study, examination of publicly available social media posts did not involve direct interaction or communication with human subjects. Additionally, posts were anonymized to uphold user privacy.

BERTopic was used to analyze 50,013 posts referencing GLP-RAs and initially explored several configurations. A target topic size of 6 with a minimum topic size of 100 posts was chosen to prevent overly granular clusters. One topic was automatically flagged as a low-density outlier, yielding a total of five topics. The 5-topic model was then selected based on the clarity of cluster separation visualized in the intertopic distance map (Multimedia Appendix 1), which is a 2D UMAP projection of topic embeddings representing relative semantic distances between topics. There was no overlap between the 5 topics, and comparable distances were demonstrated, supporting the validity of the model.

Topics were categorized using the top keywords identified using class-based TF-IDF scores (Figure 1). BERTopic modeling yielded 5 distinct topics: healthy weight management, weight loss medications, semaglutide, public figures, and Ozempic. These broad topics serve as a foundation for understanding the dominant narratives surrounding weight loss drugs on social media. They provide a structured lens to interpret public discourse, identify recurring patterns, and uncover underlying concerns and motivations.

Topic 0 (healthy weight management) emerged as a prominent theme characterized by keywords such as “weight,” “health,” and “management.” Posts in this category focused on lifestyle-based approaches to combating obesity. Topic 1 (weight loss medications) captured broader discussions about pharmaceutical interventions, featuring user experiences, medication comparisons, and commentary on side effects. Topic 2, characterized by the keyword “semaglutide,” reflected in-depth discourse on the clinical use and effectiveness of this specific medication. Topic 3 (public figures) centered on the role of celebrities and influencers in shaping perceptions of weight loss treatments. Finally, topic 4 (Ozempic) focused on its off-label use for weight loss, concerns around access and affordability, and growing public interest.

Then, a content analysis was conducted on a random sample of 2500 Facebook posts. As depicted in Table 2, 13 distinct themes emerged during the qualitative analysis, each comprising various subthemes. The most popular themes across the sample included weight management programs, neutral weight loss medication posts, harms of weight loss medications, promotion of weight loss medications, and access to weight loss medications. Table 3 provides a detailed overview of each theme and subtheme, accompanied by illustrative quotations to highlight specific examples of Facebook user statements. A total of 89 posts were excluded from the final analysis because they lacked substantive references to weight management or the population of interest.

The five BERTopic-derived topics included (1) healthy weight management, (2) weight loss medications, (3) semaglutide, (4) public figures and media, and (5) Ozempic. Across the 5 topics, several overarching themes emerged, such as natural and whole foods and supplements, physical activity, vegan or vegetarian diets, attitudes toward weight loss medications, speculation about medication use, preference for nonpharmacologic options, access to weight loss medications, weight loss programs, commercial expansion, hormone therapy, and research studies. Additionally, post counts for each theme are described across the 5 BERTopic-derived topics (Figures 2-4). The 5 most prevalent themes were as follows: “weight management programs,” “weight loss medications—neutral,” “weight loss medications—harms,” “weight loss medications—promotion,” and “access to weight loss medications.” “Weight management programs” referred primarily to program promotion, supplement sales, and company advertisements. “Weight loss medications—neutral” included informational content and regulatory updates. “Weight loss medications—harms” referred to negative aspects such as adverse events and exploitative messaging. “Weight loss medications—promotion” focused on perceived benefits, and “access to weight loss medications” included issues related to cost, insurance, and availability. Illustrative examples of posts for each theme and subtheme are presented in Table 3.

The most common themes in this group were “weight management programs,” including 279 (55.8%) posts; “natural and whole foods and supplements,” including 84 (17%) posts; and “weight loss medications—neutral,” including 31 (6%) posts. Most “weight management programs” posts (n=245) promoted programs, while fewer (n=34) offered health advice. Most posts referenced health outcomes, such as diabetes, obesity, and cardiovascular disease. “Weight loss medications—neutral” posts included information, expert opinions, or federal guidance on weight loss medications. This category also included notable references to physical activity (n=30).

In this subcategory, the most common themes included “access to weight loss medications,” comprising 106 (21.2%) posts; “weight loss medications—neutral,” comprising 99 (20%) posts; “weight loss medications—promotion,” comprising 67 (13%) posts; and “weight loss medications—harms,” comprising 67 (13.4%) posts. Access posts referenced availability (n=46), cost (n=15), insurance coverage (n=26), and lawsuits or investigations (n=19). Neutral posts included educational content, with fewer posts mapping to the “updates on regulations” subtheme. Promotional posts highlighted weight loss (n=30), health benefits (n=9), both weight loss and health benefits (n=18), and celebrity posts (n=10), allowing for further discussion regarding the weight loss benefits of these medications. With identical theme counts, posts focused on harms raised concerns about health risks, exploiting insecurities, excessive weight loss, temporary results, and celebrity posts.

The distribution of posts was dominated by “weight management programs,” including 377 (75.4%) posts, followed by “weight loss medications—neutral,” including 47 (9%) posts and “promotion,” including 31 (6%) posts. Most posts related to this topic promoted paid or supplement-based weight management programs (n=373), consistent with the commercial focus observed across the full dataset. Neutral posts covered educational content and webinar information (n=45) or regulatory updates (n=2), while promotional posts centered around weight loss success (n=23) and weight loss combined with health benefits (n=7).

This subcategory involved weight loss medication posts, with “neutral” accounting for 149 (29.8%) posts, “harms” accounting for 138 (27.6%) posts, and “promotion” comprising 126 (25.2%) posts. “Neutral” posts included educational facts or webinar information (n=148) and FDA regulation updates (n=1). Harm-related posts frequently referenced celebrities (n=114). Some discussed how weight loss medications harm one’s health (n=18), while others noted the exploitation of insecurities (n=6). This category also included posts in which celebrities were critical of weight loss medications or public complaints regarding celebrity weight loss medication usage. Promotional posts highlighted celebrities (n=115), including reports of positive results. Additional subthemes from positive posts included weight loss (n=6), health benefits with weight loss (n=3), and general health benefits (n=2). There was an almost even divide of posts about celebrities in the harm and promotion themes regarding weight loss medications. The public figures and media subcategory had the highest number of posts surrounding celebrities, with a total of 229 posts.

The most common themes were “weight loss medications—harms,” including 149 (29.8%) posts; “neutral,” including 140 (28%) posts; and “promotion,” including 99 (20%) posts. Harms of weight loss medications included adverse events (n=51), exploiting insecurities (n=5), excessive weight loss (n=1), temporary results (n=2), and celebrity posts (n=90). The discussion of the negative impacts of Ozempic on Facebook was more common than that of its positive effects. The majority of neutral weight loss medication posts were linked to education (n=134), with only a few mapping to the regulatory subtheme (n=6). For positive posts linked to weight loss medications, celebrity posts were identified most frequently, while posts highlighting weight loss, health benefits, or both appeared less frequently. Unlike other subcategories, Ozempic included the most posts linked to the harms of weight loss medications (n=149).

The final most prevalent theme included access to weight loss medications, which encompassed GLP-1 RA availability, affordability, lawsuits, and investigations. Many users were concerned about GLP-1 RA shortages: “Can’t get my Ozempic and am out. Anyone know which pharmacy has the 2 in stock?” Furthermore, the increasing costs presented another accessibility barrier: “Zepbound has a much lower sticker price than its rival, Novo Nordisk’s Wegovy.” Additionally, insurance coverage was described across numerous posts: “Some lawmakers say the United States cannot afford to keep a decades-old law that prohibits Medicare from paying for new weight loss drugs.” Finally, lawsuits and investigations often discussed regulatory pricing and other guidelines that may impact public access to the medications: “Bernie Sanders called a vote to subpoena the chief of Novo Nordisk’s U.S. division over pricing of Ozempic and Wegovy.”

Overall, social media discussions about weight management revealed a complex landscape with a myriad of themes and a wide spectrum of public sentiment. Social media users shared their weight loss successes and perceived health benefits, as well as adverse physical and psychological events they experienced. The data also revealed exposure to marketing, including a variety of promotional programs and advertisements. Public figures and celebrities were commonly featured in messaging related to endorsements and critical discourse on weight loss medications. Finally, posts that discussed medication access underscored significant concerns about GLP-1 RA shortages, high costs, limited insurance coverage, and ongoing lawsuits that could impact weight loss medication pricing and availability.

Mixed methods analysis of Facebook posts discussing GLP-1 RA medications revealed 13 distinct thematic patterns across 5 unique topics. The most prevalent themes included weight management programs, weight loss medications, and access to weight loss medications. Healthy weight management posts primarily featured weight loss programs, natural and whole foods, supplements, and physical activity. In the weight loss medications subcategory, access and educational posts were most frequently identified. Semaglutide-related posts were dominated by program promotions, while public figures and media primarily focused on weight loss medications, with a high percentage of celebrity influence in both harm and promotional contexts. Ozempic posts most frequently discussed harms and education, with fewer posts promoting the use of weight loss medications. Finally, online discourse was heavily shaped by commercial interests, personal experiences, and celebrity influence, highlighting a need for evidence-based messaging around weight loss interventions.

Our findings highlight novel themes across identified topics while also confirming current sentiments regarding weight loss medications among social media users. Facebook users often shared personal experiences, motivations for GLP-1 RA use, and research supporting the use of weight loss medications. These sentiments align with existing studies that describe the benefits of GLP-1 RA use for sustainable weight loss [30]. In addition to physical health benefits, studies have documented the mental health benefits of weight loss medications. A recent systematic review reported associations between GLP-1 RAs and reductions in depressive symptoms, suicidal ideation, alcohol use, substance use, and binge eating behaviors [31]. However, some adverse psychiatric outcomes have been documented in pharmacovigilance analyses, with varying effects across different types of weight loss medications [31].

Beyond clinical outcomes, prior research highlights the stigma surrounding GLP-1 agonists, rooted in cultural ideals favoring “disciplined” weight loss—an image embedded in fatphobic ideologies [13]. High costs, limited availability, and lack of insurance coverage exacerbate class-based fatphobia, disproportionately affecting lower-income individuals with fewer options for pharmacological treatment [13]. Scholars emphasize that sustainable cultural change requires initiatives beyond medical interventions, including public health campaigns and educational resources that foster body diversity, empathy, and an understanding of the complexity of obesity as a multifaceted health condition [13].

The widespread promotion of GLP-1 RAs has been heavily influenced by celebrity culture where the pressure to maintain one’s physical appearance and weight is critical. Weight management culture in the United States is entangled with long-standing trends of fatphobia, body image pressures, and evolving medical approaches to obesity, as evidenced by increased usage of GLP-1 RAs [13]. Celebrities’ early adoption of GLP-1 RAs amplifies the desirability of these medications. High-profile celebrities like Oprah Winfrey have acknowledged their use of weight loss medications, while others are widely suspected of doing the same [32]. The growing visibility of celebrity use of weight loss medications can further influence public attitudes toward obesity, fatphobia, and body image [13].

By analyzing 2500 randomly selected Facebook posts, this study offers insights into public perceptions and information-sharing practices surrounding weight loss medications. By analyzing diverse subcategories, we captured a wide range of content, including educational, commercial, and celebrity influences. This study addresses existing data gaps, such as social media discourse and the cultural framing of GLP-1 RA medications, which are relevant to public health communication, patient education, and policy development. Additionally, this study employed BERTopic, a modeling technique that captures contextual and semantic relationships in text and is well suited to informal social media discourse. Compared to traditional models, BERTopic enhances topic coherence, dynamic reduction, and outlier detection. After removing duplicates and outliers, we randomly sampled 2500 posts for detailed analysis, ensuring a diverse dataset spanning educational content, personal experiences, and commercial promotion.

However, several limitations merit attention. First, the analysis was limited to Facebook and may not reflect discussions occurring on other social media platforms. Additionally, Crowdtangle data did not provide information on the type of accounts that originated from the included Facebook posts. Second, while the keyword list was informed by prior literature and author expertise, it was not exhaustive, potentially missing relevant online discussions. Third, we lacked demographic and engagement metrics (eg, likes and shares), which could have offered additional insights for interpreting influence. Finally, the analysis was restricted to English-language posts, limiting generalizability to non-English-speaking communities.

This study underscores the complexity of social media discourse surrounding GLP-1 RAs and the intersection of commercial, clinical, and cultural narratives. As weight loss medications gain increasing popularity across diverse populations, it is imperative to ensure that their use is not only effective but also safe. This mixed methods study highlights the depth and complexity of social media discourse surrounding GLP-1 RAs as an interweaving narrative of commercial, clinical, and cultural perspectives. Our analysis reveals that GLP-1 RA use is shaped by individual and social contexts, including personal values, lived experiences, expectations, and stigma. Future studies should build upon this descriptive and exploratory study to examine long-term patterns of GLP-1 RA use, benefits, and side effects across diverse populations. Additionally, future studies can consider social media platform diversity and how social media users navigate sources of evidence-based information to counteract the spread of misinformation online. Research is also needed to support understanding of user experiences with non-FDA-approved products (such as compounded drugs) or medications obtained outside of traditional health care settings (such as directly from the manufacturer), as data on adverse effects would not be fully captured in electronic health records or claims. Our analysis lays important groundwork for future work focused on understanding GLP-1RAs through a more complete, person-centered lens—a critical component of evidence-based medicine.