Posttraumatic Stress Disorder Content on TikTok: Cross-Sectional Analysis of Popular #PTSD Posts.

Background: TikTok became an increasingly popular platform for mental health discussions during a major global stressor (COVID-19 pandemic). On TikTok, content assumed to promote user engagement is delivered in a hyperindividually curated manner through a proprietary algorithm. Mental health providers have raised concerns about TikTok's potential role in promoting inaccurate self-diagnoses, pathologizing normal behaviors, and fostering new-onset symptoms after exposure to illness-related content, such as tic-like movements linked to conversion or factitious disorders. The accuracy of posttraumatic stress disorder (PTSD)-related content with respect to conveying symptoms, diagnosis, and treatment deserves further investigation. Objective: This study aimed to characterize the accuracy of PTSD-related TikTok content. Methods: In February 2022, a search was conducted on TikTok using the hashtag #PTSD, and the 100 most-liked videos were retrieved. Videos were excluded if they were in a non-English language, duplicated, unrelated to PTSD, lacked audio, or contained nonfunctioning links. A publicly available Python package (TikTokPy) was used to scrape available metadata (views, shares, etc). Using the Patient Education Materials Assessment Tool for Audiovisual Materials (PEMAT-AV), videos were independently coded by 2 reviewers for the overall accuracy of the video (useful, personal experience, or misleading), whether the creator self-identified as a health care professional, symptoms mentioned, and overall video understandability and actionability. A third reviewer was consulted in the rare instances of coding disagreements. Results: Of the 100 included videos, 29 were classified as useful, 59 were classified as personal experience (subjective experience without outright inaccuracies), and 12 were classified as misleading. The degree to which PTSD-related information was accurate was not associated with its understandability, actionability, or user engagement. Besides useful videos being longer (mean 88.7, SD 63.1 seconds) than personal experience videos (mean 42.7, SD 44.5 seconds), no group differences in video metadata were observed across the number of views, likes, shares, or comments (P>.05). While self-identified HCPs were more likely to post useful content, they also contributed to 33% (4/12) of misleading videos. Changes in cognition and mood were the most frequently reported symptoms of PTSD (38/100, 38% of videos). Conclusions: Our findings were roughly consistent with previous studies of mental health-related TikTok content accuracy, although this is variable by diagnosis. TikTok's continuously adaptive algorithmic content delivery may expose users to nonspecific and potentially misleading "click-bait" mental health information, which could influence symptom interpretation and clinical presentation. Clinicians should be aware of the digital landscape shaping patients' perceptions of PTSD.