Study Finds Variation in Accuracy of Mental Health Content

Mental health and neurodivergence information across social media platforms varies widely in accuracy, with misinformation rates reaching up to 57% in some cases, according to a comprehensive new systematic review.

The review, published in the Journal of Social Media Research, analyzed 27 studies encompassing 5,057 social media posts across multiple platforms. Researchers from the University of East Anglia’s Norwich Medical School found that, on average, 26% of mental health content contained misinformation, though rates fluctuated dramatically depending on platform and topic.

TikTok emerged as particularly problematic among the platforms studied. Content related to attention-deficit hyperactivity disorder (ADHD) on TikTok showed misinformation rates of 52%, while autism-related content on the platform contained inaccuracies 41% of the time.

YouTube performed somewhat better but still showed concerning levels of misinformation, ranging from 7% for dissociative identity disorder content to a troubling 57% for videos about magnetic resonance imaging claustrophobia. Across all YouTube content examined, researchers found an average misinformation rate of 22%.

Facebook content showed a comparatively lower average misinformation prevalence of 15%, while the single study examining content on X (formerly Twitter) reported 19% inaccurate information. Notably, YouTube Kids, which is specifically targeted at younger viewers, demonstrated better accuracy with no misinformation detected for anxiety and depression content and just 9% for ADHD-related videos.

The research team, led by Alice Carter, also identified significant differences in misinformation rates across mental health topics. Content about neurodivergent conditions generally contained more inaccuracies than other mental health topics, with autism-related misinformation reported at 40-41% and ADHD at 38-52%. In contrast, postpartum depression content showed substantially lower misinformation rates of just 3-8%.

When assessing content reliability, researchers found most social media mental health information fell short. Using the DISCERN tool, which evaluates health information quality, YouTube content scored between 31 and 36, indicating poor reliability overall. Modified DISCERN scores for other platforms ranged widely, from very poor reliability (0.4 for TikTok videos on dissociative identity disorder) to relatively high reliability (3.55 for YouTube videos on agoraphobia).

Professional credentials made a difference in content accuracy. Information produced by healthcare professionals was generally more reliable and higher quality than content from non-professionals, though some studies found minimal differences between the two groups.

The review itself had limitations, with study quality averaging about 65% and ranging from 41% to 80%. Many of the analyzed studies focused on content in a single language and lacked robust interrater reliability measures. Additionally, the researchers noted significant variation in how different studies defined and identified misinformation, making direct comparisons challenging.

“There is a need for strengthened content moderation, as well as consistent definitions and measures of mental health misinformation,” Carter and colleagues concluded in their report.

The findings come at a critical time when more people, especially young adults and teenagers, are turning to social media for mental health information and self-diagnosis. Public health experts have expressed growing concern about the potential impact of mental health misinformation, particularly on vulnerable populations seeking guidance online rather than from qualified healthcare professionals.

The researchers reported no conflicts of interest in conducting this systematic review. Their work highlights the urgent need for improved standards in mental health content across social media platforms to protect users from potentially harmful misinformation.