More than half of videos about Attention Deficit Hyperactivity Disorder (ADHD) shared on TikTok contain misinformation, a new study has revealed.
Researchers found social media platforms are “awash” with misleading or unsubstantiated mental health content – and that this is the worst on TikTok.
They analysed more than 5,000 social media posts about mental health topics – including autism, ADHD, schizophrenia, bipolar disorder, depression, eating disorders, OCD, anxiety and phobias – across social media platforms, such as TikTok, YouTube, Facebook, Instagram and X (formerly Twitter), to investigate the accuracy of mental health and neurodivergence information.
According to the findings, 52 per cent of ADHD-related videos and 41 per cent of autism videos analysed on TikTok were inaccurate, with the platform frequently found to contain higher levels of misinformation in its mental health content than other platforms.
The company claimed the study is “flawed” and said it removes harmful health misinformation.
The team, from the University of East Anglia, also found that posts about neurodivergence such as autism and ADHD contained higher levels of misinformation than many other mental health topics.
Dr Eleanor Chatburn, from UEA’s Norwich Medical School, said: “Our work uncovered misinformation rates on social media as high as 56 per cent. This highlights how easily engaging videos can spread widely online, even when the information isn’t always accurate.
“Social media has become an important place where many young people learn about mental health, but the quality of this information can vary greatly. This means that misleading content can circulate quickly, particularly if there aren’t accessible and reliable sources available.”
The review, published in The Journal of Social Media Research, marks the first to examine mental health and neurodivergence information across multiple social media platforms.
YouTube averaged 22 per cent misinformation, according to the findings, while Facebook averaged just under 15 per cent.
The study found that content created by healthcare professionals was consistently more accurate, but professional voices still represent only a small share of mental health content circulating on these platforms.
The researchers are calling for health organisations and clinicians to create and promote better evidence-based content, as well as for improved content moderation, standardised tools for assessing online mental health information, and clearer definitions of misinformation.
Dr Chatburn warned: “As well as leading to misunderstanding of serious conditions and pathologising ordinary behaviour, misinformation can also lead to delayed diagnosis for people that actually do need help.
“When false ideas spread, they can feed stigma and make people less likely to reach out for support when they really need it.
“It can also make mental illness seem scary or hopeless, which creates even more fear and misunderstanding.
“On top of that, when people come across misleading advice about treatments, especially ones that aren’t backed by evidence, it can delay them from getting proper care and ultimately make things worse.”
A TikTok spokesperson said: “This is a flawed study that relies on outdated research about multiple platforms. The facts are that we remove harmful health misinformation and provide access to reliable information from the WHO, so that our community can express themselves about what matters to them and find support.”
The company added that it has launched the UK Clinician Creator Network, a group of 19 NHS-qualified clinicians who share their medical expertise on the platform to over 2.2 million followers.
YouTube, Facebook, Instagram and X have been approached for comment.
AI video maker ad banned for exposing woman’s body
Meta failed to stop illegal financial ads in UK 1,000 times in just a week
AI advert banned over claims it could expose women’s bodies
Senators demand shutdown of new app that lets users make AI versions of real people
DoorDash wants to pay you to film your ‘everyday tasks’ to help train AI
