Back

Analysis of online information available for treatment of depression

Ghezzi, P.; Clarke, Z.

2019-10-14 psychiatry and clinical psychology
10.1101/19008284 medRxiv
Show abstract

The Internet has become a prime source of health information available to the public. Our aims were to assess the content and quality of online information for the treatment of depression. We searched, "How to cure depression" on Google and analysed the first 200 websites according to the website typology and a standard health information assessment tool, the Journal of the American Medical Association (JAMA) criteria (presence of authorship, date, disclosure, and references). We also analyzed content in terms of treatments mentioned and developed a Quality Indicator Score (QIS) based on the guidelines for treating depression from the UK National Institute for Health and Care Excellence (NICE). News websites were the most frequent typology followed by health portals, non-profit, professional and government; commercial websites were the least represented. In the top ten websites, news and health portals remained first and second respectively. Antidepressants were the most mentioned treatment, followed by psychotherapy, lifestyle & exercise, social support, diet. The least mentioned interventions were sunlight & light therapy, routine, and ketamine & psychedelics. Commercial websites preferentially mentioned supplements, while ketamine and psychedelic drugs were the most covered by news outlets. Analysing webpages according to our QIS showed the median of NICE recommended treatments was 2.5 out of 5 possible treatments: antidepressants, lifestyle & exercise, psychotherapy, social support and ECT. Government websites had the highest QIS, news and commercial websites the lowest. Webpages with high QIS ranked higher in Google.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.