Observational · 2026
What online trip reports reveal about side effects of psychedelics and related drugs
- Original title
- Adverse Effect-Related Language in Recreational Psychedelic and non-Psychedelic Substance Experience Reports.
- Authors
- Hase A, Erdmann M, Calder A, et al.
- Journal
- Journal of studies on alcohol and drugs, 2026
- Molecules
- LSD, Psilocybin, MDMA, DMT, Ayahuasca, Ketamine, Other
- Conditions
- Other
- Participants
- 4226
- Evidence grade
- Low
- Source
- Read the original paper
The question
Can computer analysis of language in online drug experience reports show which unwanted effects are most associated with different psychedelic and related substances?
What they did
Researchers gathered 4,226 online reports written by people describing recreational drug experiences. They used latent semantic analysis, a technique that measures how closely text resembles other text, to compare each report with 77 reference texts about known adverse effects drawn from the research literature. Seven groups were compared: ayahuasca, isolated DMT, ketamine, LSD, MDMA, psilocybin mushrooms, and recreational antidepressant use as a non-psychedelic comparison. They also used sentiment analysis to classify whether each report was broadly positive or negative in tone.
What they found
Fifteen adverse-effect categories showed large differences between substances. The strongest were language related to vomiting, excessive drowsiness, cardiac events, sleep problems, and dissociation or depersonalization (feeling detached from oneself or one’s surroundings). Ayahuasca reports stood out for vomiting-related language, ketamine for dissociation-related language, and MDMA for drowsiness- and cardiac-related language. MDMA scored low on dissociation language, while ayahuasca and DMT scored low on sleep-problem language and DMT also low on drowsiness. A follow-up analysis found LSD and psilocybin mushroom reports scored higher than all other groups on anxiety-related language.
The fine print
This study measured word patterns in reports, not verified medical outcomes, so a high similarity score is not the same as a confirmed side effect. The reports come from people who chose to post online, which is not a representative sample, and doses, purity and other drugs taken were not controlled. The authors themselves flag the observational context as a limitation. This summary is based on the abstract only.
What it means
Language in online experience reports lines up with side effects already reported for each substance, suggesting such text may be a cheap way to compare drug risk profiles, though it cannot replace clinical safety data.
Evidence grade: Low. Observational analysis of self-selected, self-written online reports using a language-similarity measure rather than direct clinical assessment of harms.
This summary is based on the study’s published abstract.