The hallucinogen-psychosis link mostly vanished after adjusting for prior illness

Whether hallucinogen use directly causes psychosis, or whether the association reflects something else — most obviously, that people prone to psychosis are also more likely to use hallucinogens, or to be more severely affected when they do — has been a recurring, unresolved question shaping drug policy debates. A study published this month in the Journal of Clinical Psychiatry tests directly how much of the observed hallucinogen-psychosis link is explained by patients' preexisting psychiatric history [s1].

The design

Researchers conducted a retrospective analysis using MarketScan Medicaid and commercial insurance claims databases from 2015 to 2019, focused on a population-based sample of individuals who had received substance use disorder treatment following a documented substance-related emergency admission or hospitalization [s1]. They compared rates of psychosis-related admissions occurring 30 days to 6 months after the index substance-related event between two groups: people whose index event was hallucinogen-related, and people whose index event involved a different, nonhallucinogen substance [s1]. Cox regression models were used, first unadjusted and then adjusted for demographics and clinical characteristics including prior psychiatric history, to test whether that adjustment explained the observed association [s1].

What it found

Among 273,466 individuals with substance-related admissions, psychosis diagnoses were markedly more common following hallucinogen-related admissions — 16.4% — compared with 6.6% following nonhallucinogen substance-related admissions (p < .001) [s1]. In the unadjusted model, hallucinogen-related admissions carried an increased psychosis risk: hazard ratio 1.22 (95% CI 1.19–1.25) [s1]. But once the model adjusted for clinical characteristics — centrally, prior psychiatric history — that association became nonsignificant: hazard ratio 0.97 (95% CI 0.95–1.00) [s1].

Why this specific before-and-after comparison is the whole story

This study's entire value lies in the contrast between its unadjusted and adjusted results, and that contrast is stark: an association that looked real and statistically robust in the raw data (HR 1.22, clearly excluding 1.0) essentially vanished — landing right at or just below 1.0 — once baseline psychiatric history was accounted for [s1]. That pattern is close to a textbook demonstration of confounding: a variable (prior psychiatric history) that's associated both with the exposure (being more likely to use or be hospitalized for hallucinogen use) and with the outcome (being more likely to develop psychosis) can produce an apparent association between the exposure and outcome that isn't actually causal, once you know to look for and adjust for it.

What this means for the population it actually studied

It's worth being precise about who this study describes: not the general population of people who use hallucinogens, but specifically people who had already experienced a substance-related emergency admission or hospitalization severe enough to warrant it — a population already selected for significant substance use problems and, quite plausibly, elevated rates of comorbid psychiatric conditions. The 16.4% raw psychosis rate following hallucinogen-related admissions in this population is not a general-population risk estimate, and this study doesn't attempt to characterize psychosis risk among hallucinogen users broadly, including the much larger population who never has a substance-related hospitalization at all.

Why the authors frame this as a policy-relevant finding

The study's authors state their aim directly: examining "the extent to which observed associations between hallucinogen use and psychosis are mediated by preexisting psychiatric conditions, providing important evidence for psychedelic policy formulation" [s1]. Their conclusion argues that population-level epidemiologic associations between hallucinogens and psychosis — the kind of statistic that gets cited in policy debates about regulating or restricting psychedelics — may often reflect underlying vulnerability factors in the population being studied, rather than a direct causal effect of hallucinogens themselves [s1].

What this doesn't establish

This finding, while methodologically informative, doesn't mean hallucinogens carry zero psychosis risk for anyone — it specifically addresses whether the population-level association in this claims-database sample is explained by confounding, not whether hallucinogens could still meaningfully worsen or trigger psychotic symptoms in a person with an existing vulnerability to psychosis, a question this study's design isn't built to answer definitively. Claims-database diagnostic codes are an imperfect proxy for confirmed clinical status, both for the "hallucinogen-related" exposure classification and for psychiatric history. The 30-day-to-6-month window studied captures a specific timeframe that may not reflect psychosis risk over a longer horizon.

What to watch

Whether this confounding-adjusted finding is replicated in other claims databases or prospective cohorts, and how it factors into ongoing drug policy and psychedelic regulation discussions that have historically cited raw, unadjusted association statistics. This article is not medical advice.

Sources

1. Hallucinogen-Psychosis Associations Are Confounded by Baseline Psychiatric History — Journal of Clinical Psychiatry, 10 June 2026