Psychedelics and mental health: a population study

Psychedelic substances such as LSD, psilocybin and mescaline have long histories of ceremonial, religious and therapeutic use and became widely used in the 1960s. Earlier clinical studies and expert harm-assessments have not demonstrated persistent neurotoxicity, dependence, or widespread long-term psychiatric harm from these classical serotonergic psychedelics. However, despite millions of lifetime users in the United States, population-level evidence on whether lifetime psychedelic use is associated with current mental health problems remains limited. Krebs and colleagues set out to evaluate associations between lifetime use of classical psychedelics and a range of past-year mental health indicators in a large, nationally representative sample of US adults. The study aims to determine whether lifetime or recent psychedelic use is independently associated with higher rates of serious psychological distress, mental health treatment, or symptoms of eight psychiatric disorders, after adjusting for sociodemographic characteristics, other drug use, risk-taking and exposure to traumatic events. This population-level assessment is intended to complement clinical and case-based literature by providing a broad epidemiological perspective.

The investigators analysed pooled public-use data from the National Survey on Drug Use and Health (NSDUH) for 2001–2004, selecting participants aged 18 years or older. The NSDUH uses a randomly selected, nationally representative civilian non-institutionalised sample; trained interviewers conducted household interviews in a confidential, audio-assisted computer format. Half of 2004's participants were excluded because of question changes; the final analytic sample comprised 130,152 respondents and the reported response rate was 78%. Approximately 10% of respondents were omitted from the public file for data quality or anonymity reasons. Psychedelic exposure was defined as lifetime use of LSD, psilocybin, mescaline or peyote; mescaline and peyote were analysed both combined and separately. Past-year LSD use was also available. Mental health outcomes comprised multiple past-year indicators: serious psychological distress measured by the K6 scale (cut-point 13+), four measures of mental health treatment (inpatient, outpatient, psychiatric medication prescription, and felt a need but did not receive treatment), symptom-based indicators for eight DSM-IV disorders derived from the short-form CIDI (panic disorder, major depressive episode, mania, social phobia, generalized anxiety disorder, agoraphobia, posttraumatic stress disorder, and non-affective psychosis), and the seven individual psychotic symptoms (with non-affective psychosis defined as two or more psychotic symptoms). A broad set of control variables was included to reduce confounding: age, gender, race/ethnicity (seven categories), education, household income, marital status, propensity for risky behaviour, lifetime exposure to an extremely stressful event, and lifetime non-medical use of ten classes of drugs (including cannabis, opiates, cocaine, benzodiazepines/barbiturates, stimulants, MDMA, inhalants, poppers and PCP). For analyses of past-year LSD use, past-year use of these drugs was also controlled for (with a single inhalant variable). Multivariate logistic regression models produced adjusted odds ratios (aOR) with 95% confidence intervals and p-values; all analyses accounted for NSDUH sampling weights and complex design and were run using SPSS Complex Samples. The authors report checks for multicollinearity (variance inflation factors under 2.5), consideration of events-per-variable recommendations for logistic regression, handling of missing data (participants with missing data on relevant outcomes or past-year illicit drug use were excluded), and small sensitivity checks that set missing items to "no" with minimal effect.

In the pooled sample of 130,152 adults, 21,979 respondents reported lifetime use of any psychedelic, representing 13.4% weighted. Bivariate comparisons showed psychedelic users were more likely to be younger, male, white or Native American, have higher education and income, be unmarried, engage in risky behaviours, report exposure to extremely stressful events, and to have used other illicit drugs; before adjustment, psychedelic users had higher rates of mental health indicators. After multivariate adjustment, lifetime use of any psychedelic was not associated with higher odds of any past-year mental health outcome evaluated. For serious psychological distress (K6), lifetime use of any psychedelic showed no significant association; however, some specific substances were associated with lower odds: lifetime psilocybin (aOR 0.8, p = 0.009), lifetime mescaline (aOR 0.9, p = 0.04), and past-year LSD use (aOR 0.7, p = 0.01). Regarding mental health treatment outcomes, lifetime psychedelic use overall was not associated with inpatient, outpatient, medication or unmet need. Specific findings included lower odds of outpatient treatment (lifetime LSD, aOR 0.9, p = 0.002) and psychiatric medication prescription (lifetime LSD, aOR 0.9, p = 0.04; lifetime psilocybin, aOR 0.8, p = 0.00008; lifetime mescaline/peyote, aOR 0.8, p = 0.004; lifetime peyote, aOR 0.8, p = 0.01). Lifetime psilocybin use was associated with lower odds of inpatient mental health treatment (aOR 0.8, p = 0.04). On the CIDI-SF psychiatric symptom indicators, lifetime psychedelic use was not significantly associated with any of the eight disorder symptom sets (aORs ranged 0.8–1.1). There was an inverse association for one psychotic symptom — "felt a force taking over your mind" — (any psychedelic aOR 0.7, p = 0.03). Substance-specific associations included lower odds of panic symptoms with psilocybin (aOR 0.9, p = 0.006) and lower odds of agoraphobia symptoms with mescaline/peyote (aOR 0.6, p = 0.005). Psilocybin and mescaline/peyote were each associated with lower odds of the specific psychotic symptom mentioned above (psilocybin aOR 0.6, p = 0.004; mescaline/peyote aOR 0.7, p = 0.04). Stratified analyses (by sex, age group, past-year illicit drug use, and lifetime extremely stressful event) found no evidence that lifetime psychedelic use increased risk of any mental health outcome. Instead, the authors report a number of instances where psychedelic use was associated with lower odds of certain outcomes, although many were of marginal statistical significance (0.05 > p > 0.01). Notable subgroup findings included a lower odds of the psychotic symptom "felt force taking over mind" among females (aOR 0.5, 95% CI 0.3–0.7, p = 0.0005), lower odds of generalized anxiety symptoms among those aged 18–25 (aOR 0.8, 95% CI 0.6–1.0, p = 0.03), and several lower odds among past-year illicit drug users (for example, inpatient treatment aOR 0.7, 95% CI 0.5–0.9, p = 0.02). Among respondents without a lifetime extremely stressful event, psychedelic users had lower odds of psychosis symptoms (aOR 0.5, 95% CI 0.3–0.9, p = 0.03) and certain psychotic symptoms (e.g. "felt force inserting thoughts" aOR 0.4, p = 0.02; "felt force steal thoughts" aOR 0.3, p = 0.008). Native American respondents had relatively high weighted prevalence of lifetime psychedelic use (25%) and lifetime peyote use (14%), but they comprised less than 1% of lifetime psychedelic users in absolute numbers and less than 3% of lifetime peyote users. Excluding Native Americans changed adjusted odds ratios by less than 2% on average and rendered one mescaline/peyote association non-significant (p = 0.06). Missing data affected under 2% of observations per analysis, and sensitivity analyses treating missing as "no" produced minimal changes (under 4% on average) and did not alter statistical significance.

Krebs and colleagues conclude that, in this large representative sample, lifetime use of classical serotonergic psychedelics was not associated with increased odds of past-year mental health problems. The adjusted odds ratios for associations with psychedelic use were small (no aOR exceeded 1.2), and in several instances specific psychedelics were associated with modestly lower odds of certain outcomes. The authors note that these inverse associations could reflect potential beneficial effects of psychedelic use, better baseline mental health among people who try psychedelics, or chance findings arising from multiple comparisons. The findings are described as consistent with clinical trials in healthy volunteers and with prior population and case-control studies that do not implicate psychedelics as causes of lasting anxiety, depression or psychosis. The authors discuss the contested diagnosis of hallucinogen persisting perceptual disorder (HPPD), arguing that many symptoms attributed to HPPD are also present in people who have never used psychedelics and that the causal link remains unproven. Key limitations acknowledged by the investigators include the cross-sectional, retrospective design which precludes causal inference; absence of important covariates such as family psychiatric history; reliance on screening instruments rather than full diagnostic interviews; potential recall bias and under-reporting of drug use; lack of data on dose, drug purity or context of use; and omission of institutionalised populations (prisoners, some hospitalised patients, military personnel). The authors also note they did not adjust for multiple comparisons, so some marginally significant inverse associations may be false positives. Finally, the discussion addresses the limitations of case reports, emphasising that isolated case accounts often fail to exclude pre-existing psychiatric vulnerability, coincident timing of onset, other drug use, or recall and attribution biases. The authors present several reasons why case reports may overstate causal links between psychedelics and long-term psychiatric disorders, and they indicate their population-level results do not support the notion that psychedelic use, at the population level, is an independent risk factor for persistent mental health problems.