Predicting responses to psychedelics: a prospective study

Haijen and colleagues situate this study within the recent resurgence of clinical and experimental work on serotonergic psychedelics, noting that single-dose administrations (notably of psilocybin) have been reported to produce enduring improvements in psychological well-being and promising clinical effects across diverse disorders. They point out that, despite accumulating evidence of therapeutic potential, the ability to predict individual acute and longer-term responses to psychedelics remains limited. Prior work has linked acute ‘‘mystical’’ or ‘‘peak’’ experiences and trait absorption to favourable outcomes and identified some predictors of challenging experiences, but most existing studies are limited either to controlled settings (often with psilocybin only) or to prediction of acute effects rather than longer-term change. This paper therefore set out to collect prospective, naturalistic data to improve prediction of both the acute psychedelic experience and subsequent changes in psychological well-being. Using an online, longitudinal survey administered at five time points (1 week before, 1 day before, 1 day after, 2 weeks after and 4 weeks after a planned psychedelic experience), the investigators aimed to test two primary hypotheses: that subjective well-being (measured by the WEMWBS) would increase at 2 weeks post‑experience, and that features of the acute experience—particularly mystical-type experiences—would predict later changes in well-being. They also planned to examine the relative contributions of baseline trait variables, ‘‘set’’ and ‘‘setting’’ factors, intentions for the experience, and drug dose to acute and longer-term outcomes.

Haijen and colleagues used a prospective, opportunity-sampling design with web-based data collection. Inclusion criteria were age 18 or over, adequate English comprehension, and an intention to take a classic psychedelic in the near future (psilocybin/magic mushrooms/truffles, LSD/1P-LSD, ayahuasca, DMT/5‑MeO‑DMT, salvia divinorum, mescaline, or iboga/ibogaine). Participants registered on a bespoke website, provided informed consent, supplied a planned date for their experience, and received emails with unique identification links to complete five surveys timed to that date. The survey battery combined established instruments and a set of self-constructed items. Primary outcome was psychological well-being assessed by the 14-item Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS). Baseline trait measures included the Ten‑Item Personality Inventory (TIPI) for the Big Five domains, a shortened modified Tellegen Absorption Scale (MODTAS; 25 items), the short Spielberger State‑Trait Anxiety Inventory (STAI‑SF), the Short Suggestibility Scale (SSS) and a Stubborn Opinionatedness scale (SOP). Acute measures administered one day after the experience comprised the Altered States of Consciousness scale (OAV/11D‑ASC), from which a composite ‘‘visual effects’’ score was derived (elementary imagery, complex imagery, audio‑visual synaesthesia), the 30‑item Mystical Experience Questionnaire (MEQ), and the 30‑item Challenging Experience Questionnaire (CEQ). Doses were reported post‑hoc and converted into LSD-equivalent categories (low to extremely high). Several self-constructed items captured ‘‘set’’ (pre-state), aspects of the setting, and ten possible intentions for the experience (e.g. recreational, spiritual, therapeutic, to confront difficult emotions). Principal component analysis (PCA) with Varimax rotation reduced set/setting items into three components named ‘‘set,’’ ‘‘setting,’’ and ‘‘clear intentions,’’ and the intentions into three components labelled ‘‘spiritual connection,’’ ‘‘recreation,’’ and ‘‘emotional.’’ Sample sizes at each time-point were N = 654, N = 535, N = 379, N = 315 and N = 212 respectively. Statistical approaches included repeated-measures GLM ANOVA for WEMWBS change, independent t‑tests for subgroup comparisons, linear mixed models (to handle missing data) with time as a repeated effect to model well‑being over time and interactions with predictors, and multivariate GLM regression to predict acute MEQ, CEQ and visual-effects scores from traits, set/setting and dose. All predictor variables were standardised and an alpha of 0.05 was used.

Participant recruitment yielded progressively smaller samples across the five surveys: 654 at baseline down to 212 at the 4‑week follow‑up, with some participants reporting additional psychedelic experiences during follow‑up (88 at 2 weeks, a further 29 at 4 weeks). PCA of the 12 self-constructed set/setting items produced three components explaining 54.33% of variance: ‘‘set’’ (31.12%), ‘‘setting’’ (12.49%) and ‘‘clear intentions’’ (10.72%). PCA of the ten intention items also produced three components explaining 55.14% of variance: ‘‘spiritual connection’’ (26.88%), ‘‘recreation’’ (16.24%) and ‘‘emotional’’ (12.01%). Psychological well‑being (WEMWBS) increased at 2 weeks after the psychedelic experience and remained elevated at 4 weeks compared with baseline. In subsequent mixed-model analyses, a range of relationships were identified. Baseline trait absorption and higher reported drug dose were positively associated with all acute measures: MEQ (dose F(181)=16.07, p<0.001; absorption F(181)=21.11, p<0.001), CEQ (dose F(181)=8.03, p<0.006; absorption F(181)=6.82, p<0.020) and visual effects (dose F(181)=21.05, p<0.001; absorption F(181)=8.98, p<0.004). The multivariate models explained R² = 0.342 for MEQ, R² = 0.324 for CEQ and R² = 0.211 for visual effects. Predictors of mystical‑type (MEQ) scores included clear intentions (F(181)=6.30, p<0.020) and a trend for the spiritual‑connection intention (F(181)=3.75, p≈0.06). Challenging experiences (CEQ) were negatively predicted by the ‘‘set’’ component (F(181)=21.09, p<0.001) and by the ‘‘recreation’’ intention component (F(181)=7.72, p<0.007); the ‘‘setting’’ component showed a trend towards reducing CEQ scores (F(181)=3.33, p≈0.08). Clear intentions also positively predicted visual effects (F(181)=4.61, p<0.040). When modelling change in well‑being over time with a broad set of predictors, baseline trait variables emerged as particularly influential. Trait anxiety showed the strongest effect on well‑being change, evidenced by a significant time-by-trait‑anxiety interaction (F(172)=3.93, p<0.030); trait anxiety was negatively associated with baseline WEMWBS and with WEMWBS at 4 weeks but not at 2 weeks, suggesting transient attenuation of its negative association at 2 weeks. Number of lifetime psychedelic uses interacted with time (F(106)=6.11, p<0.004): more lifetime uses were negatively associated with WEMWBS at 2 and 4 weeks (t(136)=-3.29, p<0.002; t(96)=-2.61, p<0.020), indicating smaller well‑being gains in more experienced users. Individuals self‑rating as ‘‘highly experienced psychedelic users’’ had higher WEMWBS scores overall at baseline and follow‑ups. Baseline openness predicted baseline and 2‑week WEMWBS; emotional stability, clear intentions and visual effects correlated with baseline WEMWBS but did not predict change. The ‘‘setting’’ component was positively associated with WEMWBS at 2 weeks (t(139)=2.11, p<0.040) with a trend at 4 weeks. Finally, being in a self‑reported therapeutic environment did not show consistent effects on acute experience measures in these analyses.

Haijen and colleagues interpret their findings as supporting the importance of both extra‑pharmacological factors (set, setting and intentions) and stable baseline traits in shaping psychedelic responses. They highlight three central conclusions: first, psychological well‑being increased at 2 weeks and remained elevated at 4 weeks after a naturalistic psychedelic experience; second, mystical‑ or peak‑type acute experiences were positively associated with later increases in well‑being, whereas challenging experiences were associated with lower well‑being across time; third, in a more comprehensive model that included traits, set/setting, intentions, dose and acute measures, baseline trait variables—particularly trait anxiety and number of lifetime uses—explained the largest proportion of variance in well‑being change. The authors position these results alongside prior literature showing a link between mystical/peak experiences and positive longer‑term outcomes and prior reports implicating absorption and dose in intensifying psychedelic effects. They note that mystical‑type and challenging experiences are not mutually exclusive and that existing instruments may not capture whether challenging periods were ultimately resolved into therapeutic breakthroughs; they therefore advocate development of measures sensitive to ‘‘breakthrough’’ or cathartic outcomes. Haijen and colleagues also discuss the therapeutic relevance of preparing participants to have clear intentions and a positive pre‑state, and of ensuring a comfortable setting, because these factors predicted more favourable acute profiles (more mystical‑type, fewer challenging experiences). The authors acknowledge multiple limitations inherent to the web‑based observational design: reliance on self‑reported timing and dose (dose estimates were collected after the experience and may be biased by subjective intensity), uncertain compliance with scheduled surveys, sample attrition across time-points, and likely selection bias towards individuals motivated and positively disposed to psychedelic use. They also note the limited operationalisation of ‘‘therapeutic setting’’ in their items and the broader debates about the construct validity of ‘‘well‑being’’ as an outcome. For future work, they recommend refining measurements of setting, developing an ‘‘emotional breakthrough’’ questionnaire, recruiting more heterogeneous samples (to examine cultural, religious or superstitious moderators), and pursuing controlled studies to further test and refine predictive models. Ultimately, the investigators propose that iterative, empirically derived guidelines could help inform clinical decisions about whether, when, where and how to dose with psychedelics, with the aim of reducing risks and maximising benefit.

The authors conclude that, in this large prospective naturalistic sample, baseline trait variables were the most potent predictors of change in psychological well‑being after a psychedelic experience. Mystical‑type (peak) acute experiences were positively related to well‑being change, while challenging experiences showed a general negative association. Clear intentions and feeling prepared (positive ‘‘set’’), together with intentions related to spiritual connection or recreation, were conducive to peak experiences and protective against challenging experiences. Higher trait absorption and larger reported doses intensified acute experiences and increased the likelihood of both mystical and challenging elements. The authors emphasise that much variance remains unexplained and call for further controlled and prospective research to refine prediction models and to develop empirically grounded guidelines for clinical use.