The association between cannabis and psychosis is established, but the role of underlying genetics is unclear. We used data from the EU-GEI case-control study and UK Biobank to examine the independent and combined effect of heavy cannabis use and schizophrenia polygenic risk score (PRS) on risk for psychosis.

Genome-wide association study summary statistics from the Psychiatric Genomics Consortium and the Genomic Psychiatry Cohort were used to calculate schizophrenia and cannabis use disorder (CUD) PRS for 1098 participants from the EU-GEI study and 143600 from the UK Biobank. Both datasets had information on cannabis use.

In both samples, schizophrenia PRS and cannabis use independently increased risk of psychosis. Schizophrenia PRS was not associated with patterns of cannabis use in the EU-GEI cases or controls or UK Biobank cases. It was associated with lifetime and daily cannabis use among UK Biobank participants without psychosis, but the effect was substantially reduced when CUD PRS was included in the model. In the EU-GEI sample, regular users of high-potency cannabis had the highest odds of being a case independently of schizophrenia PRS (OR daily use high-potency cannabis adjusted for PRS = 5.09, 95% CI 3.08–8.43, p = 3.21 × 10−10). We found no evidence of interaction between schizophrenia PRS and patterns of cannabis use.

Regular use of high-potency cannabis remains a strong predictor of psychotic disorder independently of schizophrenia PRS, which does not seem to be associated with heavy cannabis use. These are important findings at a time of increasing use and potency of cannabis worldwide.

Childhood adversity and cannabis use are considered independent risk factors for psychosis, but whether different patterns of cannabis use may be acting as mediator between adversity and psychotic disorders has not yet been explored. The aim of this study is to examine whether cannabis use mediates the relationship between childhood adversity and psychosis.

Data were utilised on 881 first-episode psychosis patients and 1231 controls from the European network of national schizophrenia networks studying Gene–Environment Interactions (EU-GEI) study. Detailed history of cannabis use was collected with the Cannabis Experience Questionnaire. The Childhood Experience of Care and Abuse Questionnaire was used to assess exposure to household discord, sexual, physical or emotional abuse and bullying in two periods: early (0–11 years), and late (12–17 years). A path decomposition method was used to analyse whether the association between childhood adversity and psychosis was mediated by (1) lifetime cannabis use, (2) cannabis potency and (3) frequency of use.

The association between household discord and psychosis was partially mediated by lifetime use of cannabis (indirect effect coef. 0.078, s.e. 0.022, 17%), its potency (indirect effect coef. 0.059, s.e. 0.018, 14%) and by frequency (indirect effect coef. 0.117, s.e. 0.038, 29%). Similar findings were obtained when analyses were restricted to early exposure to household discord.

Harmful patterns of cannabis use mediated the association between specific childhood adversities, like household discord, with later psychosis. Children exposed to particularly challenging environments in their household could benefit from psychosocial interventions aimed at preventing cannabis misuse.

While cannabis use is a well-established risk factor for psychosis, little is known about any association between reasons for first using cannabis (RFUC) and later patterns of use and risk of psychosis.

We used data from 11 sites of the multicentre European Gene-Environment Interaction (EU-GEI) case–control study. 558 first-episode psychosis patients (FEPp) and 567 population controls who had used cannabis and reported their RFUC.

We ran logistic regressions to examine whether RFUC were associated with first-episode psychosis (FEP) case–control status. Path analysis then examined the relationship between RFUC, subsequent patterns of cannabis use, and case–control status.

Controls (86.1%) and FEPp (75.63%) were most likely to report ‘because of friends’ as their most common RFUC. However, 20.1% of FEPp compared to 5.8% of controls reported: ‘to feel better’ as their RFUC (χ2 = 50.97; p < 0.001). RFUC ‘to feel better’ was associated with being a FEPp (OR 1.74; 95% CI 1.03–2.95) while RFUC ‘with friends’ was associated with being a control (OR 0.56; 95% CI 0.37–0.83). The path model indicated an association between RFUC ‘to feel better’ with heavy cannabis use and with FEPp-control status.

Both FEPp and controls usually started using cannabis with their friends, but more patients than controls had begun to use ‘to feel better’. People who reported their reason for first using cannabis to ‘feel better’ were more likely to progress to heavy use and develop a psychotic disorder than those reporting ‘because of friends’.

Risk of suicide-related behaviors is elevated among military personnel transitioning to civilian life. An earlier report showed that high-risk U.S. Army soldiers could be identified shortly before this transition with a machine learning model that included predictors from administrative systems, self-report surveys, and geospatial data. Based on this result, a Veterans Affairs and Army initiative was launched to evaluate a suicide-prevention intervention for high-risk transitioning soldiers. To make targeting practical, though, a streamlined model and risk calculator were needed that used only a short series of self-report survey questions.

We revised the original model in a sample of n = 8335 observations from the Study to Assess Risk and Resilience in Servicemembers-Longitudinal Study (STARRS-LS) who participated in one of three Army STARRS 2011–2014 baseline surveys while in service and in one or more subsequent panel surveys (LS1: 2016–2018, LS2: 2018–2019) after leaving service. We trained ensemble machine learning models with constrained numbers of item-level survey predictors in a 70% training sample. The outcome was self-reported post-transition suicide attempts (SA). The models were validated in the 30% test sample.

Twelve-month post-transition SA prevalence was 1.0% (s.e. = 0.1). The best constrained model, with only 17 predictors, had a test sample ROC-AUC of 0.85 (s.e. = 0.03). The 10–30% of respondents with the highest predicted risk included 44.9–92.5% of 12-month SAs.

An accurate SA risk calculator based on a short self-report survey can target transitioning soldiers shortly before leaving service for intervention to prevent post-transition SA.