Cannabis use and risk of schizophrenia: a Mendelian randomization study

Abstract

Cannabis use is observationally associated with an increased risk of schizophrenia, but whether the relationship is causal is not known. Using a genetic approach, we took 10 independent genetic variants previously identified to associate with cannabis use in 32 330 individuals to determine the nature of the association between cannabis use and risk of schizophrenia. Genetic variants were employed as instruments to recapitulate a randomized controlled trial involving two groups (cannabis users vs nonusers) to estimate the causal effect of cannabis use on risk of schizophrenia in 34 241 cases and 45 604 controls from predominantly European descent. Genetically-derived estimates were compared with a meta-analysis of observational studies reporting ever use of cannabis and risk of schizophrenia or related disorders. Based on the genetic approach, use of cannabis was associated with increased risk of schizophrenia (odds ratio (OR) of schizophrenia for users vs nonusers of cannabis: 1.37; 95% confidence interval (CI), 1.09-1.67; P-value=0.007). The corresponding estimate from observational analysis was 1.43 (95% CI, 1.19-1.67; P-value for heterogeneity =0.76). The genetic markers did not show evidence of pleiotropic effects and accounting for tobacco exposure did not alter the association (OR of schizophrenia for users vs nonusers of cannabis, adjusted for ever vs never smoker: 1.41; 95% CI, 1.09-1.83). This adds to the substantial evidence base that has previously identified cannabis use to associate with increased risk of schizophrenia, by suggesting that the relationship is causal. Such robust evidence may inform public health messages about cannabis use, especially regarding its potential mental health consequences.

Figure 1

Meta-analysis of prospective observational studies reporting an association between use of cannabis and risk of schizophrenia or related disorders. Meta-analysis uses a random-effects model. Studies are sorted by type of outcome (schizophrenia only vs schizophrenia and related outcomes). Odds ratios (ORs) and 95% confidence intervals (CIs) express the risk of schizophrenia or psychotic symptoms for ever use of cannabis (compared with never use). For additional information on each study, see Supplementary Table S1. Dunedin, Dunedin Multidisciplinary Health & Development Study; ECA, Epidemiologic Catchment Area; EDSP, Early Developmental Stages of Psychopathology Study; NEMESIS, Netherlands Mental Health Survey and Incidence Study; SC, Swedish Cohort.

Figure 2

Meta-analysis of the association of genetically instrumented use of cannabis and risk of schizophrenia for the 10 single-nucleotide polymorphisms (SNPs) under analysis. Odds ratios (ORs) and 95% confidence intervals (CIs) express the risk of schizophrenia per-1-log unit increase in ever use of cannabis. Meta-analysis uses a fixed effect model. The method to derive the population-based OR of schizophrenia among users of cannabis compared with nonusers (OR 1.37; 95% CI, 1.09–1.67), as presented in the main text and Figure 3, is described in the Supplementary Information.

Figure 3

Comparison of observational (blue) and causal (red) estimates for use of cannabis and risk of schizophrenia. Two observational estimates are provided according to a stringent definition of schizophrenia (as reported in the Swedish cohort) or to an outcome comprising studies reporting risk of schizophrenia or psychotic symptoms (derived from the meta-analysis reported in Figure 1) for ever use of cannabis. Causal estimates represent population-based associations derived by conventional (Figure 2) and multivariable Mendelian randomization (MR). The total number of cases and controls in each analysis are presented.

Figure 4

Sensitivity analysis of the association of use of cannabis and risk of schizophrenia by sequentially removing each single-nucleotide polymorphism (SNP) from the analysis. The red vertical line represents the summary causal effect estimate (derived from Mendelian randomization) when including the 10 SNPs in the analysis (presented in Figure 3). Odds ratios (ORs) and 95% confidence intervals (CIs) represent the population-based risk of schizophrenia in users of cannabis (compared with nonusers).