The 94 SNPs used as a proxy for risk of schizophrenia case status provided little evidence of a causal effect of schizophrenia risk on likelihood of being a smoker (OR per 1 unit increase in log odds of schizophrenia 1.02, (95% CI 0.98–1.06, p = 0.32). This translates to an odds ratio of 1.01 (95% CI 0.98–1.04) for smoking per doubling of odds of schizophrenia (see Fig. 1). There was moderate heterogeneity (I2 = 38%, 95% CI 20–52% p = 0.0002), but the random-effects estimate was very similar to the fixed effects model (OR 1.01, 95% CI 0.98–1.05, p = 0.37). When we performed MR Egger regression, there was no evidence that the intercept (representing bias in the causal estimate due to pleiotropy) differed from the null (intercept OR 1.01, 95% CI 0.99–1.02, p = 0.49). The slope point estimate (representing the causal effect estimate of schizophrenia on smoking initiation) did not statistically differ from the null (OR 0.95, 95% CI 0.78–1.16, p = 0.61).
