Table 3.

Results of the bidirectional two-sample Mendelian randomization analysis between lifetime cannabis use and schizophrenia including results of 4 sensitivity analyses

	Cannabis-Schizophrenia (p<5e-08, 5 SNPs)				Cannabis-Schizophrenia p<1e-05, 69 SNPs*)				Schizophrenia-Cannabis (p<5e-08, 109 SNPs**)
	B	SE (B)	OR	p	B	SE (B)	OR	p	B	SE (B)	OR	p
IVW	0.039	0.158	1.04	0.806	0.091	0.051	1.10	0.074	0.151	0.046	1.16	0.001
Weighted Median	−0.048	0.105	0.95	0.649	0.069	0.049	1.07	0.156	0.163	0.049	1.17	0.001
MR-Egger SIMEX	−0.044	0.190	0.96	0.827	0.106	0.110	1.11	0.340	0.071	0.293	1.07	0.810
Weighted Mode	−0.084	0.125	0.92	0.536	0.016	0.071	1.02	0.823	0.315	0.178	1.37	0.080
GSMR after HEIDI filtering	−	−	−	−	0.192	0.080	1.21	0.017	0.237	0.038	1.27	5.36e-10

Significant results (p<0.05; tested two-sided) are shown in bold. Inverse Variance Weighted regression analysis (IVW); MR-Egger simulation extrapolation (SIMEX); Generalized Summary-data-based Mendelian Randomization (GSMR; HEIDI outlier analysis detects and eliminates from the analysis instruments that show significant pleiotropic effects on both risk factor and disease); risk coefficient representing the change in outcome for a one-unit increase in the exposure variable (B); standard error of the B (SE [B]); odds ratios represent the odds of schizophrenia for lifetime cannabis users versus non-users (when cannabis is the exposure) or the odds of lifetime cannabis use for those with a schizophrenia diagnosis versus those without (when schizophrenia is the exposure) (OR); p-value (p).

^*Number of SNPs in instrument was 74 for the GSMR analysis

^**Number of SNPs in instrument was 102 for the GSMR analysis.