Table 3. Multivariable Mendelian randomization estimates for effects of genetically mimicked statins (effect sizes of LDL-cholesterol) and of testosterone (effect size) together on IHD in men and women using the UK Biobank.

					Mendelian randomization estimates
Sex	Exposures	Instrumented by	Adjusted for	Method	OR	95% CI	P value	MR-Egger intercept p-value
Men	Statin mimic	1 Statin SNP on LDL-cholesterol	Testosterone	Inverse variance weighted	1.05	0.74 to 1.47	0.79
	Testosterone	125 SNPs on testosterone	statin	Inverse variance weighted	1.11	1.04 to 1.20	0.003
	Statin mimic	1 Statin SNP on LDL-cholesterol	Testosterone	MR-Egger	0.73	0.48 to 1.11	0.14
	Testosterone	125 SNPs on testosterone	statin	MR-Egger	1.09	1.02 to 1.17	0.02	0.005
	Statin mimic	6 Statin SNPs on LDL-cholesterol	Testosterone	Inverse variance weighted	1.02	0.72 to 1.43	0.91
	Testosterone	125 SNPs on testosterone	statin	Inverse variance weighted	1.11	1.04 to 1.20	0.003
Women	Statin mimic	1 Statin SNP on LDL-cholesterol	Testosterone	Inverse variance weighted	0.98	0.75 to 1.16	0.53
	Testosterone	254 SNPs on testosterone	statin	Inverse variance weighted	0.96	0.90 to 1.04	0.33
	Statin mimic	1 Statin SNP on LDL-cholesterol	Testosterone	MR-Egger	0.72	0.55 to 0.94	0.02
	Testosterone	254 SNPs on testosterone	statin	MR-Egger	0.96	0.89 to 1.03	0.27	0.001
	Statin mimic	6 Statin SNPs on LDL-cholesterol	Testosterone	Inverse variance weighted	0.92	0.74 to 1.16	0.49
	Testosterone	254 SNPs on testosterone	statin	Inverse variance weighted	0.97	0.90 to 1.04	0.36

One statin SNP is rs12916, and six statin SNPs additionally included rs5909, rs10066707, rs17238484, rs2006760 and rs2303152 taking into account all their correlations. The unit of LDL-cholesterol is approximately 0.83 mm/L. An effect size of testosterone is approximately, 0.23 nmol/L in women (Haring et al., 2012) and 3.1 nmol/L in men (Mohr et al., 2005).