Table 4.

Mendelian randomization estimates and sensitivity analyses for the results in HUNT.

Method	Estradiol levela on colon cancer			Bioavailable testosteroneb on lung non-adenocarcinoma			Total testosteronec on lung non-adenocarcinoma
	HRd (95% CI)	p-value	Q-statistic/ p-value	HRe (95% CI)	p-value	Q-statistic/ p-value	HRf (95% CI)	p-value	Q-statistic/ p-value
IVW	0.38 (0.16–0.88)	0.02	3.25/0.20	0.47 (0.23–0.96)	0.04	84.55/0.52	0.60 (0.37–0.98)	0.04	136.62/0.26
Weighted median	0.61 (0.19–1.92)	0.40		0.45 (0.13–1.58)	0.21		0.43 (0.19–0.95)	0.04
MR-Egger	4.73 (0.16–140.73)	0.37	1.00/0.32	0.38 (0.10–1.37)	0.14	84.40/0.50	0.66 (0.29–1.48)	0.31	135.55/0.24
MR-Egger intercept	-0.22 (-0.51–0.07)	0.13		0.008 (-0.03–0.05)	0.70		-0.004 (-0.04–0.03)	0.79

CI confidence interval, HR hazard ratio, IVW inverse-variance weighted, SD standard deviation, SNP single-nucleotide polymorphism.

^a3 SNPs/^b87 SNPs/^c127 SNPs were used as instrumental variables for estradiol level²², bioavailable and total testosterone²³, respectively.

^dPer one-SD increase in genetically predicted rank-transformed estradiol level.

^ePer one-SD increase in genetically predicted bioavailable testosterone level.

^fPer one-SD increase in genetically predicted total testosterone level.