Table 2.

Causal estimates of atrial fibrillation on ischemic stroke

Genetically predicted exposure	Outcome phenotype	MR-Egger intercept P value	Cochran’s Q statistic P value for heterogeneity	MR method	Beta	Standard error	P value
Atrial fibrillation (16 SNPs)	Ischemic stroke	0.028	0.068	IVW	0.188	0.026	1.03E−12
				MR-Egger (bootstrap)	0.251	0.047	< 0.001
				Weighted median	0.240	0.030	4.83E−16
				MR-RAPS	0.201	0.025	5.57E−16
				MR-PRESSO	NA	NA	NA

MR Mendelian randomization, IVW inverse variance–weighted, RAPS robust adjusted profile score, PRESSO pleiotropy residual sum and outlier, SNP single-nucleotide polymorphism

MR-PRESSO analysis was performed, but as the MR-PRESSO global test for heterogeneity did not identify correctable effects of outliers, the causal estimates were the same as those of the inverse variance–weighted method

The units of the causal estimates were log odds ratio for the cause (atrial fibrillation) and log odds ratio for the effect (stroke)