Table 1. Differences between the studies of Zhou and Sofianopoulou in methodology and results.

Factor	Zhou et al.6	Sofianopoulou et al.7
Choice of variants	35 variants associated with 25(OH)D levels at genome-wide significant level	Variants in 4 gene regions chosen due to biological link with vitamin D synthesis or metabolism (‘focused score’); a secondary analysis using 71 genome-wide significant variants was also performed
Pleiotropic associations	Several genetic variants had pleiotropic associations with LDL-cholesterol	Minimal evidence of pleiotropy for primary analysis; substantial pleiotropy via LDL-cholesterol and triglycerides for secondary analysis using genome-wide variants
Cardiovascular outcomes	Cardiovascular disease (combined outcome of coronary artery disease, stroke, and peripheral vascular disease)	Separate analyses for coronary heart disease, stroke, and cardiovascular mortality
Results	Strong associations of genetically predicted 25(OH)D with outcomes in vitamin D-deficient strata; weak (but non-null) associations overall	Primary analysis: associations of genetically predicted 25(OH)D with outcomes in vitamin D-deficient strata (particularly for cardiovascular mortality); null associations overall Secondary analysis: weak association with coronary heart disease in overall analysis that attenuated toward null on adjustment for LDL-cholesterol and triglycerides; stronger association with stroke in deficient stratum that persisted despite adjustment for LDL-cholesterol and triglycerides