Skip to main content
. 2019 May 22;365:l1855. doi: 10.1136/bmj.l1855

Table 2.

Mendelian randomisation (MR) sensitivity analyses

Analysis Two-sample One-sample
Estimate (95% CI) P value Estimate (95% CI) P value
Education-coronary heart disease
IVW 0.63 (0.60 to 0.67) <0.001 0.51 (0.26 to 1.00) 0.05
MR-Egger 0.68 (0.54 to 0.85) 0.001 0.54 (0.12 to 2.34) 0.41
MR-Egger intercept NA 0.37 NA 0.93
Weighted median 0.62 (0.57 to 0.67) <0.001 0.98 (0.96 to 0.99) 0.33
Education-stroke
IVW 0.71 (0.68 to 0.75) <0.001 0.46 (0.30 to 0.71) <0.001
MR-Egger 0.72 (0.60 to 0.87) 0.001 0.57 (0.22 to 1.47) 0.25
MR-Egger intercept NA 0.76 NA 0.60
Weighted median 0.71 (0.66 to 0.76) <0.001 0.99 (0.98 to 1.00) 0.002
Education-myocardial infarction
IVW 0.61 (0.58 to 0.65) <0.001 0.18 (0.08 to 0.38) <0.001
MR-Egger 0.67 (0.52 to 0.85) 0.001 0.20 (0.04 to 1.03) 0.05
MR-Egger intercept NA 0.39 NA 0.88
Weighted median 0.59 (0.54 to 0.65) <0.001 0.99 (0.98 to 1.00) 0.002
Education-cardiovascular disease
IVW NA NA 0.64 (0.51 to 0.82) <0.001
MR-Egger NA NA 0.57 (0.34 to 0.95) 0.03
MR-Egger intercept NA NA NA 0.59
Weighted median NA NA 0.96 (0.94 to 0.98) <0.001
Education-BMI
IVW −0.22 (−0.24 to −0.20) <0.001 −0.36 (−0.49 to −0.23) <0.001
MR-Egger −0.28 (−0.49 to −0.07) 0.009 −0.15 (−0.41 to 0.12) 0.29
MR-Egger intercept NA 0.99 NA 0.08
Weighted median −0.27 (−0.30 to −0.23) <0.001 −0.51 (−0.62 to −0.39) <0.001
Education-systolic blood pressure
IVW −0.15 (−0.17 to −0.14) <0.001 −0.14 (−0.24 to −0.04) 0.005
MR-Egger −0.13 (−0.21 to −0.05) 0.002 −0.10 (−0.30 to 0.11) 0.37
MR-Egger intercept NA 0.32 NA 0.65
Weighted median −0.18 (−0.21 to −0.16) <0.001 −0.12 (−0.21 to −0.03) 0.008
Education-smoking
IVW −0.32 (−0.33 to −0.31) <0.001 −0.37 (−0.47 to −0.28) <0.001
MR-Egger −0.29 (−0.36 to −0.22) <0.001 −0.40 (−0.60 to −0.20) <0.001
MR-Egger intercept NA NA NA 0.73
Weighted median −0.35 (−0.37 to −0.33) <0.001 −0.37 (−0.46 to −0.29) <0.001

IVW=inverse-variance weighted; NA=not applicable

Coronary heart disease, stroke, and myocardial infarction are in odds ratio units. BMI, systolic blood pressure, and smoking behaviour are in standard deviation units. In one-sample analyses, the weighted median was estimated on the risk difference scale and converted to odds ratio by using linear combinations.