Table 1 Comparison of methods when all genetic variants are valid instruments.

From: A robust and efficient method for Mendelian randomization with hundreds of genetic variants

 

Scenario 1: all instruments valid

Method

Mean

SD

Mean SE

Power

Null causal effect: θ = 0

 Inverse-variance weighted

0.000

0.022

0.023

4.2

 MR-Egger

0.001

0.061

0.062

4.4

 Weighted median

0.000

0.028

0.033

2.1

 MR-PRESSO

−0.000

0.022

0.022

5.1

 Weighted mode-based

0.001

0.061

0.199

0.3

 Contamination mixture

0.000

0.028

–

4.9

Positive causal effect: θ = +0.1

 Inverse-variance weighted

0.094

0.023

0.024

98.1

 MR-Egger

0.063

0.064

0.066

16.1

 Weighted median

0.091

0.030

0.035

78.8

 MR-PRESSO

0.094

0.024

0.023

98.0

 Weighted mode-based estimate

0.082

0.051

0.143

15.5

 Contamination mixture

0.096

0.029

–

91.0

  1. MR-PRESSO Mendelian Randomization Pleiotropy RESidual Sum and Outlier method of Verbanck et al.17
  2. All methods tested gave unbiased causal estimates in the simulation scenario when there were no invalid instruments. Mean, standard deviation (SD), mean standard error (mean SE) of estimates and empirical power (%) in simulation study for Scenario 1 (all 100 variants valid instruments)
Back to article page