to detect a causal effect is well controlled for J ≤ 50 variants, but for over 100 variants some type I error inflation is apparent. In summary, MR-Egger regression works well with large numbers of genetic variants (in the sense that it has an increased power to detect pleiotropy), as long as the variants are not too weak. Table 2.Performance of inverse-variance weighted and MR-Egger regression estimates ina simulation study for two-sample Mendelian randomization with a null causal effect (β = 0) and a fixed sample size, and varying the number of genetic variants (J)Inverse-variance weightedMR-Egger regressionJMean F statisticMean estimate (mean SE)Power to detect causal effectMean estimate (mean SE)Power of MR-Egger testPower to detect causal effectNo causal effect: β = 0Scenario (c) directional pleiotropy, InSIDE satisfied3407.00.042 (0.028)0.1270.003 (0.103)0.0590.0545295.00.039 (0.022)0.2480.000 (0.060)0.0850.05010172.00.038 (0.015)0.5800.001 (0.037)0.1660.05115121.00.037 (0.013)0.7800.000 (0.030)0.2480.0482093.60.037 (0.011)0.8940.001 (0.025)0.3290.0553064.40.037 (0.009)0.9800.001 (0.020)0.4750.0525039.80.036 (0.007)1.0000.002 (0.015)0.6670.05810020.70.035 (0.005)1.0000.005 (0.011)0.8770.08215014.20.035 (0.004)1.0000.007 (0.008)0.9440.150
