To ensure that the MR‐Egger analysis does not depend on the reported reference alleles, Bowden et al suggested the genetic variants in univariable MR‐Egger be orientated so the direction of association with the risk factor is either positive for all variants or negative for all variants.9 However, this may not be possible for multivariable MR‐Egger as the same reference allele must be used for associations with each risk factor and with the outcome. We suggest that the variants should be orientated with respect to their associations with the risk factor of primary interest, although we would recommend a sensitivity analysis considering different orientations if multiple risk factors are of interest. If the genetic variants are all valid instruments, then directional pleiotropy should not be detected with respect to any orientation.
