As in the univariable setting, causal estimates of the effect of each risk factor on the outcome can be obtained from individual‐level data using the 2‐stage least squares method.7 The same estimates can also be obtained using multivariable weighted linear regression of the genetic association estimates, with the intercept set to zero (referred to as the multivariable IVW method)20: (6)β^Yj=θ1MIβ^X1j+θ2MIβ^X2j+θ3MIβ^X3j+ϵMIj,weights=se(β^Yj)−2. We propose the natural extension to multivariable MR‐Egger using the same regression model but allowing the intercept to be estimated: (7)β^Yj=θ0ME+θ1MEβ^X1j+θ2MEβ^X2j+θ3MEβ^X3j+ϵMEj,weights=se(β^Yj)−2.
