A framework for assessing selection and misclassification bias in mendelian randomisation studies: an illustrative example between body mass index and covid-19

Mendelian randomisation (MR) can be implemented as an instrumental variable analysis, with genetic variants as an instrument for exposures (or modifiable risk factors) to explore their causal effects on the occurrence and prognosis of disease12 (fig 1). Genetic variants are randomly allocated at conception, and not modified throughout life, meaning MR is less likely to be biased by confounding or reverse causality (ie, confounding by prevalent disease) than conventional multivariable regression.123 In individual level MR (often referred to as one sample MR), the genetic instrumental variable-exposure and genetic instrumental variable-outcome associations are from the same sample, and individual level data are used to derive the MR estimate. In a summary data MR study (often referred to as two sample MR), the genetic instrumental variable-exposure and genetic instrumental variable-outcome associations come from two non-overlapping samples from the same underlying population.24 We focus on the summary data MR approach in this paper,…
Read Original Article: A framework for assessing selection and misclassification bias in mendelian randomisation studies: an illustrative example between body mass index and covid-19 »