MR Dictionary

An MR analysis that tries to differentiate whether factor A is a cause or a consequence of factor B or whether there is a true bidirectional causal effect between A and B.

Genetic variants for factor A (e.g., what is assumed to be the exposure) and factor B (e.g., what is assumed to be the outcome, influenced by the exposure) need to fulfil the MR assumptions. This requires genetic variants for both variables (A and B) to be used to test each of the three possibilities: A causes B; B causes A; or there is a true bidirectional causal relationship between A and B. In comparing results in both directions, caution is required to take account of differences in statistical power related to the instrumental variable (IV)-exposure association and any violations of MR assumptions.

Bidirectional Mendelian randomization. Adapted from Richmond et al. (A) In assessing whether physical activity (trait 1, represented by T1) causally influences body mass index (trait 2, represented by T2), genetic variants associated with physical activity (indicated by Z1) are used in an MR analysis. (B) In the reverse direction, genetic variants associated with body mass index (indicated by Z2) are used in an MR analysis to assess whether body mass index causally influences physical activity.
Figure 2.1 - Bidirectional Mendelian randomization. Adapted from Richmond et al. (A) In assessing whether physical activity (trait 1, represented by T1) causally influences body mass index (trait 2, represented by T2), genetic variants associated with physical activity (indicated by Z1) are used in an MR analysis. (B) In the reverse direction, genetic variants associated with body mass index (indicated by Z2) are used in an MR analysis to assess whether body mass index causally influences physical activity.

References

Other terms in 'Definition of MR and study designs':