As genetic variants are fixed at conception, in general, they instrument a life-course predisposition to a particular risk factor. This makes MR less prone to regression dilution bias, but such genetic variants might be unsuited to testing the effects of risk factors that are believed to act during critical (a time period in which the risk factor is hypothesised to solely have its effect) or sensitive (a period during which the risk factor is hypothesised to have the greatest effect) periods.
If the research question is about determining an effect in a critical or sensitive period (i.e., showing an exposure only affects an outcome when it occurs during a specific age range and not at other life times or has a stronger effect when it occurs during a specific time of life), MR may not be a useful method as the result it will provide is the mean effect of the exposure across all/most of the life course. As GWASs of trajectories (e.g., change in weight or height across infancy childhood and into early adulthood) are undertaken, they may provide genetic IVs that could be used in MR to compare change in an exposure between two ages with change in the same exposure between two later ages and compare these to see if there are differences in the effect at different ages. This is likely to need very large sample sizes.
References
- Lawlor DA, Tilling K, Davey Smith G. Triangulation in aetiological epidemiology. Int J Epidemiol 2017.
- Lawlor DA, Richmond R, Warrington N et al. Using Mendelian randomization to determine causal effects of pregnancy (intrauterine) exposures on offspring outcomes: Sources of bias and methods for assessing them. . Wellcome Open Research 2017;2:11.
Other terms in 'Sources of bias and limitations in MR':
- Assortative mating
- Canalization
- Collider
- Collider bias
- Confounding
- Dynastic effects
- Exclusion restriction assumption
- Harmonization failure (in two-sample MR)
- Homogeneity Assumption
- Horizontal Pleiotropy
- Independence assumption
- InSIDE assumption (in two-sample MR using aggregate data)
- Monotonicity assumption
- MR for testing developmental origins
- No effect modification assumption (Additional IV assumption)
- Non-linear effects
- Non-overlapping samples (in two-sample MR)
- Overfitting
- Pleiotropy
- Population stratification
- Regression dilution bias (attenuation by errors)
- Relevance assumption
- Reverse causality
- Same underlying population (in two-sample MR)
- Statistical power/efficiency
- Vertical Pleiotropy
- Weak instrument bias
- Winner's curse