“A Cautionary Note on the Use of Mendelian Randomization to Infer Causation in Observational Epidemiology”, Murielle Bochud, Arnaud Chiolero, Robert C. Elston, Fred Paccaud2007-09-19 ()⁠:

[criticism of Ebrahim & Smith2007] …Table 1 indicates the commonly acknowledged necessary conditions (1–9) for Mendelian Randomization to provide this protection in observational epidemiology.2,3 We list additional necessary conditions (10–12) that have been given little attention and may be of relevance for certain genetic variants.

Table 1: Necessary conditions for the use of Mendelian Randomization to infer causality in observational epidemiology.
N Condition
1 Th ere are enough data to establish reliable genotype-intermediate phenotype, or genotype-trait, associations
2 Th ere is no confounding due to linkage disequilibrium
3 Th ere is no confounding due to population stratification
4 Th ere is no pleiotropy
5 Th ere is no canalisation nor developmental compensation (ie. a functional adaptation to a specific genotype influencing the expected genotype-disease association)
6 A suitable genetic variant exists to study the exposure of interest
7 Th e association between gene and gene product is strong
8 Th e effects of a gene on a disease outcome acts only via the intermediate phenotype
9 Th e genetically determined exposure has a similar impact on the disease outcome as the environmental exposure investigated
10 T here is no segregation distortion at the locus of interest
11 T here is no selective survival due to the genetic variant of interest
12 T here is no parent-of-origin effect