Epidemiologists are interested in using mendelian randomization to assess causation between observed phenotypes, using genotype as an instrumental variable.
Minelli and coworkers (2004) note that metaanalysis of traditional epidemiological studies finds that a 3 uM (25%) decrease in serum homocysteine level is associated with an 11% decrease in coronary heart disease (CHD) risk.
The 677C>T transversion in the MTHFR gene is associated with higher homocysteine levels, therefore an increase in CHD risk would be expected if the direction of causation runs from homocysteine level to CHD, and MTHFR genotype has no other associations with CHD risk.
| Homocysteine level | CHD Odds Ratio | |
| MTHFR C/C v T/T | 2.7 uM (2.02-3.41) | 1.21 (1.06-1.40) |
So predicted Odds Ratio for CHD per 3 uM increase homocysteine level 1.24 (1.06-1.49).