Oddities of the odds ratio

Over at Genomes Unzipped, my esteemed colleague Carl Anderson has his first ever blog post: an exploration of the various ways in which the effects of genetic variants on disease risk can vary from person to person. 

This potential variation has been the cause of much angst among critics of the direct-to-consumer genetic testing industry. However, as Carl notes, DTC testing companies generally do a pretty good job of conveying the uncertainty associated with one source of variation (differences in population background), and can't really be blamed for not accounting for the effects of environment and age given the currently weak scientific literature in this area.
However, it's worth noting that the current literature does provide some promising hints that variation in effect sizes may not be as large as originally feared. As Carl notes, one recent study in PLoS Genetics found that genetic factors associated with type 2 diabetes show no compelling evidence for varying effect sizes between cohorts of European Americans, African Americans, Latinos, Japanese Americans, and Native Hawaiians. 
Larger samples and a broader range of diseases will be required to confirm how widely this pattern holds, but it's tentatively reassuring for individuals of non-European ancestry: in most cases, risk estimates from your 23andMe data based on European cohorts will probably still be broadly applicable even if you're one of the majority of human beings who aren't descended from pallid European ancestors.
Moving forward, we can expect very large longitudinal studies (such as the half-million strong UK Biobank) to provide more precise estimates of the interactions between genetic risk factors and environmental variables in individuals from different populations. In the meantime, the generic advice I give to all genetic testing customers applies: read everything you can, treat the caveats seriously, and take every risk estimate as provisional and uncertain (with some being far more uncertain than others, of course!). We're still at the beginning of the genetic revolution, and uncertainty is simply the price you pay for getting in early.

More like this