most self-genotyping is worthless (Score 2, Interesting)

I study associations of genotypes with diseases for a living. The article worries me a little because it does not make clear that many genotype-disease correlations are very weak or tenuous. The article begins like this:

Nearly every day, somebody in the research community claims to have found a genetic marker associated with some sort of health condition.

That is true but we must acknowledge that many if not most such claims have failed to replicate and they are probably mostly statistical errors -- do enough research and you will eventually "find" something that isn't real. Those of us in the business are highly skeptical of all original claims unless they are overwhelmingly strong. In the absence of consistent replication we tend to withhold judgment on the validity of a result.

The article continued:

If you are curious and want to check yourself for those inherited traits, there are several options at your disposal.

If you are checking yourself for a certain genotype, then we could say that you are checking yourself for an inherited trait, but we usually don't refer to genotypes as traits. I think the author of the article is presuming a strong relationship of a genotype with some trait like a form of cancer, asthma, longevity, schizophrenia, myocardial infarction or stroke. He then suggests that readers can study that trait in themselves by looking at their genotypes. Unfortunately, even if the studies that claimed to establish an association of the genotype with the trait were correct, the association might be so weak that it isn't clinically useful. For example, if schizophrenia occurs in 1% of the general population but in 1.2% of people with a certain genotype, that is a 20% increase in risk for that genotype, which might lead to a statistically significant result that is published in a journal, promoted by a university press release and then published in newspapers all over the world. But to any individual, knowing their genotype would change their schizophrenia risk by only 0.2%. Is it worthwhile to know about such a minor risk factor?

It gets worse. The person who tests himself might then decide that he is a high-risk person or a low-risk person based on this one genotype, but that is a very bad way to decide risk. With schizophrenia, for example, a positive family history can increase your risk by a factor of 10 or more (depending on which relative[s] are affected). Also, having pre-schizophrenic symptoms will imply higher risk. Age of onset is usually before age 40, so people older than 40 who are not already schizophrenic have low risk. So a 20-year-old man who is having mild hallucinatory experiences, coming up with really odd ideas and who has two schizophrenic brothers might take the test, find out he doesn't have the 1.2% genotype and conclude that he is at low risk of schizophrenia. He would be mistaken.

With regard to heart disease: Ignore your genotype but watch your blood pressure, eat well and exercise. Same for stroke, diabetes, obesity and many other traits.

Most of the value in new genetics research for so-called "complex diseases" (a.k.a. multifactorial diseases) is in the implications for breakthroughs in understanding pathophysiology. Most findings have little or no implication for individuals.