Flaws Cited in Gene Variation Studies

From "Findings," The Washington Post, 8/22/07.

Excerpt:

"Most studies claiming to show that gene variations have different impacts on illness in men and women are flawed, scientists said.

"Just 12 percent of 77 scientific papers that found a difference in the way gene variants affect disease in men and women were correctly analyzed, the researchers report in today's Journal of the American Medical Association.

"Most of the studies weren't big enough to determine whether the mutations have different effects in males and females, said John P. Ioannidis, of the University of Ioannina School of Medicine in Greece, who led the analysis."

Comment: I assume that a meta-analysis published in JAMA has been peer reviewed and is credible (and that the WP reporter got the story right). On the other hand, the papers that are criticized as having used statistics incorrectly were also, presumably published in peer reviewed journals.

One conclusion is that statistics is a tough business. It is easy to draw incorrect conclusions from data. I suspect that a lot of researchers are not nearly as well prepared in statistical methods as they should be, and have failed to develop the feel for their interpretation that is needed to do really good work.

Clearly, we are in trouble with gender issues. For far too long, our culture has assumed that gender difference existed and were important, where no such differences did exist or were not important. On the other hand, we seem to have ignored gender differences in areas such as medicine where they did exist and were important.

Still, it is hard and expensive to develop the data to isolate all the differences that make a difference in biomedical research. Perhaps with computers and the development of national health information systems, together with genomics and other techniques for reliably identifying individual differences, we will do better in the future. JAD