Corporate Research, Study Design and Genome Wide Association Studies
There’s an interesting point about methodology in this item from today’s JAMA: An Unwelcome Side Effect of Direct-to-Consumer Personal Genome Testing: Raiding the Medical Commons (by Amy L. McGuire and Wylie Burke). The article is about “personal genome” testing, based on genome-wide association studies. (Such testing has been discussed on this blog here, here, and here.) The main worry presented in the article is that such tests are sold on the basis of fairly weak association studies, and the customers of these studies will often become patients of an increasingly overburdened healthcare system. Such companies might be seen as offloading (in economic terms, externalizing) costs in the form of burdens on the “medical commons.” (It’s a decent point, though I wish the authors had said more about the nature of that commons, and what kind of obligations various kinds of agents have to preserve it, and why.)
Here’s the bit about study design:
In addition, many companies may use a scientific advisory board to determine which scientific findings to include in their service. Information about the research relied on, as well as its potential limitations, should be shared with consumers. Moreover, some direct-to-consumer personal genome companies are leveraging consumer data and using it, along with phenotypic information collected with consent, to study genotype-phenotype associations.2 For this research to be interpretable, it must conform to best practices of measurement and study design, including attention to biases in sample selection and limitations of self-reported phenotype. Although these companies are not subject to federal regulation for the protection of human subjects of research, their studies should be conducted according to accepted scientific and ethical standards. These standards may be difficult for companies to accomplish, but ad hoc, underpowered, or biased analyses cannot provide an adequate substitute. In addition, results of these studies should be provided in the form of peer-reviewed publications when possible.