Assessing candidate genes as risk factors for mental disorders: the value of population-based epidemiological studies.

Abstract

It is now possible to routinely collect DNA for the investigation of genetic risk factors in psychiatric epidemiology. However, concerns have been raised about the investigation of genetic risk factors in population surveys and in case-control studies with population controls, because of spurious associations arising through population stratification. To overcome this problem, some geneticists advocate the use of family controls and studies on isolated populations. However, such approaches may not be the best for psychiatric epidemiology. A critical analysis is made of arguments against the investigation of genetic risk factors in studies with population controls. It is argued that concerns about population stratification have been overblown. It is simply a form of confounding and can occur in studies of environmental as well as genetic risk factors. Epidemiologists have traditionally dealt with this and other types of confounding by matching on potential confounders in the design of the study, or stratification by confounders in the analysis. Family controls overcome population stratification, but not other forms of confounding. They are a good choice when the sole interest is genetic risk factors, but psychiatric epidemiologists are typically interested in both genetic and environmental factors. Family controls are poor for examining environmental risk factors because they can give rise to over-matching. Isolated populations may produce genetic associations that do not generalize to other populations, and they may have a restricted range of environments. We conclude that population surveys and case-control studies using population controls are suitable for the investigation of both genetic and environmental risk factors and have greater public health relevance than alternative designs.