Genetics in psychosomatic medicine: research designs and statistical approaches

It has become increasingly clear that genetic factors influence many of the behaviors and disease endpoints of interest to psychosomatic medicine researchers. There has been increasing interest in incorporating genetic variation markers into psychosomatic research. In this Statistical Corner article, we build on the valuable experiences gained during two workshops for "starters in the field" at the American Psychosomatic Society and the Society for Psychophysiological Research to review two common genetically informative research designs for human studies: twin and genetic association studies. We outline statistical techniques for each and, for genetic association studies, address special topics, including the treatment of race and ethnicity, gene x gene and gene x environment interaction, haplotype analysis, and power and sample size. Finally, we discuss the issue of nonreplication and interpretation of results derived from genetic association studies. We hope this overview of twin and genetic association designs will support and stimulate thoughtful applications of genetic approaches within psychosomatic medicine.