RT Journal Article
SR Electronic
T1 GL.05 Controversies in gene–environment interaction research: why all the fuss?
JF Journal of Neurology, Neurosurgery & Psychiatry
JO J Neurol Neurosurg Psychiatry
FD BMJ Publishing Group Ltd
SP e2
OP e2
DO 10.1136/jnnp-2011-300504.25
VO 82
IS 8
A1 M Munafò
YR 2011
UL http://jnnp.bmj.com/content/82/8/e2.18.abstract
AB There has been considerable interest in recent years in the role of gene×environment (G×E) interactions in the aetiology of psychiatric disorders. While the fact that genetic effects may moderate the effects of environmental exposure is widely accepted, and supported by family and twin studies, there is controversy regarding the degree of evidence for specific G×E interactions reported in the literature. Here the various reasons for this ongoing controversy, and the extent to which this debate has polarised the field of G×E research, are discussed.A number of relevant factors influencing the weight of evidence for specific G×E interactions will be discussed, including the degree of stringency employed in the definition of “replication”, the impact of different statistical models used to test for these interactions, and the statistical power of individual studies. In only a minority of studies is a replication reported that is qualitatively comparable to that in the original report, which may or may not mean that the original finding was spurious. Given reasonable assumptions regarding likely genetic and environmental effects, simulations indicate that many published studies may be underpowered, suggesting that some positive results for widely reported G×E interactions may indeed be compatible with chance findings.Many difficulties arise from the use of existing datasets, which obscures whether different findings across studies are due to characteristics of the data (eg, questionnaire vs interview measures, sample characteristics), genuine effect heterogeneity, or simply chance. A further difficulty is the temptation to mine datasets for nominally significant results, which raises the risk of spurious findings (further exacerbated when interaction effects in the absence of main effects are observed). Recommendations for future studies will be discussed.