[Skip to Navigation]
[Skip to Navigation Landing]
November 2017

Parsing Heterogeneity

Author Affiliations
  • 1Department of Psychiatry and Patrick Wild Centre, University of Edinburgh, Edinburgh, Scotland
JAMA Psychiatry. 2017;74(11):1089-1090. doi:10.1001/jamapsychiatry.2017.2953

It is commonplace for researchers to invoke the heterogeneity of psychiatric disorders as an explanation as to why the results of their study differ from others. It is quite common to hear suggestions that heterogeneity is a particular problem in psychiatry. To my knowledge, that statement has no evidence to support it. Heterogeneity is ubiquitous in medicine.

Sampling biases, methodological variabilities, insufficient power, and differential confounding are the most likely explanations for disagreements between studies. If these can be confidently excluded and true rather than artifactual heterogeneity across a population be proven, an explanation of that is required. Population stratification represents the most scientifically tractable approach to teasing out meaning from the morass of data. Clinically, the most useful response would be subtyping by biomarkers that have practical use.