Statistical support for subtypes in posttraumatic stress disorder: the how and why of subtype analysis.

A number of researchers have argued for the existence of different subtypes of posttraumatic stress disorder (PTSD). In the current paper we present criteria by which to assess these putative subtypes, clarify potential pitfalls of the statistical methods employed to determine them, and propose alternative methods for such determinations. Specifically, three PTSD subtypes are examined: (1) complex PTSD, (2) externalizing/internalizing PTSD, and (3) dissociative/nondissociative PTSD. In addition, three criteria are proposed for subtype evaluation, these are the need for (1) reliability and clarity of definition, (2) distinctions between subtypes either structurally or by mechanism, and (3) clinical meaningfulness. Common statistical evidence for subtyping, such as statistical mean difference and cluster analysis, are presented and evaluated. Finally, more robust statistical methods are suggested for future research on PTSD subtyping.