We welcome the response by Brietzke and colleagues to our report. First, they suggest that differences in illness severity between those with v. those without the comorbidities we excluded might have accounted for reductions in the prevalence rate of psychotic symptoms in our sample. To help address this point, we reanalysed our data by conducting a logistic regression of data for PTSD patients with v. without the excluded comorbidities as a predictor of the likelihood of having psychotic symptoms while controlling for Global Assessment of Functioning scores. Comorbid status remained a significant predictor. It is important to clarify we did not exclude all comorbidities from the PTSD sample: we removed only those that also allow for the presence of psychotic symptoms (e.g. schizophrenia, bipolar disorder). Other comorbid diagnoses (e.g. anxiety disorders) remained in the refined PTSD sample showing the low prevalence of psychosis. Therefore, we do not believe that differences in illness severity can adequately explain our findings.

Second, Brietzke et al point out that psychotic symptoms in PTSD probably fall along a spectrum from congruent (e.g. hallucinations related to vivid re-experiencing of the trauma) to non-congruent (e.g. bizarre, non-trauma-related hallucinations). We agree with this observation in general, but disagree that congruence criteria are likely to clarify these issues from a diagnostic standpoint. The distinction between congruent v. non-congruent psychotic symptoms in primary mood disorders is known to lack prognostic value and patients frequently exhibit characteristics of both at the same time. ^{Reference Gaudiano, Uebelacker and Miller1}

We agree with Rosen & Lilienfeld ^{Reference Rosen and Lilienfeld2} that continued conceptual confusion regarding the PTSD diagnosis suggests the need for greater caution, rather than a rush to expand the criteria to encompass larger groups of clinical presentations, until the validity of the core features of the PTSD diagnosis can be better established. We also disagree with Breitzke et al that investigations of biological correlates of PTSD are likely to shed much more light on these issues. Extensive previous research in this area has found a lack of evidence for biomarkers linked specifically to the PTSD diagnosis as opposed to those that cut broadly across diagnostic categories and clinical presentations. ^{Reference Rosen and Lilienfeld2}