Principles of diagnostic taxonomy suggest that disorders of a specific class, or spectrum, should aggregate more with each other than with disorders from another class. Results of recent comorbidity studies raise questions about whether this is true for post-traumatic stress disorder (PTSD) – which has been classified as an anxiety disorder since DSM–III – and the implications for where the diagnosis should be located in DSM–V.

Several factor analyses of diagnostic data from epidemiological and clinical samples suggest that PTSD covaries more strongly with disorders defined by anhedonia, worry and rumination (i.e. the unipolar mood disorders and generalised anxiety disorder) than with ones characterised by pathological fear and avoidance (e.g. the phobias, panic/agoraphobia and obsessive–compulsive disorder). ^{Reference Miller, Fogler, Wolf, Kaloupek and Keane1} However, classifying PTSD among these ‘anxious-misery’ disorders provides a poor fit to the data because PTSD is conditional on trauma exposure and, in new-onset cases, typically develops before its comorbid conditions. For example, when PTSD and major depression co-occur following trauma exposure, PTSD usually precedes or develops concurrently with the depression. New-onset major depression that develops in the wake of trauma rarely precedes or develops in the absence of PTSD. ^{Reference Breslau, Davis, Peterson and Schultz2} This implies a causal influence of PTSD on comorbid psychopathology and suggests a distinct phenomenology which should be reflected in its diagnostic class membership within DSM.

Developmental studies have shown that adult psychopathology is often foreshadowed by childhood and/or adolescent problems in the same domain. Along these lines, many adults with anxiety disorders report histories of juvenile anxiety disorders but they do not typically report juvenile externalising disorders. The exception to this is found among samples of individuals with PTSD where adult patients frequently have histories of childhood externalising disorders. ^{Reference Gregory, Caspi, Moffitt, Koenen, Eley and Poulton3} Twin studies align with this finding and have shown that PTSD shares genetic influences with both internalising- and externalising-spectrum diagnoses. ^{Reference Koenen, Fu, Lyons, Toomey, Goldberg and Eisen4} Other work suggests that many adults with PTSD exhibit a predominantly externalising pattern of comorbidity characterised by problems in the domain of impulse-control, antisociality and substance misuse. ^{Reference Miller, Kaloupek, Dillon and Keane5} These findings raise concern about conceptualising PTSD simply as the manifestation of a vulnerability to anxiety-related psychopathology.

Since its third edition, the DSM has taken a largely descriptive, as opposed to aetiological, approach to defining and classifying disorders. The most notable exception to this is the PTSD diagnosis, which specifies a causal relationship between trauma exposure and symptom development. We believe that the most appropriate location for PTSD in DSM–V would be among a class of disorders precipitated by serious adverse life events, i.e. a spectrum of traumatic stress disorders. Candidates for inclusion would include PTSD, acute stress disorder, adjustment disorder, a traumatic grief or bereavement-related diagnosis, and possibly complex PTSD. These disorders are the product of an environmental pathogen (i.e. a traumatic stressor) operating on individual diatheses that span the spectrum of human variation in vulnerability to psychopathology. This diathesis–stress interaction can result in extensive heterogeneity in the phenotypic expression of psychopathology – pathological anxiety being just one manifestation of the process.