An evaluation of the DSM-5 factor structure for posttraumatic stress disorder in survivors of traumatic injury.

Confirmatory factor analytic studies of the latent structure of DSM-5 PTSD symptoms using self-report data (Elhai et al., 2012; Miller et al., 2013) have found that the four-factor model implied by the DSM-5 diagnostic criteria provided adequate fit to their data. However, the fit of this model is yet to be assessed using data derived from gold standard structured interview measures. This study evaluated the fit of the DSM-5 four-factor model and an alternative four-factor model in 570 injury survivors six years post-injury using the Clinician Administered PTSD Scale (Blake et al., 1990), updated to include items measuring new DSM-5 symptoms. While both four-factor models fitted the data well, very high correlations between the 'Intrusions' and 'Avoidance' factors in both models and between the 'Negative Alterations in Cognitions and Mood' and 'Arousal and Reactivity' factors in the DSM-5 model and the 'Dysphoria' and 'Hyperarousal' factors in the alternative model were evident, suggesting that a more parsimonious two-factor model combining these pairs of factors may adequately represent the latent structure. Such a two-factor model fitted the data less well according to χ(2) difference testing, but demonstrated broadly equivalent fit using other fit indices. Relationships between the factors of each of the four-factor models and the latent factors of Fear and Anxious-Misery/Distress underlying Internalizing disorders (Krueger, 1999) were also explored, with findings providing further support for the close relationship between the Intrusion and Avoidance factors. However, these findings also suggested that there may be some utility to distinguishing Negative Alterations in Cognition and Mood symptoms from Arousal and Reactivity symptoms, and/or Dysphoria symptoms from Hyperarousal symptoms. Further studies are required to assess the potential discriminant validity of the two four-factor models.