Background: The latent structure of PTSD has been widely investigated to improve the diagnostic validity of the disorder and address issues related to heterogeneity (Amour, Müllerová, Elhai, 2016). However, extant research has been limited by exclusive use of data aggregated across individuals, neglecting intra-individual data, likely limiting our ability to make valid inferences about the individual (Molenaar, 2007). The present study uses a data.driven approach to examine the latent structure of PTSD within a single individual.
Method: The present analysis uses data from a single participant with PTSD drawn from an ongoing research study. Each participant completed a 30-day ecological momentary assessment period, wherein participants received four phone surveys per day and rated the degree to which they had experienced the extant PTSD symptoms during the preceding four hours. Data from each participant was subjected to within-person exploratory and confirmatory factor analysis to identify the latent structure of PTSD within each participant (Cattell, Cattell, & Rhymer, 1947). The fit of this final model was then compared against the fit of both the extant five factor Dysphoric Arousal model of PTSD, which has received the most empirical support to date, and the four factor DSM-5 model of PTSD using conventional fit indices (Hu & Bentler, 1999). Result: A two-factor model derived from the within-person factor analysis procedure demonstrated good fit (χ(241) = 286.68, p = .023, RMSEA = .047, SRMR = .065, TFI = .99, CFI = .99) to the participant's data, whereas both the Dysphoric Arousal model (χ(242) = 408.16, p < .001, RMSEA = .113, SRMR = .097, TFI = .82, CFI = .84) and DSM-5 model (χ(246) = 453.57, p < .001, RMSEA = .123, SRMR = .12, TFI = .78, CFI = .81) provided poor fit.
Discussion: Recently, Galatzer-Levy and Bryant (2013) demonstrated that extant diagnostic criteria result in 636,120 ways to meet criteria for PTSD, casting doubt on the ability of existing psychiatric taxonomies for understanding the individual patient. These results support the hypothesis that attempts to address heterogeneity that exclusively use data aggregated across individuals is limited because it may not capture the structure of PTSD within the individual patient. These results also provide support for the utility of idiographic research methodologies for better capturing the latent structure of psychopathology within the individual (Barlow & Nock, 2009).
University of California, Berkeley