This grant application asks for two years of support for analysis of data collected in three epidemiologically-defined, population-based samples. The proposed analysis will use a latent variable modeling strategy to examine clustering of symptoms of Posttraumatic Stress Disorder (PTSD). To date, diagnostic criteria for psychiatric disorders have been based chiefly on clinical impressions and studies of special populations. Empirical validation, using a general population sample, is critical in order to understand the association of PTSD with comorbid psychiatric disorders, as well as associations with etiologic factors, biological markers, and treatment response. The following specific aims will be addressed: 1) Using latent variable models, the investigators will measure the clustering of symptoms of PTSD and compare the empirical estimation of clustering to the diagnostic criteria for PTSD; 2) the investigators will examine the construct validity of alternative definitions of PTSD, by examining the relationship between these definitions of PTSD and external validators. Three primary data analytic strategies will be used. Latent variable models for categorical variables, a class structural equation models, and latent class analysis will be used to model the relationship of individual symptoms of PTSD to a latent construct(s). These models will be extended to investigate relationships between symptoms of disorders, course (e.g., chronicity, impairment, functional status), and suspected risk factors. Profile analysis of categorical variables, in addition to the other two strategies, will be employed to compare group differences in clustering of PTSD symptoms. This study will have multiple potential health benefits. Improved understanding of the clustering of psychiatric symptoms can guide the development of nosologies that will lead to move valid diagnoses and more effective treatment of psychiatric disorders. In addition, the proposed analyses can further the understanding of the comorbidity between psychiatric disorders and their specific and shared risk factors.
