Preliminary analysis showed that 13 of the 14 temporally primary lifetime DSM-IV/CIDI disorders had elevated univariate ORs predicting disaster-related PTSD (11 of them significant at the 0.05 level), but that only a handful were significant in a multivariate model due to high co-morbidity. (See supplementary material online.) The most parsimonious characterization of these associations used dummy variables for exactly 1 (17.6%, OR 9.8, 95% CI 0.5–192.4) and 2+ (14.1%, OR 60.0, 95% CI 21.1–170.5) prior lifetime DSM-IV/CIDI disorders as predictors. The significant OR of prior lifetime TEs in model 3 decreased substantially, while the significant OR of serious injury or death of someone close to the respondent increased substantially when mental disorders were controlled in model 4 compared to model 3.
