Diagnostic performance of the PTSD checklist and the Vietnam Era Twin Registry PTSD scale.

Abstract

Self-report questionnaires are frequently used in clinical and epidemiologic studies to assess post-traumatic stress disorder (PTSD). A number of studies have evaluated these scales relative to clinician administered structured interviews; however, there has been no formal evaluation of their performance relative to non-clinician administered epidemiologic assessments such as the Composite International Diagnostic Interview (CIDI). We examined the diagnostic performance of two self-report PTSD scales, the PTSD checklist (PCL) and the Vietnam Era Twin Registry (VET-R) PTSD scale, compared to the CIDI. Data were derived from a large epidemiologic follow-up study of PTSD in 5141 Vietnam Era Veterans. Measures included the PCL, VET-R PTSD scale and CIDI. For both the PCL and VET-R PTSD scale, ROC curves, areas under the curve (AUC), sensitivity, specificity, % correctly classified, likelihood ratios, predictive values and quality estimates were generated based on the CIDI PTSD diagnosis. For the PCL and VET-R PTSD scale the AUCs were 89.0 and 87.7%, respectively. Optimal PCL cutpoints varied from the 31-33 range (when considering sensitivity and specificity) to the 36-56 range (when considering quality estimates). Similar variations were found for the VET-R PTSD, ranging from 31 (when considering sensitivity and specificity) to the 37-42 range (when considering quality estimates). The PCL and VET-R PTSD scale performed similarly using a CIDI PTSD diagnosis as the criterion. There was a range of acceptable cutpoints, depending on the metric used, but most metrics suggested a lower PCL cutpoint than in previous studies in Veteran populations.