Posttraumatic stress disorder (PTSD) is one of the most common mental health problems seen in the Veterans Health Administration (VHA). All VHA patients are screened for PTSD annually, most often in primary care settings. The Posttraumatic Stress Disorder Checklist (PCL) is one of the most frequently used measures to assess PTSD severity, but has not been well evaluated as a screening tool.
The primary objective of the proposed project was to establish the diagnostic utility of the PCL for VHA patients. Because partial or subclinical PTSD may also require clinical attention, we examined cut-off scores based on the presence or absence of both PTSD and partial PTSD. We examined the operating characteristics of the PCL based on designated patient characteristics (age, gender, ethnicity), and setting considerations (primary care). A secondary objective of the present study was to establish the incremental predictive validity of the PCL given prior diagnostic and service use information.
The diagnostic characteristics of the PCL were calculated against the Clinician Administered PTSD scale (CAPS) as the reference gold standard. Secondary analyses were performed on four existing de-identified VHA data sets that included demographic characteristics, PCL items, CAPS items, and when available, pre-existing diagnostic and service use information. The sample was 1676 veterans.
The prevalence of PTSD in this analytic sample was 33.4%. Receiver operating characteristics (ROC) were plotted for the PCL and the corresponding area under the curve was .92. Using likelihood ratios rather than efficiency to establish cutoff scores, we found that PCL scores below 34 were suggestive of no PTSD. PCL scores between 34 and 43 were suggestive of some trauma related distress, but probably not a PTSD diagnosis. PCL scores between 44 and 54 suggested significant distress and a probable PTSD diagnosis, and scores above 54 suggested a definitive PTSD diagnosis. The large majority of patients with scores suggesting probable or definitive PTSD met criteria for either the full diagnosis or subclinical levels of symptoms that are associated with significant clinical impairment. Performance was somewhat poorer when limited only to primary care patients with undetected PTSD, suggesting greater attention to primary care screening protocols may be needed. Unadjusted demographic comparisons suggest some differences in screening performance: The AUC for the PCL was significantly better for men than for women and for older veterans as compared to younger veterans. No differences were found for race/ethnicity.
The results of this study have identified diagnostic information for the PCL when used as a screening instrument for PTSD. The instrument may provide significantly more clinical information when not limited to a single cut score, but used as ranges of scores with varying predictive validity for PTSD. The resulting predictive validity can then be used in conjunction with other relevant patient information to optimize clinical decision making
External Links for this Project
None at this time.