Post-traumatic stress disorder screening test performance in civilian primary care.

PURPOSE we determined the test performance characteristics of four brief post-traumatic stress disorder (PTSD) screening tests in a civilian primary care setting. METHODS this was a cross-sectional cohort study of adults attending a family medicine residency training clinic in the southeastern USA. Four hundred and eleven participants completed a structured telephone interview that followed an index clinic visit. Screening tests included: PTSD Symptom Checklist-Civilian Version (17 items), SPAN (four items), Breslau's scale (seven items) and Primary Care PTSD screen (PC-PTSD) (four items). A modified Clinician-Administered PTSD Scale was used to determine past month PTSD for comparison. Receiver operating characteristic analysis based on area under the curve (AUC) was used to assess diagnostic efficiency (>0.80 desired). Cut-off scores were selected to yield optimal sensitivity and specificity (>80%). RESULTS past month PTSD was substantial (women = 35.8% and men = 20.0%; P < 0.01). AUC values were PTSD Symptom Checklist (PCL) (0.897), SPAN (0.806), Breslau's scale (0.886) and PC-PTSD (0.885). Optimal cut-scores yielded the following sensitivities and specificities: PCL (80.0% and 80.7%; cut-off = 43), SPAN (75.9% and 71.6%; cut-off = 3), Breslau's scale (84.5% and 76.4%; cut-off = 4) and PC-PTSD (85.1% and 82.0%; cut-off = 3). Overall and gender-specific screening test performances were explored. CONCLUSIONS results confirm: (i) PTSD was common, especially among women; (ii) all four PTSD screening tests were diagnostically adequate; (iii) Two of four PTSD screening tests showed adequate sensitivity and specificity (>80%) and (iv) The PC-PTSD screening test (four items) appeared to be the best single screening test. There are few studies to establish the utility of PTSD screening tests within civilian primary care.