Performance characteristics of the Posttraumatic Stress Disorder Checklist and SPAN in Veterans Affairs primary care settings

Abstract

Background Posttraumatic stress disorder (PTSD) is a treatable disorder, and individuals with this condition may benefit from early detection. Many people with PTSD are not aware of its symptoms and do not seek treatment, making a brief and targeted screening program a worthwhile endeavor. For this reason, research aimed at improving screening instruments could yield substantial benefits. Objectives The primary objective of this research was to assess the diagnostic performance of two popular PTSD screening assessments, the PTSD Checklist (PCL) and the SPAN, in a Veterans Affairs (VA) primary care setting. Additionally, we compared the screening performance of these two assessments by sex and race. Methods The PCL and SPAN were compared with a gold standard, the Clinician-Administered PTSD Scale. Receiver operating characteristic curves were used in conjunction with sensitivity and specificity measures to assess the performance of each screening assessment. These analyses are based on a large database ( n=1076) that was derived from a multisite cross-sectional study conducted at four southeastern VA medical centers. Results Results for the PCL support cutoff scores lower than those previously published, whereas results for the SPAN support the previously recommended cutoff score of 5 (sensitivity of 73.68% and specificity of 81.99%). We found no significant difference in areas under the curve (AUCs) by sex and by race between the PCL and SPAN. We did find that there was a highly significant difference ( P<.0006) in overall diagnostic ability (as measured by the AUC) between the PCL (AUC=0.882) and SPAN (AUC=0.837), making the PCL the preferred screening tool, unless brevity is essential. Conclusions Clinicians and researchers should consider lower cutoff scores for the PCL, but the originally suggested cutoff score for the SPAN is appropriate.