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.