Developing an optimal short‐form of the PTSD Checklist for DSM‐5 (PCL‐5)

• Published in: Depression and anxiety Vol. 36; no. 9; pp. 790 - 800
• Main Authors: Zuromski, Kelly L., Ustun, Berk, Hwang, Irving, Keane, Terence M., Marx, Brian P., Stein, Murray B., Ursano, Robert J., Kessler, Ronald C.
• Format: Journal Article
• Language: English
• Published: United States Hindawi Limited 01-09-2019
• Subjects: Adult; Artificial intelligence; Checklist - methods; Checklist - standards; diagnosis; Diagnostic and Statistical Manual of Mental Disorders; Factor analysis; Factor Analysis, Statistical; Female; Humans; Learning algorithms; Machine learning; Male; Mass Screening; Medical diagnosis; Military Personnel; Post traumatic stress disorder; psychological tests/psychometrics; Psychometrics; Sensitivity and Specificity; Stress Disorders, Post-Traumatic - diagnosis; trauma‐ and stressor‐related disorders; diagnosis; trauma- and stressor-related disorders; psychological tests/psychometrics; military personnel
• ISSN: 1091-4269; 1520-6394
• DOI: 10.1002/da.22942

Background
Although several short‐forms of the posttraumatic stress disorder (PTSD) Checklist (PCL) exist, all were developed using heuristic methods. This report presents the results of analyses designed to create an optimal short‐form PCL for DSM‐5 (PCL‐5) using both machine learning and conventional scale development methods.
Methods
The short‐form scales were developed using independent datasets collected by the Army Study to Assess Risk and Resilience among Service members. We began by using a training dataset (n = 8,917) to fit short‐form scales with between 1 and 8 items using different statistical methods (exploratory factor analysis, stepwise logistic regression, and a new machine learning method to find an optimal integer‐scored short‐form scale) to predict dichotomous PTSD diagnoses determined using the full PCL‐5. A smaller subset of best short‐form scales was then evaluated in an independent validation sample (n = 11,728) to select one optimal short‐form scale based on multiple operating characteristics (area under curve [AUC], calibration, sensitivity, specificity, net benefit).
Results
Inspection of AUCs in the training sample and replication in the validation sample led to a focus on 4‐item integer‐scored short‐form scales selected with stepwise regression. Brier scores in the validation sample showed that a number of these scales had comparable calibration (0.015–0.032) and AUC (0.984–0.994), but that one had consistently highest net benefit across a plausible range of decision thresholds.
Conclusions
The recommended 4‐item integer‐scored short‐form PCL‐5 generates diagnoses that closely parallel those of the full PCL‐5, making it well‐suited for screening.