Diagnostic Accuracy of the Veteran Affairs' Traumatic Brain Injury Screen

Abstract

ObjectiveTo comprehensively estimate the diagnostic accuracy and reliability of the Department of Veterans Affairs (VA) Traumatic Brain Injury (TBI) Clinical Reminder Screen (TCRS). DesignCross-sectional, prospective, observational study using the Standards for Reporting of Diagnostic Accuracy criteria. SettingThree VA Polytrauma Network Sites. ParticipantsOperation Iraqi Freedom, Operation Enduring Freedom veterans (N=433). Main Outcome MeasuresTCRS, Comprehensive TBI Evaluation, Structured TBI Diagnostic Interview, Symptom Attribution and Classification Algorithm, and Clinician-Administered Posttraumatic Stress Disorder (PTSD) Scale. ResultsForty-five percent of veterans screened positive on the TCRS for TBI. For detecting occurrence of historical TBI, the TCRS had a sensitivity of .56 to .74, a specificity of .63 to .93, a positive predictive value (PPV) of 25% to 45%, a negative predictive value (NPV) of 91% to 94%, and a diagnostic odds ratio (DOR) of 4 to 13. For accuracy of attributing active symptoms to the TBI, the TCRS had a sensitivity of .64 to .87, a specificity of .59 to .89, a PPV of 26% to 32%, an NPV of 92% to 95%, and a DOR of 6 to 9. The sensitivity was higher for veterans with PTSD (.80–.86) relative to veterans without PTSD (.57–.82). The specificity, however, was higher among veterans without PTSD (.75–.81) relative to veterans with PTSD (.36–.49). All indices of diagnostic accuracy changed when participants with questionably valid (QV) test profiles were eliminated from analyses. ConclusionsThe utility of the TCRS to screen for mild TBI (mTBI) depends on the stringency of the diagnostic reference standard to which it is being compared, the presence/absence of PTSD, and QV test profiles. Further development, validation, and use of reproducible diagnostic algorithms for symptom attribution after possible mTBI would improve diagnostic accuracy.