Test–Retest Reliability and Efficacy of Individual Symptoms in Concussion Management

Abstract

Objective: (1) To determine test–retest reliability of individual Sport Concussion Assessment Tool—Third Edition (SCAT-3) symptom scores and symptom severity scores, (2) to examine the specificity/sensitivity of individual SCAT-3 symptom severity scores acutely (24-48 hours) postconcussion, and (3) to develop a model of symptoms best able to differentiate concussed from nonconcussed student athletes and cadets. Design: Prospective, longitudinal, and cross-sectional. Setting: Twenty-six civilian schools and 3 US service academies. Participants: Collegiate student athletes (n = 5519) and cadets (n = 5359) from the National Collegiate Athletic Association–Department of Defense Grand Alliance: Concussion Assessment, Research and Education Consortium, including 290 student athletes and 205 cadets, assessed 24 to 48 hours postconcussion. Independent Variables: Concussed and nonconcussed student athlete and cadet groups. Main Outcome Measures: Sport Concussion Assessment Tool—Third Edition individual symptom severity scores, total symptom scores, and symptom severity scores. Results: Results indicated poor test–retest reliability across all symptom scores (intraclass correlation coefficient = 0.029-0.331), but several individual symptoms had excellent predictive capability in discriminating concussed from nonconcussed participants (eg, headache, pressure in the head, and don't feel right had area under the curve >0.8, sensitivity >70%, and specificity >85%) regardless of baseline testing. These symptoms were consistent with Chi-square Automatic Interaction Detector classification trees with the highest mean probability. Conclusions: Findings support the excellent diagnostic accuracy of honest symptom reporting, notwithstanding the known limitations in symptom underreporting, and suggest that there may be added value in examining individual symptoms rather than total symptom scores and symptom severity scores alone. Finally, findings suggest that baseline testing is not necessary for interpreting postconcussion symptom scores.