Prediction of One-Week Sport-Related Concussion Symptom Severity Using the Sport Concussion Assessment Tool and CogState Brief Battery

BACKGROUNDThe recovery from sport-related concussion (SRC) is highly heterogenous, with many individuals experiencing symptoms that persist beyond typical recovery timeframes. The early identification of individuals at risk of prolonged symptoms is therefore critical to inform timely interventions and set realistic recovery expectations. Although acute symptom burden is one predictor of future symptom burden, reliance on self-reported measures may limit objectivity and reduce clinical utility in settings where symptom evaluation may be unreliable. In this prospective cohort study, we evaluated the discriminatory accuracy of the CogState Brief Battery, alone and in combination with the Sport Concussion Assessment Tool (SCAT), to classify Australian football players with SRC from Australian footballers without SRC at 24-hours post-injury/match. Furthermore, we examined whether CogState performance and symptom severity at 24 hours were associated with symptom outcomes at one-week post-injury. Adult amateur Australian football players (n=181) were recruited following SRC (n=109 SRC, 86% male) or after a non-injured match (n=72, 90% male). Participants completed the CogState Brief Battery, SCAT and Rivermead Post Concussion Questionnaire (RPQ) at 24-hours and one-week post-injury or match. Area under the receiver operating characteristic (AUC) analyses quantified the ability of 24-hour CogState task performance and SCAT symptom severity to distinguish SRC from controls. Linear regression models examined associations between CogState performance and symptom severity (SCAT and RPQ), within and across the 24-hour and one-week time points. Additional models evaluated whether combining 24-hour symptom severity assessments with CogState performance improved prediction of one-week symptom burden and symptomatic status. SCAT symptom severity demonstrated excellent discriminatory classification accuracy for SRC versus controls at 24-hours post-injury (AUC [95% CI]: 0.949 [0.916 - 0.981]). CogState task performance showed lower discriminatory accuracy but demonstrated fair classification and prognostic utility (e.g., Identification task AUC [95% CI]: 0.666 [0.582 - 0.750]). CogState performance at 24-hours was significantly associated with overall symptom severity at both 24-hours and one-week, as well as with symptom severity across individual symptom domains. In combined models, 24-hour symptom severity and CogState performance independently contributed to the prediction of symptomatic from asymptomatic individuals at one-week post-SRC (e.g., Identification task AUC [95% CI]: 0.721 [0.606 - 0.835] for classification based on <4 versus [&ge;]4 symptoms). These findings indicate that CogState performance at 24-hours post-SRC may serve as an objective adjunct to subjective symptom-based reporting, supporting both diagnosis and early prognostication in the clinical evaluation of SRC.