Sensitivity and Specificity of a Multimodal Approach for Concussion Assessment in Youth Athletes

Abstract

Context: Current international consensus endorses a multimodal approach to concussion assessment. However, the psychometric evaluation of clinical measures used to identify postconcussion performance deficits once an athlete is asymptomatic remains limited, particularly in the pediatric population. Objective: To describe and compare the sensitivity and specificity of a multimodal assessment battery (balance, cognition, and upper and lower body strength) versus individual clinical measures at discriminating between concussed youth athletes and noninjured controls when asymptomatic. Design: Prospective cohort study. Setting: Hospital laboratory setting. Participants: A total of 32 youth athletes with a concussion and 32 matched (age and sex) noninjured control participants aged 10–18 years. Intervention(s): Participants were administered preinjury (baseline) assessments of cognition (Immediate Post-Concussion Assessment and Cognitive Testing [ImPACT]), balance (BioSway), and upper and lower body strength (grip strength and standing long jump). Assessments were readministered when concussed participants reported symptom resolution (asymptomatic time point). Noninjured control participants were reassessed using the same time interval as their concussion matched pair. Sensitivity and specificity were calculated using standardized regression-based methods and receiver operating characteristic curves. Main Outcome Measures: Outcome measures included baseline and postinjury ImPACT, BioSway, grip strength, and standing long jump scores. Results: When asymptomatic, declines in performance on each individual clinical measure were seen in 3% to 22% of the concussion group (sensitivity = 3%–22%) compared with 3% to 13% of the noninjured control group (specificity = 87%–97%) (90% confidence interval). The multimodal battery of all combined clinical measures yielded a sensitivity of 41% and a specificity of 77% (90% confidence interval). Based on discriminative analyses, the multimodal approach was statistically superior compared with an individual measures approach for balance and upper and lower body strength, but not for cognition. Conclusions: Results provide a foundation for understanding which domains of assessment (cognition, balance, and strength) may be sensitive and specific to deficits once symptoms resolve in youth athletes. More work is needed prior to clinical implementation of a preinjury (baseline) to postinjury multimodal approach to assessment following concussion in youth athletes.