Diagnostic accuracy and reliability of sideline concussion evaluation: a prospective, case-controlled study in college athletes comparing newer tools and established tests.

ABSTRACT

OBJECTIVE: To assess diagnostic accuracy and reliability of sideline concussion tests in college athletes. METHODS: Athletes completed baseline concussion tests including Post-Concussion Symptom Scale, Standardised Assessment of Concussion (SAC), modified Balance Error Scoring System (m-BESS), King-Devick test and EYE-SYNC Smooth Pursuits. Testing was repeated in athletes diagnosed acutely with concussion and compared to a matched teammate without concussion. RESULTS: Data were collected on 41 concussed athletes and 41 matched controls. Test-retest reliability for symptom score and symptom severity assessed using control athletes was 0.09 (-0.70 to 0.88) and 0.08 (-1.00 to 1.00) (unweighted kappa). Intraclass correlations were SAC 0.33 (-0.02 to 0.61), m-BESS 0.33 (-0.2 to 0.60), EYE-SYNC Smooth Pursuit tangential variability 0.70 (0.50 to 0.83), radial variability 0.47 (0.19 to 0.69) and King-Devick test 0.71 (0.49 to 0.84). The maximum identified sensitivity/specificity of each test for predicting clinical concussion diagnosis was symptom score 81%/94% (3-point increase), symptom severity score 91%/81% (3-point increase), SAC 44%/72% (2-point decline), m-BESS 40%/92% (5-point increase), King-Devick 85%/76% (any increase in time) and EYE-SYNC Smooth Pursuit tangential variability 48%/58% and radial variability 52%/61% (any increase). Adjusted area under the curve was symptom score 0.95 (0.89, 0.99), symptom severity 0.95 (95% CI 0.88 to 0.99), SAC 0.66 (95% CI 0.54 to 0.79), m-BESS 0.71 (0.60, 0.83), King-Devick 0.78 (0.69, 0.87), radial variability 0.47 (0.34, 0.59), tangential variability 0.41 (0.30, 0.54) CONCLUSION: Test-retest reliability of most sideline concussion tests was poor in uninjured athletes, raising concern about the accuracy of these tests to detect new concussion. Symptom score/severity had the greatest sensitivity and specificity, and of the objective tests, the King-Devick test performed best.