Multifaceted concussion assessment battery: sensitivity at the expense of specificity?

Abstract

A multifaceted assessment battery is recommended for testing suspected concussed athletes; however, the individual tests have limitations and potentially may lead to false positive outcomes. Therefore, the purpose of this study was to psychometrically evaluate concussion assessment tools used for intercollegiate student-athletes, with a focus on the time interval between baseline and subsequent assessments. 92 collegiate student-athletes matched between concussion (N = 46. F32/M14) and non-concussion (N = 46, 32F/14 M) completed the standard assessment of concussion, balance error scoring system, symptom questionnaire, and computerized neurocognitive tests at baseline and acutely (<48 hours) following a concussion. Test outcomes were compared between time points with three approaches, 1) vs baseline, 2) reliable change scores (RCI), and 3) vs American College of Rehabilitation Medicine (ACRM). Concussion assessment sensitivity, specificity, likelihood ratios, predictive values, and diagnostic odds ratio were calculated. A receiver operator characteristic compared area under the curve (AUC) for the overall battery as well as between academic years. The sensitivity of the battery was high (78.3 - 95.7%), but specificity was low to moderate (6.5 - 52.2%) with comparison to the RCI typically performing best. The three approaches yielded AUC values between 0.51 - 0.63 which is below the discriminatory threshold (0.70) with comparison to RCI performing best. By academic year, Juniors was the only year in which the ROC exceeded the threshold (0.75). The number of tests failed did not improve any AUC values (0.51 - 0.69) to exceed the threshold. The outcomes of this study support the premise that concussion must be evaluated using a comprehensive clinical examination including a variety of assessments to evaluate each clinical domain. Clinicians need to recognize that the multifaceted assessment battery has high sensitivity, but the overall psychometrics do not exceed the threshold for group discrimination and caution must be used in their interpretation.