The aim of this study was to analyze the best combination of clinical variables associated with concussion subtypes using a multidomain assessment comprising medical history; symptoms; and cognitive, ocular, and vestibular impairment in a cohort of patients presenting to a concussion specialty clinic.
Adolescent patients (n = 293) completed demographics and medical history, Concussion Clinical Profiles Screening, Immediate Post-Concussion Assessment and Cognitive Testing, and vestibular ocular motor screening at their first visit (mean 7.6 ± 7.8 days postinjury) to a concussion specialty clinic. Each participant was adjudicated to have one or more subtype (anxiety/mood, cognitive, migraine, ocular, and vestibular) by a healthcare professional based on previously published criteria. A series of backward, stepwise logistic regressions were used to identify significant predictors of concussion subtypes, and predictive probabilities from the logistic regression models were entered into area under the receiver operating characteristic curve (AUC) models.
Each of 5 logistic regression models predicting primary subtypes accounted for 28%–50% of the variance (R2 = 0.28–0.50, p < 0.001) and included 2–8 significant predictors per model. Each of the models significantly differentiated the primary subtype from all other subtypes (AUC = 0.76–0.94, p < 0.001).
These findings suggest that each concussion subtype can be identified using specific outcomes from a multidomain assessment. Clinicians can employ such an approach to better identify and monitor recovery from subtypes as well as guide interventions.