ABSTRACT
OBJECTIVE: To determine which components from a multidomain assessment best predict protracted recovery in pediatric patients with a concussion. STUDY DESIGN: A prospective cohort of patients aged 5-9 years who presented within 21 days of concussion to a specialty clinic were categorized into normal (≤30 days) and protracted (>30 days) recovery. Participants provided demographic and medical history information, and completed the Child Sport Concussion Assessment Tool-5 symptom report and balance assessment, the Vestibular/Ocular Motor Screen-Child (VOMS-C), and the Pediatric Immediate Post-concussion Assessment and Cognitive Testing. Univariate logistic regressions (LR) were used to inform a follow-up forward stepwise LR to identify the best predictors of protracted recovery. Receiver operating characteristic analysis of the area under the curve (AUC) was used to identify which predictors retained from the LR model best discriminated recovery. RESULTS: The final sample included 68 patients (7.52 ± 2.3 years; 56% male), 36 (52.9%) with normal and 32 (47.1%) with protracted recovery. Results of the LR to identify protracted recovery were significant (P < .001) and accounted for 39% of the variance. The model accurately classified 78% of patients, with days to first clinic visit (OR, 1.2; 95% CI, 1.1-1.4; P = .003) and positive VOMS-C findings (OR, 8.32; 95% CI, 2.4-28.8; P < .001) as significant predictors. A receiver operating characteristic analysis of the AUC of this 2-factor model discriminated protracted from normal recovery (AUC, 0.82; 95% CI, 0.71-0.92; P < .001). CONCLUSIONS: Days to first clinic visit and positive findings on the VOMS-C were the most robust predictors of protracted recovery after concussion in young pediatric patients.