Classification analyses of pediatric concussion assessment battery

Abstract

Objective Demonstrate classification statistics (OR, Likelihood Ratios) of a pediatric multi-component battery of neurocognitive tests, child and parent symptom report to identify the presence or absence of concussion. Design Prospective cohort design. Setting Outpatient clinic, teaching hospital. Participants Two groups of children, ages 5–7 years (n=22 per group) and ages 8–12 years (n=67 per group) (≤7 days post-concussion) were compared to gender/age-matched groups of non-injured children. Interventions Predictor variables: Response Speed Composite, Child and Parent Post-Concussion Symptom Inventory (PCSI) symptom scores. Main Outcome Measurements Group membership—concussion, control. Results MANOVAs revealed significant group differences for the Response Speed Composite, Child and Parent PCSI symptom scores for both age groups. Discriminant function analyses conducted on the Child and Parent PCSI scores and Neurocognitive Composite scores independently and in combination indicate parent and child PCSI symptoms and Response Speed demonstrate adequate classification independently; in combination offered greatest rates. For the 5–7 year age group, 73% were correctly classified by the three combined scores (Positive Likelihood Ratio=3.5, Negative Likelihood Ratio=0.58, OR=6.0, 80% Area Under Curve (AUC)). 84% of 8–12 age group were identified correctly by the three combined scores (LR+=47.0, LR−=.30, OR=155, 94% AUC). Relative differences between scores will be demonstrated. Conclusion The multi-component pediatric battery demonstrates appropriate capability to classify children referred for a suspected concussion. Increased parent and child PCSI symptom ratings and decreased neurocognitive performance (Response Speed) increases diagnostic accuracy, assisting the clinician in their clinical decision-making. Competing interests Drs Gioia, Isquith and Vaughan are co-authors of Pediatric ImPACT battery.

Ähnliche Arbeiten