Reliability of the balance error scoring system in a population with protracted recovery from mild traumatic brain injury.

OBJECTIVE: This study aimed to identify the Balance Error Scoring System's (BESS) intraclass reliability in a cohort of patients with prolonged symptoms using variance component analysis and intraclass correlation coefficient (ICC). SETTING: Outpatient sports medicine/concussion clinic. PARTICIPANTS: A total of 241 paediatric and 102 adult patients with symptoms lasting longer than 10 days. INTERVENTIONS: BESS testing. DESIGN: Retrospective review. MAIN OUTCOME MEASURES: Percent variance for each BESS component and intraclass reliability. A five-component model (including all components except for firm double-leg) and a four- component model (including all components except for firm and foam doubleleg) were also performed to compare ICC values. RESULTS: The largest source of variance came from stance (41.1%). The BESS components firm single (25.5%) and foam tandem (27.5%) stances accounted for the largest percentages of variance, while firm double (1.1%) and foam double (6.9%) accounted for the smallest percentages. The ICC for all patients was 0.800, and increased both if the firm double stance was excluded, or if both double-leg stances were excluded. CONCLUSION: BESS reliability appears to be high in a concussed cohort, regardless of age. Removing the two double-leg stance portions increases the ICC of the test without failing to identify balance deficits.