External validation of the minimal important change in the Berg Balance Scale scores of patients hospitalized for subacute stroke.

Abstract

To investigate the external validation of several reported minimal important change (MIC) values in the Berg Balance Scale (BBS) scores of patients with subacute stroke in a rehabilitation unit, in the overall cohort and walking-ability subgroups. Longitudinal study. A rehabilitation unit. Patients with subacute stroke (n=286). Not applicable. The BBS, Functional Ambulation Categories (FAC), and Functional Independence Measure (FIM) were assessed at 30 days (baseline) and 60 days (follow-up) after stroke onset. BBS score changes between the baseline and follow-up were identified and categorized based on ≥2-, ≥5-, and ≥9-point thresholds. We divided the patients into pairs of groups based on FAC ≥1 or FIM ≥22 improvement. External validation was performed using the LR (likelihood ratio): LR+ >2.0 and LR- <0.5 were considered valid. We also divided the patients according to walking ability (FAC ≤2: assisted-walking group, FAC ≥3: unassisted-walking group), and external validation was repeated using the 5-point BBS MIC for the assisted-walking group and 4-point BBS MIC for the unassisted-walking group. The 5- and 2-point BBS MICs met the LR- criteria for both FAC and FIM improvements as anchors (5 pts: LR- = 0.48 [FAC], 0.46 [FIM]; 2 pts: LR- = 0.21 [FAC], 0.22 [FIM]); however, the 9-point BBS MIC did not. In the assisted-walking group, the 5-point BBS MIC also met the LR- criteria (LR- = 0.27 [FAC], 0.35 [FIM]). In the unassisted-walking group, the 4-point BBS MIC did not meet the criteria. The 5-point BBS MIC was applicable for patients with subacute stroke in a rehabilitation unit.