External validation of Berg Balance Scale cutoff values for walking among patients with stroke: a retrospective multicenter study

PURPOSE: To externally evaluate previously reported Berg Balance Scale (BBS) cutoff values related to walking independence and walking speed in patients with stroke. MATERIALS AND METHODS: In this multicenter cross-sectional study, we investigated 349 patients with stroke.We classified study participants as above or below each published BBS cutoff and evaluated each cutoff against two outcomes: independent walking (Functional Ambulation Categories [FAC] ≥4) and community walking speed (comfortable walking speed [CWS] ≥0.8 m/s). Cutoffs developed for comparable target conditions were evaluated as external-validation analyses, whereas cutoffs developed using different walking definitions were evaluated in exploratory cross-outcome analyses. We classified participants as above or below each prespecified BBS cutoff and calculated sensitivity, specificity, positive likelihood ratio (LR+), and negative likelihood ratio (LR-) for each cutoff. A cutoff was considered useful when the LR+ was > 5.0 and the LR- was < 0.2. RESULTS: For independent walking (FAC ≥ 4), the 48-point cutoff was the only FAC-derived cutoff that met both prespecified LR benchmarks; the 46- and 49-point cutoffs met only one benchmark each. For community walking (CWS ≥ 0.8 m/s), none of the gait-speed-based cutoffs met both LR benchmarks. CONCLUSIONS: In an independent stroke cohort, the 48-point BBS cutoff showed potentially useful discriminative performance for identifying FAC ≥ 4. No gait-speed-based BBS cutoff met both prespecified LR benchmarks for CWS ≥ 0.8 m/s.