Discriminative accuracy of the 42-point modified Mini-Balance Evaluation Systems Test for identifying Expanded Disability Status Scale-defined disability in relapsing-remitting multiple sclerosis: an exploratory diagnostic accuracy study.

Abstract

Balance impairment is common in multiple sclerosis, but the discriminative performance of clinical balance measures for Expanded Disability Status Scale (EDSS)-defined disability remains unclear. I compared the 42-point modified Mini-Balance Evaluation Systems Test (modified Mini-BESTest) and the Single-Leg Stance Test (SLST) for EDSS greater than or equal to 4.0 in adults with relapsing-remitting multiple sclerosis. This secondary, single-center, cross-sectional diagnostic accuracy analysis used prospectively collected data from 54 ambulatory participants with static or dynamic balance impairment (EDSS = 1.0-6.0). Both balance tests were administered during the same outpatient visit, while EDSS was assessed separately by a neurologist. Areas under the receiver operating characteristic curve (AUC) were compared using the DeLong test; associations were estimated with Firth penalized logistic regression, and the Youden-derived threshold was assessed by bootstrapping. Twenty-five participants (46%) had EDSS greater than or equal to 4.0. The modified Mini-BESTest showed an AUC of 0.95 [95% confidence interval (CI) = 0.90-1.00], exceeding the SLST (0.77, 0.65-0.90; DeLong Z = 3.18, P = 0.002). An exploratory threshold of less than or equal to 31/42 yielded 92% sensitivity, 90% specificity, a positive likelihood ratio of 8.9 and a negative likelihood ratio of 0.09, and recurred in 79% of 2000 bootstrap resamples. Each additional point was associated with lower adjusted odds of EDSS greater than or equal to 4.0 (odds ratio = 0.75, 95% CI = 0.61-0.86). The modified Mini-BESTest showed excellent discrimination and outperformed the SLST. The less than or equal to 31/42 threshold is exploratory, requires external validation, and must not be applied to the standard 28-point Mini-BESTest.