Insights Into the Mini-BESTest Scoring System: Comparison of 6 Different Structural Models

Objective. The Mini-Balance Evaluation Systems Test (Mini-BESTest) is a balance scale common to clinical practice, but different scoring has been proposed, that is, total score and/or subsections. This study aimed to investigate Mini-BESTest validity by comparing 6 structural models and to establish the best model for discriminating fallers from nonfallers, that is, those who did or did not report at least 2 falls in the 6 months before evaluation. Methods. In this cross-sectional validation study, data from 709 individuals with idiopathic Parkinson disease (Hoehn and Yahr stages 1-3) were analyzed. Individuals were evaluated with the Mini-BESTest, and fall history was recorded. Construct, convergent, and discriminant validity and reliability of the 6 models were analyzed. The ability of the models to adequately identify individuals with or without a history of falls was tested with receiving operating characteristic curves. Results. Confirmatory factor analysis showed that the unidimensional models and the 4-factor solutions showed the best fit indexes. Conversely, second-order models, which allowed reporting of both total and subsections, did not converge. Most models and factors showed a low convergent validity (average variance extracted values <0.5). Correlations among the anticipatory postural adjustments factor with both the sensory orientation and the dynamic gait factors of multidimensional models were high (r >= 0.85). Unidimensional model reliability was good, whereas low values were found in one-half of the subsections. Finally, both unidimensional models showed a large area under the receiving operating characteristic curve (0.81). Conclusion. The original unidimensional Mini-BESTest model-with a total score of 28-showed the highest validity and reliability and was best at discriminating fallers from nonfallers. Conversely, its 4 subsections should not be reported separately, because they were highly correlated and had low reliability; therefore, they are not actually capable of measuring different aspects of balance. Impact. This study shows that the Mini-BESTest should be used only with the original unidimensional scoring system in people with Parkinson disease.

Automated summary

The original unidimensional Mini-BESTest model with a total score of 28 showed the highest validity and reliability and was best at discriminating fallers from nonfallers. However, its 4 subsections should not be reported separately, because they were highly correlated and had low reliability.