Abstract: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.

Improving equivalent scores for clinical neuropsychology: a new method for regression model selection

Modeling of Correlated Cognitive Function and Functional Disability Outcomes with Bounded and Missing Data in a Longitudinal Aging Study

Criterion validity and sensitivity to change of the Early Rehabilitation Index (ERI): results from a German multi-center study

Generating normative data from web-based administration of the Cambridge Neuropsychological Test Automated Battery using a Bayesian framework

Nothing wrong about change: the adequate choice of the dependent variable and design in prediction of cognitive training success

Evaluating treatment effectiveness under model misspecification: A comparison of targeted maximum likelihood estimation with bias-corrected matching

On the Selection of Item Scores or Composite Scores for Clinical Prediction

Modules for Automated Validation and Comparison of Models of Neurophysiological and Neurocognitive Biomarkers of Psychiatric Disorders: ASSRUnit - A Case Study

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

Modelling the Distribution of Cognitive Outcomes for Early-Stage Neurocognitive Disorders: A Model Comparison Approach

A tutorial on regression-based norming of psychological tests with GAMLSS.

Data-driven decision making for the screening of cognitive impairment in primary care: a machine learning approach using data from the ELSA-Brasil study

Examining the Measurement Equivalence of Alzheimer Disease Research Centers' Neuropsychological Test Battery (Version 3) Between Singapore and US Samples

Methodology for the generation of normative data for the U.S. adult Spanish-speaking population: A Bayesian approach

Appraisal of Gene Expression-Based Classifiers for Neuropsychiatric Disorders: A Meta-Regression

Comparing Stochastic Optimization Methods for Variable Selection in Binary Outcome Prediction, With Application to Health Policy

Reducing Attenuation Bias in Regression Analyses Involving Rating Scale Data via Psychometric Modeling

Assessing Clinical Significance: Proposed Extensions to Method

Validation of regression-based change formulae for mild cognitive impairment and Alzheimer's disease

Correlation-based tests for the formal comparison of polygenic scores in multiple populations

A comparison of methods for model selection when estimating individual treatment effects