Enhancing the usefulness of the Mini-BESTest for measuring dynamic balance: a Rasch validation study

BACKGROUND: Recently, a new balance scale, the Mini-BESTest, was introduced. This scale can be administered in about 15 min, and focuses on "dynamic balance". In spite of the recently increased use of this scale, further psychometric studies seem called for to enhance confidence in its use in different fields of clinical practice and research. AIM: To re-examine through Rasch analysis the metric properties of the Mini-BESTest and provide a nomogram that allow to quickly transform raw scores of the scale into linear estimates of dynamic balance. DESIGN: Observational cross-sectional study. SETTING: Rehabilitation hospital. POPULATION: A total of 234 patients were consecutively admitted with a variety of neurological diseases causing balance impairment. METHODS: Internal construct validity was assessed by determining how well data fit the Rasch model. Reliability was estimated for both persons and items. Scale unidimensionality and local independence of items were analysed performing a principal component analysis (PCA) on the standardized residuals. Also, a differential item functioning (DIF) analysis was run to assess the stability of item calibration across subsamples of patients. RESULTS: All 14 items of Mini-BESTest fitted the "dynamic balance" construct, i.e., the mean of the squared residuals for both the infit and outfit was between 0.8 and 1.2. The person abilities-item difficulty matching was very good. The reliability indices of the Mini-BESTest were as follows: person separation index=3.24 and person reliability=0.91; item separation index=12.00 and item reliability=0.99. The PCA of standardized residuals showed that the variance attributable to the Rasch factor was good (68%) and the eigenvalue of the unexplained variance in the first contrast was just 1.9, thus confirming the unidimensionality of the scale. No DIF was found across gender and age groups. CONCLUSION: The reliability indexes confirmed their high values, giving a high degree of confidence in the consistency of both person-ability and item-difficulty estimates. Results allowed to transform the ordinal summed raw scores of the Mini-BESTest into interval-level measurements using a nomogram. Since no significant local dependence between items was found, this means that each Mini-BESTest item is appropriate for measuring the variable of interest (dynamic balance) and not redundant. DIF analysis showed the stability of item hierarchy and difficulty among subsamples of patients of different gender and age. CLINICAL REHABILITATION IMPACT: This study further increases confidence in use of the Mini-BESTest for clinical assessment of dynamic balance in patients undergoing rehabilitation