Análise das propriedades psicométricas da versão brasileira do Balance Evaluation Systems Test e do MiniBESTest em indivíduos com hemiparesia decorrente de acidente vascular encefálico

Abstract

Understanding how balance is controlled directs us in the way we assess and treat its dysfunction. The ability to balance arises from the interaction of the visual, vestibular, cognitive, biomechanical and somatosensory systems integrated within the central nervous system. Individuals with hemiparesis often presents balance impairments. The Balance Evaluation Systems Test (BESTest) is an instrument for clinical assessment of balance that aims to help identify the subsystems of postural control that may account for the change in functional balance, so that treatment can be directed specifically to the subsystems involved in the balance deficit. Its reduced version, the MiniBESTest, is an instrument aimed at the screening of balance disorders. The BESTest and MiniBESTest were translated and culturally adapted into Portuguese-Brazil in a study with elderly and individuals with Parkinson's disease and the Brazilian versions have not yet been studied in other populations. Thus, the present study aimed to evaluate the reliability and construct validity of the Brazilian versions of the BESTest and MiniBESTest in subjects with chronic hemiparesis due to stroke. The test-retest and interrater reliability were assessed using the weighted kappa coefficient. The Rasch model was used for analysis of construct validity in 40 chronic hemiparesis (21 men, mean age = 58.8 ± 12.84). Moderate to almost perfect agreement was observed for the BESTest for test-retest (0,55 Kw 1, 0; p<0,05) and interrater (0,48 Kw 1,0; p<0,05). Strong to almost perfect agreement was observed for MiniBESTest for test-retest (0,62 Kw 1,0; p<0,05) and interrater reliability (0,71 Kw 1,0; p<0,05). The items of the BESTest and MiniBESTest divided people in about four and three skill levels respectively. The individuals' measurement and items calibration stability ranged from 0.89 to 0.96 for the instruments. The instruments were multidimensional. In BESTest, the most difficult item was "sit on the floor and stand up" and in MiniBESTest "compensatory stepping correction lateral to the right." The easier item for both instruments was "stand up on firm surface, eyes open." Two items of the BESTest were erratic and one individual showed erratic pattern of responses, being within the expected range for the instrument. In MiniBESTest were not observed items or individuals with erratic pattern. The level of difficulty of the items of both instruments was appropriate for the skill level of individuals. We concluded that the Brazilian versions of the BESTtest and MiniBESTest proved reliable and with adequate construct validity, being able to discriminate different skill levels among individuals, without ceiling or floor effect. It is recommended, however, caution in interpreting the total score of the instrument and attention in relation to erratic items, checking whether factors associated with unexpected behaviors are related to the balance or the lack of movement commonly found in these individuals. Since the multidimensionality does not preclude the use of an instrument as a clinically valid tool, the two instruments appear to be useful to evaluate the balance in individuals with chronic hemiparesis.