Brazilian version of the Functional Gait Assessment: translation, reliability, and validity for use on stroke patients

Objective. Translate and adapt the Functional Gait Assessment (FGA) to Brazilian Portuguese as a measure of balance during gait in stroke patients and determine its validity and reliability based on evidence of its measurement properties. Method. A convenience sample of 45 stroke survivors (average of seven months since diagnosis) was included (mean age: 55 years; 51% women). The instrument under consideration was translated and back-translated. The performance of the FGA was assessed by two raters to determine intrarater and interrater reliability. Concurrent and discriminant validity were investigated using the Berg Balance Scale (BBS) as well as normal and fast walking speed. Results. No difficulties with the translation were found during the application of the correlation coefficient=0.93) and interrater reliability (intraclass correlation coefficient=0.90) were almost perfect for the total scores. The reliability of single items was also strong, ranging from 0.74 to 0.95. Concurrent validity with other measures of gait and balance was moderate to substantial. The FGA was correlated (p<0.001) with the BBS (0.71), normal walking speed (0.66), and fast walking speed (0.70). Conclusion. The Brazilian version of the FGA is a reliable, valid instrument for assessing functional gait performance in stroke survivors. y la fiabilidad entre evaluadores (coeficiente de correlación intraclase=0,90) fueron casi perfectos para la puntuación total. La fiabilidad de los artículos individuales también fue fuerte, con un rango de 0,74 a 0,95. La validez concurrente con otras medidas de marcha y equilibrio fue de moderada a sustancial. El AGP mostró una correlación (p<0.001) con el BBS (0.71), la velocidad de caminata normal (0.66) y la velocidad de caminata rápida (0.70). Conclusión. La versión brasileña de la FGA es un instrumento válido y confiable para evaluar el rendimiento funcional de la marcha en los sobrevivientes de un accidente cerebrovascular.

correlation coefficient=0.93) and interrater reliability (intraclass correlation coefficient=0.90) were almost perfect for the total scores. The reliability of single items was also strong, ranging from 0.74 to 0.95. Concurrent validity with other measures of gait and balance was moderate to substantial. The FGA was correlated (p<0.001) with the BBS (0.71), normal walking speed (0.66), and fast walking speed (0.70). Conclusion. The Brazilian version of the FGA is a reliable, valid instrument for assessing functional gait performance in stroke survivors. Keywords. Stroke; Balance; Gait; Validation Resumen Objetivo. Traduzca y adapte la Functional Gait Assessment (FGA) al portugués brasileño como una medida de equilibrio durante la marcha en pacientes con accidente cerebrovascular y determine su validez y fiabilidad en función de la evidencia de sus propiedades de medición. Método. Se incluyó una muestra de conveniencia de 45 sobrevivientes de accidente cerebrovascular (media de siete meses desde el diagnóstico) (edad media: 55 años; 51% mujeres). El instrumento en cuestión fue traducido y retrotraducido. El desempeño de la FGA fue evaluado por dos evaluadores para determinar la confiabilidad intra e interevaluador. Se investigó la validez concurrente y discriminante utilizando la Escala de equilibrio de Berg (BBS), así como la velocidad de caminata normal y rápida. Resultados No hubo dificultades con la traducción durante la aplicación de las pruebas. Por lo tanto, no fueron necesarios cambios estructurales o conceptuales en la versión traducida para obtener equivalencia cultural. El intra-evaluador (coeficiente de correlación intraclase=0,93) y la fiabilidad entre evaluadores (coeficiente de correlación intraclase=0,90) fueron casi perfectos para la puntuación total. La fiabilidad de los artículos individuales también fue fuerte, con un rango de 0,74 a 0,95. La validez concurrente con otras medidas de marcha y equilibrio fue de moderada a sustancial. El AGP mostró una correlación (p<0.001) con el BBS (0.71), la velocidad de caminata normal (0.66) y la velocidad de caminata rápida (0.70). Conclusión. La versión brasileña de la FGA es un instrumento válido y confiable para evaluar el rendimiento funcional de la marcha en los sobrevivientes de un accidente cerebrovascular. as a modification of the DGI to improve reliability and diminish the ambiguity of several items that caused scoring problems for evaluators 9,10 . The FGA items address several aspects of gait and are used to measure balance and gait disorders. It is also easily performed in the clinical setting with a single evaluator 9 . The 10-item FGA has seven of the eight items on the DGI and three additional items: "gait with a narrow base of support", "ambulating backward" and "gait with eyes closed". Each of the 10 items is scored on an ordinal scale with four levels (0 to 3 points). The maximum score is 30 points. Although the scale involves important functional aspects of gait, such as quality of movement, deviation from the intended walking path, the need for a gait-assistance device and the time required to perform the tasks, there is no consensus on a cutoff point that defines individuals with a high risk of falls 10 .
In a systematic review performed by Weber et al. 10  Low-cost assessment tools that are easy to administer have been developed to enable the quantitative and/or qualitative description of gait in the clinical setting and involve functional tasks during gait that require dynamic balance 5,11,12 . For these tools to have clinical usefulness, however, they must be valid, reliable, and capable of detecting changes in the gait pattern of an individual. The

Initial translation and evaluation of initial translation (backtranslation)
The researchers first contacted the authors of the FGA to obtain authorization for the translation and validation of physiotherapists, an occupational therapist, and a physician.

Evaluation of cultural equivalence
The Portuguese version of the FGA was administered to 12 stroke survivors. The expression "I don't understand" was added to the instructions. Items that received more than

Correlation instruments
The BBS and walking speed test were administered on the first evaluation day by the second physiotherapist for the determination of correlations with the FGA and the establishment of concurrent validity.
The BBS consists of 14 items involving dynamic and static balance in the sitting and standing positions. Each item is graded from 0 (poorest balance) to 4 (best balance). The maximum score is 56 points, with higher scores indicating better balance. 9 The BBS has been used as a reference tool to establish construct validity in studies involving novel walking balance assessment tools for stroke patients 15 .
Normal and fast walking speed were measured (in meters per second) using the Ten-Meter Walk Test, which was administered by marking a starting and ending point at a distance of 10m. The participants were asked to stand behind the starting point and walk at a comfortable pace (1) and at a fast pace (2) until crossing the end point. Each participant was given two trials with each pace and the average of the two trials was used for analysis. If a participant felt tired during the assessment, he or she could have a five-to-tenminute rest.

Data analysis
Data analysis was performed using the Statistical  15,16 . All significance tests were two-sided and were conducted with a 5% significance level (95% confidence interval).

Participants
The sample was composed of 45 individuals with a diagnosis of stroke. All participants were able to walk independently (with or without a gait-assistance device) and

Translation and cultural adaptation
The Brazilian version of the FGA can be found in the Appendix. No difficulties occurred during the application of the tests because of the translation. Therefore, no structural or conceptual changes to the versions (translation and back-translation) were needed to achieve cultural equivalence. The test was understood well by the multidisciplinary committee (specialist in physical medicine, occupational therapist, and specialized physiotherapists) and the participants. None of the items was considered incomprehensible by more than 15% of the participants.

Reliability
The median scores for the FGA items are presented in Table 2. The most difficult items to perform were walking with eyes closed, during head rotations, gait with narrow base of support, ambulating backward and going up and down steps. The impairment score ranged from 2 (mild impairment) to 3 (normal).
The ICCs for pooled intrarater and interrater reliability for each question and the total FGA score are displayed in to 1.0 (items 6 and 10), demonstrating strong reliability.
Pooled interrater reliability was almost perfect, with an ICC of 0.90 for the total score. Single item values ranged from 0.74 (Items 2 and 5) to 0.95 (Item 10).

Concurrent validity
Concurrent validity was substantial for the BBS and fast walking speed and moderate for normal walking speed ( Table 3)   To ensure a process free of bias, the translation and cross-cultural adaptation of the FGA followed internationally accepted guidelines 19,20 . To obtain the best accuracy and   15 . Tests with normal and fast walking speeds were used to determine the concurrent validity of the FGA measure, which involves the evaluation of functional skills, such as walking normally and as fast as possible.
The mean total FGA score was 20 (range: 0 to 30). The participants received a mean score of 2 (mild impairment) on seven items: gait on level surface, gait with horizontal or vertical head turns, gait with eyes closed, gait with narrow base of support, ambulating backward and going up steps (Table 2). These activities depend not only on balance but also the visual, somatosensory, and vestibular systems, which are often affected in cases of stroke 14 . The FGA items enable the assessment of activities performed by stroke survivors when walking in the community that involve balance and other sensory systems. This instrument can be used for functional gait evaluations in the outpatient setting as well as in hospital wards. It can also be easily integrated into the day-to-day clinical practice of physiotherapists and is easily understood.
One limitation of the present study was the failure to measure walking balance and the determination of the validity of the FGA with regards to predicting falls after a stroke (sensitivity). Future studies are needed to determine the predictive validity and responsiveness to change of the FGA in stroke survivors. Another limitation was the fact that only patients in the subacute and chronic stages of a stroke (within the first three months and six or more months after the stroke event) were included. As the FGA items assume that the patient is ambulatory, only those with at least minimum ability for independent walking with or without a gait-assistance device can be assessed. It is unclear whether the FGA exhibits similar psychometric properties for acute stroke patients with minimal ability for independent walking with or without a gait-assistance device.

CONCLUSION
In this study, the Brazilian version of the Functional Gait Assessment was translated without the need for crosscultural adaptation, achieving almost perfect interrater and intrarater reliability for total score. Moreover, good concurrent validity was found when compared to other balance and gait measurement tools. Therefore, the FGA is a promising, useful tool for clinicians and researchers to evaluate functional gait status in stroke survivors. Further