Longitudinal robustness of variables predicting independent gait following severe middle cerebral artery stroke: a prospective cohort study

Objective: To determine within the first 10 weeks post onset the most robust variables in the prediction of recovery of independent gait at six months post stroke. Design: A prospective cohort study. Subjects: One hundred and one first ever ischaemic middle cerebral artery stroke patients. None of these patients were able to walk at onset and all suffered from a marked hemiplegia. Setting: Twenty-four determinants, possibly related to recovery of gait at six months, were measured within 14 days following stroke onset. Based on Functional Ambulation Categories (FAC) independent gait was classified into present (FAC >= 4) or absent (FAC < 4). Bivariate logistic regression analysis was used to select determinants. Only significant determinants during the entire 10-week period were used for further weekly multivariate logistic prediction modelling of independent gait at six months post stroke. Results: After six months post onset 62% (N = 63) regained independent gait. Age, Barthel Index, Trunk Control Test, Motricity Index of arm and leg, Brunnstrom Fugl-Meyer stage of leg motor recovery, and type of intervention were significant determinants in bivariate analysis, but age of patient and Barthel Index were the most robust determinants in the final prediction model. Weekly re-evaluation produced sensitivity values between 89% and 96% and specificity values between 53% and 62%. Conclusion: In initially non-ambulatory stroke patients age and Barthel Index were the most robust variables during the first 10-week poststroke period in the prediction of independent walking at six months. However, prediction of non-ambulation at six months proved to be less accurate.