Predicting Home and Community Walking Activity Poststroke.

BACKGROUND AND PURPOSE Walking ability poststroke is commonly assessed using gait speed categories developed by Perry et al. The purpose of this study was to reexamine factors that predict home and community ambulators determined from real-world walking activity data using activity monitors. METHODS Secondary analyses of real-world walking activity from 2 stroke trials. Home (100-2499 steps/d), most limited community (2500-4499 steps/d), least limited community (5000-74 999), and full community (≥7500 steps/d) walking categories were developed based on normative data. Independent variables to predict walking categories were comfortable and fast gait speed, 6-minute walk test, Berg Balance Scale, Fugl Meyer, and Stroke Impact Scale. Data were analyzed using multivariate analyses to identify significant variables associated with walking categories, bootstrap method to select the most stable model and receiver-operating characteristic to identify cutoff values. RESULTS Data from 441 individuals poststroke were analyzed. The 6-minute walk test, Fugl Meyer, and Berg Balance Scale combined were the strongest predictors of home versus community and limited versus unlimited community ambulators. The 6-minute walk test was the strongest individual variable in predicting home versus community (receiver-operating characteristic area under curve=0.82) and limited versus full community ambulators (receiver-operating characteristic area under curve=0.76). A comfortable gait speed of 0.49 m/s discriminated between home and community and a comfortable gait speed of 0.93 m/s discriminated between limited community and full community ambulators. CONCLUSIONS The 6-minute walk test was better able to discriminate among home, limited community, and full community ambulators than comfortable gait speed. Gait speed values commonly used to distinguish between home and community walkers may overestimate walking activity.