Background: Mobility recovery after stroke is highly variable, yet is typically described using average patterns that obscure meaningful differences between individuals. Identifying distinct recovery trajectories may improve prognostication and guide rehabilitation strategies. Methods: We conducted a retrospective cohort study of adults admitted for stroke to a large health system between 2016 and 2024. Mobility was assessed using Activity Measure for Post-Acute Care (AM-PAC) Basic Mobility, which was collected during routine clinical care. Growth mixture modeling was used to identify subgroups with distinct mobility recovery trajectories during the first 180 days after stroke. Subgroups were then characterized with baseline personal and clinical characteristics. Results: Seven hundred and fifty individuals contributed 3,389 mobility observations (median 4 per person). A five-class solution was selected based on model fit and classification quality. Distinct trajectories were identified: low stable (n=127), low rapidly improving (n=29), mid declining (n=169), mid improving (n=365), and high stable (n=60). Subgroups differed in both baseline mobility and patterns of change over time, with some demonstrating improvement, others remaining stable, and one declining.
Background: Mobility recovery after stroke is highly variable, yet is typically described using average patterns that obscure meaningful differences between individuals. Identifying distinct recovery trajectories may improve prognostication and guide rehabilitation strategies. Methods: We conducted a retrospective cohort study of adults admitted for stroke to a large health system between 2016 and 2024. Mobility was assessed using Activity Measure for Post-Acute Care (AM-PAC) Basic Mobility, which was collected during routine clinical care. Growth mixture modeling was used to identify subgroups with distinct mobility recovery trajectories during the first 180 days after stroke. Subgroups were then characterized with baseline personal and clinical characteristics. Results: Seven hundred and fifty individuals contributed 3,389 mobility observations (median 4 per person). A five-class solution was selected based on model fit and classification quality. Distinct trajectories were identified: low stable (n=127), low rapidly improving (n=29), mid declining (n=169), mid improving (n=365), and high stable (n=60). Subgroups differed in both baseline mobility and patterns of change over time, with some demonstrating improvement, others remaining stable, and one declining. Individuals in improving subgroups were generally younger, more likely to be independent before stroke, received physical therapy on a greater proportion of hospital days, and were more frequently discharged to inpatient rehabilitation. In contrast, those in low or declining trajectories had lower baseline function, longer hospital stays, and were more likely to be discharged to skilled nursing facilities. Conclusions: The distinct mobility recovery trajectories identified in this work reflect the heterogeneity present in routine clinical practice. Subgroups differed in both recovery patterns and characteristics. Early identification of trajectory membership may improve prognostication and inform more targeted rehabilitation strategies.