Dynamic task–based rehabilitation and time to loss of toilet ambulation in a palliative care unit: a retrospective cohort study

PURPOSE: To determine whether dynamic-task rehabilitation, compared with bedside-only or no rehabilitation, is associated with longer time to loss of toilet ambulation among patients with advanced cancer in a palliative care unit (PCU). METHODS: This retrospective study included patients with cancer admitted to a PCU who died during hospitalization (January 2020-March 2024). Eligible patients had walked to an in-room fixed toilet and stayed ≥ 14 days. Patients were classified as dynamic-task rehabilitation (Ex; ≥ 2 sessions), bedside rehabilitation (BS; ≤ 1 dynamic-task session), or no rehabilitation (NI). The primary outcome was time from PCU admission to loss of toilet ambulation. Kaplan-Meier and multivariable Cox analyses were used; Fine-Gray analyses treated death as a competing event. RESULTS: A total of 205 patients were analyzed (Ex, n = 78; BS, n = 65; NI, n = 62). Median time to loss of toilet ambulation was 27 days in Ex, 20 days in BS, and 14 days in NI. In adjusted Cox models (NI reference), Ex was associated with a lower hazard of ambulation loss (HR, 0.462; 95% CI, 0.320-0.667), whereas BS was not (HR, 0.828; 95% CI, 0.559-1.226). Results were consistent in competing-risk analyses (Ex vs NI: sHR, 0.461; 95% CI, 0.328-0.647). CONCLUSION: Patients able to receive dynamic-task rehabilitation had a longer time to loss of toilet ambulation during PCU hospitalization. This association may reflect rehabilitation content, patient characteristics, and the clinical feasibility of dynamic-task rehabilitation. Prospective studies are needed to clarify causality.