Heterogeneity in the Effectiveness of Rehabilitation Dose on Activities of Daily Living in Post-Acute Stroke: Target Trial Emulation

OBJECTIVES: To identify patient factors associated with heterogeneity in functional benefits from higher-dose in-hospital rehabilitation therapy intensity during the post-acute phase after ischemic stroke. DESIGN: Nationwide observational cohort study using routine medical claims data. The "intervention" of interest was higher-dose in-hospital rehabilitation therapy, defined as an average of ≥2 h/d during the 30 days after transfer to a rehabilitation ward (vs < 2 h/d). Treatment effects were estimated using G-computation [conditional average treatment effect (CATE) and average treatment effect (ATE)], and heterogeneity was examined using hierarchical clustering and a doubly robust augmented inverse probability weighting approach. SETTING AND PARTICIPANTS: Adults aged ≥18 years with ischemic stroke admitted to acute-phase wards and transferred to rehabilitation wards within 30 days, with pre-admission Modified Rankin Scale (mRS) 0-2 and walking dependence at admission (N = 6259). METHODS: Exposure: Higher-dose rehabilitation therapy (≥2 h/d) during the 30 days posttransfer. PRIMARY OUTCOME: Favorable functional outcome, defined as mRS 0-2 at 60 days after transfer. G-computation was used to estimate the ATE and individual CATEs; hierarchical clustering and augmented inverse probability weighting were used to evaluate and corroborate between-patient heterogeneity in treatment response. RESULTS: Higher-dose rehabilitation therapy improved functional status at 60 days after transfer (ATE, 15.3%; 95% CI, 12.6%-18.0%). Heterogeneity analyses indicated larger benefits among subgroups with higher levels of activities of daily living (ADLs) at transfer, particularly in self-care (CATE, 23.7%; 95% CI, 19.6%-27.6%), mobility (CATE, 23.8%; 95% CI, 18.9%-28.7%), and excretion control (CATE, 22.8%; 95% CI, 18.9%-27.2%). CONCLUSIONS AND IMPLICATIONS: Patient characteristics contributing to heterogeneity in response to higher-dose post-acute rehabilitation included ADL domains, sex, age, pre-admission mRS, and acute hospital stay.In practice, these findings support targeted intensification of rehabilitation (≥2 h/d) for patients with relatively preserved ADL at transfer, where gains appear greatest. From a policy perspective, stratified approaches may help optimize resource allocation in rehabilitation wards.