Objective: To identify individual-, hospital-, and community-level predictors of home health and outpatient physical and/or occupational therapy (PT and/or OT) following stroke. Design:The COMPASS study was a pragmatic trial examining the effectiveness of a transitional care model for stroke.We conducted an observational, secondary analysis of COMPASS data linked to insurance claims.Setting: 40 acute care hospitals in North Carolina (NC).Participants: 6,754 patients with stroke or transient ischemic attack discharged directly home from the hospital and who had linked NC insurance claims data. Interventions: Not applicable.Main Outcome Measures: 30-day use of PT and/or OT in the home or outpatient setting and time to first visit, number of visits, and receipt of both PT and OT in 90 days.Multivariable, generalized estimating equations examined individual-, hospital-, and community level predictors of therapy use, controlling for clinical measures.Results: 28 and 35 percent of the sample used PT and/or OT in the home or outpatient setting within 30 and 90 days, respectively.Among those who received therapy within 90 days, the mean (SD) number of visits was 9.6 (9.4) and 48 percent received both PT and OT.Individuallevel factors associated with 30-day use and/or a shorter time to first visit included: female sex, living in a metropolitan area, living in a more educated community, having dual Medicare/Medicaid or Medicare Advantage insurance (relative to traditional Medicare), having access to an informal caregiver, and having a primary care provider.Having Medicaid insurance was associated with lower therapy use and a longer time to first visit.Few hospital-level characteristics were predictive of therapy use or time to first visit.Community-level measures of therapist supply were associated with therapy use and shorter time to first visit.Conclusions: These findings highlight opportunities for targeted interventions to improve access to community-based rehabilitation following stroke.
Freburger et al. (Wed,) studied this question.