Characteristics of Acute Stroke Patients Readmitted to Inpatient Rehabilitation Facilities: A Cohort Study.

Background: Reducing acute care readmissions from inpatient rehabilitation facilities (IRFs) is a healthcare reform goal. Stroke patients have higher acute readmission rates and persistent impairments, warranting second IRF hospitalization consideration.

Objective: To provide evidence-based information to justify IRF readmission for patients with post-stroke impairments.

Main Outcome Measure: Variables that increase the likelihood of a second IRF hospitalization.

Design: Retrospective cohort study.

Setting: Seven-center rehabilitation network.

Participants: Stroke patients, readmitted to acute care, who returned or did not return to an in-network IRF between 1 October 2014-31 December 2017(n = 380).

Interventions: Univariable analyses (Returned/Did Not Return to IRF) described demographics, stroke type and risk factors. Between group differences in readmission causes, motor impairments and functional independence measure (FIM) scores were examined. Return to IRF logistic regression model included variables with P < .1. Odds ratio and 95% CI were calculated; Relative risk was calculated for categorical variables. P < .05 equaled statistical significance.

Results: One hundred ninety-two stroke patients returned to IRF, 188 did not. Returned to IRF patients were younger (60.6 vs. 66 years; P < .001), sustained hemorrhagic strokes (22.4 vs. 14.2%; P = .01), had lower cardiac disease prevalence (41.7 vs. 55.3%; P = .008) or non-Medicare insurance (59.9 vs. 39.4%; P < .001). Did Not Return to IRF patients had higher admission and discharge motor and total FIM scores. Per point decrease in discharge FIM, second IRF hospitalization odds increased 4% (OR 1.04; 95% CI 1.01-1.07; P = .02). Hemorrhagic stroke patients had 33% increased odds or a 15% higher relative risk of second IRF hospitalization than patients with ischemic stroke [OR 1.33; 95% CI 1.21-1.47; RR 1.15; 95% CI 1.1-1.2; P < .001]. Non-Medicare insurance was associated with 39% increased odds or a 20% higher relative risk of second IRF hospitalization than Medicare [OR 1.39; 95% CI 1.01-1.92; RR 1.2, 95% CI 1.006-1.404; P = .04).

Conclusions: Hemorrhagic stroke, non-Medicare insurance or lower discharge FIM score during the first IRF hospitalization predict a second IRF stay. Further work is needed to establish the validity of within IRF stay readmission measures.