2017 HSR&D/QUERI National Conference

4026 — Stroke Readmission and Mortality: A Comparison between VA Community Living Centers and Community Nursing Homes

Lead/Presenter: Huanguang Jia, COIN - North Florida/South Georgia and Tampa
All Authors: Jia H (COIN-North Florida/South Georgia and Tampa) Pei Q (COIN-North Florida/South Georgia and Tampa) Hale-Gallardo JL (COIN-North Florida/South Georgia and Tampa) Bates BE (Aleda E. Lutz VA Medical Center) Wu SS (COIN-North Florida/South Georgia and Tampa) Vogel WB (COIN-North Florida/South Georgia and Tampa) Wang X (COIN-North Florida/South Georgia and Tampa) Cowper Ripley DC (COIN-North Florida/South Georgia and Tampa)

Objectives:
Hospital readmission and mortality are commonly used for quality improvement and cost controls. VA Community Living Centers (CLC) and Community Nursing Homes (CNH) are two major sources of post-acute stroke care for Veterans; however, limited information is available about hospital readmissions and mortality rates for Veteran residents at these locations. This study assessed differences in stroke readmissions and mortality between CLC Veterans and CNH Veterans nationally and by region.

Methods:
This retrospective study included all Veterans diagnosed with stroke (N = 18,272) and residing in CLCs (n = 12,660) or VA-contracted CNHs (n = 5,612) between 2006 and 2009. CLC and CNH Minimum Datasets (MDS2.0) were linked with Veteran vital status data and VA and Medicare inpatient data. Stroke readmission and mortality referred to the event incidence occurring within the 12 months post nursing home admission date. The 5 MyVA regions were used to group the CLCs and CNHs. Poisson regression for readmission and logistic regression for mortality were applied adjusting for propensity score and rehabilitation care utilization.

Results:
The 12-month readmission rate was 27% for CNH and 15% for CLC Veterans with 90% of CNH Veterans' readmissions occurring under Medicare vs. 58% for CLC Veterans. Poisson regression results showed that CLC (vs. CNH) Veterans were 50% less likely to be readmitted for stroke (Coefficient±SD: -0.65±0.08, p < 0.001). The 12-month crude mortality rate was 25% for CLC and 23% for CNH Veterans. Logistic regression results demonstrated that CLC (vs. CNH) Veterans were more likely to die (AOR = 1.13, CI = 1.12, 1.14, P = 0.009). Significant regional variations in the readmission and mortality outcomes were observed in our regression models: Pacific CLC Veterans and Continental CNH Veterans had more rehospitalizations; Continental CLC Veterans and Pacific CNH Veterans had higher chances of dying than their counterparts from other regions, respectively.

Implications:
During the 12-month follow-up time period, Veterans with stroke at VA-contracted CNHs were twice as likely as being rehospitalized for stroke but less likely to die than their CLC counterparts, with significant regional variations.

Impacts:
Our findings about Veterans' post-stroke readmissions and mortality at CLCs and CNHs and insights regarding geographic variations of these outcomes would help VA policy makers and clinicians in their evidence-based policy and clinical decision making.