• Cari Levy, Evelyn Hutt, and Lauren Pointer.
• University of Colorado Denver, Division of Health Care Policy and Research, Denver Veterans Affairs Medical Center, 1055 Clermont, Mailstop 151, Denver, CO 80220, USA. cari.levy@va.gov
• J Am Med Dir Assoc. 2012 Mar 1; 13 (3): 199-201.

ObjectivesTo determine predictors of dying in VA nursing homes, community living centers (CLCs), compared with dying in a hospital.DesignRetrospective cohort study.SettingVA CLCs.ParticipantsIncluded were 7408 CLC decedents from FY2005 to FY2007.MeasurementsOutcomeSite of death obtained from VA Vital Statistics files. Predictors of Death Site: VA-MDS variables defining patient demographics, functional status, cognitive status, major diagnostic categories, and care planning documentation.AnalysisLogistic regression was used to estimate the odds ratio of death in the CLC relative to the hospital for patient and facility characteristics.ResultsAmong decedents, 87% died in the CLC and 13% in a hospital. More than half of all decedents were neither enrolled in hospice nor designated as having end-stage disease. The strongest predictor of site of death in a CLC relative to a hospital was being enrolled in hospice (OR = 20.94; 95% CI: 12.38, 35.44). A designation of end-stage disease increased the odds of death in a CLC by 3.9 times (95% CI: 2.78, 5.47) compared with death in a hospital. Advance directive rates in CLCs were high (73.4%); having any advance directive increased the odds of death in a CLC by 1.57 times (95% CI: 1.35, 1.82).ConclusionRecognition of end-stage disease and documentation of advance directives are powerful determinants of site of death for CLC residents. Receipt of hospice care in a CLC is a strong predictor of site of death in a CLC even in the absence of collaboration with community-based hospice and financial incentives to avoid hospitalization.Published by Elsevier Inc.

20 keywords...

hide…