LAUSR

OBJECTIVES Nursing homes (NHs) collaboration with hospices appears to improve end-of-life (EOL) care among dying NH residents. However, the potential benefits of NH-hospice collaboration may vary with the patterns of this collaboration. This study examines the relationship between the attributes of NH-hospice collaboration, especially the exclusivity of NH-hospice collaboration (ie, the number of hospice providers in a NH), and EOL hospitalizations among dying NH residents. DESIGN This national retrospective cohort study linked 2000-2009 NH assessments (ie, the Minimum Data Set 2.0) and Medicare data. A linear probability model with facility fixed-effects was estimated to examine the relationship between EOL hospitalization and the attributes of NH-hospice collaborations, adjusting for individual and facility characteristics. We also performed a set of sensitivity analyses, including stratified analyses by volume of hospice services in a NH and stratified analyses by rural vs urban NH locations. SETTINGS All Medicare and/or Medicaid certified US NHs with at least 8 years of data and at least 30 beds. PARTICIPANTS NH decedents resided in Medicare and/or Medicaid certified NHs in the US between 2000 and 2009. We restricted the analyses to those continuously enrolled in Medicare fee-for-service in the last 6 months of life and those who were in NHs for the last 30 days of life. In total, we identified 2,954,276 NH decedents over the study period. MEASUREMENTS The outcome variable was measured as dichotomous, indicating whether a dying NH resident was hospitalized in the last 30 days of life. The attributes of NH-hospice collaboration were measured by the volume of hospice services (defined as the ratio of number of hospice days to the total NH days per NH per calendar year) and the number of hospice providers in a NH (defined as the number of unique hospice providers in a NH per year). We categorized NHs into groups based on the number of hospice providers (1, 2 or 3, and ≥4) in the NH, and conducted sensitivity analysis using a different categorization (1, 2, and 3+ hospice providers). RESULTS The pattern of NH-hospice collaboration changed significantly over years; the average number of hospices in a NH increased from 1.4 in 2000 to 3.2 in 2009. The volume of NH-hospice collaboration also increased substantially. The multivariate regression analyses indicated that having more hospice providers in the NH was not associated with lower risks of EOL hospitalizations. After accounting for individual and facility characteristics, increasing hospice providers from 1 to at least 4 was associated with an overall 1 percentage point increase in the likelihood of EOL hospitalizations among dying residents (P < .01), and such relationship remained in NHs with moderate or high volume NHs in the stratified analyses. Stratified analysis by rural vs urban NHs suggested that the relationship between the number of hospice providers and EOL hospitalizations was mainly in urban NHs. CONCLUSIONS More hospice providers in the NH was not associated with lower EOL hospitalizations, especially among NHs with relatively high volume of hospice services.