Evaluating Hospice Effect on Hospitalizations of Nursing Home Residents

Study Objectives

US nursing home residents are frequently hospitalized at the end-of-life. Medicare pays about $1.6 billion annually for these end-of-life hospitalizations. Hospice enrollment has been shown to reduce nursing home residents' risk of hospitalization. We examined whether residing in facilities with higher hospice concentration can 1) reduce the risk of hospitalization in residents not receiving hospice (a spill-over effect); and 2) further decrease hospice residents' risk of hospitalization (an incremental effect).

Data Sources

Medicare Denominator File, Inpatient Claims and Hospice Claims were linked with the Minimum Data Set. Provider of Services File and Area Resource File were also used.

Study Design

We conducted cross-sectional analysis of secondary data. We used a national sample of 747,641 long-term-care residents (505,851 non-hospice and 241,790 hospice) in 14,030 nursing homes who died in 2005-2007.

Method

We applied instrumental variable models to address the potential endogeneity between hospice concentration and hospitalization. The distance between each nursing home and the closest hospice was the instrumental variable.

Principal Findings

In the last 30 days of life, 37.6% of non-hospice and 23.2% of hospice residents were hospitalized. For every 10% increase in hospice concentration, non-hospice residents' risk of hospitalization decreased by 5.1%; hospice residents' risk of hospitalization decreased by 4.8%.

Conclusions

Increasing hospice concentration at the facility level reduces hospitalizations of non-hospice residents (spill-over effect) and further diminishes hospice residents' risk of hospitalization (incremental effect). Future evaluation of hospice effect on government expenditures should take into account these additional effects of hospice use.