Impact of nursing home characteristics on place of death among residents receiving post-acute care

Post-acute (PA) residents typically stay in nursing homes (NHs) for a short period of time and are not expected to die there. Nevertheless, they account for 38% of NH decedents. While many PA residents die in NHs, a significant number die in hospitals, with wide variation across facilities and states in place-of-death (POD). We examined the relationship between NH organizational characteristics and POD among PA residents.

Data: Minimum Data Set (MDS); Medicare beneficiary file; inpatient claims; Online Survey Certification and Reporting System for all US Medicare certified NHs, in CY2005-2007. Analytical sample: 337,535 decedents; 12,543 NHs.

We estimated logistic regression model with state fixed effects. POD is the dependent variable, defined as deaths occurring in-hospitals or NHs. POD risk factors, derived from the MDS, were included as independent variables. Facility characteristics, e.g. staffing, size, ownership, case mix, were included.

68% of PA decedents die in NHs. Controlling for individual risk-factors, several facility characteristics were significant predictors of POD. Odds of death in NH were significantly higher for residents of hospital-affiliated NHs (OR=1.40, p<0.001), those with higher RN staffing (OR=1.14, p<0.001), and government-owned (OR=1.08, p<0.001). However, PA residents in rural facilities had lower odds of in-NH death (OR=0.93, p<0.001), as did residents in facilities with higher average case mix (OR=0.55, p<0.001).

The dependence of POD on facility characteristics (controlling for residents' risk factors) suggests that practice styles differ systematically across NHs. Further research is needed to determine if such variations offer opportunities for improvements in both quality of care and cost-efficiencies.