Objectives: Recent public attention has focused on quality of care for the dying. Where one dies is an important individual and public health concern.
Materials and methods: The 1993 National Mortality Followback Survey (NMFS) was used to estimate the proportion of deaths occurring at home, in a hospital, or in a nursing home. Sociodemographic variables, underlying cause of death, geographic region, hospice use, social support, health insurance, patients' physical limitations, and physical decline were considered as possible predictors of site of death. The relationship between these predictors and site death with multinomial logistic regression methods was analyzed.
Results: Nearly 60% of deaths occurred in hospitals, and approximately 20% of deaths took place at home or in nursing homes. Decedents, who were black, less educated, and enrolled in an HMO were more likely to die in the hospital. After adjustment, functional decline in the last 5 months of life was an important predictor of dying at home (for loss of 3 or more ADLs [OR, 1.57; 95% CI, 1.11-2.21]). Having functional limitations 1 year before death, and experiencing functional decline in the last 5 months of life were both associated with dying in a nursing home.
Conclusions: Rapid physical decline during the last 5 months was associated with dying at home or in a nursing home, whereas earlier functional loss was associated with dying in a nursing home.