Abstract
To better understand determinants and potential disparities in end of life, we model decedents' place of death with explanatory variables describing familial, social, and economic resources. A retrospective cohort of 204,041 decedents and their family members are drawn from the Utah Population Database family caregiving dataset. Using multinomial regression, we model place of death, categorized as at home, in a hospital, in another location, or unknown. The model includes family relationship variables, sex, race and ethnicity, and a socioeconomic status score, with control variables for age at death and death year. We identified the effect of a family network of multiple caregivers, with 3+ daughters decreasing odds of a hospital death by 17 percent (OR: 0.83 [0.79, 0.87], p < 0.001). Place of death also varies significantly by race and ethnicity, with most nonwhite groups more likely to die in a hospital. These determinants may contribute to disparities in end of life.