Predicting place of death in the Program of All-Inclusive Care for the Elderly (PACE): Participant versus program characteristics

OBJECTIVES: To assess the relationship between participant-specific and program-specific characteristics and the place of death in the Program of All-Inclusive Care for the Elderly (PACE). DESIGN: A retrospective, population-based, cross-sectional study. SETTING: PACE is a community-based managed care model serving a frail, nursing home-eligible population of Medicare beneficiaries. PARTICIPANTS: Two thousand two hundred sixty-three PACE decedents from 12 sites that were fully capitated for Medicare and Medicaid by mid-1997. MEASUREMENTS: Participant-specific characteristics include the range of demographic, functional, and cognitive status; skilled care needs; medical conditions; and advance care directives measures. The PACE site-specific indicator was also included to assess the effect of program sites on place of death. Bivariate, logistic, and multinomial logit models were employed. RESULTS: The probability of death at home is twice as great (45%) for PACE program participants as for the general population of older Americans. Twenty-one percent of PACE participants die in hospitals, compared with 53% of Medicare beneficiaries. Participants who are older and live in the community have a significantly greater (P < .05) probability of dying at home than does an average PACE enrollee. Those with a do-not-resuscitate order are 7.4% less likely to die in the hospital than are those without. PACE participants with a live-in informal caregiver are 10.3% (P < .05) less likely to die at home than those without a caregiver. When all participant characteristics were accounted for, there remained significant variation in the place of death by PACE plan. We found the least amount of site-specific variation in deaths occurring at home and the most variation in deaths occurring in hospitals. CONCLUSION: Nationally, the low rate of home death has prompted some to conclude that "it is unlikely that we will, as a society, decide to take our dying older people back into our homes." The PACE experience suggests that it is possible to modify this trend. The results indicate that patient and program characteristics are both important in predicting the place of death. The variation across PACE sites suggests that an opportunity for narrowing these gaps, and reducing the proportion of hospital deaths, exists. The PACE end-of-life "practice style" could potentially be applied to other populations and may have important public policy implications.