Hospitalization in the Program of All‐inclusive Care for the Elderly (PACE): Rates, Concomitants, and Predictors

BACKGROUND: The Program of All-Inclusive Care for the Elderly (PACE) replicates the model of comprehensive, community-based geriatric care pioneered by On Lok, that enrolls frail older adults who meet states' criteria for nursing home care, and that uses interdisciplinary teams to assess the participants and to deliver care in appropriate settings. As managed care, PACE receives capitated payment from Medicare and Medicaid. Thus, PACE's fiscal incentives are thought to be aligned with the goals of optimizing health, function, and quality of life through the delivery of effective primary, preventive, restorative, supportive, and palliative care and through the avoidance of inappropriate and expensive hospital and nursing home utilization. OBJECTIVES: To describe short-term hospital utilization, hospital discharge diagnoses, time from enrollment to first hospitalization and its clinical predictors, and hospitalization in relation to mortality among PACE participants. METHODS: Data on short-term hospitalization and participants were recovered from PACE's minimum data set. Bed use was evaluated in annual cross-sections of current participants. Primary hospital discharge diagnoses were available for discharges from September 1, 1993 through March 31, 1997. The time from enrollment to hospitalization was calculated for the participants (n = 5478) who were admitted between January 1, 1990 and March 31, 1997. The characteristics of this inception cohort were used to develop a Cox regression model of hospitalization. All PACE deaths were identified and the place of death was recovered, together with the medical records used in the hospital during PACE enrollment or 6 months before death. RESULTS: Bed-days per 1,000 PACE participants per year were comparable with the general Medicare (fee-for-service) population, at 2,046 (in 1998) versus 2014 (in 1997) despite the greater morbidity and disability for PACE participants, as reflected in their enrollment characteristics and primary hospital discharge diagnoses. The time to hospitalization was 773 days (median); 95% confidence interval, 725, 814, and was predicted by disease, treatment, social and demographic factors. Whereas 8% of PACE deaths occurred in acute hospitals, less than one-third of the decedents spent any time in the hospital in the 6-month interval before death. CONCLUSIONS: Overall, short-term hospital utilization among PACE participants is low in contrast with that for other older and disabled populations. Participant predictors of hospitalization in PACE are generally consistent with other studies in older clinical and community populations. Both utilization and risk vary considerably across PACE sites, independent of participant-level risk factors, hence suggesting that further investigation is required to study PACE's management of acute illness and hospitalization decisions. Critical to maintaining PACE's success is an understanding of the independent impact of the organization and the environment of health care on this management.