Disenrollment From an Acute/Long-Term Managed Care Program (PACE)

Abstract

The Program of All-Inclusive Care for the Elderly (PACE) is an acute/long-term managed care plan designed to care for the most frail and vulnerable Medicare beneficiaries. To our knowledge, this is a first study to examine patterns and predictors of disenrollment from PACE. PACE, with its comprehensive delivery system, dual capitation, and a focus on the most vulnerable population, may be expected to achieve low rates of exit and little selective dissenrollment. This study examines whether these goals have been accomplished. The study includes 30 PACE programs and 14,657 individuals enrolled in them. Individual-level records, obtained from an administrative database, contain information on sociodemographics, caregiver support, health status and disability, medical history, service utilization, and disenrollment. Program-level variables also were included. Cox proportional hazard models, with time-varying and time-invariant covariates, were employed to predict time to disenrollment. Our findings show a low level of disenrollment. We find no increase in disenrollment risk by age, functional or cognitive impairment, Medicaid eligibility, or diagnoses. Certain characteristics (eg, nursing treatments) appear to reduce the disenrollment hazard, whereas others (eg, hospital admissions, private pay status) significantly increase it. The risk of disenrollment also increases with longer nursing home stays, until 80-90 days, whereupon it begins to decline. Both enrollee and program-level attributes predict program disenrollment. Programmatic and quality of care improvements may be needed to further minimize disenrollment, particularly in programs experiencing rates that are substantially greater than the average.