Background: In rebalancing from nursing homes (NHs), states are increasing access of NH-certified dually eligible (Medicare/Medicaid) patients to community waiver programs and Programs of All-Inclusive Care for the Elderly (PACE). Prior evaluations suggest Medicaid's PACE capitation exceeds its spending for comparable admissions in alternative care, although the latter may be underestimated. We test whether Medicaid payments to PACE are lower than predicted fee-for-service outlays in a long-term care admission cohort.
Methods: Using grade-of-membership methods, we model health deficits for dual eligibles aged 55 or more entering waiver, PACE, and NH in South Carolina (n = 3,988). Clinical types, membership vectors, and program type prevalences are estimated. We calculate a blend, fitting PACE between fee-for-service cohorts, whose postadmission 1-year utilization was converted to attrition-adjusted outlays. PACE's capitation is compared with blend-based expenditure predictions.
Results: Four clinical types describe population health deficits/service needs. The waiver cohort is most represented in the least impaired type (1: 47.1%), NH entrants in the most disabled (4: 38.5%). Most prevalent in PACE was a dementia type, 3 (32.7%). PACE's blend was waiver: 0.5602 (95% CI: 0.5472, 0.5732) and NH: 0.4398 (0.4268, 0.4528). Average Medicaid attrition-adjusted 1-year payments for waiver and NH were $4,177 and $77,945. The mean predicted cost for PACE patients in alternative long-term care was $36,620 ($35,662 and $37,580). PACE's Medicaid capitation was $27,648-28% below the lower limit of predicted fee-for-service payments.
Conclusions: PACE's capitation was well under outlays for equivalent patients in alternative care-a substantial savings for Medicaid. Our methods provide a rate-setting element for PACE and other managed long-term care.