The Arizona Long-Term Care System is the first capitated, long-term care Medicaid program in the nation to operate statewide. It promotes an extensive home and community-based services program intended to lower long-term care costs by substituting home care for institutional care. Because the program is statewide, finding a suitable control group to evaluate it was a serious problem. A substitute strategy was chosen that compares actual costs incurred to an estimate of what costs would have been in the absence of home and community-based (HCB) services.
To estimate the likelihood of institutionalizing clients in the absence of HCB services, coefficients for institutionalization risk factors were estimated in a logistic regression model developed using national data. These were applied to characteristics of Arizona clients. The model assigned approximately 75 percent of the program’s clients to a category with traits that were determined to resemble nursing home residents’ traits. A similar methodology was used to estimate lengths of nursing home stays. Lengths of stay by the program’s nursing home patients were regressed on their characteristics using an event history analysis model. Coefficients for these characteristics from the regression analysis were then applied to HCB services clients to estimate how long their nursing home stays would have lasted, had they been institutionalized. These estimated nursing home stays were generally shorter than these same patients’ observed home and community stays.
Risk of institutionalization was then multiplied by estimated length of stay and by monthly nursing home costs to estimate what costs would have been without the HCB services option. The expected costs were compared to actual costs to judge cost savings. Home and community-based services appeared to save substantial amounts on costs of nursing home care. Estimates of savings were very robust and did not appear to be declining as the program matured. Savings probably came from several sources: the assessment teams that judged client eligibility were employed by a state agency and thus were independent from the program contractors; clients were required to be in need of at least a three-month nursing home stay; a cap was placed on the number of HCB services clients contractors were allowed to serve each month; the capitated payment methodology forced managed care contractors to hold down average HCB services costs or lose money; and the HCB services and nursing home costs were blended in the capitated rate, so that plans that failed to place clients in HCB services would lose money by using more nursing home days than their monthly capitated rate allowed.