Balancing resources and risk: Selecting home care clients in Florida’s cares program

Selecting home care patients who would otherwise go into a nursing home always involves error: serving too many or too few. To clarify the choices program and case managers must make, we propose a risk-based alternative to current selection methods that involves scientifically-derived variable weighting and conscious choice of cut-off score for bestowing home care eligibility. We illustrate our proposal with data from Florida’s Comprehensive Assessment and Review of Long-term Care Services (CARES) program. Using logistic regression we identify characteristics that distinguish clients recommended for nursing home placement from those referred to the community and use these results to estimate the risk of nursing home recommendation for each client. An approach to using these risk scores to determine eligibility is demonstrated along with assessment of the impact of alternative risk score cut-offs on denying care to as many as half or as few as 5% of clients served.