We describe two approaches to case-mix for Medicare-covered stays in skilled nursing facilities (SNFs) and compare them in terms of accuracy, complexity and intuitive appeal. Data on resident characteristics and per diem costs (including staff time and ancillary costs) for Medicare-covered SNF stays are used to compare two case-mix systems. A simple index-based system, developed using a linear regression model, is compared to Resource Utilization Groups-III (RUG-III), a category-based system. RUG-III is the case-mix system currently used in the Medicare SNF prospective payment system (PPS). Statistical measures of case-mix accuracy, other than R-squared, are explored. We find that a category-based system offers greater intuitive appeal for clinicians, while an index-based system offers greater accuracy and simplicity. R-squared, as a measure of case-mix accuracy, is overly sensitive to the choice of analytic sample and model specification. We conclude that both the category- and index-based approaches offer significant advantages. Future research should focus on the effective blending of the two approaches.
Learning Objectives: Participants in this session will 1. learn two approaches to case-mix design 2. understand new developments in Medicare's PPS for nursing facilities
Keywords: Nursing Homes, Financing
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.