Does the RUGs III case-mix system adequately adjust for the cost of Medicare patients in nursing homes?

Medicare pays for Medicare patients in nursing homes using the RUGs III case-mix methodology. To avoid perverse incentives, i.e. incentives to both impede access for high severity patient and to lower quality, it is important that the case mix adjustment capture the cost of Medicare patients appropriately. This study used a cost function approach to examine this question.

Medicaid cost reports data for 946 free standing (87% of all) nursing homes in California in 2005 was augmented with MDS and economic census data. We estimated hybrid cost functions in which the dependent variable was log of total costs and independent variables included outputs, wages, and the percent Medicare days. Outputs were measured by RUGs case-mix adjusted days. Because Medicare days are likely to be endogenous with costs, we used instrumental variables. To assess which specific services contribute to the incremental costs due to Medicare patients, we repeated the analysis for specific cost centers.

On average, 11% of nursing homes days were Medicare, post-acute. The IV performed well, with an incremental F(1,939)=16.6. Medicare days were significantly associated with costs (p<0.000) even when controlling for RUGs case-mix. The marginal cost of a one percentage point increase was estimated at $116,123 or about 2.0% of annual costs for the average facility. Analysis by cost center found that the increase was due to rehabilitation, pharmacy, RNs, laboratory, and administration.

This study confirms concerns raised by MEDPAC that the RUGs III case-mix system does not adequately captures costs and should be replaced by RUGs IV.