105. Adjusting VA Nursing Home Costs for Case-Mix
W Yu, VA HSR&D Health Economics Resource Center; TH Wagner, VA HSR&D Health Economics Resource Center; S Chen, VA HSR&D Health Economics Resource Center; PG Barnett, VA HSR&D Health Economics Resource Center
Objectives: The Millennium Bill (HR 2116) passed in January, 1999, requires the VA to "establish a program of extended care services for veterans." The VA is considering whether to provide these services in-house or through contractual arrangements with non-VA providers. However, in either regard, the VA needs better methods for tracking nursing home costs that account for case-mix. This study (1) developed a method to estimate nursing home cost with case-mix adjustment; (2) examined the cost variation within each of the top 20 most common reasons for VA nursing home admissions; and (3) evaluated the impact of case-mix adjustment on the average daily rate of a facility.
Methods: The sample contains 39,792 VA nursing home stays in FY1998. We calculated an average wage-weighted work unit (MWWU) for each nursing home stay using up to 3 wage-weighted work units (WWU) recorded in the patient assessment files. For patients whose last assessment before discharge was longer than 3 months, we developed a regression model to estimate a WWU at discharge to reflect the change in resource use since the last assessment.
To examine the case-mix variability within homogeneous medical conditions, we selected the 20 most common reasons for nursing home admission and examined the variations of MWWU within each medical condition. To evaluate the impact of case-mix on nursing home average cost, we calculated a case-mix index for each facility by dividing the case-mix adjusted total days by the unadjusted total days. We then examined the difference between the average daily costs with and without case-mix adjustment for 109 VA nursing homes that had more than 100 nursing home stays.
Results: In our sample, the MWWU varies substantially among patients (mean=698, std=199, min=413, max=1800). When we explored case-mix within homogeneous medical conditions, we found that the 20 most common reasons for nursing home admission accounted for 96.3% of the total admissions. Within each medical condition, the MWWU still varied substantially, suggesting that controlling for diagnosis does not adjust for case-mix.
We examined the daily costs before and after case-mix adjustment. After case-mix adjustment, the average daily costs decreased from 24 to 5% among the 30 most severe case-mix nursing homes. In contrast, the average costs of the 30 least severe case-mix nursing homes increased from 5 to 28%. This indicates that case-mix is crucial for understanding the costs of a particular VA nursing home.
Conclusions: This study shows that using a simple average daily rate to determine VA nursing home costs, even when controlling for medical conditions, can miss the large variations of resource use among patients.
Impact: We developed a method to calculate an average resource use index for each nursing home stay. The MWWU can be used to adjust for resource use of individual nursing home stays as well as to measure average resource use for a study cohort and case-mix for a facility. This method is easy to use, suggesting that other researchers should include such measures in nursing home cost determination.