The Dartmouth Atlas project brought great focus to the presence of geographic variation in health spending, at least in the Medicare program. The authors of that series of reports believe that the variation in spending is not associated with quality of care and that differing practice styles account for much of the variation, inferring that the low-cost states can be used as a model to reduce spending in higher cost states. Other researchers have disputed these findings, both on methodological and data bases. A working paper from staff at the Federal Reserve explores these questions in some depth, making some interesting findings. (FR Working Paper) The main data source for the paper is the CMS national health spending database, supplemented with a variety of data sources for socioeconomic and health behavior factors. The analysis focuses on acute health care spending–that for doctors, hospitals, testing, etc.–and ignores long-term care spending, which is affected by other factors.
The author concludes that there is little support for the notion that variation in spending is attributable to different practice patterns. Instead she suggest that most of the variation in Medicare spending is due to socioeconomic factors that affect the need for care and to interactions between the Medicare and Medicaid program and with private health spending. The researcher further suggests there is little correlation between Medicare spending and overall spending per capita in a state. Perhaps most importantly, there is a negative correlation between the level of Medicare spending in a state and its growth rate; that is low-cost states are not low-spending growth, which suggests some regression to the mean is at work. From a methodological perspective, the author suggests that analyses which control for health attributes at a state level do a better job of explaining variation than do those which control those attributes at an individual level, as the Dartmouth work does. The research has strong practical implications, as CMS is beginning to base some reimbursement on this alleged inappropriate variance in geographic spending, and apparently may be making mistake in doing so.