Variation in Medicare Expenditures Across States

By Philip SusserPublished May 1, 2013

null — Large geographical variation in Medicare reimbursements per enrollee exists within the United States. Although, when accounting for acceptable variation such the overall health of a population, these differences are not as dramatic. A geographically based index that would penalize certain inefficient regions with histories of high spending is inefficient; the main problem being that variation exists within regions. A more individualized approach must be taken.

There is large variation in healthcare spending in the United States with low relative quality of care in higher spending areas. Specifically, Medicare reimbursements vary drastically across states and certain regions within states. The Dartmouth Atlas was a project developed by the Dartmouth Institute for Health Policy and Clinical Practice to plot Medicare expenditures and quality of care across the United States by geographic region. A map of 2009 Medicare reimbursement payments per enrollee across the U.S. show large disparities; Medicare spending is growing at different rates across geographical regions. While some areas are offering lower cost care, with payments as low as $6,632 per enrollee, other areas are cashing in $16,125 per Medicare enrollee. Although, the data provided by the Dartmouth Atlas does not account for differences in prices across regions, which are adjusted by Medicare according to the input prices that different doctors face. While differences suggested by Dartmouth might be slightly inflated because of this, there are other, more important factors at play that account for variation in spending. With large disparities in spending, there are some questions left to be asked: What accounts for these differences? Is there any soluble way that the health care system can reward low cost, high quality areas through a geographic index?

In a 2013 report, the Institute of Medicine (IOM) assessed the quality of care across geographic regions of the US based on Medicare reimbursements and the efficacy of a geographically based reimbursement index. In assessing differences in spending, they proposed two explanations for the stark variation. The “acceptable” reason was differences in the health status of a population. So, if a certain hospital was located in a neighborhood with a sick and old population, of course the amount of Medicare reimbursement payments per enrollee would be higher for that region. This means that even in the most efficient, ideal system, there would still be variation across geography. The “unacceptable” source of variation was a high volume of low quality care, which can be enticing for doctors in a fee-for-service dynamic. An example of inefficient, low quality care is duplication of medical tests. The success of a geographic value index would be measured by the extent to which it reduced this sort of inefficient overprovision of care.

To demonstrate just how much “unacceptable” variation there was across the United states, the IOM looked at a high spending region and low spending region, accounting for “acceptable” differences such as different population health. The ratio of the 90th percentile spending areas to 10th percentile spending areas in 2008 was 1.44 before adjusting for acceptable variation and 1.23 after adjusting for some acceptable variation, a significant drop. This indicated that differences in health spending across regions might not be as high as they originally seemed through figures such as the Dartmouth Atlas.

There are some downsides to a geographic value index and things to remain wary of in its application. As the IOM notes, a geographically based reimbursement index must adjust payments based on health outcomes as a result of clinical care, not the overall health of a patient because there are other factors at play in accessing the overall health of a patient that do not necessarily come as a result of the provision of health care. For example, it would be difficult to penalize a dermatologist in a certain region through Medicare reimbursement rates for having patients with high levels of heart disease. Another problem with the geographic value index is that there are differences in the quality of care within regions. If a certain practice that provided high quality care happened to be located next to a large hospital with highly inefficient care, the index would affect the reimbursement payments for both parties because they are deemed to be in the same geography. Furthermore, a decrease in payments for inefficient areas can cause an offsetting behavior in which physicians increase the volume treatment, in turn exacerbating inefficiency of treatments.

For a geographic value index to be most fair, there needs to be very little variation in behavior of the “sub-regions” whose reimbursement payments would be affected uniformly. The IOM found that within the highest spending regions, the highest spender spends at least 36% more than the lowest spender, a significant difference. These data is important in understanding that regional differences in health care tendencies are averages, not patterns. The IOM also found that even within practices, doctors had different prescription tendencies, with some prescribing more expensive medication and others choosing the cheaper counterparts. Thus, the geographic value index’s assumption that all physicians and hospitals act in a similar fashion is flawed.

The geographic value index is seemingly a watered down response to the fee-for-service system in its emphasis on quality over quantity. Its intentions of quality driven care are good, but the indexing parameters are imperfect. What’s more, it does not effectively eliminate the fee-for-service system, but rather it adjusts prices in response to the average outcomes of regions with diverse physician patterns – “It would reward low value providers in high level regions and punish high value providers in low value regions.” If a government directed, financed, and coordinated health care system continues, the most effective policy to monitor physician patterns would have to be at the individual level. Tracking patient outcomes as a result of a hospital visit would make doctors feel financially responsible for the care that they provide. If doctors were found to be exhibiting patterns of overspending without positive patient results, they would have to pay a fine to the government. The revenue from this fine could then be directed to more efficient doctors and hospitals as a reward for optimal care. There would be many oversight costs in this system, but it would create the appropriate incentives for doctors to treat patient care with more frugality.