Geographic Variation in Medicare Spending Still a Mystery

March 07, 2014

Like Spanish conquistadores in the New World searching for a mythical city of gold, healthcare academics have explored another landscape with a fortune in mind: the reason there are wide geographic differences in Medicare spending.

Consider that Medicare spent, on average, $15,357 per beneficiary in Miami, Florida, in 2012 compared with $6569 in Grand Junction, Colorado, according to the Centers for Medicare & Medicaid Services (CMS). Find the source of the variation, make the rest of America similar to Grand Junction, and the United States could save lots of gold in terms of healthcare spending, the argument goes. The potential savings could reach $700 billion a year, the White House estimated in 2009.

Nobody has yet to find the El Dorado of healthcare, however, states a literature review of this research, published online yesterday in Health Affairs.

"Studies indicate that there is no single answer to addressing variation in Medicare spending by region," write lead author Amanda Cassidy, principal of Meitheal Health Policy in Richmond, Virginia, and coauthors. "Even after multiple factors are considered, some geographic differences remained unexplained."

The Dartmouth Institute for Health Policy and Clinical Practice put the subject of geographic variation in Medicare spending on the map in 1996 with the debut of its Dartmouth Atlas of Healthcare. It drills down to the level of "hospital referral regions," (HRRs) which describe where patients go locally for major cardiac and neurological procedures.

Themes on a Variation

Describing the variation is the easy part. Explaining it is another matter. Cassidy and colleagues go down the list of possible reasons:

  • Variation in Medicare payment: Medicare adjusts its payments to providers by region according to differences in their overhead costs and other factors such as physician–patient ratios. These payment differentials explain spending differentials, but only partially, studies indicate.

  • Variations in patient health status: Could it be that high-spending areas have sicker patients? After all, the percentage of Medicare beneficiaries with diabetes in 2012 ranged from 15% in Grand Junction to nearly 44% in Harlingen, Texas, which ranked 15th in unadjusted per capita spending, according to the CMS. In addition, a study published in 2013 said health status could account for up to 85% of spending variation. Deflating this theory, however, is the notion that providers in some regions churn up more illness because they perform more tests and procedures than their colleagues elsewhere.

  • Variation in the use of services: The research in this category has tried to determine whether higher spending in a particular region simply reflects the tendency of its hospitals, physicians, and other providers to deliver more services, or more intense services, perhaps in response to patient preferences. Where do patients go for postacute care after a hospitalization, for example? A skilled nursing home? A home-health agency? Or outpatient providers? The choices, region by region, would contribute to spending differences.

Table 1. HRRs With the Highest and Lowest per Capita Medicare Spending in 2012

Highest per capita HRRs  
 Miami, Florida $15,537
 Bronx, New York $14,699
 Manhattan, New York $13,699
 Los Angeles, California $13,319
 Chicago, Illinois $13,059
Lowest per capita HRRs  
 Honolulu, Hawaii $6790
 Dubuque, Iowa $6716
 Bend, Oregon $6667
 Missoula, Montana $6633
 Grand Junction, Colorado $6569

Source: CMS

Academics also have hypothesized "induced demand" as a driver for healthcare services. Rather than demand generating supply, supply generates demand. Hospitals and physicians naturally want to stay busy, and the higher their concentration in a region, the higher spending will go. "However, others note that physicians may locate where patients are sicker, so this correlation could be misleading," write Cassidy and colleagues.

That is one fly in this theory's ointment. Another is that even in areas where spending is higher, the volume and intensity of services vary from service to service. Medicare reformers, therefore, should not tweak every provider's payment in a given area as if they were all high-users.

What further complicates the research into the causes of geographic variation in Medicare spending is that private insurance and Medicaid do not follow the same patterns, write the authors. They cite a study published in Health Affairs in 2010 that looked at 2 Texas communities with big differences in Medicare spending, but not so big differences in spending by Blue Cross and Blue Shield of Texas, perhaps because of prior authorization and other managed care tools.

Maybe it is not so important to solve the mystery of geographic variation spending after all. That is what the Institute of Medicine (IOM) suggested last year when it recommended against increasing or decreasing Medicare payments on the basis of a geographic value index. Such an index would account for the relative cost and quality of care in a given region. The IOM shied away from this idea because in a region deemed high-cost and low-quality, some providers manage to be low-cost and high-quality, and they should not be penalized like the rest, according to the IOM.

"Instead, the IOM recommended that CMS continues investigating possible payment reforms that focus on providing incentives for clinical decisions makers to improve care coordination and health outcomes," note Cassidy and colleagues. In other words, spending variation can be reduced by shifting from fee-for-service to pay-for-performance across the country. Getting physicians and hospitals to participate in accountable care organizations, medical homes, and other new ways to deliver and reward high-quality care may be the path toward the elusive city of gold.

Cassidy has disclosed no relevant financial relationships.

Health Affairs. Published online March 6, 2014. Full text


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.