Geographic Variation In Health Care Spending: Fill & Download for Free

Regional variations in health care spending have been studied for more than three decades, primarily by John Weinberg and colleagues at the Dartmouth Institute for Health Policy and Clinical Practice through the Dartmouth Atlas Project. The Project analyzes the costs for services provided to beneficiaries in traditional fee-for-service Medicare (excluding Medicare Advantage enrollees) in geographic areas known as “hospital referral regions” and the factors that may contribute to these varying costs, including the supply of resources (such as hospital beds and specialist physicians), utilization of services, quality of care, and patient health status. The Project defines hospital referral regions as regional health care markets based on where patients are referred for major or specialized inpatient medical care, each having a minimum population size of 120,000 and containing at least one hospital that performs major cardiovascular procedures and neurosurgery. The Dartmouth Atlas Project uses Medicare data because of the availability of billing records for Medicare’s fee-for-service patients, data that is not available for the total privately insured population. The Project’s website indicates that several state-based studies of all health insurance claims (both Medicare and commercial) have found that variations in resources and quality in the non-Medicare population closely resemble those in the Medicare population.2 Data from the Dartmouth Atlas Project show that nationally, Medicare spent an average of $8,304 per enrollee in 2006. However, considerable variation in spending occurred among the 306 U.S. hospital referral regions, with the highest-cost regions spending more than three times the amounts spent in the lowest-cost regions. The highest-spending regions were Miami, Florida ($16,351) and McAllen, Texas ($14,946), compared to the lowest-spending regions of Honolulu, Hawaii ($5,311) and Minot, North Dakota ($5,542). State spending also showed variation, from a high of $9,564 in New York to just over half that amount ($5,311) in Hawaii.3 Trends in the spending rates of increase from 1992–2006 also varied considerably from region to region. The per capita rate of increase in inflation-adjusted Medicare spending averaged 3.5% nationally, but ranged from a low of 1.6% in the Honolulu, Hawaii hospital referral region to almost four times that rate (6.2%) in the Lincoln, Nebraska region.4 A February 2008 Congressional Budget Office (CBO) study of geographic variation in health care spending also found large differences across the country in spending for the care of similar patients, using primarily Medicare data but also total health care spending data. CBO found that the geographic variation in Medicare spending from the lowest to the highest spending areas has narrowed in recent years, while the variation in total health care spending has increased. Geographic variation in spending by the Department of Veterans Affairs has also increased in recent years so that it is similar to Medicare’s. Spending variation in the U.S. has been much larger than in Canada and somewhat larger than in the United Kingdom, countries where the financing of health care is more centralized than in the United States.5A December 2009 report by the Medicare Payment Advisory Commission (Med PAC) finds wide regional variation in Medicare spending per beneficiary, and less regional variation in Medicare per beneficiary utilization of services, which Med PAC says are not equivalent measures and should not be confused. Med PAC found that Medicare spending per beneficiary in areas at the 90th percentile of national average Medicare spending was 55% higher than at the 10th percentile (or, looking at the extremes, the highest spending area was more than two and a half times the lowest spending area). To analyze Medicare service use, Med PAC adjusted Medicare spending data for differing regional Medicare payments (e.g., wages, special payments to teaching hospitals, rural add-on payments, etc.) and for average health status, and found that service use in areas at the 90th percentile of national average Medicare service use was 30% higher than at the 10th percentile (looking at the extremes, the area with the greatest service use—Miami-Dade County—had twice the level of service use as the lowest area—nonmetropolitan Hawaii). Another finding was that areas with high levels of service use are not always the areas with high growth rates.