Skip to Content

Geographic Differences in Health Care Spending

'How are geographic differences in health care spending evaluated?

Much of the discussion around geographic variation in health care expenditures is based on data compiled in the Dartmouth Atlas of Health Care. Researchers have compared institutional spending on Medicare services at the end of life and compared spending across cities and states. For instance, the Atlas' "Hospital Care Intensity Index" (HCI) compares Medicare utilization and spending in the last two years of life based on where services are delivered (not where patients are from). Hospitals are grouped into 306 Hospital Referral Regions (HRR) which, on average, serve populations of approximately 1 million people.

The Dartmouth Atlas External Link also aggregates HRRs by state, with New Jersey having the highest HCI and Utah the lowest .However, within states, there is significant variation in HCI; in Florida,Miami spends the most in the last two years of life, Tallahassee the least. Within each HRR, there is also variation, with teaching hospitals often spending more than average even in "low intensity" regions.

Conclusions from the Atlas showing large regional difference in medical spending indicate 30 percent health care "waste," suggested by Orszag and others; however:

  • Medicare beneficiary spending is a poor proxy for overall spending in a state or region.
  • Variation has two principle components: costs/prices and utilization rates; MedPAC has found that the majority of spending variation is eliminated when costs are accounted for.
  • Health status, influenced by poverty and other factors, accounts for much of the variation in utilization rates.
  • When patients are properly risk adjusted and are studied from time of admission forward (instead of starting at death and working backward), higher utilization rates often lead to improved outcomes, and are not considered "waste" by those patients.

Conclusions and Recommendations

  • Geographic differences in Medicare spending are not a good measure of "efficiency," or of health care spending overall.
  • Is there over-utilization? Yes. Is there under-utilization? Yes. Do we know exactly why? No.
  • Once price inputs are adjusted, some variation in utilization still exists; additional studies of geographic differences by the IOM are needed because poverty, race, and physician behavior all play a role.
  • Current "value" or "efficiency" adjustment proposals would represent the largest change in the Medicare payment system in 25 years with unknown effects on patients-and propose a blunt policy solution for a problem whose cause is not fully understood. These changes may harm the sickest, most vulnerable Medicare patients.
  • There is significant risk of unintended consequences when the results of aggregate behavior (e.g., states, regions, counties) are applied to individual patients.
  • Major Medicare payment changes are too important to relegate to one IOM study, a brief review by HHS, and whole-cloth acceptance or rejection by members of Congress.