U.S. healthcare nations

June 25th, 2018 / By Paul LaBrec

While reading one of the news summary services I regularly review, I came across an article based on the book American Nations: A History of the Eleven Rival Regional Cultures of North America by Colin Woodard.1 The book was published in 2011 and I’ve read several summaries and discussions of topics raised in the book since then. Having started my collegiate career with an undergraduate degree in Cultural Anthropology, the topic of culture always piques my interest. According to a map presented in the book the cities of Huntington, NY and Chippewa Falls, WI are part of the “nation” of Yankeedom, and Tucson, AZ and San Diego, CA are part of the “nation” of El Norte. I’ve lived in each of these cities (nearly two decades each in three of them) and I’m a little skeptical of considering these pairs part of the same “nation.” But that’s another discussion for another forum. The article, however, did get me thinking about regionality in the U.S. healthcare system.

I first encountered the concept of regional market variation in U.S. health care when I left a university setting for private sector employment with a health care data startup venture in San Diego. Coincidentally, the year was 1996 coinciding with the first publication of the Dartmouth Atlas of Health Care based of the work of John Wennberg at Dartmouth University. Wennberg and colleagues posited, and later demonstrated with Medicare data, that there is regional variation in the utilization and price for healthcare services that cannot be fully explained by traditional demand side factors such as demographics, health status or disease incidence.2

Wennberg conducted his geographic analysis by constructing Hospital Service Areas (HSAs) based on local markets for hospital services defined by ZIP Code where most of the Medicare residents of an area received hospital services. This process resulted in 3,436 HSAs. HSAs were aggregated into 306 Hospital Referral Regions (HRRs) reflecting regional markets for tertiary medical care. Each HRR has at least one city where both major cardiovascular procedures and neurosurgery are performed.3 Examples of regional variation among Medicare enrollees from the first Dartmouth Atlas include a fourfold variation in per capita rates of coronary artery bypass surgery, an eightfold variation in rates of radical prostatectomy, and 33-fold variations in the use of breast sparing surgery in the treatment of breast cancer.4

One of my main functions at the startup venture I mentioned was to generate estimates and forecasts of various measures of health services utilization by health care setting for small area geographies. We started by applying cohort-specific utilization rates from various national datasets to cohort population estimates and forecasts by ZIP Code and Census Tract. We accounted for regional variation in utilization by using Medicare and commercial claims databases to calculate utilization rates by Metropolitan Statistical Areas and non-MSA areas by state and compare those rates against a national average resulting in a “regional adjustment factor,” and also applied a trend factor to selected forecasts. After accounting for demographic differences between area populations we still saw regional variation, as did Wennberg, with some areas having double-digit percentage differences from other areas.

In many of its publications, the National Center for Health Statistics (NCHS) includes regionally-specific utilization data for four regions of approximately 12 states each: Northeast, Midwest, South, and West. Generally, the Northeast and South have highest utilization rates followed by the Midwest, while the West has the lowest use rates.5

The Centers for Medicare and Medicaid Services (CMS) began adjusting Medicare payments for geographic market variation in 1992 with the addition of Geographic Practice Cost Indices (GPCIs) in the calculation of component Relative Value Unit (RVU) input prices. The formula and definition of geographic areas for adjustment was recently reviewed by the Institute of Medicine which made recommendations about the structure of the formulas as well as the definitions of geographic areas for adjustment.6 The 2018 Medicare Physician Fee Schedule includes GPCI adjustments for 112 state or sub-state areas, reflecting the second year of a two-year update transition.7

In the twenty years since the first Dartmouth Atlas was published, more evidence has mounted documenting the regional variation in healthcare practice. This variation, driven by differences in health care supply, practice patterns of the local medical culture, and other factors highlights the importance of recognizing the influence of local factors on medical practice. I wouldn’t consider local healthcare markets as unique “nations,” but variation between markets should be recognized as important. As national healthcare policy mandates greater uniformity in benefit packages, quality and value measurement, and performance targets we may see a reduction in this variation. For now, that remains to be seen.     

Paul LaBrec is research director for Populations and Payment Solutions with 3M Health Information Systems.


References

1 Woodard, C. American Nations: A History of the Eleven Rival Regional Cultures of North America. (Viking Press, 2011).

2 The Dartmouth Atlas of Health Care. “Atlases and Reports.” Accessed June 21, 2018.

3 Dartmouth Atlas of Health Care. “About Our Regions.” Accessed June 21, 2018.

4 Wennberg, J et al. The Dartmouth Atlas of Health Care. (The Trustees of Dartmouth College, 1996), p. 7.

5 Centers for Disease Control. “National Ambulatory Medical Care Survey: 2015 State and National Summary Tables (Table 1).” National Ambulatory Medical Care Survey, 2015. October, 2017. Accessed June 22, 2018.

6 MaCurdy, T. et al. Geographic Adjustment of Medicare Payments to Physicians: Evaluation of IOM Recommendations. (Burlingame, CA. Acumen LLC), July, 2012. Accessed June 21, 2018.

7 Centers for Medicare and Medicaid Services. “Physician Fee Schedule – January 2018 Release.” Accessed June 22, 2018.