There is more money spent on health care for each citizen of Massachusetts and Pennsylvania than for a citizen in Utah, Arizona or Nevada. In fact, per capita health spending was 59% lower in Utah than Massachusetts in 2004. The latest state-by-state spending variations are highlighted in Health Affairs’ web-exclusive feature.

Welcome to the statistical phenomenon in health care known as “regional variation.” The guru-researcher of regional variation is John Wennberg, who has detailed these trends in fhe Dartmouth Atlas.

New regional health spending data were published in Health Affairs, which we health economists and policy wonks eagerly anticipate each bi-monthly issue. The September/October issue covers “Caring for the Vulnerable.” The regional variation report is a web exclusive, which you can access through this link. This map from the New York Times illustrates the variability.

Why are there regional variations in health spending across states in the U.S.? There are many factors that contribute: the proportion of Medicare enrollees (i.e., elderly) and the proportion of Medicaid (which is the source of Alaska being a big per capita health spender) are big contributors to spending differences.
Another factor to consider is Roemer’s Law: this was one of my earliest lessons in health economics back in the day. The concept is known as “supply-induced demand.” In 1961, Milton Roemer (a fellow alum of the University of Michigan School of Public Health, although we graduated several decades apart) wrote a seminal paper on the relationship between the supply of hospital beds in a community and spending. Roemer found that, if a bed was built in a community, it was used and paid for. This may seem like a tautology. But the way Professor Paul Feldstein, health economist extraordinaire, explained it to me:
“A built bed is a filled bed is a billed bed.”
According to Roemer’s research, if you add a second hospital to a one-hospital town, the new one fills and the existing hospital stays filled. Similarly, when new digital technologies are introduced into a local community (such as MRI, CT, and PET scanning equipment), use of these technologies grows but they do not replace existing modalities. This creates a never-ending cost-spiral which contributes to regional spending variations (regions that become the technology “haves” vs. the “have-nots”).
I’m absolutely not arguing against the adoption and use of new medical technology–just the rational, appropriate use of it based on evidence. Medical innovation is one of America’s few areas of global competitive advantage (along with entertainment and fast food, a topic for another post). We must continue to nurture innovation, but do so in the context of evidence-based medicine.
Health Populi’s Hot Points: As we add new health resources to a community, we often duplicate capital resources and add costs into the system. Medical technology is generally additive, not replacing. Adopting and using technology based on what will work for a specific patient — the principles of evidence-based medicine — will help to address variations, which will impact costs and improve quality.
Sources: Health Spending By State Of Residence, 1991-2004, Health Affairs, (; The Dartmouth Atlas,; Bed Supply and Hospital Utilization: a natural experiment, Milton Roemer, Hospitals, November 1, 1961, 35:36-42.