Document Type


Publication Date



Geisel School of Medicine


In 1967, after finishing my postgraduate training at Johns Hopkins in internal medicine and epidemiology, I took a job at the University of Vermont in Burlington as Director of the Northern New England Regional Medical Program (RMP). The Vermont program was one of some fifty RMPs that blanketed the country as part of President Johnson’s Great Society program. The idea behind the RMP was that advances in biomedicine had so improved the outcomes of major killer diseases that it was critical to ensure that all Americans, not just those fortunate enough to live near an academic medical center, had access to these services.

Epidemiologists are interested in what happens to groups of people: for example, how many get heart attacks (incidence) and what happens to patients (outcomes) according to the treatment they receive (prognosis). It was thus quite natural for me to want to use the tools of epidemiology to provide population-based information about the distribution of health care resources and the utilization of services among Vermont communities; after all, good planning for improvement requires knowledge about the current status of the health care system.

Alan Gittelsohn, a biostatistician from Johns Hopkins who had also been my teacher, and I developed a strategy we called the small area analysis of health care delivery. The method defined the geographic boundaries of local health care markets, based on where patients actually went for their care, and described the per capita use of resources and services for the resident populations.

Our first Vermont small area analysis (which we published in a 1973 article in Science) brought a big surprise. While we had expected to find a rural health care system characterized by underservice, we found instead a typology of care characterized by vast variations in the deployment of resources and the utilization of services among neighboring communities, without apparent rhyme or reason. The results forced me to re-examine the policy assumptions behind the RMP and a good deal else about the U.S. health care economy. It was evident that the problems facing regional and local health care markets were much more profound than the barriers to diffusion of new technology the RMP was designed to overcome.

I have spent most of my career studying the variation phenomenon. In a recent book, “Tracking Medicine: A Researcher’s Quest to Understand Health Care”, I chronicled this research and the evolution of thinking on the causes and remedies for unwarranted variation. Several of my colleagues and students have asked if I would make our research papers—some of which are hard to find and some not yet published—available to a wider audience. This website is my effort to do this. I provide introductory remarks for each paper in an attempt to clarify the context, relationship to previous work and the role the ideas in these papers played in building an understanding of practice variation. Over the years I have had the good fortune of working together with colleagues who have shared my fascination with the story of practice variation, and who have contributed in fundamental ways to the ideas and concepts presented in these papers.

The ordering is by topic, arranged in roughly chronological order:

  • The Early Papers
  • The Prostate Outcomes Studies
  • The Epidemiology of Medical Care
  • The Policy Papers
  • Editorials, Commentaries & Perspectives

Included in

Epidemiology Commons