Discovering the epidemic of overtreatment

Similar populations living in different regions of the United States get exposed to wildly different amounts of medical care.

If that sounds like an old story, it is. It’s now four decades old. But it is an important story to reflect on as we consider the path forward for our medical care system.

In the late 1960s, a nephrologist trained in epidemiology was sent to Burlington, Vt., to run the state’s regional medical program. The program was part of the Lyndon B. Johnson administration’s effort to bring the advances of modern medicine to all parts of the nation. Its goal was to identify which areas of Vermont were underserved.

The task required that the state be subdivided into geographic units within which residents received most of their medical care (not surprisingly, the 13 units were constructed around towns with hospitals). And it required an emerging technology: a mainframe computer.

It was the dawn of the era of big data. Electronic medical records, though quite primitive, were analyzed to reveal the pattern of medical care delivery in the entire state.

The findings were — so to speak — all over the map. Depending on where she lived within the state, a woman’s chances of having her uterus removed varied as much as threefold. A man’s chances of having his prostate removed varied as much as fourfold. And the children of Morrisville were more than 10 times as likely to have their tonsils removed as their counterparts in Middlebury.

Because the population of the state was so homogenous, it sure looked like the variations were driven by the medical care system, not its patients.

No medical journal would publish the findings, and so they instead appeared in the journal Science.

Despite his original motivation to identify underservice, the nephrologist-cum-epidemiologist concluded the 1973 Science article with a decidedly different take: “the possibility of too much medical care and the attendant likelihood of iatrogenic illness is as strong as the possibility of not enough.”

He was a radical.

His name is John E. Wennberg, M.D., M.P.H; but at Dartmouth, we all call him Jack. He is the reason Dartmouth is on the health policy map.

Jack went on to document similarly wildly variable medical practices in the other New England states. But it wasn’t until he compared two of the nation’s most prominent medical communities — Boston and New Haven, Conn. — that the major medical journals took notice. In the late 1980s, both the Lancet and the New England Journal of Medicine published the findings that Boston residents were hospitalized 60% more often than their counterparts in New Haven. Oh, by the way, the rate of death — and the age of death — in the two cities were the same.

It was an alternative version of the Harvard-Yale game — and Yale won.

In the 1990s, Jack led the effort to catalog the patterns of medical care for the entire nation, and the Dartmouth Atlas of Health Care was born. The atlas data were central to the contention, made by the Obama administration, that there was substantial waste in U.S. medical care.

You probably knew that already. And Jack’s work is a big part of the reason you do.

OK, it’s interesting history. But how is it relevant today?

Because this work represents the genesis of a new science — medical care epidemiology, a science we are about to need a lot more of.

Classically, epidemiology examines exposures relevant to infectious disease: think water supplies as the source of cholera epidemics in the mid-1800s to food supplies as the source of recent Salmonella outbreaks. In the mid-1900s, epidemiology began to tackle exposures relevant to chronic disease — discovering, for example, how cigarette smoking increases the risk of dying from lung cancer 20 times.

Medical care epidemiology examines the effect of exposure to medical care: how differential exposure across time and place relates to population health outcomes. It acknowledges that medical care can produce both benefits and harms, and that conventional concerns about underservice should be balanced by concerns about overdiagnosis and overtreatment. Think of it as surveillance for a different type of outbreak: outbreaks of diagnosis and treatment.

Medical care epidemiology is not a substitute for traditional clinical research. Instead, it is a complement, because there are many questions that cannot be studied in randomized trials. How do new diagnostic and treatment technologies affect clinical practice? Do specialists better spend their time doing procedures or providing support for primary-care practitioners? How frequently should patients be seen? Do patients do better taking more medicines or fewer?

Jack is the father of this new science and the inspiration that led the next generation of physicians to enter the field. His colleagues will honor his contribution this month at Dartmouth on this 40th anniversary of his Science paper.

But all Americans have benefited from his contributions: bringing science to bear on the practice of medical care, recognizing that too much medical care is a problem, and arguing that medical care should serve the needs of the patient, not the needs of the system.

H. Gilbert Welch is a professor of medicine at the Dartmouth Institute for Health Policy and Clinical Practice and an author of “Overdiagnosed: Making People Sick in the Pursuit of Health.” He is currently on sabbatical at Montana State University in Bozeman.