found to be as strongly influenced by subjective factors related to the attitudes of individual physicians as by science. These subjective considerations, which I call collectively the 'practice style factor,' can play a decisive role in determining what specific services are provided a given patient as well as whether treatment occurs in the ambulatory or the inpatient setting. As a consequence, this style factor has profound implications for the patient and the payer of care.
"For example," he continued, "the practice style factor affects whether patients . . . with mild angina, or with a host of other ailments, receive conservative treatments in an ambulatory setting or undergo a surgical operation in a hospital. . . . The practice style that favors inpatient treatment greatly affects the demand for hospital care and has serious implications for efforts to constrain costs.
"These implications become clear when one recognizes that, within a region or state, different opinions held by physicians . . . are the most important determinant of variations in per capita costs for the treatment of specific diseases. . . . Some of the differences in opinion arise because the necessary scientific information on outcomes ismissing. . . . To resolve the differences in opinion—and to learn whether high or low rates of admission reflect appropriate care—more scientific information must be obtained.
"For other conditions, the practice style factor appears unrelated to scientific controversies. Physicians in some hospital markets practicemedicine in ways that have extremely adverse implications for the cost of care, motivated perhaps by reasons of their own or their patients' convenience, or because of individualistic interpretations of the requirements for 'defensive medicine.' Whatever the reason, it certainly is not because of adherence to
Wennberg, says Michael Zubkoff, "was just very persistent. He was able to stick with the theme: 'There are these variations in how we care for very similar patients. Why?' Wennberg peeled the onion back and kept going deeper and deeper."
medical standards based on clinical outcome criteria or even on statistical norms based on average performance."
Wennberg went on to demonstrate, once again, that the supply of hospital beds and specialists in a particular market often determines how medicine is practiced—even at the most prestigious medical centers. Comparing the health-caremarkets in New Haven, Conn., home to Yale's teaching hospital, and in Boston, home to Harvard, Wennberg found that per-capita health-care costs were about twice as high in Boston, "largely because" Boston had more "beds and medical personnel per capita."
Basically, because the beds were there, physicians used them. To this day, Dartmouth studies show that in areas with more specialists, patients get seen more
often—simply because the specialists have more time open in their appointment books. Yet doctors in these markets aren't consciously aware that they are hospitalizing or seeing patients more often. In the 1984 article, Wennberg reported that when he "asked clinicians who have practiced in both Yale and Harvard teaching hospitals to estimate the per-capita expenditures in each market, their answers indicate[d] they had no awareness of the magnitude of the difference. What is more surprising, many [did] not accurately guess which of the two markets is the more expensive."
Back in 1984, Wennberg was reluctant to assert that patients in Boston were receiving too much care. As he pointed out, "to learn whether high or low rates of admission reflect appropriate care—more scientific information must be obtained."
But, persistent as always, Wennberg pursued the "outcomes research" that would make it clear whether patients were benefiting (or not) from more care. In 1988, he established the Center for the Evaluative Clinical Sciences (CECS) at Dartmouth. And he soon collected around him a group of like-minded researchers—statisticians, economists, clinician-investigators, epidemiologists. They published more and more papers, reinforcing and expanding on Wennberg's findings. And they refined and improved their research methodologies.
By 1993, the CECS cadre had attained sufficient critical mass to begin teaching those methodologies. Today, the program offers two master's degrees (M.S. and M.P.H.) and a Ph.D.
In 1996, CECS published the first Dartmouth Atlas of Health Care—taking the concept Jack Wennberg had developed in Vermont and Maine, then used in New Haven and Boston, and applying it to the whole country. (See more about the Atlas.) That and