surprise at the finding. "It was implausible that the need for tonsillectomies varied by that degree."

In Pediatrics in 1977, he noted that physicians defended geographic variation as "inevitable" in cases where there is "professional uncertainty" and physicians "do not know which level of use is 'appropriate.'" Wennberg was not satisfied with the explanation: "This defense, while explaining the past, cannot justify the future," he declared. "For one thing, the dollar costs of uncertainty are too great."

By then, Wennberg had branched out to study geographic variations in the delivery of care in Maine. There, he calculated that in communities where few children had their tonsils out, the cost per capita for tonsillectomy was 85c. In areas where tonsillectomies were popular, the cost per capita was $4.55. "Projected nationally, the lowrate strategy costs less than $200 million; the high-rate strategy costs nearly $1 billion," he observed. And that was in 1975 dollars.

"But the costs, of course, are not only in dollars,"Wennberg continued. They could also be measured in lives. In Vermont from 1969 to 1973 and in Maine in 1973, "three postoperative deaths followed tonsillectomy," he wrote. "For so costly a procedure, ambiguity concerning its value will likely become increasingly intolerable."

For years, even medical students had recognized the dearth of scientific evidence behind the procedure. "When I was in medical school, I remember one professor—a particularly arrogant Park Avenue practitioner—asking: 'What are the indications for a tonsillectomy?'" recalls Dr. James Strickler, who was dean of Dartmouth Medical School when Wennberg joined the Dartmouth faculty. "And one of my more provocative classmates responded: 'A hundred dollars and a pair of tonsils.'"

Wennberg went on to show that tonsillectomies were just one of many procedures done far more often in some locales than in others. He identified geographic variation in the rate of other common operations, including

It was his children--specifically their tonsils—from whom Wennberg got his first insight into health-care variations. This photo dates from his Vermont years.

When Wennberg looked at the health-care data for the two neighboring Vermont towns, he realized that 70% of the Stowe children had had their tonsils out by the time they were 15 years old, as opposed to only 20% of those in Waterbury.

appendectomies, hysterectomies, prostate surgeries, and gall bladder removals.

Moreover, Wennberg had begun to demonstrate a clear connection between a community's supply of health-care resources—such as surgeons and hospital beds—and how much surgery the residents of that community received. More surgeons and more beds equaledmore procedures. Put simply, supply seemed to drive demand.

Four years before his article appeared in Pediatrics, Wennberg had published, in the journal Science, his first study about geographic variations in health care. There, he and his coauthor, Dr. Alan Gittelsohn, established a firm connection between the supply of health-care resources in a community and how much care its citizens received. At the same time, they acknowledged that they could not say which rate of care represented "a better allocation of resources." For a given surgery, it was not clear whether patients in communities where surgeons

were in relatively short supply received too little care, or whether those in areas with a surfeit of surgeons got too much.

Few studies had been done "comparing outcomes under controlled circumstances,"Wennberg explained at the time. Sadly, the situation hasn't improved much. "Jack's first paper in Science was so on target," says Dr. James Weinstein, an orthopaedic surgeon at Dartmouth. "This is what we are working on today."Weinstein, in fact, heads the largest randomized surgical trial ever funded by the National Institutes of Health—a comparison of surgical and nonsurgical treatments for various kinds of back pain. And it is Weinstein who has been named to succeed Wennberg as the leader of the people and projects Wennberg assembled at Dartmouth.

More harm than good
But before those people and projects existed, even before Wennberg's arrival at Dartmouth, he began to realize just how much was at stake: "Given the magnitude of the variations," he wrote in 1973 in Science, "the possibility of too much medical care and the attendant possibility of iatrogenic illness [illness that is caused by medical care] is presumably as strong as the possibility of not enough service and unattended morbidity and mortality." In other words, in towns blessed with many surgeons, the extra surgeons actually might be doing more harm than good.

This is not what Wennberg's peers in the medical profession wanted to hear. In fact, he almost didn't get that first paper published. "We tried the New England Journal of Medicine, we tried the Journal of the American Medical Association. . . . We tried all of the medical journals," he recalls.

What did the reviewers at those journals say?Wennberg chuckles: "They didn't say anything. We got form rejection letters." ("This still happens," he noted in an interview in Health Affairs in 2004. "Generally we don't bring good news.")

Back in 1973, he recalls, "Science was the journal of last resort, but we were delighted to get the paper accepted."