Taking the guesswork out of medicine

David Eddy '59 turned a mathematical eye to health care. 

By
Genny Beckman Moriarty
July 30, 2018

By his own admission, David Eddy ’59 — referenced by many as a father of evidence-based medical decision-making — was a mediocre student who struggled to scrape by with C’s. Although the diagnosis did not exist back then, Eddy feels certain he had attention deficit disorder.

“My mind could not stay put on anything,” he says. “When the assignment was something like [John Galsworthy’s] The Forsyte Saga, I would start at the top of the first page, but before I could get to the bottom my mind would go out the window. Although my eyes would zigzag down the page, nothing on that page went into my brain. In four years at Exeter I was unable to complete a single reading assignment.”

To survive, Eddy had to develop other skills – primarily critical listening and questioning. By paying close attention to his classmates’ comments and asking questions to stealthily fill in the gaps, he became able to reverse engineer virtually any text. He got so he could do this sufficiently well to join the discussions, write essays, pass tests, and survive the “eyeball-to-eyeball contact” of the Harkness method. These skills, along with a naturally inquisitive and disruptive spirit, would eventually lead Eddy to instigate a seismic shift in how medical decisions are made, and to pioneer the use of mathematical models of diseases and treatments.

After Exeter, Eddy went to Stanford. (“I only got in on the strength of Exeter’s name,” he quips.) There, the ADD continued to limit his ability to study, which wasn’t helped by his choice of history as a major. Eddy decided on medical school next — for no better reason, he says, than his father, grandfather and great-grandfather were physicians. After a “thoroughly unremarkable” performance at the University of Virginia, Eddy earned his MD. Then, “by some failure in the selection process” he secured a prestigious internship and post-doctoral fellowship in cardiovascular surgery back at Stanford. It was during that period that his passion and talents finally began to take shape.

Ever curious, Eddy began to ask about the evidence that underpinned what he was being taught. He was startled to discover there was little if any good evidence behind many of the tests and treatments that were widely accepted by physicians. Instead, he found oversimplifications, overreliance on anecdotes and a propensity to believe that if something had been done in the past, it should be done in the future.

“It was astounding,” he recalls. “Some of the most time-honored maxims, like ‘Once a C-section, always a C-section,’ had absolutely no research to back them up.” Eddy bristled at the thought of doing something simply because it was considered “standard and accepted” practice, and he was beginning to realize he had to get out of clinical medicine. Then luck struck. 

By a fluke, Eddy walked in on some engineering students who were working out a math problem on a chalkboard. The mathematical symbols excited him, so he purchased Quick Calculus, a self-guided textbook, from the campus bookstore. “With incredible naiveté I thought, ‘That looks interesting. I think I’ll learn some math,’ ” he recalls. The same student who had struggled to get through his books at Exeter devoured the text in two days. From there, he churned through a five-volume set — typically taught over two years — in the span of one month. Eddy believes his hyper-focus and interest in the subject area overrode his other learning disabilities. But he also makes a nod to maturity: “I still had ADD, but by then I had the mental machinery to focus it better.”

Convinced mathematics could help improve health care decision-making (“although I didn’t yet know how”), Eddy dropped out of his fellowship and “begged” his way into Stanford’s engineering graduate school. While working toward his doctorate in a subset of engineering mathematics, Eddy continued to comb through medical journals like an investigative reporter. He wanted to know exactly which procedures had been proved to be effective, under what circumstances, and for whom. He also began to build mathematical models to help answer those questions.

Eddy received his doctorate in 1978. Two years later, his thesis, published in book form as Screening for Cancer: Theory, Analysis and Design, won the Frederick W. Lanchester Prize, the top international honor in the field of operations research. Tenure and a full professorship in engineering at Stanford, with a joint appointment in surgery, promptly followed.

Luck struck again when Eddy’s work on a mathematical theory of screening came to the attention of the American Cancer Society, and they asked him to help write guidelines for cancer screening. Released in 1980, the ACS’s new recommendations were the first guidelines by a major national organization to be rigorously based on evidence and mathematical analysis. They were also quite controversial, because they overturned common beliefs about such things as the ideal frequency for cervical and colon cancer screenings, the best age to begin mammograms, and the effectiveness of chest X-rays in screening for lung cancer.

Not surprisingly, the guidelines angered a lot of radiologists and physicians, who believed that early detection always saves lives. “But those concerns were easy to counter because of the strong evidentiary basis for the new recommendations,” Eddy says. This wouldn’t be the last time Eddy’s work courted controversy, but he doesn’t see himself as taking sides. As he views it, evidence-based medicine is neutral; it simply lets the evidence determine the policy. “All I was proposing,” he says, “is that we apply to clinical medicine the principles of the scientific method that were introduced back in the 1600s.”

While Eddy was pushing the need for high-quality research to support guidelines, coverage policies and performance measures, he also had to come to grips with the limitations of clinical trials: “They are expensive, can take decades to complete, and the technologies being evaluated can change while the trial is underway. It is not possible to answer all the important questions with clinical trials.”

True to form, Eddy didn’t let those limitations stop him. With the belief that mathematical models could answer questions much more sophisticated than the human mind could broach, in the 1990s he moved beyond disease-specific models. Eddy began to build a large-scale, clinically realistic model that spanned multiple diseases and included pertinent aspects of physiology, populations and health care systems. The work eventually culminated in a groundbreaking model called Archimedes, which he developed in collaboration with particle physicist Leonard Schlesinger and a team of mathematicians and physicists. Archimedes proved capable of simulating long-term clinical trials and performing prospective, blinded predictions of their results with startling accuracy. Supported by Kaiser Permanente, Eddy and Schlesinger founded a health care modeling company to make the model available to clinicians.

It’s been 25 years since Eddy began work on Archimedes and nearly 40 since he first helped write the ACS screening guidelines. Today there are entire departments devoted to evidence-based medicine. And while Eddy believes there is still progress to be made in how it’s practiced, he is gratified to see such a widespread commitment to its principles.

Reflecting on his own contributions to the field, Eddy traces a direct line back to his struggles in Exeter’s classroom — and he offers some advice to parents of children with ADD. “Take heart,” he says. “When you have a learning disability, you have to develop survival skills that go beyond anything that can be taught in a classroom. You become a listener, a questioner, a scrambler. You take risks. You try things that might not work. Later, when your brain catches up, you are likely to be more creative, more willing to take on disruptive challenges. Think of your child as learning non-traditional skills that will prove invaluable later in life.”

So said the man who helped start a revolution in medical decision-making. 

Editor's note: This article first appeared in the summer 2018 issue of The Exeter Bulletin.

More to Explore

Exonians Blane Zhu, Anne Brandes, Ginny Little and Mai Hoang (left to right) stand with poet Jill McDonough.

Poetic Promise

The Lamont Younger Poets Prize program expands the Exeter tradition of excellence in writing.