New ModelBy Karen Southwick
Catalyst: Dr. David Eddy
In 1972, David Eddy, then 31 years old and halfway through a residency in cardiovascular surgery at Stanford University, made a rather uncomfortable personal discovery, one that would change his life forever. Though his father, grandfather and great-grandfather all were physicians, Eddy, himself, realized that while he loved the idea of healing people, the actual business of surgery didn't excite him at all. In fact, it repelled him. It was too imperfect an art, he thought, too grounded in arbitrary judgments rather than hard facts.
After witnessing a particularly grueling, unsuccessful surgery as an intern that year, Eddy recalls, he made the bold decision to drop out of his residency program. At best, life-and-death judgments were far too subjective, Eddy reasonedyet rarely did people question a doctor's decisions, regardless of how uninformed by data. "I would go into the [medical] literature to get the probabilities of how various treatments would turn out, but I couldn't find any numbers whatsoever," Eddy recalls. Instead, he says, he'd find "a series of dictatorial statements suggesting that if a patient has a particular ailment such as a breast lump, then you do treatment X. There was no weighing of benefits or harms. What physicians were doing back then was following simple if-then rules."
And that was just the half of it, Eddy says: Too little was known about patientshow the combination of their unique histories, their lifestyles, their hereditary backgrounds, along with the quality of physician care and specific medical interventions, all had a bearing on one other. Yet doctors were making serious treatment decisions without fully recognizing these interactions. "The lack of accountability to data was astounding," says Eddy. There had to be a better way, he thought. "I realized, quite dreadfully," he says, "that this haphazard approach made it depressing for me to be around sick people." Indeed, Eddy told a local medical society: "The intellectual foundation of medical care is based on the assumption that whatever a physician decides is, by definition, correct. While many decisions are, no doubt, correct, many are not, and elaborate mechanisms are needed to determine which are which."
Today, some 30 years later, Eddy has built just such an elaborate mechanisma complex software program he calls Archimedes, named for the ancient Greek mathematician who boasted he could move the world with a single lever. The software model, for the first time in medical history, uses mathematical algorithms and equations to translate the beat of a heart or the twitch of a muscle into the ones and zeros of the Digital Age, replicating in numbers the behavior of a disease and creating a "virtual reality" in which patients, providers and institutions interact as they would in the real world.
Put simply, Archimedes is to doctors what airline cockpit simulation software is to pilotsa way to manipulate the variables of human behavior and environment to mimic real-life outcomes. The model can analyze everything from the onset of disease in the cells of the pancreas to the number of parking lots in the medical center complexand the need for more lots five years from now. In other words, it's a complete model of systems as well as patients and diseases.
For example, using Archimedes, physicians can see the impact of age, sex, race, income, employment, behaviorto name just a few variableson human health, along with the influence of various treatment decisions, and do it all at the accuracy level of traditional clinical trials that can last years and involve thousands of real-life patients. Want to know the chances a 29-year-old male with cancer and who smokes will respond better to one treatment over another? Now, Eddy says, some trials that used to take years can be run by Archimedes in a half-houror less.
But it's not just about potential savings for cash-strapped hospitals and medical institutions. Eddy believes evidence-based care, as represented by Archimedes, may one day help patients determine the affect of a drug, itself, over time on individual patientsand help prolong the lives of those with heart disease or HIV or diabetes. Ultimately, he says, it's all about delivering the best treatment to the right people at the right price, a mission analogous to the original dream of managed care that HMOs like Kaiser Permanente are still struggling to provide.
Naturally, Eddy's groundbreaking project is raising eyebrows. "What David has done is really important," says Alain Enthoven, Ph.D., senior fellow at Stanford's Center for Health Policy, and one of the pioneers of managed care. Enthoven praises Archimedes' accuracy at mimicking actual outcomes of clinical trials; the American Diabetes Association (ADA) is already using it to devise treatment guidelines so as to prove whether lifestyle changessuch as improved diet and weight lossoffer better results than merely popping pills.
But not all physicians are sold on Eddy's workafter all, it's their authority that the Archimedes program is targeted to topple. Even Richard Smallwood, a longtime admirer of Eddy's and his adviser at Stanford, says most people don't think "as logically as Dave does," particularly when it comes to a field as emotionally charged as medical care. "Dave might have a really good system, but selling it to the rest of the world will be tough," he says. Peggy O'Kane, president of the nonprofit National Committee for Quality Assurance (NCQA), which rates HMOs based on quality, agrees that political opposition to Archimedes could be its Achilles' heel. Resistance will come from many quarters, she says, including physicians who have their own way of doing things, medical manufacturers who sell high-priced technologies, and consultants promoting their own pricey systems for healthcare reform.
Healthcare CIOs, too, question the politics of it. Says Terry Jackson, corporate director of information systems at Trover Foundation, a healthcare services organization in Madisonville, Ky.: "Any time you get down to a clinical system like that, the physicians are going to dictate whether or not they're going to use it. They are partners in providing care. You don't go to your partners and force them to do things that they don't want to."
Eddy acknowledges he's stepping on toesfrom those of pharmaceutical manufacturers whose pills Archimedes may judge to be no better than less-expensive generics, to trial lawyers sniffing out medical malpractice suits.
But Eddy has always been a bit of an upstart: Friends recall that, as a Ph.D. student in mathematics at Stanford, Eddy wrote a monograph that pointed out errors in physician decision-making. None of the leading medical journals would publish it, but others who had been nurturing ideas that would develop into the concept of managed care sat up and took notice. At a cocktail party, Enthoven gave a copy of Eddy's monograph to an advisory board member at the National Cancer Institute, who later invited Eddy to speak at an NCI meeting. Eddy's impassioned, outraged case against the unquestioned judgment of doctors so impressed Arthur Holleb, then executive director of the American Cancer Society (ACS), that a few weeks later, the ACS sought Eddy's help in designing new cancer-screening guidelines, foreshadowing his later work. In the research for his thesis, he had built a general mathematical model for "intermittent inspections," essentially a screening regimen for potential health problems. His work for the ACS expanded upon that, inspiring changes in cancer screening that are still in use today, such as when to do Pap smears for cervical cancer and sigmoidoscopies for colon cancer. For this research, the Operations Research Society of America awarded Eddy its prestigious Lanchester Prize in 1980, in recognition of the most important contribution to the field from applied mathematics.
In the years that followed, Eddy kept up his pressure on traditional medicine, lecturing at Stanford and Duke universities about the application of mathematics to medicine (he was a full professor at Stanford and the J. Alexander McMahon Professor of health policy and management at Duke), and in 1990, he was asked to join the nonprofit Kaiser Permanente system, headquartered in Oakland, Calif.the largest HMO in the country. It proved to be a critical alliance: Eddy, with his desire to create practice guidelines, and Kaiser, spearheading the drive to deliver better care at better prices.
To help him develop the sophisticated mathematical models he'd need, he turned to Len Schlessinger, an assistant vice president of applied physics at a local consulting firm. Though Schlessinger's work involved mathematical modeling of physical phenomenon like radio waves and electrical fields, Eddy convinced Schlessinger that he could apply the same thinking to healthcare. During their first meeting in 1993, Schlessinger recalls, he was so excited by Eddy's vision that he joined Kaiser on the spot. "I had this long experience in building mathematical models and applying them to physical situations," he says. "Here was an opportunity to apply that to a brand-new field where it could have a gigantic impact."
Together, Eddy and Schlessinger, often spreading out their work across the broad, empty tables of Kaiser's first-floor cafeteria in Pasadena, began hashing out the concept of Archimedes and how it would incorporate the physical characteristics of people with a disease, plus the progression of the disease itself, creating a virtual representation of an affected populationsay, diabetics or heart patients.
There were obstacles. For starters, they needed massive computer power to run the complex model, but supercomputers leased from Sandia National Laboratories weren't up to the job, and the lab's personnel didn't have the modeling skills needed. They tried Argonne National Laboratory next, but it, too, could not produce the needed algorithms. At that point, Eddy and Schlessinger decided to go it alone. They made two critical decisions: In place of a supercomputer, they would employ a series of PCs to run the mathematical modelusing so-called grid computingand they would bring the simulation research completely inside Kaiser and do it themselves. "I'd heard at one of the defense labs about a mathematical model of the European war theatre used by the U.S. Defense Department in WWII," Eddy says. "They'd model every jeep, every town, every hill, even every round of ammunition and whether the roads were asphalt or concrete because that would affect the speed of vehicles in the rain or shine. It was incredibly detailed, and Len and I said to ourselves, 'That's exactly where we want to go with healthcare.'" Schlessinger did the math, and the pair was on its way.
Bleeding Red Ink
Bleeding Red Ink
Some years later, the team faced another challenge: Kaiser, like the rest of the managed care industry, was bleeding red ink and looking for ways to cut costs. The Southern California region balked at continuing to pay about $500,000 annually for the Eddy-Schlessinger project. Recalls Les Zendle, associate medical director for the Southern California Permanente Medical Group: "There were several people at the time who thought the project was too long-term. They said it was wonderful work but the payoff was too far in the future."
Fortunately for Eddy, Kaiser's then-newly formed Care Management Institute, devoted to the very ideals of evidence-based medicine that Eddy had been preaching for years, came to the rescue. Thanks to the backing of all-important physician champions like CMI director Paul Wallace and Jed Weissberg, the associate executive director for quality for the medical groups in all of Kaiser's seven regions, Archimedes was back on track. By 2000, Eddy and Schlessinger had translated into mathematical equations the progression of four chronic diseases that are among the most expensive for doctors to treat and patients to pay for: diabetes, asthma, coronary artery disease and congestive heart failure. The following year, in 2001, the ADA, with funding from Bristol-Myers Squibb, signed a multiyear contract to use Archimedes to develop new treatment guidelines for diabetes.
What's next? In two years, Eddy says the team hopes to make Archimedes available on the Web. Eddy plans to cover the cost of maintaining and updating the system by offering it on a pay-per-use basis to a variety of organizations like HMOs, government organizations and charities, which could, in turn, offer it to individual patients. Institutional users like health systems might pay $50,000 to $100,000 a year for broader access. Drug companies and government organizations that want help in designing trials might pay even more. "The big money is on the clinical trials side," Eddy says, and could be used to help subsidize low-cost or free access for patients and doctors.
But Eddy's longstanding disdain for the healthcare system remains his key motivator. Despite spending an average of $5,000 for healthcare for every person in the U.S., he says, the system continues to keep many from receiving quality care, such as the 40 million uninsured Americanswhile at the same time pushing others toward needlessly expensive, sometimes harmful treatments. "We've got a $1.5 trillion tank rolling down the road with its windshield fogged over," Eddy says. What Archimedes aims to do, he says, is harness the power of information to give the system " a windshield wiper." To be sure, as long as Eddy is driving reforms, change can't be far behind.
Karen Southwick is a San Francisco-based technology and health writer and a former executive editor of Forbes ASAP magazine.