Without comprehending all of the outcomes data for different treatments, and adjusting for individual patients’ co-morbidities, physicians rely on personal judgment, what their peers do, along with rules and traditions received from their training. Simple habit and tradition explain, for instance, why so many doctors continue to order annual screening chest X-rays long after epidemiological studies have proved them worthless.

Ideally, EBM would deal with the many problems of collecting, interpreting, and disseminating data, serving as a kind of GPS for clinicians as they navigate the complexities of patient care. But in early 2008, physicians and the medical technology industry remain skeptical. While in principle they agree that the applications of treatments, procedures, and products should be based on evidence, they diverge on what this would mean in practice. Thanks to the Internet, scores of new clinical studies appear daily. How are these to be sorted, evaluated, correlated, and analyzed for quality and applicability? Which studies are valid? What constitutes sufficient evidence to dictate practice guidelines? What is one to do when competing medical interests introduce plausible evidence for conflicting claims? How can evidence from population-based studies be applied to individual patients? And, perhaps most important of all, how does one deal with factors that have nothing to do with evidence—financial interests, culture, values, politics—that support a medical practice?

For a cautionary example about how other factors can trump science, one need look no further than the debacle of high-dose chemotherapy and autologous bone marrow transplants for breast cancer. During the late 1980s, oncologists imagined they could treat advanced breast cancer by delivering extremely high doses of chemotherapy along with bone marrow transplants to mitigate toxicity. Dying women had no alternatives, and in the absence of any well-designed studies or controlled trials, experts insisted that such a regimen offered the only hope for patients. Insurers that refused to cover the $100,000-plus “experimental” treatments were successfully sued in court and pilloried in the press. It took a decade for government and private insurance companies to complete a series of definitive clinical trials that proved the procedure not only offered no benefit, but actually increased women’s risk of premature death. In the meantime, 20,000 women were treated at a cost of at least $2 billion. In this instance, the desire to do something for desperate patients combined with financial incentives for doctors and hospitals to push aside questions of evidence. Physicians who advocated this painful, ultimately useless, procedure, David Eddy remarked, “owe this country an apology.”

The need to improve the quality and cost of American medicine will drive the adoption of EBM. But resistance will be intense, which is easy to see from looking at areas where there are lots of high-quality data. One recent example: In May 2005, the U.S. Agency for Healthcare Research and Quality (AHRQ) published a comprehensive study of episiotomy (surgical resection of the skin between the vaginal opening and the anus). Obstetricians use this procedure in at least one-third of vaginal births in America, ostensibly to help their patents. But the AHRQ could find no evidence of clinical benefit, and much evidence of harms to mothers, from infection to incontinence. Unfortunately, anyone waiting for evidence that the AHRQ study has influenced obstetrics has been waiting in vain.

Perhaps the main obstacle to changing the current system is that it isn’t in most doctors’ interest to try to change it. Most episiotomies are done for the convenience of surgeons, who get out of the hospital faster and also bill for an extra procedure. As long as no major physician group or healthcare organization makes a dedicated effort to distinguish itself from the pack based on EBM, superior outcomes, and transparency, everyone has an economic incentive to provide services, regardless of benefit. More disturbing still, in purely economic terms, clinicians have real incentive not to know whether or not lucrative, albeit dubious, treatments work. The importance of fixing this broken system is widely acknowledged. Ultimately, doing so will require reinventing the medical information economy for the 21st century.