Many medical schools are adopting policies to address thorny issues between them and the pharmaceutical industry. I believe that the guidelines of most institutions fall short in a couple of ways.

Virtually no policies discuss relationships with research physicians who work in the pharmaceutical industry, many of whom worked in academia. It is important to give trainees access to clinician-researcher role models in the industry and guidelines on how to work within and with the industry. Policies on controlling access to drug company representatives and on research transparency should not preclude contact between trainees and physicians working in the industry.

Rules on conflicts of interest and openness about potential biases should be evenhanded; that is, physicians and other researchers from pharmaceutical industries should be treated the same as physicians and researchers on academic faculties with financial ties to the industries, which may also create conflicts of interest. Policies should address the fact that having a pharmaceutical company as a sponsor is not the only circumstance that may cause scientists to become biased about a drug—there is academic advancement, self-promotion, and self-aggrandizement—and there should be guidelines for research faculty who teach medical students and residents, just as there are guidelines for clinician-researchers who teach physicians at continuing medical education conferences.

Academic policies I am familiar with fail to address an essential issue: teaching trainees to read and evaluate research critically so they can be sensitive to many potential sources of bias, no matter who the sponsor is. At the very least, this type of analysis requires a thorough understanding of biostatistics and clinical study design. Trainees should be taught how to recognize features of the protocol that bias the results in a particular fashion, conclusions that go beyond what the data warrant, compelling reasons to think the trial results extend beyond the particular patient population in the trial, meta-analyses that combine results for different patient populations, and authors who are strong proponents of one side of a medical controversy.

There must be a mechanism to ensure that trainees' exposure to clinical trials is not limited solely to trials that authors decide to report and that journal editors decide to publish. Even the informed reader, skilled in critical assessment of the literature, will not detect problems unless a minimum standard of information is available—for example, inclusion and exclusion criteria, study type, sample size, primary outcome measure and key secondary outcomes, and the specific name of the drug or investigational compound. Requiring the public registration of all clinical trials would be a good policy for the medical community to endorse.