trolled clinical trials; the scientists who review those data are well trained in evaluating them. Postmarket data are collected from a broader array of sources, including controlled clinical studies, as well as from single-arm observational studies that lack comparator data. Further, because it is difficult to determine how many people received a drug, benefit–risk calculations are complex, and therefore postmarket data are more likely to reflect safety problems than beneficial results. Like the scientists who review premarket data, those who review postmarket data are well trained in evaluating them. However, integration of the two datasets is difficult, and there have been reported tensions between reviewers in the FDA’s Office of New Drugs (OND) and Office of Surveillance and Epidemiology (OSE). As Dr. FitzGerald said, “We are all conscious of the limitations that apply to other people’s work. We are a little less conscious of those that apply to our own.” Discussion of the integration of pre- and postmarket review was framed by the recommendations of the IOM report listed above.


Dr. Tilson identified and discussed four major sets of operational challenges to the implementation of a life-cycle approach to drug review:

  • Methodological challenges—Knowledge of product safety does not readily build over time and in a linear fashion; rather, gaining such knowledge is a complex and nuanced process. The methodological challenges include understanding how to determine a calculus for benefit-to-risk balance, how to recalculate benefit–risk balance on an ongoing basis (see IOM Recommendation 4.5), how to monitor effectiveness (Recommendation 5.4), and how to manage risk and evaluate Risk Minimization Action Plans (RiskMAPs) (Recommendation 4.4).

  • Human resource challenges—Dr. Tilson echoed earlier comments stressing the need for more experts trained in epidemiology while also pointing out that improving the U.S. drug safety system will require an expanded workforce in all areas, not just epidemiology. He urged that a companion study be undertaken to consider not only who should do the work, but also where they should work (e.g., see Recommendation 3.4), what their competencies should be, what the mix of staffing should be, and what it will take to educate and train these new personnel, including the required academic infrastructure.

  • The need for evidence—Benefit–risk data from preclinical, clinical, and postmarket spontaneous reports are all limited with respect to their


This section is based on the presentation of Dr. Tilson.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement