drugs, especially to identify rare or unsuspected safety problems. Controlled phase 4 studies will remain important for verifying that drugs approved on the basis of limited exposures and surrogate end points actually have health benefits and for assessing whether common adverse events can be attributed to a drug when such events (such as heart attacks in older adults) emerge as a potential safety signal. This chapter underscores the importance of generating strong science to support regulatory decision-making about the risks and benefits associated with drugs and the importance of ensuring that the decisions made throughout a drug’s lifecycle are credible and transparent.

GENERATING THE SCIENCE

Understanding Risk and Benefit for Approval Decisions

As has been described in Chapter 2, a New Drug Application (NDA) and the reviews of an NDA by CDER staff contain thousands of pages of information about the effects of a drug. CDER clinical reviewers are expertly trained to analyze the efficacy and safety data from clinical trials. Individual case reports of adverse events from the trials are reviewed, as are comparisons of event rates of many safety outcomes in the overall product database, including those in uncontrolled safety studies. The reviewers also consider the statistical methods used by the company to generate the results. CDER has issued many guidances and documents of policies and procedures outlining the best ways to review and analyze such data (DHHS et al., 2005; FDA, 2005c). Clinical trials are designed to test hypotheses that are the agreed-on bases for determining efficacy. Trials designed to test hypotheses about serious safety outcomes would in most cases require many more subjects than are needed for an efficacy endpoint. For some conditions, the efficacy outcomes may be surrogate endpoints, which are expected to capture the information about efficacy but are usually not informative about safety.

Safety information can emerge from clinical trials, but rare events may not surface at all; if they do, it is at a rate so low that one cannot distinguish a drug-caused event from one expected by chance (background incidence). Safety information is usually limited to reports of common adverse events, the relation of which to drug exposure can be assessed by comparing rates between study treatment groups, or adverse events already predicted by results of animal studies or in connection with other drugs in the same class. Safety information also includes abnormalities in clinical laboratory test values seen during preapproval trials that may portend occasional clinically significant events. That set of suspected adverse events serves as a starting point for decisions about postmarket surveillance and drug safety research. The safety profile of a new molecular entity (NME) is especially uncertain, because of a lack of information on similar drugs already on the market.



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