or millions of people use the drug (GAO, 1990; Okie, 2005; Racoosin, 2006). For example, bromfenac, a non-steroidal anti-inflammatory drug (NSAID) marketed for 11 months in 1997–1998, was found to have serious and sometimes fatal liver toxicity in about one in 20,000 people who used the drug (Friedman et al., 1999); the NDA clinical trial base would have had to include 60,000 patients to detect such an effect before marketing (Friedman et al., 1999).

Preapproval clinical trials also have little information on the effects of long-term exposure to the drug due to their often short duration. Furthermore, clinical trials usually do not represent the full array of patients who will use the product once it is approved. Trials often exclude patients with comorbidities or those taking other medications, although both may be common among future users of the marketed drugs. Elderly patients, ethnic and racial minorities, and the very sick are underrepresented, and pregnant women are generally excluded from trials. Drugs generally have not been tested in children as part of the NDA, although patent-extension incentives are aimed specifically at encouraging pediatric testing in children (Meadows, 2003).

Those limitations are inherent in the system and cannot be changed without adding considerably to the time and expense of drug approvals, which would delay patient access to potentially beneficial drugs. It is generally understood that it is not routinely realistic to require premarket trials on tens or hundreds of thousands of subjects. Thus, inherent in the fundamental design of the drug approval system is the delayed availability of important safety data until a drug is used in larger and more diverse populations after marketing. That approach means that the initial postmarket period is a critical time for developing a fuller understanding of a drug’s safety profile.

Premarket clinical trials are designed primarily with efficacy. Safety issues sometimes surface, but the challenge is the possibility of unusual, unexpected, undocumented risk. If sponsors and CDER reviewers are not vigilant about identifying and pursuing safety signals in the trials, the opportunity to evaluate safety in the premarket trial period may be lost.

In the premarket period there usually is a shortage of information on how a new drug compares with other treatments for the same indication. Sponsors are not routinely required to submit such comparative trials to obtain approval. Once a drug is on the market, it can be difficult to compel sponsors or others to undertake appropriate comparative trials. Sponsors usually do not initiate such trials unless they believe that their product has a readily identified or demonstrable advantage. A postmarket comparative trial of newer hypertension agents—angiotensin-converting enzyme (ACE) inhibitors—against older diuretic drugs, for example, found the older drugs to be more effective in reducing blood pressure (Appel, 2002). In addition, comparative trials are expensive, and cost-benefit considerations are not