Up to one fifth of all new prescription drugs may ultimately be recalled or produce potentially harmful side effects. The safety of new agents cannot be known with certainty until a drug has been on the market for many years.

Harvard researchers based on an evaluation of the 548 drugs that were first marketed between 1975 and 1999. They looked up all drug recalls and scanned the Physician's Desk Reference, a commonly used source of drug information, for new warnings on side effects.

During the study period, 10% of new drugs either received new warnings or were withdrawn, with half of those developments occurring within 7 years after the drug first appeared on the market. Based on these results, a new drug has a 20% chance of being withdrawn or producing previously unknown side effects over a 25-year period.

Side effects from new drugs can have a widespread impact. Almost 20 million Americans took one or more of the five drugs that were withdrawn between September 1997 and September 1998.

The researchers found that 56 of 548 new drugs approved by the agency during the 25-year period were later subjected to so-called "black box" safety warnings or banned from the market altogether. FDA uses "black box" warnings on drug labels to warn physicians of potentially dangerous side effects or drug interactions.

In some cases, those side effects can be deadly: Since 1993, seven drugs that were approved, then later withdrawn, may have contributed to over 1,000 deaths.

Several high-profile drugs have been pulled from the market by regulators over the last few years. One drug, the antihistamine terfenadine, also known as Seldane, spent nearly 13 years on the market before being banned in 1998. Another, the gastrointestinal drug cisapride, was available for over 6 years. Both drugs were pulled because researchers discovered high rates of heart toxicity associated with their use.

Fewer than 1 in 10 adverse drug reactions are reported to the US Food and Drug Administration (FDA). As such, new drugs may be causing more harm than this study illustrates.

So the study is definitely an underestimate of what is going on.

Pre-approval studies to evaluate the safety and effectiveness of drugs may not use enough patients to detect all possible adverse effects. The studies may also exclude some patients, such as children or those with other diseases, who will ultimately receive the drug, thereby potentially failing to identify their specific risks.

JAMA May 1, 2002;287:2215-2220, 2273-2275