Adverse drug events: identification and attribution

The definition of an adverse drug event should be tailored to one's purpose in examining the incident. Although the more specific of these definitions is required for scientific evaluation of the link between drug and event, other less stringent definitions are usually adequate for clinical purposes. Knowledge about the safety profile of a drug in humans is limited at the time of marketing. The mechanisms for supplementing safety data during postmarketing include (1) the Spontaneous Reporting System maintained by the Food and Drug Administration, (2) formal projects to assemble safety data on larger or more complex populations, and (3) formal projects designed to answer specific research questions. Judgments about attribution can be no better than the data that support them. The criteria applied by the clinician to the individual adverse drug experience to determine association differ from those required to establish causation based on epidemiologic evidence. In most situations, regulatory action on drug recall should be based on epidemiologic evidence. This article will discuss the choice of a definition for an adverse drug event, examine the extent and nature of the safety data assembled on a drug at the time it is marketed, propose the best methods for collecting additional information after marketing, and designate factors to be considered in judging a drug to be causally related to an adverse event.