In 1971, the AMA Council on Drugs announced the discontinuation of its Registry on Adverse Reactions.1 Although this approach to the collection and evaluation of clinical data was well conceived, the Registry did not serve as an early warning system for the detection of previously unknown adverse reactions to drugs, and, with a few exceptions, it did not bring to light useful new information on known adverse reactions. A parallel effort by the Food and Drug Administration has been similarly unproductive.
The failure of the Registry is attributable to a number of factors. Without doubt, underreporting was of primary importance. Since relatively few reports were received, only a small fraction of the total number of significant reactions became available for study. A second important factor was the inability to determine the number of patients who had received a drug, a deficiency that precluded making any reasonably accurate estimate of