and so on, can constitute an important safety signal. Spontaneous reporting is subject to certain limitations, including underreporting, the influence of bias in reporting, lack of denominator data, and difficulties in attribution of association between reported event and drug exposure.

Little has been done to optimize the usage of AERS for drug safety signal detection until recently. The work of DuMouchel and others raised the real possibility of doing automated searches of AERS to identify possible associations worthy of further followup. These “data mining” techniques greatly increase the value of AERS data, and that of other spontaneous reporting systems. Developing more rigorous systems in which to investigate AERS signals or any other possible risks of interest is warranted and long overdue; such systems have the potential to improve the ability to develop safety information in a more rapid and more reliable manner. The Centers for Education and Research on Therapeutics (CERTs) are assessing the potential use of health care databases for enhanced identification of adverse drug events. But the addition of new tools such as the use of health care databases does not mean we should abandon the old, especially now that we have methods to substantially enhance the value of these older tools.

In 2004, FDA received 422,889 reports of suspected drug-related adverse events. Of those reports, 21,493 were MedWatch reports directly from individuals (about 15 percent of which came directly from consumers), 162,107 were manufacturer 15-day (expedited) reports, 89,960 were reports of serious events included in manufacturer periodic reports, and 149,329 were reports of other events included in manufacturer periodic reports. As described in Chapter 2, safety evaluators in ODS/OSE review case reports in their drug-class portfolios. That is necessarily very time-consuming. Electronic submission of adverse event (AE) reports makes the system more efficient and timely, although it is reported that only half of the AE reports are submitted electronically, so the AERS contractor must spend time in performing data entry before the information can be reviewed by the safety evaluators.

A safety evaluator receives about 650 electronic reports per month. Review of AE reports can sometimes identify rare or unusual events that require additional research to understand. The following are some drugs for which AEs were identified through AERS: terfenadine (torsade de pointes and sudden death), cisapride (torsade de pointes and sudden death), troglitazone (hepatic failure), infliximab (tuberculosis and opportunistic infections), and cerivastatin (rhabdomyolysis) (Wysowski and Swartz, 2005). Statistical approaches available for the analysis and display of AERS data (such as the WebVDME program) have received only limited use by CDER until recently. CDER staff have recently described how the use of a Bayesian statistical analysis would have confirmed the cerivastatin, rhabdomyolysis, and renal failure association after 6 months of postapproval use if it had



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