Empirical estimation of under-reporting in the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS)

Abstract

Background: To examine how closely reporting rates in the FDA Adverse Event Reporting System (FAERS) reflect expected rates of known adverse drug events (ADEs).

Methods: We selected three groups of drugs to reflect hypothesized variation in sensitivity to reporting, including statins, biologics, and narrow therapeutics index drugs (NTI). The numbers of ADEs in FAERS were divided by utilization estimates from ambulatory health care data (NAMCS/NHAMCS) to calculate a reported proportion. One sample z-test for proportions compared the proportion of ADEs reported to an expected ADE proportion derived from drug labels, reference databases, and peer-reviewed papers.

Results: The majority of drug-ADE pairs showed significant under-reporting. For example, roughly 0.01% to 44% of statin events were reported (z-test p < 0.0001). Biological (0.002% to >100%) and NTI (20% to >100%) drugs had relatively higher reporting rates. Roughly 20% to 33% of the minimum number of expected serious events were reported with biologics and NTI drugs.

Conclusions: This study supports previous evidence of under-reporting of ADEs in spontaneous reporting data. But, under-reporting varies considerably by the type of drug and the type of ADEs, and this variability in under-reporting should be considered when interpreting safety signals.

Keywords: Adverse event; FAERS; FDA; pharmacoepidemiology; under-reporting bias.