Researchers Suggest FAERS Improvements

Share

Researchers from the Novartis Institutes for Biomedical Research say that while the FDA Adverse Event Reporting System (FAERS) remains the primary source for drug postmarketing pharmacovigilance, it is largely un-curated, unstandardized, and lacks a method for linking drugs to the chemical structures of their active ingredients. They attempted to correct those problems by mapping drugs to their ingredients and using natural language processing to classify and correlate drug events.

Writing in an online eLife article, they identify four key observations from their work: 

  •          much of the potential signal in FAERS and related databases is obscured by chemical name redundancy that introduces false association and hides associations that would be significant on aggregation;
  •          FAERS reports tilt toward serious outcomes, partly owing to a confusion of adverse drug reports and outcomes;
  •          FAERS suffers from several forms of conflation; and
  •          once biases and conflations are corrected, the molecular mechanism of previously hidden adverse drug reports can be revealed.

An improvement that the researchers recommend is introducing automatic mapping of drugs and synonyms to ingredients, as they did in their work. “Alerts could be issued for indications where serious outcomes are comment and hard to distinguish from adverse drug reports,” they say. “Definitive drug-adverse drug report associations would require information on exposure…. The introduction of reporting automatic into FAERS would serve the dual purpose of reducing errors, such as misclassifying adverse drug reports and indications, and making the tool more interactive and rewarding for health professionals.”

Read more