Characterizing the FDA Adverse Event Reporting System (FAERS) as a Network to Improve Pattern Discovery.

Area of Science:

• Pharmacovigilance and Drug Safety
• Network Science and Data Mining
• Computational Biology and Bioinformatics

Background:

• Adverse events (AEs) in drug-safety monitoring often present as complex clinical patterns.
• Network analysis (NA) has been applied to systems like VAERS, but FAERS network characterization remains incomplete.

Purpose of the Study:

• To characterize the network properties of the FDA Adverse Event Reporting System (FAERS).
• To leverage network characteristics for improved adverse event pattern discovery in FAERS.
• To provide insights into the application of NA in drug safety research.

Main Methods:

• Represented FAERS data using preferred terms (PTs) and drugs as nodes, with interconnections as edges.
• Analyzed network properties including degree distribution, clustering, and diameter.
• Applied edge weighting algorithms based on report co-occurrence and clustering.

Main Results:

• The FAERS network comprised over 20,000 nodes (PTs and drugs) with millions of interconnections.
• FAERS subnetworks exhibited heavy-tailed degree distributions, high local clustering, and low diameters.
• A log-normal model better described the degree distribution than a power law.

Conclusions:

• Network-based techniques successfully identified known adverse drug reactions and clustering patterns.
• FAERS network characteristics show similarities to other AE reporting systems like VAERS.
• This systematic NA of FAERS provides a foundational understanding for its network structure and utility in drug safety.