A Data-Driven Reference Standard for Adverse Drug Reaction (RS-ADR) Signal Assessment: Development and Validation

Area of Science:

• Pharmacovigilance and Drug Safety
• Real-World Data Analysis
• Computational Epidemiology

Background:

• Real-world data (RWD), including multicenter electronic health records (EHRs), generate numerous adverse drug reaction (ADR) signals.
• Lack of a validated reference standard hinders the assessment of computationally detected ADR signals.

Purpose of the Study:

• To develop a comprehensive reference standard for ADRs (RS-ADR).
• To facilitate systematic detection, assessment, and understanding of drug-ADR associations from RWD.

Main Methods:

• Integrated established drug-ADR reference sets (e.g., Side Effect Resource, OMOP, EU-ADR).
• Developed a pharmacovigilance dictionary and annotated EHR data.
• Utilized MetaLAB, MetaNurse, and FDA Adverse Event Reporting System (FAERS) data for cross-institutional validation.

Main Results:

• The RS-ADR includes 1344 drugs, 4485 ADRs, and over 6 million drug-ADR pairs with consensus votes.
• Successfully validated novel ADR signals, such as famotidine-related hepatic dysfunction.
• A web-based interface is available for querying RS-ADRs.

Conclusions:

• The RS-ADR, enhanced by pharmacovigilance dictionaries and RWD, can streamline ADR signal detection and evaluation.
• This standard is crucial for improving the causality assessment of ADRs identified through computational methods.
• Future iterations will improve with increased RWD participation and advanced algorithms.