A Drug Similarity-Based Bayesian Method for Early Adverse Drug Event Detection

Area of Science:

• Pharmacovigilance
• Computational Toxicology
• Biostatistics

Background:

• Biochemical drug similarity aids in predicting adverse drug events (ADEs) and enhancing real-world data mining signals.
• Existing methods require improvement in false-positive control and time-to-detection evaluation.

Purpose of the Study:

• To test a drug similarity-based Bayesian method for early ADE detection with false-positive control.
• To evaluate the method's performance against reference methods using real-world data and simulations.

Main Methods:

• Employed a Bayesian approach where prior ADE probabilities for less frequent drugs are derived from biochemically similar, more frequent drugs.
• Utilized posterior probability of the null hypothesis for signal detection and false-positive control.
• Evaluated methods using the US FDA's Adverse Event Reporting System (FAERS) and a simulation study.

Main Results:

• The tested Bayesian method achieved ADE detection 2 years faster (5 years vs. ≥7 years) than reference methods in FAERS data.
• The method showed higher AUC values (0.57-0.79 vs. 0.32-0.71) and improved detection probabilities (0.31-0.60 vs. 0.11-0.41) in simulations.
• Demonstrated effective false-positive control and comparable performance across different drug similarity types in high-throughput ADE mining.

Conclusions:

• The drug similarity-based Bayesian method shows potential for accelerating ADE detection.
• This approach offers improved control over false-positive rates in pharmacovigilance.
• The findings support the utility of this method for early identification of adverse drug events.