Multivariate generalized mixed-effects models for screening multiple adverse drug reactions in spontaneous reporting...

Area of Science:

• Pharmacovigilance
• Computational toxicology
• Biostatistics

Background:

• Spontaneous reporting systems are crucial for drug safety surveillance.
• Traditional methods for detecting adverse drug reactions (ADRs) require repeated analyses for each drug-ADR pair.
• Existing quantitative measurements like proportional reporting rate (PRR) and reporting odds ratio (ROR) are computationally intensive for large datasets.

Purpose of the Study:

• To develop a novel, efficient method for estimating PRR and ROR for multiple ADRs simultaneously.
• To apply a generalized mixed-effects model for streamlined drug safety signal detection.
• To enable the simultaneous evaluation of multiple drugs and their associated ADRs.

Main Methods:

• A generalized mixed-effects model was developed for single-analysis estimation of PRR and ROR.
• The method was designed to analyze associations between drugs and numerous ADRs concurrently.
• The approach was extended to detect drug-drug interactions from concurrent medication use.

Main Results:

• Simulation studies demonstrated that the proposed method achieved comparable false-positive rates and sensitivity to traditional PRR and ROR.
• The method successfully identified known ADRs when applied to the FDA Adverse Event Reporting System database.
• The generalized mixed-effects model facilitated the simultaneous assessment of multiple ADRs across various drugs.

Conclusions:

• The proposed generalized mixed-effects model offers a simplified and efficient approach to drug safety signal detection.
• This method enhances the ability to analyze multiple drug-ADR relationships and drug-drug interactions simultaneously.
• The novel method improves the efficiency of pharmacovigilance by reducing the need for repeated individual analyses.