Comparison of Data Mining Methods for the Signal Detection of Adverse Drug Events with a Hierarchical Structure in Postmarketing Surveillance.

Goeun Park,Seok-Jae Heo,Inkyung Jung,Heesun Jung

doi:10.3390/life10080138

Abstract

There are several different proposed data mining methods for the postmarketing surveillance of drug safety. Adverse events are often classified into a hierarchical structure. Our objective was to compare the performance of several of these different data mining methods for adverse drug events data with a hierarchical structure. We generated datasets based on the World Health Organization’s Adverse Reaction Terminology (WHO-ART) hierarchical structure. We evaluated different data mining methods for signal detection, including several frequentist methods such as reporting odds ratio (ROR), proportional reporting ratio (PRR), information component (IC), the likelihood ratio test-based method (LRT), and Bayesian methods such as gamma Poisson shrinker (GPS), Bayesian confidence propagating neural network (BCPNN), the new IC method, and the simplified Bayesian method (sB), as well as the tree-based scan statistic through an extensive simulation study. We also applied the methods to real data on two diabetes drugs, voglibose and acarbose, from the Korea Adverse event reporting system. Only the tree-based scan statistic method maintained the type I error rate at the desired level. Likelihood ratio test-based methods and Bayesian methods tended to be more conservative than other methods in the simulation study and detected fewer signals in the real data example. No method was superior to the others in terms of the statistical power and sensitivity of detecting true signals. It is recommended that those conducting drug‒adverse event surveillance use not just one method, but make a decision based on several methods.

Highlights

It is critical to detect signals of adverse drug reactions from real-world data early enough to protect public health
It is recommended that those conducting drug-adverse event surveillance use not just one method, but make a decision based on several methods
The type I error rates of the reporting odds ratio (ROR), proportional reporting ratio (PRR), and information component (IC) methods were relatively high for the standard cutoff and for all total sample sizes, which means that spurious detection could frequently occur even when there are no actual signals

Summary

Introduction

It is critical to detect signals of adverse drug reactions from real-world data early enough to protect public health. In Korea, for example, the Korea Institute of Drug Safety and Risk Management (www.drugsafe.or.kr) collects the information through a spontaneous reporting system. Through this system, anyone, for example, a patient who has taken the drug, a doctor, or the manufacturer, can report an AE. Anyone, for example, a patient who has taken the drug, a doctor, or the manufacturer, can report an AE They report information such as the symptoms of the AE, the date of onset, the name of the drug, the frequency and duration of the dose, patient information, and causality assessment information. As the causality can only be reported by medical experts, the information

Objectives

Methods

Results

Discussion

Conclusion