Hypergeometric hypothesis test-based quantification of drug and adverse drug reaction association

摘要

The article is intended to launch a discussion on a new method to quantify associations between drugs and adverse drug reactions (ADRs). Based on the contingency table for a drug and ADR, we proposed a method to evaluate the strength of a pairwise association using a hypergeometric distributions-based hypothesis test. In contrast to existing frequentists methods, the main advantage is that the interpretation of the drug and ADR inter-relationships is established upon a theoretical statistical distribution without any arguable assumptions. Based on results from experiments with Adverse Event Reporting System (AERS) data, a threshold with P-value = 0.01 is suggested for the hypothesis test, which is comparable to α = .05 for PRR and ROR.