Meta-Analysis of Drug Safety Data with Logistic Regression

This article reviews some basic principles of meta-analysis techniques for comparative clinical trials data and also discusses the use of logistic regression when comparative data are not easily available. In comparing a treatment drug and a control drug, a common problem in meta-analysis is that not many comparative clinical trials are available in the literature on the selected pair of drugs. There are, however, many clinical studies done separately on each of the drugs of interest. The situation is discussed where conventional meta-analysis results might not produce significant findings due to relatively few comparative studies involving the control drug of interest, while the logistic regression done by pooling all available individual and comparative trials could provide some insights on how the treatment and control drugs are different after controlling on covariates such as study design, patient age groups, disease groups, and so forth.