Premium
Analysis of safety data in clinical trials using a recurrent event approach
Author(s) -
Gong Qi,
Tong Barbara,
Strasak Alexander,
Fang Liang
Publication year - 2014
Publication title -
pharmaceutical statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.421
H-Index - 38
eISSN - 1539-1612
pISSN - 1539-1604
DOI - 10.1002/pst.1611
Subject(s) - censoring (clinical trials) , event (particle physics) , clinical trial , computer science , safety monitoring , statistics , log rank test , contingency table , data mining , medicine , survival analysis , mathematics , bioinformatics , physics , quantum mechanics , biology
As an important aspect of the clinical evaluation of an investigational therapy, safety data are routinely collected in clinical trials. To date, the analysis of safety data has largely been limited to descriptive summaries of incidence rates or contingency tables aiming to compare simple rates between treatment arms. Many have argued that this traditional approach failed to take into account important information including severity, onset time, and multiple occurrences of a safety event. In addition, premature treatment discontinuation due to excessive toxicity causes informative censoring and may lead to potential bias in the interpretation of safety events. In this article, we propose a framework to summarize safety data with mean frequency function and compare safety events of interest between treatments with a generalized log‒rank test, taking into account the aforementioned characteristics ignored in traditional analysis approaches. In addition, a multivariate generalized log‒rank test to compare the overall safety profile of different treatments is proposed. In the proposed method, safety events are considered to follow a recurrent event process with a terminal event for each patient. The terminal event is modeled by a process of two types of competing risks: safety events of interest and other terminal events. Statistical properties of the proposed method are investigated via simulations. An application is presented with data from a phase II oncology trial. Copyright © 2014 John Wiley & Sons, Ltd.