Online Program Home
My Program

Abstract Details

Activity Number: 660
Type: Contributed
Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 AM
Sponsor: Biopharmaceutical Section
Abstract #321103
Title: Evaluation of Logistic and Cox Regression Models Using Simulated Survival Data and Clinical Practice Research Datalink
Author(s): Jessica Kim * and Chenyi Pan and Clara Y. Kim and Esther Zhou
Companies: FDA and University of Virginia and FDA/CDER and FDA/CDER
Keywords: logistic regression ; Cox proportional hazards model ; Monte Carlo simulation ; post marketing safety assessment ; Clinical Practice Research Datalink

In safety analysis, binary outcomes are typically analyzed with logistic regression models, where a logistic link function is used to relate outcome to the response variable. It is well known that this model is applicable when the follow up duration is fixed. On the other hand, when the follow up varies across patients, the outcome of interest often becomes the time to event. Such outcomes can be analyzed with Cox proportional hazards regression models. In this article, the logistic model and Cox regression model are compared to provide a general guideline of how to choose between them in the post-marketing safety assessment. We evaluate the performance of these two models empirically based on Monte Carlo simulations using the parameter specifications described in the paper by Bender et al. and real data from Clinical Practice Research Datalink (CPRD) under different scenarios. We assess the strength and weaknesses of each model when both models are applicable to the data by comparing the models with regard to biasness of the estimates, confidence intervals, and power using simulation.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association