Application of nonlinear mixed models in Logistic regression with random effect in clinical trials.

Author: Dai-jing YUAN ¹ ; Zhi-xiong YANG
Author Information

1. School of Finance and Statistics, East China Normal University, Shanghai 200241, China. daijing.yuan@hotmail.com
Publication Type:Journal Article
MeSH: Binomial Distribution; Clinical Trials as Topic; methods; Humans; Logistic Models; Models, Biological; Nonlinear Dynamics
From: Journal of Southern Medical University 2010;30(8):1923-1929
CountryChina
Language:Chinese
Abstract:
OBJECTIVETo explore the application of nonlinear mixed models fitting logistic regression in clinical trials.
METHODSTwo clinical trials were selected to exemplify the method for fitting nonlinear logistic regression using nonlinear mixed models by running NLMIXED procedure in SAS.
RESULTSAll the parameters and their standard errors were estimated, and each factor could be properly interpreted.
CONCLUSIONNonlinear mixed models in which both fixed and random effects enter nonlinearly can fit nonlinear logistic regression. These models provide effective methods to analyze the binary data in clinical trials.