Estimating Time-Varying Treatment Effect for Recurrent Childhood Diseases

Public Deposited
Analytics

Downloadable Content

Download PDF
Request Version for Screen Reader
Last Modified
  • March 19, 2019
Creator
  • Amorim, Leila
    • Affiliation: Gillings School of Global Public Health, Department of Biostatistics
Abstract
  • Many medical studies involve the occurrence of recurrent events,such as times to opportunistic infections among AIDS patients. In particular, this doctoral research has been motivated by the need for analyzing the effect of vitamin A supplementation on recurrent diarrheal episodes from a randomized community trial conducted in a cohort of 1,240 children, aged 6-48 months at baseline, in Brazil. Rate models have been used to analyze such type of data, where the rate of recurrence is modeled as a function of observed covariates and the effect of the covariates is assumed to be constant. Preliminary analysis of the vitamin A study suggested that the effect of vitamin A supplementation on diarrhea may change over time. It is important to develop methods to estimate such time-varying effects. Hence, the main purpose of this research is to develop statistical methods that incorporate time-varying coefficients in modeling recurrent time-to-event data. Rate models with time-varying coefficients are proposed to analyze recurrent time-to-event data. B-splines are used for the estimation of regression time-varying coefficients using two approaches: regression and penalized splines. Estimation of smoothing parameter, number and placement of knots is discussed. The small sample properties of the estimators are studied via simulation. Data from the vitamin A study is analyzed using the proposed methods. Another focus of the dissertation research is on the comparison of statistical methods for recurrent event data with dependent or informative censoring. Many statistical methods assume independent censoring. However, this assumption may not hold in some studies. Two methods have recently been proposed to account for dependent censoring for marginal rate models with recurrent event data. The first approach was developed by Wang, Qin and Chiang (2001), who proposed to model the occurrence of recurrent events by a subject-specific nonstationary Poisson process via a latent variable. In the second approach Miloslavsky, Keles, van der Laan and Butler (2004) proposed inverse probability of censoring weighted (IPCW) estimators for the regression parameters in the proportional rate model in order to obtain consistent estimators in the presence of dependent censoring. These two methods are critically compared through extensive simulation studies.
Date of publication
Keyword
DOI
Resource type
Rights statement
  • In Copyright
Advisor
  • Cai, Jianwen
  • Zeng, Donglin
Degree
  • Doctor of Public Health
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2006
Language

Relations

Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
file details umi-unc-1045.pdf 2019-04-08 Public Download