Statistical Methods for Assessing the Effect of Mortality on Rates of Change and Variability in a Longitudinal Study of the Elderly Public Deposited

Analytics

Downloadable Content

Download PDF
Request Version for Screen Reader
Last Modified
  • March 19, 2019
Creator
  • Douglas, Christian
    • Affiliation: Gillings School of Global Public Health, Department of Biostatistics
Abstract
  • Despite the benefits of longitudinal analysis for describing the aging process, it is not absent of complications. Failing to account for nonrandom attrition and other mechanisms that affect the ability to acquire follow &ndash up measurements may result in estimates on a relatively healthy or advantaged sample in terms of health and economic means. In modeling the process of aging in older adults, handling of attrition requires careful attention, since attrition can affect the interpretation of the conclusions. Longitudinal studies of older adults are particularly sensitive to the truncation due to death, which is usually the largest category of nonresponse in studies of older adults. We examine the effect of death on rates of change and variability on a well &ndash established data set of older adults leaving in the community. Our assessment utilizes models proposed to analyze data with outcomes truncated due to death. Using proposed methods, we analyzed an imputed NC EPESE dataset allowing only truncation due to death. Simulations were completed to evaluate the models &rsquo ability to estimate the rates of change under varying burdens of death. Additionally, the use of these methods in presence of non-participation and death was examined using the original NC EPESE. Allowing the missing mechanisms to depend on the outcomes of interest, simulations were conducted to describe the methods behavior in estimating rates of change for non-missing completely at random data. Finally, an assessment of the variability about the parameter estimates was completed. Sample size and missing completely at random burdens of death were not extremely impactful on the models' ability to estimate the rates of change. However, this was not true for not missing at random data for estimates of rates of change or variability.
Date of publication
Keyword
Subject
DOI
Identifier
Resource type
Rights statement
  • In Copyright
Advisor
  • Preisser, John
  • Suchindran, Chirayath
  • Sen, Pranab Kumar
  • Edwards, Lloyd
  • Dilworth-Anderson, Peggye
Degree
  • Doctor of Public Health
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2014
Language
Publisher
Place of publication
  • Chapel Hill, NC
Access
  • There are no restrictions to this item.
Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
Douglas_unc_0153D_14890.pdf 2019-04-10 Public