Skip to main content
Article
Covariate adjustment using propensity scores for dependent censoring
Technical Report (2016)
  • Youngjoo Cho, University of Rochester
  • Chen Hu, Johns Hopkins University
  • Debashis Ghosh
Abstract
In many medical studies, estimation of treatment effects is often of primary
scientific interest. Standard methods for evaluating the treatment
effect in survival analysis typically require the assumption of independent
censoring. Such an assumption might be invalid in many medical studies,
where the presence of dependent censoring leads to difficulties in analyzing
covariate effects on disease outcomes. This data structure is called 'semi-
competing risks data'. In marginal modeling under semicompeting risks
data, an artificial censoring technique is a promising approach to handle
dependent censoring. However, continuous covariates with large variabil-
ities may lead to excessive artificial censoring, which subsequently results
in numerically unstable estimation. In this paper, we propose a strategy
for weighted estimation of treatment effects in the accelerated failure time
model. Weights are based on propensity score modeling of the treatment
conditional on confounder variables. This novel application of propensity
scores avoids excess artificial censoring caused by continuous covariates and
simplifies computation. Monte Carlo simulation studies and application to
AIDS and cancer research are used to illustrate the methodology.

Keywords
  • Observational study,
  • Perturbation,
  • Causal Inference,
  • Resampling
Publication Date
2016
Citation Information
Youngjoo Cho, Chen Hu and Debashis Ghosh. "Covariate adjustment using propensity scores for dependent censoring" Technical Report (2016)
Available at: http://works.bepress.com/debashis_ghosh/76/