Multiple imputation approaches to regression analysis of interval-censored failure time data

Abstract

[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT REQUEST OF AUTHOR.] This dissertation discusses regression analysis of interval-censored failure time data, which occur in many fields including demographical, epidemiological, financial, medical, and sociological studies (Sun, 2006). It consists of three parts. The first part considers regression analysis of current status data under the additive hazards model and in particular, we considered the situation where the observation times depend on covariates. The second part considers regression analysis of interval-censored failure time data under the additive hazards model and time-dependent covariates. The third part considers regression analysis of interval-censored failure time data under the linear transformation model. For these situations, we proposed a general semiparametric method based on multiple imputation for inference under the regression models. This multiple imputation converts the analysis of interval-censored failure time data to that of right-censored failure time data. A major advantage of the approach is its simplicity and it can be easily implemented by using the existing software packages for right-censored failure time data. Extensive simulation studies are conducted and indicate that the approaches perform well for practical situations and are comparable to the existing methods. Real data applications are provided and model checking is discussed.