Regression Analysis of Doubly Censored Failure Time Data with Applications to AIDS Studies

Abstract

In many epidemiological studies, the survival time of interest is the elapsed time between two related events, the originating event and the failure event, and the times of the occurrences of both events are right or interval censored. We discuss the regression analysis of such studies and a simple estimating equation approach is proposed under the proportional hazards model. The method can easily be implemented and does not involve any iteration among unknown parameters, as full likelihood approaches proposed in the literature do. The asymptotic properties of the proposed regression coefficient estimates are derived and an AIDS cohort study is analyzed to illustrate the proposed approach.

A modified estimating equation for a binary time varying covariate with an interval censored changing time

“…Zhao et al (2004) also considered an estimating equation for a case where both response failure time and covariate are interval censored. Their approach was based on the estimating equation approach which Sun et al (1999) suggested to analyze doubly interval censored data. In a context of recurrent event data, Chen and Cook (2005) applied a marker process to time dependent interval censored data and proposed a joint analysis of marker process and recurrent events time process.…”

Imputación del instante de inicio de seguimiento en estudios longitudinales: aplicación a la infección por VIH

“…La alternativa de utilizar modelos de doblecensura comporta la necesidad de cálculos algo más complejos y no poder utilizar software estándar 10,11,12 . En el caso de disponer de información sobre un marcador de progresión como el CD4, lo que ocurre en una parte importante de los sujetos seroprevalentes, se puede imputar la fecha de seroconversión a partir de un modelo paramétrico, en nuestro estudio el propuesto por Muñoz et al 17 .…”

“…Existen formas de estimar directamente el período de incubación utilizando datos doblemente censurados, aunque conlleva ciertas complejidades de cálculo y no pueden utilizarse programas de análisis están-dar [10][11][12] .…”

“…Una es imputar, para cada individuo, la fecha de la seroconversión a partir de la estimación de la distribución de la incidencia de infección condicionada a sus fechas de VIH+ y VIH- [10][11][12][13][14][15]16 . La otra alternativa es utilizar imputaciones basadas en la correlación con marcadores de la progresión como los CD4 o la carga viral.…”

Estimation and goodness-of-fit for the Cox model with various types of censored data

“…For instance, the 'doubly censored data' considered by Kim et al (1993), Sun et al (1999) are completely different censoring scheme from what we consider here. In Kim et al (1993), the HIV data from heavily treated group are interval censored Case 2 data as described in Groeneboom and Wellner (1992) (see more discussion in Ren, 2003), while the HIV data from lightly treated group are interval censored Case 2 data combined with some exact observations, thus it is a partly intervalcensored data set as described in Huang (1999).…”