The assumption of non-informative censoring is widespread in epidemiologic literature. In most cohort studies, this assumption is not verifiable. In the case where censoring is informative use of standard approaches to describe survival data such as Kaplan-Meier yield biased estimates of the survival function. An expanded form of the redistribute-to-the-right algorithm represents a new method that could potentially deal with informative censoring. This study will: 1) expand redistribute-to-the-right algorithm methods to estimate the survival function in the presence of informative, right censoring; 2) under informative, right censoring, use bootstrap methods to estimate the standard error of the survival function estimated using the expanded algorithm; and 3) in MACS and ALIVE cohorts, assess the distribution of time from HIV seroconversion to death under five distinct censoring scenarios using the expanded algorithm. MACS and ALIVE represent ideal settings to apply this expanded method given that the bulk of time-to-event analyses performed in these cohorts assume non-informative censoring. As such, application of the expanded approach to data from these cohorts offers the opportunity to refine prior inferences made in these studies. ? ? ?
|Howe, Chanelle J; Cole, Stephen R; Chmiel, Joan S et al. (2011) Limitation of inverse probability-of-censoring weights in estimating survival in the presence of strong selection bias. Am J Epidemiol 173:569-77|
|Smith, Lorie B; Fadrowski, Jeffrey J; Howe, Chanelle J et al. (2010) Secondary hyperparathyroidism and anemia in children treated by hemodialysis. Am J Kidney Dis 55:326-34|