Kaplan-Meier estimation of the cumulative incidence of a competing risk

Antolini, Laura (2009) Kaplan-Meier estimation of the cumulative incidence of a competing risk. Biomedical statistics and clinical epidemiology, 3 (3). pp. 129-134. ISSN 1973-2430

[img] PDF
portiere.pdf - Published Version
Restricted to Repository staff only

Download (1MB) | Request a copy
Official URL: http://www.bmscej.com/


In this note we show as the nonparametric maximum likelihood estimator of the crude incidence of a competing risk can be obtained by applying the Kaplan-Meier estimator to the sub-distribution hazard. The proposed approach starts from expressing the crude incidence of the failure type of interest as the incidence of a new random variable where the time to the competing events are artificially moved to in nity. The cumulative incidence of this new variable is then estimated by the Kaplan-Meier estimator after proper definition of the risk sets. The number of subjects at risk for the type of failure of interest is adjusted accounting for the presence of competing events and for the presence of censoring by a suitable weighting. The presented developments provide a unifying vision of the problem of estimating overall and crude incidences in the presence of competing risks. Owing to this result, nonparametric estimation and inference on the crude cumulative incidence of a competing risk can be handled by the estimation and inference tools commonly used in the absence of competing risks provided that risk sets are properly adjusted.

Item Type: Article
Uncontrolled Keywords: Survival analysis, Competing risks, Nonparametric estimation, Gray's test, Sub-distribution hazard, Kaplan-Meier.
Subjects: 600 Tecnologia - Scienze applicate > 610 Medicina e salute (Classificare qui la tecnologia dei servizi medici) > 610.7 Medicina - Educazione e Insegnamento; Medicina - Ricerca > 610.72 Statistica medica
Depositing User: Emanuele Zinevrakis
Date Deposited: 07 Nov 2011 16:26
Last Modified: 29 Mar 2017 16:08
URI: http://eprints.bice.rm.cnr.it/id/eprint/3506

Actions (login required)

View Item View Item