A global test for competing risks survival analysis
Statistical Methods in Medical Research, vol. 29, pp. 3666–3683
Abstract
Standard tests for the Cox model, such as the likelihood ratio test or the Wald test, do not perform well in situations, where the number of covariates is substantially higher than the number of observed events. This issue is perpetuated in competing risks settings, where the number of observed occurrences for each event type is usually rather small. Yet, appropriate testing methodology for competing risks survival analysis with few events per variable is missing. In this article, we show how to extend the global test for survival by Goeman et al. to competing risks and multistate models[Per journal style, abstracts should not have reference citations. Therefore, can you kindly delete this reference citation.]. Conducting detailed simulation studies, we show that both for type I error control and for power, the novel test outperforms the likelihood ratio test and the Wald test based on the cause-specific hazards model in settings where the number of events is small compared to the number of covariates. The benefit of the global tests for competing risks survival analysis and multistate models is further demonstrated in real data examples of cancer patients from the European Society for Blood and Marrow Transplantation.
Authors 4
-
Dominic Edelmann corresponding
Affiliation as printed
Division of Biostatistics, German Cancer Research Center, Heidelberg, Germany
-
Affiliation as printed
Division of Biostatistics, German Cancer Research Center, Heidelberg, Germany
-
Affiliation as printed
Biomedical Data Sciences, Leiden University, Leiden, The Netherlands
-
Affiliation as printed
Biomedical Data Sciences, Leiden University, Leiden, The Netherlands
Cited by 2 stored of 2
2 results
No patents citing this paper on Lens.org (checked 2026-10-11).