Comparison between classic survival analysis methods and competing risks for analysis of competing events

Autores

  • Flávia Silva Corrêa Tomaz UNIVERSIDADE FEDERAL DE OURO PRETO
  • Sebastião Martins Filho Universidade Federal de Viçosa.
  • Leandro Roberto de Macedo Universidade Federal de Juiz de Fora
  • Cristiane Márcia dos Santos Universidade Federal de Ouro Preto

DOI:

https://doi.org/10.15361/1984-5529.2019v47n2p183-190

Resumo

This study compares two different methods for survival data analysis in the presence of competing events. The first method focused on standard survival analysis, more specifically on obtaining cumulative incidence by using the Kaplan-Meier estimator, modeling the effect of covariates by fitting the Cox proportional hazards model. Com­peting events were treated as censoring events. The second method, called competing risks, emphasized the achievement of cumulative incidence, modeling the effect of covariates based on the cumulative incidence func­tion and the Fine and Gray model, respectively. To illustrate and compare these two methods, we used data on racehorse injuries. This study considered the following events: injuries due to claudication (main event) and inju­ries due to other causes (competing event). The results indicated that the incidence for each of the events was overestimated when using the Kaplan-Meier estimator. Moreover, the modeling of covariate effects on specific risk fitted by the Cox model did not correspond to the effect on the incidence of this event fitted by the Fine and Gray model.

Biografia do Autor

Flávia Silva Corrêa Tomaz, UNIVERSIDADE FEDERAL DE OURO PRETO

Departamento de Ciências Administrativas, Área: Métodos quantitativos.

Publicado

04/06/2019

Edição

Seção

Estatística - Statistics