BACKGROUND AND OBJECTIVE: Breast cancer is one of the most common cancers in women. The statistical methods in the survival analysis of these patients are accelerated time models and Cox model. The purpose of this study is to evaluate two models in determining the effective factors in the survival of breast cancer.
METHODS: The study was an analytical and cohort study of survival analysis. The 538 of the patients referred to Ramezanzade Radiotherapy Center who had breast cancer and recorded survival status as a census from the April 2005 until March 2012 in Yazd. and survived by phone call. The Kaplan-Meier estimate was used to describe the survival of the patients. The research variables included clinical and demographic factors. The choice of final variables in the model was done by the methods of diminishing the dimension and all possible Cox regressions by the acaian criterion. Then, the best accelerated time model was considered Getting different distributions was also determined by the Akayake criteria.
FINDINGS: The most effective Cox model among all Cox models was variables including Age, Her2 and Ki67 variables (AIC = 30270). The generalized gamma model was the most optimal accelerated time model (AIC 463.966). Her2 was significant in both accelerated and cox models(p-value<0.05), but the Ki67 variable was not significant. (p-value>0.05).
CONCLUSION: In both accelerated time- Generalized Gamma- models and Cox Models, the Her2 variable was identified as a risk factor for breast cancer and There is a positive impact on the risk of death and reduced survival.