To Compare Survival Models And Their Application In Breast Cancer Patients

Atefeh TALEBI, Medical Sciences of Shahid Beheshti University, Iran

1 Phd Student. Department of Biostatistics , Shahid Beheshti University of Medical Sciences, Tehran, Iran
2 Department of Biostatistics & Epidemiology, School of Public Health, Hamadan University of Medical Sciences, Hamadan, Iran
3 Department of Biostatistics & Epidemiology, School of Public Health, Hamadan University of Medical Sciences, Hamadan, Iran
4 Department of Epidemiology, School of Health and Nutrition, Research Center for Health Sciences, Shiraz University of Medical Sciences, Shiraz, Iran
5 Graduate of Civil Engineering, Isfahan University, Isfahan, Iran
6 Department of Mathematics, Isfahan (Khorasgan) Branch, Islamic Azad University, Isfahan, Iran

Currently, cancer is one of the most important of problems in worldwide. Due to increasing of incidence in types of cancer, it is important to deal with them. Breast cancer is the most common cancer of women in worldwide. The cancer is the major cause of cancer deaths in women, so that more than one million new cases of cancer are diagnosed in women and 3/37% of cases lead to death in 2000. Asian women's risk of breast cancer of women in North America or Western Europe is 0.1 to 0.2. Breast cancer is not common under the age of 25; however, its incidence is rapidly increased to 50 years. To model the censored survival data is usually done with Cox regression in clinical research. One of the most widely used Cox proportional hazard model analysis of survival data. The Cox proportional hazards model is one of the most common methods to analyze of survival data. In short follow-up studies the assumption of a constant ratio is very reasonable. It means that an important assumption in this model is proportional of risk over time. However, it may not be very appropriate assumption in long-term follow-up studies and the time somehow influences the risk ratios. When the proportional hazards assumption is violated then Cox model is not reliable and other models should be considered such as extended Cox model with time-dependent variables, the frailty and cure models. We use a data set 1148 women having BC referred to Shiraz Namazi Hospital, south of Iran. The main object of this study was to evaluate and compare different models on the breast cancer survival in southern Iran and to find the key factors in the disease. The researchers used R-3.1.2 software to carry out statistical analyses.