Assessment of death risk of breast cancer patients with joint frailty models

Breast cancer is among the most common cancers in women and prevalence of disease has currently been increasing. Therefore, the number of studies on this area has been increasing. In most studies heterogeneities between the individuals, the dependency between recurrence and death are not considered while investigating the risk factors affecting the survival time of patients. However, conducting standard survival analyses by evaluating the death, recurrence and metastases independently will result biased parameter estimates. The use of joint frailty models are recommended in such cases. The heterogeneities between the individuals are considered, and recurrent events may also be evaluated in modelling the death risks of individuals with joint frailty models. The aim of the present study was to investigate the relationship between recurrence and death in breast cancer patients considering the heterogeneity between individuals and to investigate the factors effective on death and recurrence using the joint frailty model. The secondary aim of the study was to predict the death risks using the dynamic estimation methods after evaluating all history of the patient. The data of 465 patients who underwent surgery in Oncology Institute of Istanbul Faculty of Medicine between 2009 and 2016 were used in the study. A positive correlation was detected between recurrence and death at the end of the study. The effects of tumor type, the number of lymph node after surgery, neoadjuvant chemotherapy receiving status, clinical stage, and surgery type factors were found statistically significant for recurrence, and the effects of Cerb-B2, tumor type, the number of lymph nodes after surgery, and neoadjuvant chemotherapy receiving factors were found significant for death. Brier score values used in the evaluation of the obtained estimation using the dynamic estimation method were found below 0.30. In conclusion, the death risks of even the individuals with the same characteristics are different from each other in patients when the frailty effect are significant. The use of joint frailty models is recommended when the recurrence, and death cases are associated, and heterogeneity effects are detected.