Competing Risk Models for Predicting Prognosis of Cervical Cancer Patients based on Surveillance, Epidemiology and End Results (SEER) Database

Abstract Background: The objective of this investigation was to ascertain precise prognostic determinants for cervical cancer through the utilization of a competing-risks model that relied on data procured from the Surveillance, Epidemiology, and End Results (SEER) database. Methods: This study abstracted data related to cervical cancer patients from 2000 to 2018 from the Surveillance, Epidemiology, and End Results (SEER) database. The univariate analysis used a cumulative incidence function with the Gray test and a Fine-Gray specific cause (CS) and Cox proportional risk model. Results: Among the 11424 eligible cervical cancer patients, 2603 patients were found to have died of cervical cancer, while 1153 patients were found to have died from other causes. Meanwhile, a univariate Gray test established age, race, marital status, pathological type, primary site, degree of differentiation, American Joint Committee on Cancer (AJCC) staging, T stage, lymph node involvement, metastasis, tumor size, regional lymph nodes examined, regional lymph nodes positive, surgical status, regional lymphadenectomy, radiation status, and chemotherapy status all significantly influenced the amassed incidence of events of interest (P<0.05). Multifactorial competing risks analysis demonstrated that age, race, marital status, pathology type, Grade, AJCC stage, T stage, lymph node involvement, metastasis, surgery, regional lymphadenectomy, and chemotherapy status were independent risk factors affecting postoperative prognosis in patients with cervical cancer (P<0.05). Multifactorial Cox regression results differed: lymph node involvement was not an independent risk factor. Conclusions: It was found that prognostic factors for cervical cancer were identified more accurately using competing risk models than traditional methods.