Competitive Risk Analysis of Thymic Carcinoma Based on the Surveillance, Epidemiology, and End Results Database

Abstract

In general, we use the classical Cox proportional hazards model to derive factors that affect the prediction of patients diagnosed with thymic carcinoma (TC); however, when competing risks exist, the results can be biased. This study aimed to build a competing risk model for patients with TC to explore a more accurate method for assessing the relevant factors affecting patient prognosis. We obtained data on patients with TC who met the inclusion criteria between 2004 and 2016 (with additional treatment fields) in the Surveillance Epidemiology, and End Results database. The cumulative incidence function and Gray’s test were used for univariate analysis, followed by the fine-Gray and Cox proportional hazards models for multivariate analysis. Of the 478 subjects with TC who were finally included, 284 (170 died from TC, and 114 died from other causes) (59.41%) died, and 194 (40.59%) patients were alive. Univariate Gray’s test results indicated that age, marital status, tumor size, summary stage (localized, regional, or distant), chemotherapy status, and surgery status significantly affected the cumulative incidence of the target event (P < 0.05). Multivariate competing risk analyses indicated that tumor size, marital status, summary stage, and surgery status were independent risk factors for the prediction of subjects (P < 0.05). This study explored a more accurate method to assess the prognostic factors of patients with TC. Our findings can contribute to the clinical development of more scientific and accurate treatment methods, providing benefits to the majority of patients with TC.