Analysis of Proportional Hazards Assumption in Cox Regression Model Using Residuals and an Application on Lung Cancer Patients

Abstract

Objective: To aim of this study was to investigate proportional hazards assumption in Cox regression model with application to patients with lung cancer. Material and Methods: Two-hundred twenty-seven patients with lung cancer were included in this study. Cox regression model was used to determine the prognostic factors that affect survival time in patients. Results: Using the Cox regression model we found that the factors that had an impact on survival with a statistical significance (p< 0.05) were clubbing, aortic invasion, staging of the tumour, T, N, M, factors according to the TNM classification, division of NSCLC patients into squamous cell carcinoma and others, and brain metastases. Conclusion: Cox regression model was investigated in terms of graph using statistical test and Schoenfeld residuals. The results suggest that the model holds proportional hazards assumption. The SAS macro was written for the analysis. Copyright © 2005 by Türkiye Klinikleri.