[Grouped Cox regression model and its application in study of prognostic factors on cancer]

Cox proportional hazards regression model is the most popular multivariate regression model for analysis of survival data in medical follow-up studies and clinical trials, but it is unable to handle grouped survival data or large data sets with many tied failure times adequately. This paper explores the grouped proportional hazards regression model (GPH model) and its use in analysis of large data sets presented in life tables. By use of the data in a lung cancer follow-up study conducted in urban area of Shanghai, the authors give an example in detail for analysing prognostic factors of lung cancer by using GLIM.