Statistical comparison of survival models for analysis of cancer data

The Cox Proportional Hazard model is the most popular technique to analysis the effects of covariates on survival time but under certain circumstances parametric models may offer advantages over Cox's model. In this study we use Cox regression and alternative parametric models such as: Weibull, Exponential and Lognormal models to evaluate prognostic factors affecting survival of patients with stomach cancer. Comparisons were made to find the best model.

To determine independent prognostic factors reducing survival time for stomach cancer, we compared parametric and semi-parametric methods applied to patients who registered in one cancer registry center located in southern Iran using the Akaike Information Criterion.

Of a total of 442 patients, 266 (60.2%) died. The results of data analysis using Cox and parametric models were approximately similar. Patients with ages 60-75 and >75 years at diagnosis had an increased risk for death followed by those with poor differentiated grade and presence of distant metastasis (P<0.05).

Although the Hazard Ratios in the Cox model and parametric ones are approximately similar, according to Akaike Information Criterion, the Weibull and Exponential models are the most favorable for survival analysis.