Evaluation of Parametric and Semi-parametric Models in Survival Analysis of Patients with Gastric Cancer

Background and One of the most common methods used to estimate the effects of explanatory variables on survival time, is Cox semi parametric model. However, under certain circumstances, accelerated failure time parametric models are superior to the Cox model. The purpose of this study was to assess the efficiency of parametric and semi-parametric models in survival analysis of patients with gastric cancer. In this retrospective study, we analyzed 249 medical records of patients attending Tooba clinic affiliated to Mazandaran University of Medical Sciences. We obtained information on the final status of patients viaphone calls.Parametric methods including Weibull, loglogistic and log-normal and semi-parametric Cox model was fitted on the data to identifythe factors reducing survival time The results showed that patients with primary progress of disease, surgery as a treatment and patients without metastasis had higher survival rate than patientsin otherstages of disease and treatment (P<0.05). According to the value of Akaike Information Criterion (AIC) in univariate and multivariate analyses, lognormal model had best fitness in these data. However, parametric regression model had fitted better than Cox semiparametric regression. In this study lognormal model had highest fitness in our data. However, there were no significant differences between values of AIC of these models.According to the results, applying parametric model is suggested instead of semi parametric model if there is enough information about survival time and trend of variation.