Comparison of Cox regression and Weibull parametric model for survival analysis of patients with thalassemia

Beta-thalassemia major is a very severe blood disease, its Clinical signs are premature and appear from 3 to 6 months of age. It is one of the most common monogenic diseases in the world and in Iran, and if it is not diagnosed and treated during the first years of life, it will lead to death. In this study, to check the factors affecting the survival of patients with beta-thalassemia major the semi-parametric Cox and parametric Weibull models were used and their efficiency was compared.

In this longitudinal retrospective study, using the available sampling method, data from 300 patients with beta-thalassemia major referred to Zafar's thalassemia clinic in Tehran during the years 1994 to 2017 were used. To compare the models and select the best model, we used the Akaike information criterion (AIC). Data analysis was performed using Stata software version 13.

The values of the Akaike information criterion (AIC) for semi-parametric Cox model and the parametric Weibull model were 277.600 and 107.211, respectively. Based on both Cox and Weibull models, the variables of blood transfusion, age of desferal and iron deposition in the liver were significantly associated with the survival of patients. Based on Akaike's criterion, the efficiency of the Weibull model was better than the Cox model.

Considering the obtained values, the parametric Weibull model is more efficient than the semi-parametric Cox model for survival analysis of patients with beta-Thalassemia major