Survival analysis of the thalassemia major patients using parametric and semiparametric survival models

Thalasemia Major is one of the most common anemia diseases that can be fatal if not promptly diagnosed. The survival analysis of these patients can be an appropriate strategy in determining risk factors for death in these patients. The purpose of this study was to choose the best model to determine the risk factors for death in patients with the thalasemia major using common methods in the survival analysis.

The data of this retrospective cohort study, with 296 patients with thalassemia major, was collected in 2004- 2013 in Zafar Clinic in Tehran. Akaicke information Criterion was used for comparison of the models and the choice 351of the best model. Data analysis was carried out with R3.0.2 software at the significant level of 0.1.

The values of the Akaicke information criterion (AIC) for the parametric weibull, frailty weibull, log-normal, log-logistic, Gompertz, gamma and the semiparametric Cox were computed and found to be 27.56, 29.56, 18.73, 23.39, 26.26, 68.10, 24.73, respectively. The mean survival time for men and women were 40.2 and 39.7 years, respectively. The Log-normal model showed that age, age at the first desferal injection, onset of blood injection, the patient's birthplace, mother's education variables were significantly correlated with patient survival.

According to the values of AIC, the parametric log-normal model was chosen and suggested as the best model.