Estimating and Analyses the hemodialysis Patients from Date of Diagnosis until Death Using Parametric Models -The Governmental Hospitals in Khartoum State (2015 -2005)

Abstract

This research aimed at comparing different models of parametric Proportional Hazards (PH) models mainly (Weibull, Exponential, Gompertz) and Accelerated Failure Time (AFT) models mainly (Lognormal and log logistic) in patients with hemodialysis to determine the best model for assessing the survival of patient and identify significant risk factors for mortality. Recently in Sudan the end –stage renal disease (ESRD) has become a major health problem .The Study consists of 325 hemodialysis patients who were collected from the records at governmental hospitals in Khartoum State in the period from December 2005 to December 2015. Data was used to estimate the survival function with view to identify risk factors such as (age ( date of diagnosis of the disease) , Sex , Marital Status, Education status, occupation, Address, regular , Dialysis frequency per week, Hospitals , Diabetes Mellitus, Hypertension , polycystic kidney disease, Renal obstructions, Shrunken kidneys, Uncertain, Other) influencing among the end-stage renal disease (ESRD) population. The result show that the univariate and multivariate analysis, According to hazard ratio and time ratio, the variables including age, diabetes mellitus, diabetes mellitus +hypertension, urea and serum creatinine were considered to be highly significant factors and increased the risk of death in patients (shorter survival) so that they could influence survival in hemodialysis patients in the five models used in this research. Whereas other factors, such as regular, hospital, hypertension, shrunk kidneys, dialysis frequency per week, other have decreased the risk of death (longer survival) and have a direct effect on the survival of the hemodialysis patient. The median overall survival time was estimated at 84 months. Based on the log rank test, the variables considered to be important with p-value < 0.05 were regular, dialysis frequency per week, hospitals, diabetes mellitus , hypertension, diabetes mellitus and hypertension, shrunken kidneys, other. The Gompertz model, which is based on Cox-Snell Residuals, Akaike Information criterion (AIC) and the Bayesian Information Criterion (BIC),is useful among others models. Furthermore, hypertension (HR=0.612, p-value=0.039), regular (HR=0.485, p-value=0.003), urea (HR=1.004, p-value=0.045), and hospitals (HR=0.842, p-value=0.003) were found to have a significant impact on survival (P<0.05). The Gompertz model was found to have the smallest BIC values in multivariate analysis, so it was chosen as the best model for hemodialysis patients.