Dissertation (MSc Statistics)
The survival patterns and their influencing covariates (factors) for hemo-dialysis patients at Muhimbili National Hospital (MNH) in Dar es Salaam Tanzania are explained in this dissertation. Number of hemo-dialysis that a patient had, alcohol intake history and blood transfusion status of patients were significant variables for failure. The study explored that, the survival times for hemo-dialysis patients at Muhimbili National Hospital follow lognormal distribution and therefore, parametric model fitted based on this distribution. The mean and median survival time of hemo-dialysis patients observed to be 468 and 260 days respectively as suggested by parametric model (log-normal). Semi-parametric method of analysis (Cox-proportional hazard model) was used to determine the covariates for failure for hemo-dialysis patients and compare hazard for different class variables. Under this model, number of dialysis and alcohol intake history were significant variables for failure of patients. Non-parametric methods such as Kaplan-Meier estimator used to obtain the mean and median survival and compare the survival for group variables. The mean and median survival from KM-estimator observed to be 221 and 206 days respectively. Log-Rank test used to assess the significance variation in survival for class variables. A Significant variation observed for cigarette smoking and HIV/AIDs status of patients. Complications and co-morbidities are also discussed. Low blood pressure, body weakness and urine syndrome are common complications. Some co-morbidities influenced some complications. For example, 37% of patients with negative blood group had more than 2 complications.