Modeling Mortality by Causes of Death in South Africa Using Log-linear Analysis

Abstract This study aimed at identifying underlying patterns in mortality due to causes of death in South Africa using mortality statistics from 2005 to 2015 obtained from Statistics South Africa. Log-linear analysis was used in this study on mortality by causes of death dataset having three variables, cause of death (C), province (P) and year (Y). Log-linear analysis was preferred because of its capability to tease out relationships among variables. Results revealed that there are variations in mortality due to causes of death. Mortality was found to differ widely across the country, among provinces. It is recommended that prevention and management policies for HIV and TB be intensified since they still remain South Africa's major causes of death. A replication of the study could be done in another developing country using latest data to see if it will yield the same results. A multi-population mortality modeling could also be carried out using the same approach.