Infant and child mortality in South Africa in the context of a high HIV prevalence : an investigation into changing mortality patterns at a fine age resolution

Includes bibliographical references (leaves 98-111).South Africa has very high levels of HIV prevalence, with some provinces having among the highest levels in the world. Within this context it is imperative to have a clear understanding of how the epidemic is affecting infants and children in the population and to what extent interventions are affecting mortality. However, establishing accurate estimates of infant and child mortality levels is very difficult in South Africa because the data available is nearly a decade out of date. Demographic modelling techniques and extrapolations from out of date data provide the closest estimates but are less than ideal in the middle of an HIV epidemic. What is needed is a surveillance method that can provide rapid, up to date information on infant and child mortality, within an environment of high HIV prevalence that can inform health policy for South Africa's youngest citizens. This study utilized routinely collected national vital events data to describe trends in infant and child mortality from 1990 to 2006. Mortality was examined by age of death in months, a finer age resolution than has been previously published. Data used in this study consisted of unpublished mortality statistics collected by Statistics South Africa from 1990 to 2002, and data extracted from the Population Register database maintained by the Department of Home Affairs for the years 1998 to 2006. The Population Register database was investigated for use as a potential mortality surveillance tool to measure current trends in infant and child mortality and to measure any effects by HIV/IAIDS interventions at a population level. Several new and unique findings were revealed in this study. First, a new and increasing all-cause peak in mortality was discovered centring at 3 months of age - a new, previously unpublished, demographic phenomenon. Second, a coding error was found in causes of death of infants under 1 year of age in the data recorded by Statistics South Africa (Stats SA) resulting in the incorrect coding of the majority of deaths in this age group. Despite the problems with coding of cause of death, the peak in mortality at 3 months was shown to be due to HIV/AIDS mortality and was the third finding in this study. The fourth finding was that the Population Register data could further be used with the 3 month peak in mortality as a rapid surveillance tool to measure trends in infant mortality. By utilizing the up to date data from the Population Register database and monitoring the peak in mortality at 3 months, this study was able to show that HIV/AIDS interventions such as the ARV rollout and PMTCT programs are beginning to have a positive effect at a population level. This method of surveillance was able to examine changes in mortality at 3 -months at both a national and provincial level