What is the infant mortality rate in South Africa? The need for improved data

To review recent infant mortality and birth registration data in South Africa and to investigate geographical differences.

Estimates of infant mortality rates, proportion of births not registered, and proportion of births recorded in health services.

1. Published infant mortality data for South Africa were collated. Demographic data from national household surveys (1993 and 1994 October Household Surveys and the 1993 Poverty Survey by the Southern African Labour and Development Research Unit (SALDRU) at UCT) were analysed using the indirect method developed by Brass. 2. Birth registration data were analysed and compared with the estimated number of births to identify regions with greater under-registration. The number of births recorded in the health services was analysed by province in order to assess and explore alternatives within health authorities that could complement the existing system.

1. Published estimates of infant mortality for the period from 1990 range from 40 to 71/1,000 births and estimates based on national household surveys conducted in this period from 11 to 81/1,000 births. 2. Completeness of birth registration in the nine provinces ranges from less than 10% in the Eastern Cape, North West and Northern Province to 60% in the Western Cape. An overall improvement from 19% to 60% could be achieved if births recorded through the health services were included in the vital registration system.

The infant mortality rate in South Africa is not known with any certainty. The extent of completeness of the birth registration system was 19%, which indicates a need for urgent improvement in order to provide key health status indicators. This study indicates that there is some potential for improving the extent of birth registration if it could be facilitated through the health service. However, this alone would not achieve complete registration.

Surveys will have to be relied upon until such time as routinely available statistics are accurate. The October Household Survey conducted annually by the Central Statistical Service as potentially an important source of health status information. It is imperative that either the design of the birth history questionnaire be improved or that it be replaced by a less frequent but more specialised demographic and health survey.