South African death data are known to be incomplete and the level of mortality is
unknown for Blacks. This research attempts firstly to estimate both the level and
shape of the mortality curve of the Black South African population group circa 1985
by the application of indirect demographic techniques to the reported deaths for the
period 1984-86 and an estimate of the Black population in mid-1985. The life tables
thus produced are then combined with the official South African Life Tables for
Whites, Coloureds and Asians to produce a weighted average national life tables circa
1985. These national life tables are then compared with those produced by applying
the Bennett and Horiuchi method to national data for the 1984-86 period in order to
decide on suitable adjustments to make when applying the Bennett and Horiuchi
method to national data for the period 1989-91 and beyond.
The analysis of the 1984-86 data shows that only 56% of male and 44% female deaths
were recorded for the Black population group. Adjusting for this under-reporting the
life expectancy at birth for Black South Africans was 56,1 and 63,3 for men and
women respectively. Adult mortality as measured by 45Q15 was estimated to be 39%
for Black men and 24% for Black women. These estimates suggest higher levels of
mortality than previously derived. Comparison with the estimates derived for 1990-
91, suggest that there was no improvement in national mortality over the period
between the two investigations, both prior to the expected impact of the AIDS
epidemic.
The need to improve death registration in general and to obtain more reliable
estimates of child mortality were highlighted by the results of this investigation as
well as the need to develop improved methods of estimation for such data.