NDOH Annual Report 2008/09

Methodology
The 2008 Survey was conducted in 52 health districts. Antenatal HIV sentinel surveillance involved
collection of 33 927 intravenous blood samples from pregnant women at their f irst antenatal visit
served in 1 457 public health clinics over a 4 week period. The blood samples were screened using
the Enzyme Linked Immuno Sorbent Assay (ELISA) test and the Rapid Plasma Reagin (RPR) card
test.

Given that the sentinel sites were chosen on a probability proportional to size basis, the districts
were self-weighting and the estimates of provincial level prevalence were simply calculated as the
total of the results from the districts in the provinces. The national prevalence was then estimated as
a weighted average of the provincial prevalence estimates weighted according to the total number of
women aged 15 - 49 years in each province using the 2008 mid-year population estimates.
Findings

The overall national HIV prevalence among ante-natal women aged 15 - 49 years in the 2008
ANC survey, measured using the parallel test algorithm, was 29.3% (95% CI: 28.5% - 30.1%). The
occurrence of the HIV infection nationally has stabilized at around 29.0% from 2006. In 2007, the
HIV prevalence estimate among first visit antenatal attendees was 29.4% (95% CI: 28.5 - 30.1).

EXECUTIVE SUMMARY
Nationally, women in the age group 30 - 34 years still have the highest prevalence, with a prevalence
of 40.4% in 2008 compared to 39.6% in 2007. The HIV prevalence among the 15 - 24 years old
(which is the Millennium Development Goal 6, Target 7 indicator 18) was 21.7% in 2008 compared
to 22.1 % in 2007 a decline of 0.4%. There is a slight increase in HIV prevalence among young
women in the age group 15 - 19 years from 13.1% in 2007 to 14.1% in 2008. The HIV prevalence has
remained stable among women aged 25 years and above.

The highest HIV prevalence of 38.7% (CI: 37.2% - 40.1%) in 2008 was seen in the province of
KwaZulu-Natal and the lowest estimate of 16.1% (CI: 12.6% - 20.2%) was noted in the Western
Cape Province. Free State, Mpumalanga and the Western Cape provinces showed a slight increase
in HIV prevalence, while KwaZulu-Natal, North West (which had prevalence above 30%) Northern
Cape and Limpopo provinces remained static. Gauteng province is showing a tendency towards a
decrease, although this is not statistically signif icant. Mpumalanga province is the only province in
the country that continues to show some evidence of an increase in HIV infection from 32.1% in
2006 to 34.6% in 2007 to 35.5% in 2008.

District HIV prevalence has only been reported since 2006 when the geographic coverage of
sentinel sites was increased and the sample size doubled. There was a considerable variation in
HIV prevalence between the 52 health districts observed over the three year period 2006 - 2008,
particularly where the sample size in a district is small, making it diff icult to discern any trends.
However, the following inferences can be made viz: Fezile Dabi and Xhariep districts in the Free
State are showing an increase, whereas Amajuba in KwaZulu-Natal is showing a decrease in HIV
prevalence over the past three years, Dr Ruth S. Mompati in North West, Sekhukhune in Limpopo,
Sisonke and uThukela in KwaZulu-Natal showed a slight increase, while Bojanala in North West,
Waterberg in Limpopo and West rand in Gauteng showed a slight decrease over the last three years.
In addition the districts are clearly heterogeneous with respect to the epidemic, with prevalence
ranging from a high of over 45% to a low of around 5%. When data are pooled over the three years
this heterogeneity persists.

A regression analysis of determinants of HIV positive status in the survey participants using the
demographic and laboratory information showed that the most signif icant determinant factor
was age. Splitting the sample at an age of 21 years, the women less or equal to 21 years have HIV
prevalence of 16.8% compared the 34.8% of women 22 years and older. This split of the overall
group with prevalence of 29.2% has identif ied a younger subgroup that has a much lower prevalence.
No further splits were identif ied in this group of young women. In the older age-group the next split
was on race. An African subgroup (37.6%) is identif ied versus the rest (6.8%) of White, Asian and
Coloured women which has a low prevalence and no further splits were identif ied in this subgroup
of participants. One important observation from the regression analysis was that women having a
syphilis co-infection is not a strong predictor for HIV status.
Conclusion

The HIV prevalence of 29.3% in 2008 is in line with the prevalence observed in the two previous
years. To avoid a resurgence of the HIV and AIDS epidemic in South Africa, HIV prevention
efforts need to be urgently strengthened and sustained. Furthermore, ecological correlations
between the trends in HIV prevalence and behavioural changes that will focus on reducing the
incidence of infection exposure factors, especially in districts that record more than 30% HIV
prevalence, is warranted. Further in-depth epidemiological investigations on what could be causing
the interjectory between the districts and between provinces in the identif ied epicentres could assist
in understanding the different patterns of the transmission potential of the virus.