Yesterday the Minister of Health, Manto Tshabalala-Msimang, released the 2006 National HIV and Syphilis Prevalence Survey of South Africa. The minister said in a statement that the report "shows a trend towards a decrease in the prevalence of HIV amongst pregnant women who use public health facilities (29.1% in 2006 compared to 30.2% in 2005), suggesting that this may be the beginning of a decline in the HIV prevalence rates. This is positive news for our country."

According to the report there was a "significant decline in HIV prevalence amongst participants under the age of 20 years. In this group HIV prevalence was estimated at 13,7% in comparison to 15,9% in 2005." It comments:

"A decline in prevalence in this age group is suggestive of a decline in HIV incidence (new infections) and is a good indicator of the impact of intervention programmes... The findings suggest a sustained change in behaviour among young people including in engaging in safer sexual practices such as being in mutually faithful relationships may be taking place."

There are reasons however to treat this news with some caution, as it is not clear whether the results of the latest survey are strictly comparable with those of previous years. The number of pregnant women who participated in the latest (2006) survey was 33 034, an increase from 16 510 in 2005.

The 2005 report [PDF] stated that the "same clinics as had participated in the previous surveys were used". But for the latest survey there has been a major change, with the number of "health facilities" participating being increased from 399 in 2005 to 1 415 in 2006.

Lawrence Schlemmer notes that "since each clinic has a particular profile of attendees the sample must be assessed as a sample of clinics rather than women tested." Increasing the size of the sample to 1415 would have dramatically strengthened it.

However, the increase "must mean that the internal characteristics of the sample changed" which would make it non-comparable to 2005. This problem would be compounded if they now included different kinds of "health facilities" - other than ordinary clinics - in their sample.

For Schlemmer, "the very slight declines in infection rates do not appear to be statistically significant and given the change in the composition of clinics the slight changes mean even less. I am glad to see any decline in infection rates but I do not think that these results can tell us that."

The significance of the latest survey is not that they (necessarily) signal a decline, but that they are more accurate and comprehensive than in previous years.

The report states that, extrapolating to the population as a whole using the latest models, the estimate of the number of people with HIV in South Africa is in the region of 5,41m - of whom 2,29m are men and 2,86m women. HIV prevalence among those aged 15 to 49 is estimated at 18,34%. There are an estimated 257 900 children (0-14 years) living with HIV.