Modelers Leigh Johnson of the Centre for Infectious Disease Epidemiology and Research   and Rob Dorrington of the Centre for Actuarial Research at the University of Cape Town were discussing the ASSA 2008 model, which is to replace the ASSA 2003 model for estimating HIV prevalence, HIV-related deaths, the numbers of those in need of ARVs and the impact of HIV interventions. The new model will be officially released in the next three weeks.

According to Johnson, the reason the model needed to be updated was because the prevalence data projected was no longer correct because of the new data that emerged from South Africa’€™s antenatal HIV-prevalence survey. The survey increased the number of women who were tested for HIV and was thus more representative, although there was very little difference in the HIV prevalence results across the board from the ASSA 2003 and 2008 models. The ASSA 2008’€™s female prevalence matches that of the Human Sciences Research Council Household Survey, while the same was not found for males.

ASSA 2008 takes into account new epidemiological data to allow for more accurate projections of HIV prevalence and impact of interventions. It includes the ARV rollout data for up to the end of 2008. Because data shows that two-thirds of people starting ARVs are females, the ASSA 2008 model allows for different rates of ARV initiation in males and females, as well as for children and adults. ASSA 2008 also assumes a greater decrease in viral load when patients begin ARVs from a 1,76 to 2,8 unit decrease per log of viral load.

Another difference in the ASSA 2008 model is that it recognises that the attrition rate is much higher in the Western Cape than in other provinces. The cumulative rates of attrition are lower in the 2008 model than in the 2003 model.

Data has also shown that the rollout of prevention of mother to child transmission (PMTCT) of HIV has been slower than what was predicted in the 2003 ASSA model. The 2008 model takes into account the slower pace of the PMTCT rollout and the lower than expected uptake of single-dose Nevirapine. The 2008 model has also factored in the provision of dual PMTCT prophylaxis which is more effective at preventing vertical transmission.

Adult survival rates pre-ARVs have also been adjusted in the 2008 model to allow for longer survival time. There is a significant difference in the survival time of children infected at or before birth who are not receiving ARVs.

Condom usage has also been shown to be higher than that of 2003’€”reported condom use at last sexual intercourse was shown to be 70% in 2008 compared to between 30 and 40% in 2003, a factor that impacts on HIV incidence figures.

Some of the issues that the new model does not address is the impact of the new ARV guidelines to be released next month, including the provision of ARVs to people with a CD4 below 350 who are pregnant or have TB. The assumed increase in the number of people needing ARVs due to provider-initiated HIV testing has not been factored into ASSA 2008. The impact of male circumcision on decreasing HIV transmission has also not been taken into account. Studies have also found that women on ARVs are 70% more fertile than HIV-infected women not on treatment. This affects the interpretation of the antenatal clinic HIV prevalence data, which has not been considered in the ASSA 2008 model.

Another limitation of the ASSA model is that people are divided into four categories of risk of infection which does not change over the course of their lives, which is unrealistic.

The difficulty with the ASSA model is that one cannot accurately determine the numbers of people in need of ARVs because the ASSA 2008 model is based on CD 4 count as the marker for ARV-need, while the new ARV guidelines determine a patient’€™s need for ARVs based on criteria other than CD 4 count. Johnson hopes that this will only be a temporary limitation and that South Africa will adjust its guidelines to initiating ARVs to all people who have a CD 4 of less than 350.

‘€œIn order for us to project forward HIV prevalence and the impact of interventions, we will need a much more sophisticated model than ASSA will ever be able to provide. This model will need to incorporate interventions such as the impact of male circumcision, the new ARV guidelines and the other limitations that the new model does not address,’€ said Johnson.