Thomas Rehle of the Human Sciences Research Council (HSRC) and colleagues published an article in PLoS One that estimated the change in HIV incidence in South Africa. [1]

The study found a non-significant reduction in HIV incidence in the period 2005 to 2008 compared to the period 2002 to 2005. However there was a significant reduction in incidence amongst women aged 15 to 24.

The HSRC conducted country-wide surveys in 2002, 2005 and 2008. These surveys measured HIV prevalence for the whole population, as well as by gender, age group and race. By using a recently developed methodology that involved examining differences in prevalence across surveys and correcting for deaths in people with HIV the authors measured incidence from 2002 to 2005 and also from 2005 to 2008.

The methodology worked as follows: Ideally incidence would be measured by a longitudinal study, ie following the same set of HIV-negative people over a period of time and determining how many become infected. But this is impractical on a countrywide scale. However, such a study can be approximated using what is called the synthetic cohort principle. This assumes that individuals of age x in the first survey are represented by individuals aged x+t in the second survey, where t is the interval between surveys, even though the surveys do not include the same individuals. The change in HIV prevalence across this group of individuals is assumed to be due to new infections less deaths of people with HIV. Deaths in people with HIV in an age group cohort can be determined by estimating the rate of AIDS deaths based on historical distributions of survival after infection.

The interval between the 2002 and 2005 surveys was 2 years and 8 months, and the interval between the 2005 and 2008 surveys was 3 years and 7 months. But survey participants reported only their age in years (not date of birth) and so prevalence had to be interpolated in the synthetic cohorts.

The calculation of the rate of death in HIV-positive people was complicated by the scale up of antiretroviral treatment that began in South Africa in 2004. The authors accounted for this by subtracting the proportion of HIV-positive people who were alive in the 2008 survey that would have died without treatment. They were able to approximate this because HIV-positive blood samples in the 2008 survey were tested for the presence of antiretroviral medicines thereby giving an estimation of the proportion of HIV-positive people on treatment.

The researchers further assumed that on average people initiated treatment a year before they would have otherwise have died and that people starting treatment have a 10% annual mortality rate in the first years of treatment. They consequently calculated that 58% of those receiving treatment in 2008 were alive at the time of the survey due to being on antiretroviral treatment.

They estimated HIV incidence in people aged 15 to 49 in the period 2002 to 2005 to be 2.0 per 100 person years (95%CI: 1.2-3.0). This was compared to 1.3 per 100 person years in the 2005 to 2008 period (95%CI: 0.5-2.1), but this difference was not significant. There was however a significant decline in incidence across the two periods between women and men aged 15 to 24 years (5.5/100 py [95%CI: 4.3-6.6] v 2.2/100 py [95%CI: 1.3-3.1]).

The authors, in order to account for the uncertainty of the effect of treatment on the death rate of people with HIV, stated that they estimated incidence under a range of assumptions. They concluded that there remains clear evidence for a reduction in incidence among women under all credible assumptions for the effect of antiretroviral treatment.

The authors discussed how a decline in incidence could have occurred. They stated this was unlikely to have been a consequence of the natural course of the epidemic in which those groups at most risk of infection become saturated with infection and die. They also argued it was unlikely due to antiretroviral treatment because “access to treatment has only increased significantly in recent years, it is expected that such an effect would take longer to develop and require higher levels of [antiretroviral treatment] coverage for an extended period of time.” Instead, they give greater weight to the possibility that the decline was due to increased condom usage reported across the three surveys.

comment

The HIV National Strategic Plan 2007-2011 (NSP) set a target of halving HIV incidence between 2007 and 2011. [2] Therefore methods of measuring incidence and changes in incidence are needed to determine progress towards this target. However, calculating country-wide incidence is extremely difficult. This study is an ambitious effort to do so. Although its estimates do not coincide with the NSP period, if the limitations of this study can be overcome, it is conceivable that a practical way of measuring the NSP prevention target could be found. The study’s limitations are:

• The 2002 HSRC survey was widely criticised for having a low response rate and anomalous results. The accuracy of the results of this study is dependent on the accuracy of the 2002 survey.
• The method used to calculate the 95% confidence intervals assumes the data were collected in a simple random sample, but the data was from clustered samples. Therefore, the confidence intervals around the estimates of incidence should be larger.
• The researchers make several assumptions about the effects of antiretroviral treatment, such as the length of time people would have lived if they did not access ARVs and the speed and size of the scale up of the antiretroviral treatment rollout from 2004 to 2008. Therefore their calculation of the number of people with HIV who died between surveys has a wide margin of error.

Furthermore, Rehle and colleagues arguments for what could and could not be causing a possible decline in incidence are speculative. Only further research can provide more assured answers. To facilitate this and other research, several organisations have released a statement calling for the HSRC to make their data from the 2002, 2005 and 2008 surveys public. [3]

References:

1. Rehle TM et al. 2010. A Decline in New HIV Infections in South Africa: Estimating HIV Incidence from Three National HIV Surveys in 2002, 2005 and 2008. PLoS ONE 5(6): e11094. doi:10.1371/journal.pone.0011094
2. Department of Health. 2007. HIV and AIDS and STI Strategic Plan for South Africa, 2007-2011 .
3. TAC et al. 2010. HIV Incidence in South Africa- what is really happening?

Thanks to Anna Grimsrud, Rob Dorrington and Leigh Johnson for extensive assistance particularly the discussion on limitations.