Projections for overall declines in HIV epidemics might have been optimistic, while future treatment and HIV prevention needs might be greater than previously forecasted, found a comparison of mathematical models used to analyse the South African HIV epidemic.
A new version of the Thembisa mathematical model, arguably the most reliable of those that analyse the SA HIV epidemic, has been made available, reports Groundup.
Just under 150,000 people are expected to die of Aids in South Africa this year, about a quarter of all deaths. As of mid-2015 about 3.4m people were on antiretroviral treatment, and life-expectancy continues to rise, but at approximately 63 years in 2016 it is still considerably lower than the world average of over 71 years.
About 7.1m people live with HIV, about 13% of the nearly 56m South Africans. KwaZulu-Natal has the most number of infections at over 1.9m, and the highest prevalence: 18%. Mpumalanga is next at 15%, while about 12% to 14% of people in
Nathan Geffen writes in Groundup that these statistics were published last week by University of Cape Town researchers Leigh Johnson, Rob Dorrington and Haroon Moolla. He says a new version of the Thembisa mathematical model, arguably the most reliable of those that analyse the South African HIV epidemic, has been made available at www.thembisa.org.
Geffen writes that for data fundis, the spreadsheets on the website offer a treasure trove of information. The model’s estimates are calculated using a multitude of studies on HIV and the South African population. The estimates up to 2012 and 2013 are matched – calibrated is the more accurate but technical term – against the annual HIV surveys of pregnant women conducted by the Department of Health and the household surveys carried out every few years by the Human Sciences Research Council.
Geffen says the researchers have a reputation for being meticulous and a track record of producing models that closely match data from surveys. In a study published last year that compared the leading mathematical models to actual data, one of Johnson’s models generally outperformed the competition, making estimates that were for the most part close to survey data. The Thembisa model is based on this one.
Geffen writes: “Though the researchers have a track record of reliable estimates, mathematical models cannot take into account every last ebb and flow of how the world works. They are simplifications of reality that give us a useful idea of what’s going on. So for example, one of the spreadsheets calculates that 7,104,796 people had HIV as of mid-2016. It would be a mistake for a reporter to write, ‘There are 7,104,796 people with HIV’. That would be using false precision; we cannot know exactly how many people live with HIV at any point in time. Better to write: ‘Researchers estimate that about 7.1m people live with HIV in 2016’.
“The estimate of about 7.1m people living with HIV is quite precise; it’s unlikely to be much lower or higher than that. However, the estimate that 3.8m people are on antiretroviral treatment in 2016 is less precise (which is why I’ve used the 2015 estimate in the first paragraph). The real number of people on treatment could be quite a bit higher or lower than that. Scientists use confidence intervals to show the range in which the real number is likely to be. Some of these confidence intervals have not yet been published in the new version of Thembisa, but the lead researcher, Johnson, says they will be published soon.
“Although the number of HIV infections and Aids deaths remains very high, the model confirms that we are slowly becoming healthier. Life expectancy is increasing – it was about 54 years in the mid-2000s – and the number of annual Aids deaths is declining.
“In 2004 over 45,000 babies were born with HIV. This has declined to nearly 6,000, thanks to the mother-to-child transmission prevention programme. The number of new infections (adults and children) each year is coming down, but at about 270,000 this year (well over 700 per day on average), it remains far too high.
“Next month public health facilities will start offering antiretrovirals to everyone with HIV when they are diagnosed, irrespective of how advanced their disease is. It’s hoped that besides improving the health of people with HIV, this will go some way towards reducing new infections because people on treatment whose virus is no longer detectable cannot transmit HIV. The success of that effort will depend on how many people with HIV can be reached.”
Background: Mathematical models are widely used to simulate the effects of interventions to control HIV and to project future epidemiological trends and resource needs. We aimed to validate past model projections against data from a large household survey done in South Africa in 2012.
Methods: We compared ten model projections of HIV prevalence, HIV incidence, and antiretroviral therapy (ART) coverage for South Africa with estimates from national household survey data from 2012. Model projections for 2012 were made before the publication of the 2012 household survey. We compared adult (age 15–49 years) HIV prevalence in 2012, the change in prevalence between 2008 and 2012, and prevalence, incidence, and ART coverage by sex and by age groups between model projections and the 2012 household survey.
Findings: All models projected lower prevalence estimates for 2012 than the survey estimate (18•8%), with eight models’ central projections being below the survey 95% CI (17•5–20•3). Eight models projected that HIV prevalence would remain unchanged (n=5) or decline (n=3) between 2008 and 2012, whereas prevalence estimates from the household surveys increased from 16•9% in 2008 to 18•8% in 2012 (difference 1•9, 95% CI −0•1 to 3•9). Model projections accurately predicted the 1•6 percentage point prevalence decline (95% CI −0•3 to 3•5) in young adults aged 15–24 years, and the 2•2 percentage point (0•5 to 3•9) increase in those aged 50 years and older. Models accurately represented the number of adults on ART in 2012; six of ten models were within the survey 95% CI of 1•54–2•12 million. However, the differential ART coverage between women and men was not fully captured; all model projections of the sex ratio of women to men on ART were lower than the survey estimate of 2•22 (95% CI 1•73–2•71).
Interpretation: Projections for overall declines in HIV epidemics during the ART era might have been optimistic. Future treatment and HIV prevention needs might be greater than previously forecasted. Additional data about service provision for HIV care could help inform more accurate projections.
Jeffrey W Eaton, Nicolas Bacaër, Anna Bershteyn, Valentina Cambiano, Anne Cori, Rob E Dorrington, Christophe Fraser, Chaitra Gopalappa, Jan A C Hontelez, Leigh F Johnson, Daniel J Klein, Andrew N Phillips, Carel Pretorius, John Stover, Thomas M Rehle, Timothy B Hallett