HIV incidence starting to fall in the hardest hit communities in South Africa

High school students in KwaZulu-Natal, South Africa. © 2000 Patrick Coleman, Courtesy of Photoshare. Image is for illustrative purposes only.

Two separate studies from KwaZulu-Natal in South Africa have been published in the last two months, both showing falls in new HIV infections in a region where over 20% of men and 40% of women are living with HIV.

One study finds a 43% decline in HIV incidence in recent years in both men and women, with a larger and earlier decline in men. However, the other finds evidence only of a decline in young women aged 15 to 19, and none so far in men or older women.

Both show increased rates of antiretroviral treatment and viral suppression among HIV-positive men and women in the last few years and an increasing rate of HIV testing in the survey where it was not done routinely as part of a clinical trial. Medical male circumcision rates also increased in men in both studies.


virological suppression

Halting of the function or replication of a virus. In HIV, optimal viral suppression is measured as the reduction of viral load (HIV RNA) to undetectable levels and is the goal of antiretroviral therapy.

voluntary male medical circumcision (VMMC)

The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.


The surgical removal of the foreskin of the penis (the retractable fold of tissue that covers the head of the penis) to reduce the risk of HIV infection in men.

prospective study

A type of longitudinal study in which people join the study and information is then collected on them for several weeks, months or years. 

statistical significance

Statistical tests are used to judge whether the results of a study could be due to chance and would not be confirmed if the study was repeated. If result is probably not due to chance, the results are ‘statistically significant’. 

There is one variable that was significantly different in the two studies: condom use. Condom use has not increased since 2013 in either study but 64% of women in the first study said their male partners used condoms, while in the other study 18% did. While this would not explain declining incidence, it might be a context in which an increase in the proportion of people who were virally suppressed could have a bigger impact.

However, the main difference is probably that they were two very different studies. The study that documented falling incidence in all groups is part of a long-standing prospective cohort study, first established in 2005, in the largely rural district of Hlabisa in the north of KwaZulu-Natal. This study, because it has been estimating HIV incidence for years, was able to document HIV incidence on a year-to-year basis. The second study, on the other hand, only took two measurements from each participant.

The Hlabisa study

"There was a 43% decline in HIV incidence in both men and women, with a larger and earlier decline in men."

The first study was conducted by Dr Alain Vandormael, Professor Frank Tanser and colleagues, analysing data from an established cohort of all households in a specific area who are contacted annually and tested for HIV during researcher visits. Of note, the researchers restricted their documentation of HIV incidence to repeat testers who missed no more than two annual HIV tests between their last negative HIV test and their first positive one. Three-quarters of those tested were included in this definition. HIV incidence was higher in those who spent longer periods without study visits – an indication that just being in such a study had a positive effect.

Between 25,000 and 36,000 eligible participants aged over 15 in 11,000 households live in the Hlabisa area, and they are visited by trained field workers two to three times a year. The study shows that only about 40% were tested for HIV within any one year, but cumulatively 80% have ever tested (53,167 people) and, as already said, 76% have had at least two tests. During the long duration of the study, antiretroviral therapy (ART) eligibility has been expanded from availability only to people with CD4 counts below 200 (in 2004) to everyone (in 2016).

Between 2005 and 2017 the annual number of people testing HIV positive declined from 7896 in 2005 (4557 women and 3239 men) to 4984 in 2017 (3005 women and 1979 men). The peak incidence year for both men and women was 2012. In that year annual HIV incidence was 4.95% in women – meaning that just over one in 20 received a new diagnosis of HIV during that year – and 2.49% in men.

After 2012, it started falling in men, to reach 1.01% in 2017 – a fall of 59%. HIV incidence in women did not start falling significantly until 2014, reaching 3.06% in 2017 – a fall of 37%. HIV incidence in young women aged 15-29 fell “markedly” from 2014 onwards but only declined significantly in older women in 2017.

The proportion of women with HIV taking ART increased from under 25% in 2010 to over 50% in 2017, with 45% virally suppressed in 2017; in men it increased from 21% to 38% during the same period, with 33% virally suppressed in 2017. More women virally suppressed implies fewer infections in men.

Condom use in this population increased during the first decade of the century and by 2011 had reached over 60% (according to women) and 70% (according to men), with little change thereafter.

The proportion of men reporting they were circumcised increased from 3% in 2009 to 33% in 2016. HIV incidence in circumcised men decreased by 59% and in uncircumcised men by 42%.

The Greater Edendale study

"The second study found evidence only of a decline in incidence in young women aged 15-19."

The second study is a population cohort study, conducted by Dr Ayesha Kharsany, Professor Quarraisha Abdool Karim and colleagues. Within the peri-urban Greater Edendale and rural Vulindlela areas near Pietermaritzburg in southern KwaZulu-Natal, the researchers randomly selected households and invited a single inhabitant aged between 15 and 49 years from each to take part. The respondents provided a range of data and were anonymously tested for HIV, with the results compared with their knowledge of their status.

One survey was conducted in 2014-2015. A follow-up survey, solely of those who were HIV negative in the original survey and below the age of 35, was conducted 14 to 24 months later to establish HIV incidence. This enrolled 9812 participants, with 4539 taking part in follow-up.

A second survey was run from 2015 to 2016, again with follow-up some months later to calculate incidence. This enrolled 10,236 people (almost all of them different from the first survey), with 5307 taking part in follow-up.

There was a big increase in people reporting they had ever tested for HIV, from 69% of men and 82% of women in the first survey to 82% of men and 91% of women in the second. Similarly, there was an increase in knowledge of HIV status, especially in men. In the 2014 survey, 52% of men and 62% of women who had HIV knew their status; in the 2015 survey it was 63% and 73% respectively.

The proportion reporting they were on ART also expanded, from 37% of men and 46% of women, to 49% of men and 59% of women. Interestingly, viral suppression rates – as measured by testing – were actually higher than the proportion saying they were on ART. The proportion of people with HIV who were virally suppressed in the second survey was 50% of men and 62% of women. While this may include a few viral controllers, blood tests showed that 12% of those who said they were not on ART actually were.

Medical male circumcision increased from 32% to 36% between the two surveys. The proportion of women who said their partners were circumcised rose more, from 36% to 48%, which may indicate that circumcised partners are seen as more desirable – or are simply younger. However, condom use declined at the same time, from 20% to 16% of women saying their partners used condoms.

Eighty-seven per cent of the HIV-negative participants were re-tested in the follow-up surveys. This allowed annual incidence rates to be calculated. The incidence rate in young women aged 15 to 19 fell from 4.6% in 2016 to 2.7% in 2017, and this 43% decline in HIV incidence was statistically significant. There was a 42% incidence rate fall in young men too, but because HIV infections in men under 20 are rather rare, this was not significant (incidence fell from 0.4% to 0.24%). There no significant drop in incidence in the cohort in general, in men, or in women aged over 20.

One interesting finding was that people were asked about age differences between them and their partners. In general they were not great, with women, for instance, reporting that their first sexual partner was two years older than them. However, the researchers were able, by means of genetic testing, to establish linked pairs of HIV infections and it was found that in the 2014 survey the male partner in a linked pair was on average 13.5 years older than the female. By 2016 this had fallen by to 9.3 years.


"HIV testing, treatment and prevention measures are beginning to have a population-level effect on HIV incidence in this ‘hyperendemic’ part of the world."

The authors of the Hlabisa study attribute the earlier and larger fall in incidence in men to two factors: the higher proportion of HIV-positive women that were on ART and virally suppressed, and rising circumcision rates.

In the Greater Edendale study on the other hand, a significant fall in incidence has only been seen in young women. The authors comment that “increases in the number of older HIV-seropositive men in age-disparate sexual partnerships who were using ART and had viral suppression may have led to the reduction observed in HIV incidence.”

However, it is also possible that the different methodology of the two studies may explain the difference in results. The Hlabisa study documents falling incidence over a period of several years, whereas the Greater Edendale study essentially only took a single pair of measurements to calculate incidence. The Hlabisa study also followed an established cohort who were already part of a research programme and may have been affected by closer attention and monitoring, whereas the Greater Edendale study involved random sampling of the population at large.  

Either way, these studies both indicate that HIV testing and treatment, and prevention methods such as circumcision, are at least beginning to have a population-level effect on HIV incidence in this ‘hyperendemic’ part of the world.


Vandormael A et al. Declines in HIV incidence among men and women in a South African population-based cohort. Nature Communications, 10: 5482, 2019.

Kharsany ABM et al. Trends in HIV prevention, treatment and incidence in a hyperendemic area of KwaZulu Natal, South Africa. Journal of the American Medical Association Network, 2: e1914378, 2019.