Is HIV elimination a pipe dream?

Is HIV elimination a  pipe dream?A full version of the special edition of Spotlight will be released during the International AIDS Conference in Durban which starts on 17 July. This is an article which will be published in Spotlight.
 by Dr Leigh Johnson –


A number of UNAIDS publications have promoted the idea that it is possible to “end the AIDS epidemic by 2030”. There are several encouraging signs to suggest that this may be true. It is now well-established that patients who are on antiretroviral treatment (ART) and with suppressed viral loads are virtually non-infectious.With the recent revisions to WHO treatment guidelines, which now recommend ART for all HIV-positive individuals regardless of CD4 count or clinical stage, it is theoretically possible that a high proportion of HIV-positive individuals will be treated. This could drive down HIV transmission rates to very low levels. UNAIDS has stated that critical to HIV elimination will be the achievement of the ‘90-90-90’ targets: by 2020, 90% of the HIV-positive population needs to be diagnosed, 90% of diagnosed individuals need to be on ART, and 90% of patients on ART need to be virologically suppressed.

But how likely is HIV elimination? Central to answering this question are mathematical models, which attempt to predict the future based on observed historical trends in HIV prevalence, and based on assumptions about the effect of different HIV prevention and treatment strategies on HIV transmission. This article briefly reviews some of the recent modelling studies that have attempted to answer this question, and discusses some of the limitations and uncertainties associated with modelling.

The first point to note is that the term ‘HIV elimination’ is a misnomer. Most frequently, modelling studies refer to ‘virtual elimination’, which is conventionally defined as an adult HIV incidence rate of less than 0.1% per annum. But even if HIV incidence among 15-49 year olds were constant at 0.1% per annum, the long-term HIV prevalence in 15-49 year olds would not drop below 1.7% (assuming a relatively high ART coverage and a near-normal life expectancy on ART). This is still an appreciable HIV prevalence, even by the standards of many countries in West Africa.

Even if we accept this definition of HIV elimination, mathematical models are not conclusive about whether universal ART eligibility and high rates of HIV testing and ART uptake (so-called ‘test and treat’ strategies) would lead to elimination. In a systematic comparison of several different models that were applied to South Africa, Eaton et al found that out of nine models, six suggested that ‘test and treat’ strategies would not be sufficient to achieve virtual elimination by 2050. Notably, the models were generally quite optimistic about the extent to which ART would reduce HIV infectiousness (in all cases by 90% or more), though a subsequent systematic review of observational data estimated an average reduction of only 64%. It was also subsequently found that almost all the models had under-estimated the HIV prevalence that was measured in a South African household survey, conducted after the initial model projections were published. Although this points to the fallibility of mathematical modelling, it is perhaps more important to note that the models generally did not show that HIV elimination was a likely outcome, despite erring on the side of optimism.

A question that naturally follows is whether a ‘test and treat’ strategy might achieve HIV elimination when combined with other HIV prevention strategies. In a recent study, we attempted to address this question for South Africa by projecting future HIV incidence trends using a wide range of different intervention scenarios. This study predicted that – given the current uncertainty around HIV prevention and treatment programmes in South Africa – the virtual elimination target of 0.1% would be reached by 2035 in only 2% of scenarios (Figure 1). The model also predicted that although South Africa would probably reach the first 90% target by 2020, the second and third 90% targets were quite unlikely: in only 0.4% of scenarios were all three targets met.

Figure 1: HIV incidence trends in South African adults aged 15-49
Figure 1: HIV incidence trends in South African adults aged 15-49

Solid lines represent mean of model estimates. Dashed lines represent 95% confidence intervals (taking into account uncertainty regarding future epidemiological parameters). Shaded grey area represents virtual elimination threshold. Source: Johnson et al

This study also assessed which epidemiological parameters it would be most important to focus on in order to reduce HIV incidence. The most important parameter was the rate of virological suppression in ART patients: for every 10% increase in the fraction of ART patients who are virologically suppressed, it was predicted that there would be a 14% reduction in the average annual number of new HIV infections. This implies that increasing rates of virological suppression in South African ART patients from the current level of around 77% to the 90% target would achieve an 18% reduction in HIV incidence. Other parameters that were significant included the rate of condom use in non-cohabiting relationships, the introduction of intensified risk-reduction counselling for HIV-positive adults, and the uptake of medical male circumcision. Interestingly, the timing of the change to universal ART eligibility was only the 5th-most important parameter. This suggests that the recent change to universal ART eligibility is not by itself likely to have as dramatic an impact on HIV incidence as many other interventions.

The case of South Africa stands in stark contrast to the case of Denmark, the subject of another recent modelling study. In this study, it was estimated that in 2009 the HIV incidence among Danish men who have sex with men (MSM) was 0.14% per annum, very close to the virtual elimination threshold of 0.1%. Although the fraction of HIV-positive adults in Denmark who were diagnosed was very similar to that estimated in South Africa (around 80% in 2013), the fraction of diagnosed individuals on ART was 92%, and the fraction of ART patients who were virologically suppressed was 98% – both well ahead of South Africa (Figure 2).

Figure 2: Treatment cascades in Denmark and South Africa in 2013
Figure 2: Treatment cascades in Denmark and South Africa in 2013

As the authors of this study note, Denmark is exceptional. In many other high income countries, there has been a resurgence in HIV incidence among MSM, despite increasing levels of ART coverage. This resurgence is often attributed to risk compensation and ‘disinhibition’, i.e. increased levels of sexual risk behaviour due to reduced fear of HIV in the era of highly effective therapy – and perhaps also reduced public messaging around safe sex as the HIV response has become increasingly medicalised.

Taken together, these results suggest that treatment alone is not going to end the HIV epidemic, although it might be possible in concentrated epidemic settings with exceptionally high levels of virological suppression. It will be important not to neglect the ‘traditional’ HIV prevention strategies in the pursuit of the 90-90-90 targets, and the allure of new prevention approaches should not detract from the need to sustain and improve existing programmes. But even with co-ordinated strengthening of existing programmes and introduction of new prevention approaches (such as universal ART eligibility and pre-exposure prophylaxis), it is unlikely that virtual elimination will be achieved in hyper-endemic settings such as South Africa within the next 20 years. Even if ‘elimination’ is achieved, there would need to be continued high levels of HIV testing and HIV prevention messaging over the longer term if a resurgence in HIV were to be avoided. Achieving true elimination will require fundamentally new technologies such as HIV vaccines. Until we have these in place, HIV elimination needs to be seen as an aspirational ideal rather than a practical target.

DR Leigh Johnson is a Researcher at the Department of Public Health & Family Medicine for the University of Cape Town.