HIV treatment as prevention: systematic comparison of mathematical models of the potential impact of antiretroviral therapy on HIV incidence in South Africa

Abstract

Background: Many mathematical models have investigated the impact of expanding access to antiretroviral therapy (ART) on new HIV infections. Comparing results and conclusions across models is challenging because models have addressed slightly different questions and have reported different outcome metrics. This study compares the predictions of several mathematical models simulating the same ART intervention programmes to determine the extent to which models agree about the epidemiological impact of expanded ART.

Methods and findings: Twelve independent mathematical models evaluated a set of standardised ART intervention scenarios in South Africa and reported a common set of outputs. Intervention scenarios systematically varied the CD4 count threshold for treatment eligibility, access to treatment, and programme retention. For a scenario in which 80% of HIV-infected individuals start treatment on average 1 y after their CD4 count drops below 350 cells/µl and 85% remain on treatment after 3 y, the models projected that HIV incidence would be 35% to 54% lower 8 y after the introduction of ART, compared to a counterfactual scenario in which there is no ART. More variation existed in the estimated long-term (38 y) reductions in incidence. The impact of optimistic interventions including immediate ART initiation varied widely across models, maintaining substantial uncertainty about the theoretical prospect for elimination of HIV from the population using ART alone over the next four decades. The number of person-years of ART per infection averted over 8 y ranged between 5.8 and 18.7. Considering the actual scale-up of ART in South Africa, seven models estimated that current HIV incidence is 17% to 32% lower than it would have been in the absence of ART. Differences between model assumptions about CD4 decline and HIV transmissibility over the course of infection explained only a modest amount of the variation in model results.

Conclusions: Mathematical models evaluating the impact of ART vary substantially in structure, complexity, and parameter choices, but all suggest that ART, at high levels of access and with high adherence, has the potential to substantially reduce new HIV infections. There was broad agreement regarding the short-term epidemiologic impact of ambitious treatment scale-up, but more variation in longer term projections and in the efficiency with which treatment can reduce new infections. Differences between model predictions could not be explained by differences in model structure or parameterization that were hypothesized to affect intervention impact.

Figure 1. No-treatment counterfactual epidemic trends.

Male (left) and female (right) HIV prevalence (top) and incidence (bottom) amongst 15- to 49-y-olds for counterfactual HIV epidemics with no ART. The STDSIM model is calibrated to a more severe epidemic in the Hlabisa subdistrict of KwaZulu-Natal Province, South Africa. The CD4 HIV/ART and Granich models do not stratify by sex, and the same prevalence and incidence curves are plotted for both sexes for these models. PYs, person-years.

Figure 2. Impact of treatment for a scenario with eligibility at CD4≤350 cells/µl, 80% access, and 85% retention.

(A) The percentage reduction in HIV incidence in the years 2020 and 2050 when eligibility for treatment is at CD4 count ≤350 cells/µl, 80% of individuals are treated, and 85% are retained on treatment after 3 y. (B) The cumulative number of person-years of ART provided per infection averted for the same scenario. Horizontal lines indicate 95% credible intervals (CI). For the Bendavid model, results in year 2040 are reported in the right panels.

Figure 3. Proportion reduction in HIV incidence in year 2020.

For each model, the proportion reduction in HIV incidence in year 2020 for increasing access levels from 50% to 100% (horizontal axis). ART eligibility thresholds are indicated by line colour; 85% retention is indicated by solid lines, and perfect 100% retention is indicated by dashed lines.

Figure 4. Cumulative number of person-years of ART provided per infection averted through year 2020.

The cumulative person-years of ART provided per infection averted through the year 2020 for increasing access levels from 50% to 100% (horizontal axis), assuming 85% retention after 3 y. ART eligibility thresholds of are indicated by line colour. Varying retention did not affect trends between access and efficiency for any models.

Figure 5. Impact of treatment by transmission in each CD4 category.

(A) The percentage of all HIV transmissions from individuals in each CD4 cell count category in year 2012, in the no-ART counterfactual simulation. (B) The reduction in incidence in year 2020, for the 80% access and 85% retention scenario, according to the cumulative proportion of transmission that occurs after eligibility (A). For the scenario where all HIV-positive adults are eligible (“all HIV+ eligible"), the percentage of transmission after ART eligibility is the percentage of transmission that occurs after the end of primary HIV infection. Colours for models are the same as in Figures 1 and 2. The BBH, Bendavid, and STI-HIV Interaction models do not estimate the proportion of transmission in each CD4 category and are not included in this figure.

Figure 6. The impact of the existing ART programme in South Africa on HIV prevalence and incidence.

The percentage increase in HIV prevalence (top) and the percentage reduction in HIV incidence rate (bottom) compared to what would have occurred in the absence of any ART for years the 2006 to 2011. These are estimated by comparing HIV prevalence and incidence in a model calibrated to the existing scale-up of ART in South Africa from 2001 to 2011 with a model simulation with no ART provision. The CD4 HIV/ART, Eaton, Goals, Granich, and STI-HIV Interaction models use the same estimates of the number starting ART each year (from [45]). Fraser uses an existing calibration to the ART scale-up in the Western Cape Province. STDSIM is calibrated to the number of people on ART in the Hlabisa subdistrict. Vertical lines on the Eaton and STI-HIV Interaction models indicate 95% credible intervals (CI).