Although there have been great improvements in the number of people living with HIV (PLHIV) who now know their status and are receiving HIV treatment, there are still an estimated 19 million of the 35 million PLHIV who do not know that they are HIV positive and 22 million not accessing antiretroviral therapy (ART).1 The HIV treatment cascade is used as a focal framework for improving the delivery of services across the entire continuum of care—from diagnosis of HIV infection and linkage to care, to initiation of ART, retention in care, and ultimately viral suppression.2 With large global direction for HIV responses now toward increased ART scale-up, for therapeutic and prevention benefits, increased efforts are being made to monitor and report data on the HIV treatment cascade (otherwise known as the HIV care and treatment continuum) at the global, regional, national, and subnational levels.1,3–8
UNAIDS has recently proposed the ambitious targets by 2020 of 90% of PLHIV knowing their status, 90% of people diagnosed with HIV on treatment, and 90% of people on treatment with suppressed viral loads. These “process” targets are being linked with “outcome” targets of a 90% reduction in HIV incidence by 2030 compared with 2010 levels.9 However, achieving the 90-90-90 objectives is no guarantee that settings will see the desired outcomes of large reductions in HIV incidence. This is because the starting point for every setting is different. There will be relatively greater reductions where there are greater improvements from baseline levels; some countries are tending toward saturation of testing and treatment gains and should not expect to achieve the same relative gains.
In this study, we used the HIV diagnosis and treatment coverage levels reported from released HIV treatment cascades, integrated into a flow diagram risk model with current estimates of annual incidence and prevalence in a population to estimate the potential reduction in HIV incidence that is possible through increases in testing and/or treatment coverage.
We extracted information on HIV diagnoses and ART coverage levels from the HIV treatment cascades released for select settings: Australia,3 Brazil,6 China,5 Columbia,6 Denmark,10 France,11 Georgia,12 India,13 Indonesia,14 Kenya,15 the Netherlands,16 the United Kingdom,4 the United States,17 and Vietnam,5 as well as for British Columbia, Canada,8 sub-Saharan Africa,1 and at the global level.1
We constructed an HIV diagnosis and treatment flow diagram as follows (Fig. 1). Consider the total number of PLHIV (N); according to the HIV cascade estimates, a proportion (d) is categorized as diagnosed, leaving a proportion (1 − d) as undiagnosed (1 − d). People diagnosed with HIV may be on ART (a proportion, τ) or not on ART (1 − τ). A proportion of those on ART have suppressed virus (σ) and the remaining (1 − σ) do not have suppressed virus. For simplicity and broad linking of incidence to the cascade, it is assumed that there is a single “average” number of transmissions resulting from all PLHIV who do not have suppressed virus (denoted by β per person per year) and that among people who have suppressed virus this transmission rate is reduced by a factor φ on average. Accordingly, this translates to a simple risk equation relating the average population incidence, representing the number of new infections per year in a population, with the average transmission rate per person per year and parameters of the HIV cascade
Algebraically, this simplifies to
A systematic review of viral load monitoring across 6 retrospective and 2 prospective observational studies in 8 countries of low-, middle-, and high-income revealed an average 70.5% viral suppression for people on ART.18 Therefore, for global calculations, we take σ = 0.705; otherwise, if available, we apply country-specific average viral suppression (Table 1). Based on the HPTN-052 trial, we assume that people with suppressed virus are 96% less infectious (ie, φ = 0.96).33 The other parameters of the risk equation are obtained from, or derived for, the setting of application: given the average number of new infections in a population per year (incidence), estimated number of PLHIV (N), estimated coverage of ART (τ), and proportion of PLHIV who are diagnosed (d), the average number of transmissions per virally unsuppressed PLHIV per year (β) can be calculated for that setting to ensure consistency in the equation.
Plotting the proportion of PLHIV who are aware of their HIV status versus ART coverage among people diagnosed with HIV reveals that there is considerable variability in these important HIV cascade indicators across different settings (Fig. 2A). National, subnational, regional, and global diagnosis and treatment levels currently fall broadly into 1 of the 4 quadrants, split by whether the proportion of PLHIV who are diagnosed, and the coverage of ART among diagnosed PLHIV, are less or greater than 60%. Of the settings for which we obtained HIV cascade information, the United Kingdom, Australia, Denmark, France, the Netherlands, and British Columbia, Canada are in quadrant 1, already achieving high levels of diagnosis and treatment; the United States, Vietnam, and Brazil are in quadrant 2, with high HIV diagnosis levels but low treatment coverage among diagnosed PLHIV; Indonesia, China, Georgia, and India are in quadrant 3 with relatively low diagnosis levels and low treatment coverage; and global estimates, Kenya, Columbia, and broadly sub-Saharan Africa are in quadrant 4, with high treatment coverage among those diagnosed but low diagnoses levels.
Using the risk equation formula, with known values for all parameters for the setting, it is possible to estimate how incidence may change if the proportion of people who are diagnosed and/or the proportion of diagnosed people on ART increases—assuming everything else [eg, risk-related behavior or the proportion of all on ART who have viral suppression remains the same (for illustrations of the reduction in incidence when viral suppression among PLHIV on ART is increased to 90%, see Figure S1A-S1D, Supplemental Digital Content, http://links.lww.com/QAI/A677)]. For each of the settings for which we obtained cascade information, we derived the combinations of increases in diagnoses and ART coverage that could result in relative reductions in HIV incidence of 5%, 10%, 15%, through 70%. Examples of these are shown for global estimates (Fig. 2B), sub-Saharan Africa (Fig. 2C), the United States (Fig. 2D), Australia (Fig. 2E), and Indonesia (Fig. 2F).
The greatest reductions in HIV incidence can be achieved in regions with the greatest gaps in ART coverage and HIV testing. For example, in Indonesia, only an estimated 32% of all PLHIV aware of their infection, and there is 23% ART coverage among those diagnosed,1 a large programmatic scale-up is required if the UNAIDS targets of 90% diagnosis and 90% ART coverage are to be realized. However, if these large increases can be attained, then there is potential for large relative reductions in HIV incidence: for example, around 50%–55% reduction in annual incidence could be expected (see contour curves on Fig. 2F). However, if Indonesia is also able to attain the third “90,” 90% viral suppression among all people on ART, then annual incidence could reduce even further to 65% (see Figure S1D, Supplemental Digital Content, http://links.lww.com/QAI/A677). Similarly, all countries in quadrant 3 with currently low levels of HIV diagnosis and treatment could achieve large incidence reductions with scale-up of diagnosis and ART coverage. In contrast, countries in quadrant 1 already have relatively high levels of HIV diagnosis and ART treatment and therefore, if they attain the UNAIDS targets, then they could not expect to see the same relative reduction in HIV incidence. For example, in Figure 2E, we present that Australia could expect up to a 30% reduction in incidence compared with current levels, if they increased ART coverage by 13% and diagnosis by 4% to reach the 90-90 UNAIDS targets. Countries in quadrant 2 need to prioritize increasing ART coverage among diagnosed PLHIV (see Fig. 2D for the United States as an example) and countries in quadrant 4 need to prioritize increasing HIV diagnoses (see Fig. 2C for sub-Saharan Africa as an example) to maximize incidence reductions. For sub-Saharan Africa, achieving the 90-90 targets, which would require a 45% increase in diagnosis and a 3% increase in ART coverage among diagnosed—in addition to covering 90% of the newly diagnosed, could lead to an approximate 40%–45% reduction in HIV incidence; by also achieving 90% viral suppression among those on ART, the reduction in annual incidence could extend to 50% (see Figure S1B, Supplemental Digital Content, http://links.lww.com/QAI/A677).
The national HIV strategies of most countries involve targets of reductions in HIV incidence. They also generally involve process targets, importantly including increased coverage of ART and reductions in undiagnosed HIV. However, programmatic target levels and what is required to achieve epidemiologic outcome targets can be disconnected. By plotting the proportion of people diagnosed with HIV versus the proportion of diagnosed people on ART, governments can visualize in which quadrant they fall and determine whether to focus their resources on scaling up HIV testing, treatment coverage, or both to maximize incidence reductions. They can then apply the risk equation formula presented herein to determine by how much they need to scale-up HIV testing and treatment coverage to achieve incidence reduction targets or to determine what are attainable incidence reduction targets based on scale-up levels. This approach can support strategic planning and evaluation and lead to realistic target setting. The approach we present here can be applied at the global, regional, national, or subnational levels or may be applied to key-affected populations.
Various barriers exist in scaling up HIV diagnosis and treatment toward achieving incidence reduction targets. For instance, in many settings, universal treatment access is not possible because of a lack of supply of antiretroviral drugs and/or health service infrastructure, challenges in scale-up to meet more inclusive treatment eligibility guidelines, and the large finances required to support lifelong therapy. Many settings around the world have increased HIV testing coverage and frequency; however, stigma and discrimination along with infrastructure, cost, and convenience are some of the barriers to its scale-up. As a result, saturation of diagnosis levels and/or treatment coverage may occur in certain hard-to-reach populations or in certain settings regardless of increased efforts. One limitation of our model is that it assumes linear changes in diagnosis levels and/or treatment coverage and therefore does not capture saturation effects. It is important to note that scaling up HIV testing has an added benefit as people diagnosed with HIV can initiate ART earlier, and once viral suppression is achieved, rates of sexual transmission and adverse clinical events are reduced, which has personal and public health benefits.33
Our study is also limited, most specifically in that it is a simplistic representation of the population statistics and HIV cascade indicators. It is important to take into account that incidence reduction predictions are only as reliable as model inputs, which rely on surveillance and modeling estimates and are subject to considerable bias. In addition, although not performed herein, we strongly recommend performing sensitivity analyses surrounding input values (d, τ, and σ) and transmission rate reductions (φ and θ) to gauge the uncertainty of the incidence reduction estimates when this model is applied by countries in setting realistic targets. The range of uncertainty may be quite large, based on the reliability of the input estimates. A large benefit of our approach is that it provides a rough linkage between indicators in the cascade (importantly, diagnosis levels and treatment coverage) and population incidence and prevalence.
This study only provides HIV cascade estimates for a sampling of countries and regions, and it is not a full systematic review. Therefore, quadrant cut-points identified herein of 60%–60% may not be fully representative of the actual set of quadrants if a full review were undertaken. The current quadrant cut-points were arbitrarily selected based on clustering of the selected countries and regions rather than the relationship between current diagnosis and treatment levels and expected reductions in incidence.
Our model assumes homogeneity of scale-up, which is likely not representative of the true impact of how and in which subpopulations diagnosis levels and/or treatment coverage scale-up is achieved. If interventions for scale-up are targeted at subpopulations with low coverage, they are likely to be more effective than a population-wide increase in testing and treatment. Because measuring the impact of scale-up that was targeted at a subpopulation or regional area is best performed within the specific context, the more narrow the application of this model, that is, in a subpopulation or regional area, the more accurate the predictions are expected to be.
Our model assumes homogeneity among people and only uses very broad average transmission rates across populations of people; it also does not account separately for the primary HIV infection stage of higher infectiousness. Our approach also assumes a viral suppression level of 70% for the global estimate based on a systematic review; however, we applied viral suppression levels for specific regional, national, or subnational contexts where available. If one wishes to change the relative infectiousness for people on suppressive therapy, then the parameters in the risk equation may be changed.
HIV infectiousness in a population can be represented by the basic reproduction number, R0. This is the average number of secondary cases 1 case would cause in a fully susceptible population.34 An R0 of less than 1 signifies that an epidemic is in a phase tending toward elimination, whereas if R0 is greater than 1 then the HIV epidemic is likely to grow. Our analyses revealed that in some settings the average number of new infections per person per year is relatively high, for example, it is 0.143 in Indonesia—such that after just 7 years each person would likely infect 1 additional person; because PLHIV are expected to live longer than this, R0 is greater than 1. This is consistent with reports of increasing incidence in Indonesia. However, in British Columbia, the average number of new infections per PLHIV who does not have suppressed virus is 0.049, but across all PLHIV, including those on ART, the average number of new cases per PLHIV per year is 0.032. Thus, it would take 31 years for a typical PLHIV in British Columbia to cause 1 new infection. It is not known precisely how long PLHIV will live and continue engaging in sexual and injecting behavior, but this duration is suggestive of an R0, which may potentially be less than 1. For the United States, we estimated the average number of transmissions per undiagnosed PLHIV for 2009 as 0.043 and 0.03 across all PLHIV. Skarbinski et al35 reported a rate of transmission of 0.039 per person living with HIV for the same year, which is similar to our estimate.
To reach the UNAIDS 90-90-90 targets by 2020, 90% of all PLHIV need to know their HIV status, 90% of all people with diagnosed HIV infection need to receive sustained ART and 90% of people receiving ART need to have durable viral suppression.6 Worldwide, if diagnosis levels were increased by 44% and ART coverage increased by 10%, it could lead to an approximate 40% reduction in annual HIV incidence at the global level with current viral suppression rates (Fig. 2B), and this could increase to a 50%–55% annual reduction if 90% viral suppression is also achieved (see Figure S1A, Supplemental Digital Content, http://links.lww.com/QAI/A677). These annual reductions could add to substantial cumulative reductions over time, but it would be necessary to implement substantial change early.36
Finally, it is important to consider the reduction of absolute incidence in addition to the relative reduction of incidence projected in this study. In settings with high disease burden, smaller relative reductions could require large investment and effort but result in large absolute reductions in incidence. For countries where HIV diagnosis and treatment data are available, the straightforward approach presented here may be useful to determine by how much HIV testing and treatment coverage must be scaled up to achieve HIV incidence reduction targets. This is important because scale-up amounts are context specific and relative attainable incidence reductions are inversely related to baseline levels. Greatest reductions in HIV incidence can be achieved in regions where the greatest gaps in ART coverage and HIV testing exist. In other settings where diagnosis and ART coverage are already relatively high, the same relative reduction in HIV incidence could not be expected.
1. UNAIDS. The Gap Report. UNAIDS, Geneva, Switzerland; 2014.
2. WHO. Meeting Report on Framework for Metrics to Support Effective Treatment as Prevention. Geneva, Switzerland: WHO; 2012.
3. Kirby Institute, University of New South Wales. HIV in Australia. Annual Surveillance Report Supplement. Kirby Institute, University of New South Wales; 2014, Sydney, Australia.
4. Delpech V, Brown A, Conti S, et al.. Reducing Onward Transmission: Viral Suppression Among Key Population Groups Living With HIV in the United Kingdom. Presented at: Annual Conference of the British HIV Association; 2013; Manchester, United Kingdom.
5. Lo YR, Kato M, Phanuphak N, et al.. Challenges and potential barriers to the uptake of antiretroviral-based prevention in Asia and the Pacific region. Sex Health. 2014;11:126–136.
6. UNAIDS. Ambitious Treatment Targets
: Writing the Final Chapter of the AIDS Epidemic. UNAIDS, Geneva, Switzerland; 2014; Discussion paper.
7. Hall HI, Frazier EL, Rhodes P, et al.. Continuum of HIV Care: Difference in Care and Treatment by Sex and Race/Ethnicity in the United States. Presented at: XIX International AIDS Conference; 2012; Washington, DC.
8. Nosyk B, Montaner JSG, Colley G, et al.. The cascade of HIV care in British Columbia, Canada, 1996–2011: a population-based retrospective cohort study. Lancet Infect Dis. 2014;14:40–49.
9. Sidibe M. The Last Climb: Ending AIDS, Leaving No One Behind. UNAIDS Speech. Presented at: 20th International AIDS Conference Opening Session; July 20-25, 2014, Melbourne, Australia.
10. Helleberg M, Häggblom A, Sönnerborg A, et al.. HIV care in the Swedish-Danish HIV cohort 1995-2010, closing the gaps. PLoS One. 2013;8:e72257.
11. Supervie V, Costagliola D. The spectrum of engagement in HIV care in France. Presented at: 20th Conference on Retroviruses and Opportunistic Infections; March 6, 2013; Atlanta, GA.
12. Chkhartishvili N, Chokoshvili O, Sharvadze L, et al.. The cascade of care in the eastern European country of Georgia. Presented at: International AIDS Society (IAS); June 30–July 3, 2013; Kuala Lumpur, Malaysia.
13. Alvarez-Uria G, Pakam R, Midde M, et al.. Entry, retention, and virological suppression in an HIV cohort study in India: description of the cascade of care and implications for reducing HIV-related mortality in low- and middle-income countries. Interdiscip Perspect Infect Dis. 2013;2013:384805.
14. Tarmizi S. Introduction of the strategic use of ARVs for treatment & prevention in Indonesia: from policy development to implementation. Presented at: International AIDS Conference; July 20–25, 2014; Melbourne, Australia.
15. National AIDS and STI Control Programme, Ministry of Health. Kenya AIDS Indicator Survey. Kenya, Africa: National AIDS and STI Control Programme, Ministry of Health; 2012.
16. Stitching HIV Monitoring (SHM), Dutch Minister of Health, Welfare and Sports. Monitoring Report 2013, Human Immunodeficiency Virus (HIV) Infection in the Netherlands. Stitching HIV Monitoring (SHM), Dutch Minister of Health, Welfare and Sports; 2013, Amsterdam, The Netherlands.
17. Zhang L, Chow EPF, Jahn HJ, et al.. High HIV prevalence and risk of infection among rural-to-urban migrants in various migration stages in China: a systematic review and meta-analysis. Sex Transm Dis. 2013,40:136–147.
18. Bonner K, Mezochow A, Roberts T, et al.. Viral load monitoring as a tool to reinforce adherence: a systematic review. J Acquir Immune Defic Syndr. 2013;64:74–78.
19. Raymond A, Hill A, Pozniak A. Large disparities in HIV treatment cascades between eight European and high-income countries—analysis of break points; 2014.
20. Hall H, Frazier E, Rhodes P, et al.. Differences in human immunodeficiency virus care and treatment among subpopulations in the United States. JAMA Intern Med. 2013;173:1337–1344.
21. Brazilian Ministry of Health, Secretariat of Health Surveillance, Department of STD, AIDS and Viral Hepatitis. Global AIDS Response, Progress Reporting, Narrative Report, Brazil. Brazilian Ministry of Health, Secretariat of Health Surveillance, Department of STD, AIDS and Viral Hepatitis; 2014.
22. UNAIDS. AIDSinfo Database. UNAIDS, Geneva, Switzerland; 2013.
23. UNAIDS. Global Report: UNAIDS Report on the Global AIDS Epidemic 2013. UNAIDS, Geneva, Switzerland; 2013.
24. Public Health England. HIV in the United Kingdom. London, United Kingdom: Public Health England; 2012.
25. World Health Organization in partnership with UNICEF and UNAIDS. Global Update on HIV Treatment 2013: Results, Impact and Opportunities. World Health Organization in partnership with UNICEF and UNAIDS, Geneva, Switzerland; 2013.
26. Cuong DD, Agneskog E, Chuc NTK, et al.. Monitoring the efficacy of antiretroviral therapy by a simple reverse transcriptase assay in HIV-infected adults in rural Vietnam. Future Virol. London, UK, 2012;7:923–931.
27. He N, Duan S, Ding Y, et al.; for the China National, HIV. Prevention Study Group. Antiretroviral therapy reduces HIV transmission in discordant couples in rural Yunnan, China. PLoS One. San Francisco, CA, 2013;8:e77981.
28. BC Centre for Disease Control. Control BCD. HIV in British Columbia: Annual Surveillance Report 2012. BC Centre for Disease Control; 2013.
29. National AIDS Control Council of Kenya. Kenya AIDS Response Progress Report: Progress towards Zero. National AIDS Control Council of Kenya; 2014, Nairobi, Kenya.
30. Stichting HIV Monitoring (SHM), Dutch Minister of Health, Welfare and Sport. Monitoring Report 2013, Human Immunodeficiency Virus (HIV) Infection in the Netherlands—Web Appendix. Stichting HIV Monitoring (SHM), Dutch Minister of Health, Welfare and Sport; 2013.
31. Public Health England. HIV in the United Kingdom. London, United Kingdom: Public Health England; 2013.
32. Centers for Disease Control and Prevention. Diagnosed HIV infection among adults and adolescents in Metropolitan Statistical Areas—United states and Puerto Rico, 2011. In: HIV Surveillance Supplemental Report. Revised ed. Centers for Disease Control and Prevention; 2013, Atlanta, GA.
33. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
34. Dietz K. The estimation of the basic reproduction number for infectious diseases. Stat Methods Med Res. 1993;2:23–41.
35. Skarbinski J, Rosenberg E, Paz-Bailey G, et al.. Human immunodeficiency virus transmission at each step of the care continuum in the United States. JAMA Intern Med. 2015.
36. UNAIDS. 90-90-90 an Ambitious Treatment Target to Help End the AIDS Epidemic. UNAIDS, Geneva, Switzerland; 2014.