Exploring the population-level impact of antiretroviral treatment: the influence of baseline intervention context
Mishra, Sharmisthaa,i; Mountain, Elisaa; Pickles, Michaela; Vickerman, Peterb; Shastri, Sureshc; Gilks, Charlesd; Dhingra, Nandini K.e; Washington, Reynoldf,j; Becker, Marissa L.g; Blanchard, James F.g; Alary, Michelh; Boily, Marie-Claudea
aDepartment of Infectious Disease Epidemiology, Imperial College London
bSocial and Mathematical Epidemiology, London School of Hygiene and Tropical Medicine, London, UK
cGovernment of Karnataka, Bangalore, India
dSchool of Population Health, University of Queensland, Brisbane, Australia
eUNAIDS India, New Delhi
fKarnataka Health Promotion Trust, Bangalore, India
gDepartment of Community Health Sciences, Centre for Global Public Health, University of Manitoba, Winnipeg
hCentre de recherche du CHU de Québec, Département de médecine sociale et préventive, Université Laval, Québec City, Québec, Canada.
iDivision of Infectious Diseases, St. Michael's Hospital, University of Toronto, Toronto, Canada
jSt. John's Medical College Research Institute, Bangalore, India.
Correspondence to Dr Sharmistha Mishra, Department of Infectious Diseases Epidemiology, Imperial College London, St Mary's Campus, Norlfolk Place, London W2 1PG, UK. E-mail: firstname.lastname@example.org
Received 30 April, 2013
Revised 12 September, 2013
Accepted 7 October, 2013
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website ( http://www.AIDSonline.com).
To compare the potential population-level impact of expanding antiretroviral treatment (ART) in HIV epidemics concentrated among female sex workers (FSWs) and clients, with and without existing condom-based FSW interventions.
Mathematical model of heterosexual HIV transmission in south India.
We simulated HIV epidemics in three districts to assess the 10-year impact of existing ART programs (ART eligibility at CD4+ cell count ≤350) beyond that achieved with high condom use, and the incremental benefit of expanding ART by either increasing ART eligibility, improving access to care, or prioritizing ART expansion to FSWs/clients. Impact was estimated in the total population (including FSWs and clients).
In the presence of existing condom-based interventions, existing ART programs (medium-to-good coverage) were predicted to avert 11–28% of remaining HIV infections between 2014 and 2024. Increasing eligibility to all risk groups prevented an incremental 1–15% over existing ART programs, compared with 29–53% when maximizing access to all risk groups. If there was no condom-based intervention, and only poor ART coverage, then expanding ART prevented a larger absolute number but a smaller relative fraction of HIV infections for every additional person-year of ART. Across districts and baseline interventions, for every additional person-year of treatment, prioritizing access to FSWs was most efficient (and resource saving), followed by prioritizing access to FSWs and clients.
The relative and absolute benefit of ART expansion depends on baseline condom use, ART coverage, and epidemic size. In south India, maximizing FSWs’ access to care, followed by maximizing clients’ access are the most efficient ways to expand ART for HIV prevention, across baseline intervention context.
The clinical benefits of antiretroviral treatment (ART) soon after HIV infection has led some high-income countries to recommend ART when CD4+ cell counts fall below 500 cells/μl or immediately after diagnosis [1,2]. Evidence of the preventive benefits of viral suppression within partnerships also supports early ART initiation to prevent HIV spread . As a result, international guidelines recently changed the ART eligibility from CD4+ cell count 350 cells/μl or less to CD4+ cell count 500 cells/μl or less (herein referred to as CD4+ cell ≤350 or CD4+ cell ≤500), while continuing to advocate for expanded access to screening and pre-ART care . However, when faced with finite (and often diminishing) resources, low-income and middle-income countries must consider how best to maximize the preventive potential of ART for their local context and improve on their existing HIV prevention strategies . For example, could prioritized ART expansion to key populations be an efficient strategy ? The recent guidelines noted a general lack of evidence to support prioritizing eligibility or access to key populations . Since then, one modeling study found that prioritized eligibility for key populations could be highly effective and cost-effective, but this strategy remains underresearched, even in concentrated HIV epidemics .
South India's heterosexual HIV epidemics are driven largely by commercial sex, but vary in epidemic size and in the nature of sex work (including rates of turnover) [8–12]. Many regional epidemics are now declining due to condom-based, targeted female sex worker (FSW) interventions [13–18]. But their reach varies across south India, and self-reported condom use among FSWs ranged from 32 to 98% in 2007 [10,15,16,19–23]. A simultaneous, though slower roll-out of ART also began in 2004, and eligibility criteria changed from CD4+ cell count of 250 or less to CD4+ cell count of 350 or less in 2012 . Although an estimated 19.1% of India's 2.3 million adults with HIV are currently on ART (as of January 2012) , treatment coverage by 2010 varied between 4 and 30% in the state of Karnataka . Pre-ART and ART care is delivered uniformly to all segments of society within the public healthcare system, but HIV screening is prioritized for FSWs via referral for 6-monthly screening in the targeted intervention program [13,15,16]. Thus, in several states, HIV screening among FSWs is higher than in the overall population [15,16]. For example, in Belgaum, a district in the south Indian state of Karnataka, 74% of HIV-negative FSWs were screened for HIV in 2010, whereas only 8.0% of HIV-negative general population women had ever received an HIV test by 2010 . Like condom use, rates of 6-monthly HIV testing among FSWs vary widely across south Indian states, from 6.7 to 57% [15,16]. The extent to which sex work turnover and baseline condom use influence CD4+ cell distribution and eligibility across risk groups, and thereby influence the future impact of the existing ART program, remains unknown. Given that baseline ART coverage vary across districts, its influence on the future impact of expanding ART strategies in south India also requires study.
In this article, we simulate HIV epidemics from three representative but distinct regions of Karnataka to examine the potential impact of the existing ART program (medium-to-good ART coverage, CD4+ cell eligibility ≤350) beyond that achieved by existing condom-based interventions. We then estimate the incremental benefit of expanding ART with strategies that include changing CD4+ cell eligibility criteria (from ≤350 to ≤500, or to any CD4+ cell count); increasing access to care (maximize HIV screening, linkage, and retention in pre-ART care); and prioritizing expansion to FSWs or FSWs and clients. A systematic examination of ART impact across settings with different epidemic contexts and existing interventions will provide decision-makers with more nuanced evidence to support alternate ART strategies.
As of 2009, Karnataka had a population of 53 million adults over age 15, including an estimated 244 000 people living with HIV, and 134 691 FSWs . The modeled districts include urban Belgaum (adult population, 914 115), Mysore (adult population, 909 204), and Shimoga (adult population, 492 000) . Table 1[9,10,24,25,27–31] summarizes key epidemic, condom use, and ART-related features of each district. ART roll-out in Belgaum, Mysore, and Shimoga began in 2006, 2007, and 2008, respectively. Individuals receive HIV screening or symptom-directed testing via government integrated counseling and testing centers (ICTCs) [13,27]. FSW screening is facilitated by the targeted intervention programs [15,16]. Belgaum has a larger overall HIV epidemic with little turnover in sex work, medium baseline ART coverage (13.5%). Nearly half of individuals registered in care in Belgaum were on treatment by 2010 [24,27]. Mysore has an intermediate-size epidemic with rapid turnover in sex work, good baseline ART coverage (20.3%), with most individuals registered in care on treatment by 2010 [24,27]. Shimoga has the smallest HIV epidemic of the three districts, with intermediate turnover in sex work, medium baseline ART coverage (11.9%), and nearly half of individuals registered in care on treatment by 2010 [24,27]. Across the three districts, an estimated 4.8–10.9% of adults living with HIV in 2010 were registered in care, but were not on ART [24,27]. Across the state, retention within the first 12 months of entering HIV care was 88.5% in 2011 .
District-specific behavioral and HIV screening data were obtained from FSW mapping and enumeration (2010 [24,32]), serial biological and behavioral surveys of FSWs (2004–2011 [10,17]) and clients (2005–2012), general population surveys (2007, 2010) [25,33,29], serial polling booth surveys , a behavioral cohort study among FSWs [35–37], a cross-sectional [38,39] study among HIV-positive FSWs, a cohort study among HIV-positive people in the general population , district-specific FSW intervention program records (FSW HIV testing and pre-ART registration), and a comprehensive care program for HIV-positive adults . Data on ICTC test volume and positivity rates, ART coverage (2010), linkage to care, pre-ART registration, and retention in care (until August 2012) were obtained from the Karnataka State AIDS Prevention Society, the UNAIDS Country Office, and the Karnataka Health Promotion Trust [24,27,30]. Biological and other India-specific data (CD4+ cell distribution at diagnosis and ART initiation) were drawn from the literature.
We developed a deterministic, compartmental model to simulate heterosexual HIV transmission across the following risk groups: high-volume FSWs, low-volume FSWs, clients, former FSWs, former clients, low-risk individuals, and individuals who are no longer sexually active (Supplementary Digital Content 1, Figure S1, http://links.lww.com/QAD/A431). Individuals could engage in one or more partnerships (occasional commercial, regular commercial, or main). Former clients/FSWs were assumed to have the same number and type of partnerships, and the same level of HIV screening, as low-risk individuals without a history of commercial sex.
Individuals become susceptible to HIV when they become sexually active. Following HIV infection, untreated individuals progress through acute HIV, CD4+ cell count more than 500, CD4+ cell count 350–500, CD4+ cell count 250–350, and CD4+ cell count 250 or less stages, each of which is associated with HIV-attributable mortality (Supplementary Digital Content 2, Figure S2, http://links.lww.com/QAD/A432). In the model, HIV-infected individuals can enter pre-ART care after HIV screening and linkage into pre-ART clinics. A fraction of HIV-infected individuals are retained in pre-ART care, and initiate treatment when they meet CD4+ cell eligibility criteria. In addition, a fraction of HIV-positive individuals at any CD4+ cell count can initiate treatment directly to reflect the development and detection of WHO stage 3–4 diseases (for example, active tuberculosis), including individuals not in pre-ART care. Following ART initiation, it is assumed that an average of 4–6 months is required to achieve viral suppression . ART reduces HIV-attributable mortality by 20–50% in the first year of treatment, and by 50–90% thereafter, depending on the pretreatment CD4+ cell count [43–55]. After achieving ‘viral suppression’, HIV infectivity is reduced by 80–97% to reflect different levels of adherence (which is not explicitly modeled) . Individuals may discontinue ART or experience virological treatment failure and re-initiate effective ART if they develop WHO stage 3–4 disease [13,27].
For the existing baseline, condom use by FSWs was assumed to increase linearly in a piece-wise fashion between three rounds of FSW surveys (Table 1), and to remain constant after the last available survey. The rate of increase in condom use prior to the first FSW survey was extrapolated from the rate of increase between the first two FSW surveys. HIV screening and linkage to pre-ART care increased from ART roll-out to the last year for which data were available (Table 1), and remained constant thereafter. We assumed that screening tests cannot detect infection during the first 30 days of acute HIV .
We sampled over a prior range of parameters and calibrated the model to the best available regional data using a Bayesian framework [22,57]: group-specific HIV prevalence trends, population size (2001 and 2011), ICTC positivity (2011), % adults living with HIV currently in care (2009, 2010), overall ART coverage (2010), cumulative loss to follow-up, and CD4+ cell distribution at entry into pre-ART care and ART initiation (up to 2011), to obtain 50 different parameter sets per district [17,22]. Parameter sets were retained if the above model outputs reproduced the observed data within the 95% confidence interval of each data point. The calibrated parameter sets reflect potential ‘realizations’ of the observed HIV epidemic in each district, and form our baseline scenario (Table 2, ‘existing targeted intervention, existing ART’). See text, Supplementary Digital Content 3, http://links.lww.com/QAD/A433 for model equations, and calibration details, including prior and posterior parameters.
Past impact of the existing antiretroviral treatment program
We assessed the impact achieved to date of the existing condom use and ART programs from their inception until 2014 (Table 2), by comparing to a scenario in which condom use remained low at pre-2004 levels (‘no targeted intervention, existing ART’), or HIV screening remained equally low across risk groups at levels currently observed in the general population (‘existing targeted intervention, poor ART’), or an alternate baseline with ‘no targeted intervention, and poor ART’. Outcomes include absolute reduction in HIV prevalence (across risk groups) and the fraction of HIV infections averted (prevented fraction) in the total population compared with the alternate baseline (Table 2).
Future impact of the existing antiretroviral treatment program
We assessed the future impact of the existing ART program from 2014 to 2024, beyond that achieved by high condom use, by comparing against a scenario with no new ART initiation after 2014 (Table 2). We repeated this assessment with low condom use (‘no targeted intervention, existing ART’). We calculated the prevented fraction and program efficiency (the prevented fraction per 100 person-years of ART) in the total population. To interpret results across districts and the influence of existing condom use, we compared the CD4+ cell distribution in 2012 across districts and risk groups in the absence of ART (i.e., if there had never been an ART program, Table 2).
Incremental benefit of expanding antiretroviral treatment compared with sustaining the existing antiretroviral treatment program
We compared the incremental benefit of expanding ART in 2014 with strategies outlined in Table 2, versus maintaining the existing ART program. If ART eligibility changed, it was assumed to take place immediately. To maximize access to care, we linearly increased annual HIV screening between 2014 and 2016, such that by 2016, 80% of the risk group(s) underwent routine screening while linkage and retention in pre-ART care rose to 100%. Thus, we assumed that 20% of HIV-infected adults remain unreached by HIV screening efforts. The 10-year outcomes include the prevented fraction, and program efficiency [the fraction (relative impact) and number of HIV infections averted (absolute impact) for every additional 100 person-years of ART] in the total population, beyond gains achieved by the existing targeted interventions. An expansion strategy was considered resource saving if it required less person-years of ART than sustaining the existing ART program.
Sensitivity analysis: influence of baseline intervention context on the incremental impact of expanding antiretroviral treatment
We performed a sensitivity analysis to examine the extent to which baseline condom use and ART coverage influence the incremental benefit of expanding ART. We used the alternate baseline (Table 2) for each district to represent regions with low condom use and poor ART coverage (3–4% ART coverage in 2010 ) due to the absence of an FSW intervention, such that FSWs and clients are screened for HIV at the same level as the general population, consistent with observed levels of low FSW HIV testing in some states [15,16].
Past impact of the existing antiretroviral treatment program
Figure 1 suggests that without existing targeted intervention and poor ART coverage, FSW HIV prevalence in 2014 would have been 30–39, 25–34, and 11–15% by 2014 in Belgaum, Mysore, and Shimoga, respectively, compared with 7–13, 3–5, and 3–4% respectively, with existing condom use and medium-good ART coverage (see Figure S3, Supplementary Digital Content 4, http://links.lww.com/QAD/A434, for other risk groups). The model suggests that condom use played a larger role than ART in reducing HIV prevalence across risk groups (Fig. 1, Figure S3). Compared with ‘no targeted intervention, poor ART’, the targeted intervention alone prevented 27–47% (2004–2014), 29–55% (2004–2014), and 31–48% (2004–2014) of HIV infections in the total population, whereas ART alone prevented 5–11% (2006–2014), 6–18% (2007–2014), and 5–9% (2008–2014) of HIV infections in Belgaum, Mysore, and Shimoga, respectively. However, their combination prevented 30–50, 32–58, and 33–50% of HIV infections, respectively by 2014. That is, existing ART averted an additional 2–3% on top of what was achieved with the targeted intervention alone, suggesting potential redundancy in their combination.
Future impact of the existing antiretroviral treatment program
In the presence of existing targeted interventions, the existing ART program is expected to prevent 11–28, 13–23, and 12–28% of HIV infections over 10 years in Belgaum, Mysore, and Shimoga respectively, translating into 292–1690, 180–2198, and 78–320 new HIV infections averted compared with a scenario in which there is no new ART after 2014 (Fig. 2a). The slightly smaller impact in Mysore, compared with Belgaum, can be explained by fewer FSWs being treatment eligible (CD4+ cell count ≤350, Fig. 2b) due to faster turnover in sex work. Hence, despite similar HIV screening among FSWs across districts (Table 2), a smaller fraction of Mysore FSWs were on treatment by 2024 (13 versus 37% in Belgaum). However, this effect on ART impact was mitigated by higher HIV screening rates among clients in Mysore (Table 2), resulting in higher ART coverage among clients (64 versus 51% in 2024).
Compared with the scenario in which no new ART is initiated after 2014, the prevented fraction for every 100 person-years of ART was greatest in Shimoga, which represented the smallest epidemic (Fig. 2c). If condom use was assumed to remain at 2004 levels (‘no targeted intervention, good ART’), the efficiency of existing ART was reduced (Fig. 2c). Because condom use reduces HIV incidence, a larger fraction of HIV-infected adults are treatment eligible and higher ART coverage is achieved (Fig. 2d). For example, in Shimoga, ART coverage with the existing ART program in 2024 was predicted to reach 36–74 versus 24–54% with and without a targeted intervention, respectively.
Incremental benefit of expanding antiretroviral treatment compared with sustaining the existing antiretroviral treatment program
Supplementary Digital Content 5, Table S1, http://links.lww.com/QAD/A435 summarizes ART coverage across risk groups, and outcomes from each proposed strategy. Figure 3a–c shows the 10-year incremental impact of changing eligibility criteria versus sustaining the existing ART program in Belgaum (see Figure S4 and S5, Supplementary Digital Content 6 and 7, for Mysore and Shimoga, respectively). Figure 3d–f shows the 10-year impact of maximizing access with eligibility maintained at CD4+ cell count 350 or less in Belgaum (Figure S4d-f and Figure S5d-f, Supplementary Digital Content 6, http://links.lww.com/QAD/A436 and 7, http://links.lww.com/QAD/A461 for Mysore and Shimoga respectively). Overall, maximizing access to all risk groups was predicted to have a larger impact than increasing eligibility to all risk groups, because higher ART coverage could be expected with the former by 2024 (for example, a median of 82 versus 66%, respectively, in Belgaum; Table S1). This is due to ‘less than maximal’ HIV screening, linkage, and retention in pre-ART care at baseline (Table 1 and Supplementary Digital Content 3, http://links.lww.com/QAD/A433).
Increasing eligibility averted 1–20, 1–12, and 1–15% of infections in Belgaum, Mysore, and Shimoga, respectively (Figures 3a, S4a, S5a; Table S1). The most efficient approach when changing ART eligibility was to prioritize FSWs for ART initiation at CD4+ cell count of 500 or less (Fig. 3b–c; Figure S4b-c; Figure S5b-c).
Maximizing access to all risk groups averted an additional 29–46, 32–53, and 30–46% of infections in Belgaum, Mysore, and Shimoga, respectively (Fig. 3d, Figure S4d, Figure S5d). Because HIV screening was already high among FSWs, further prioritizing their access prevented a small fraction and number of HIV infections, yet also required fewer person-years of treatment than sustaining the existing ART program (Table S1) – that is, this was a ‘resource-saving’ strategy. Prioritizing access to both FSWs and clients prevented a median of 35, 74, and 39 HIV infections per additional 100 person-years of ART in Belgaum, Mysore, and Shimoga, respectively (Fig. 3c-d, S4c-d, S5c-d; Table S1).
Across all expansion strategies, for every additional person-year of treatment, the prevented fraction was largest in Shimoga (the smallest epidemic), while the largest number of HIV infections were averted in Belgaum (the largest epidemic). Maximizing access to FSWs, followed by maximizing access to FSWs and clients, was resource-saving or the most efficient approach to ART expansion in the existing baseline.
Sensitivity analysis: influence of baseline intervention context on the incremental impact of expanding antiretroviral treatment
In similar settings but in which HIV screening rates are low (alternate baseline: ‘no targeted intervention, poor ART’), changing eligibility criteria for ART had a limited impact (Fig. 3a–b, Figure S4a-b, Figure S5a-b, Table S1). Although maximizing access prevented a larger fraction in the alternate baseline than the existing baseline, it was less efficient except when FSWs’ access was prioritized (Table S1). In the alternate baseline, the large increase in annual HIV screening among FSWs (4–8 to 80%) prevented a median of 21, 13, and 22% of HIV infections, and required 530, 491, and 95 fewer person-years of ART compared with maintaining its ‘poor ART’ program (Table S1b). In the existing baseline, a smaller increase in FSW screening was required to achieve our 80% target, which prevented a median of 0.7, 0.5, and 0.8% of HIV infections with 33, 20, and 8 fewer person-years of ART, compared to maintaining its existing ART program. Across expansion strategies, a larger number of infections were averted for every additional person-year of ART (Fig. 3f, Figure S4f, Figure S5f) in the alternate, compared to the existing baseline, because the former represented a larger HIV epidemic (Fig. 1, Figure S3). Like the existing baseline, the most efficient approach involved maximizing access to FSWs (resource-saving), followed by maximizing access to FSWs and clients.
We examined the potential impact of the existing ART program and the expected benefit of improving it across different settings in south India. In addition to the preventive impact achieved by existing targeted interventions, the existing ART program could avert 11–28% of remaining HIV infections between 2014 and 2024 across the three districts studied here. Compared to existing ART program, maximizing access to all risk groups could prevent an additional 29–53% of HIV infections, whereas increasing eligibility (without increasing access) could prevent an additional 1–15% of HIV infections. Across expansion strategies, and for every additional person-year of ART, the relative impact (prevented fraction) was largest in Shimoga (smallest epidemic), and the absolute impact (number of infections averted) was largest in Belgaum (largest epidemic). The incremental benefit of ART expansion was highly sensitive to the presence or absence of targeted interventions and baseline ART coverage. Across baseline conditions, the most efficient approach could be maximizing access to FSWs (resource-saving), followed by maximizing access to FSWs and clients.
Our modeling suggests that while the existing ART program in south India likely reduced HIV prevalence across risk groups and prevented new HIV infections in the total population, condom use probably played a larger role than ART. The estimated impact of condom use alone in the study districts was similar to that reported from previous models .
The future impact of the existing ART program is expected to vary across settings, partly due to differences in baseline ART coverage and turnover in sex work, and mostly due to differences in baseline condom use. High condom use leads to a declining epidemic wherein a larger fraction of the population has late-stage HIV  and is eligible for ART (CD4+ cell count ≤350). Moreover, high condom use enables each person-year of ART to prevent a larger fraction of HIV infections because epidemics are already closer to elimination [59,60]. Thus, the existing ART program could be more efficient if baseline condom use is high. Similarly, the existing ART program is most efficient in smaller epidemics, like in Shimoga. Given the diversity in epidemic size and level of FSW interventions across India [13–16,61], the efficiency of existing ART in preventing HIV is expected to vary considerably across India.
Key determinants of the impact of increasing eligibility criteria are early diagnoses and high linkage and retention in care. In Karnataka, 80% of adults diagnosed with HIV are linked to care within a year; of those who enter pre-ART care, 11% are lost to follow-up in their first year before ART initiation . Regional data from 2010 suggest that less than 11% of adults living with HIV were in pre-ART care but not yet on ART, and this was before national guidelines changed to CD4+ cell count of 350 cells/μl or less [24,27]. Based on these data, our modeling suggests that overall, maximizing access to all (at CD4+ cell count ≤350 cells/μl) could potentially achieve a larger impact than further expanding eligibility (without increasing access).
The added benefit of expanding ART also depends on the reach of existing ART programs, which varies across India [24,62]. In six of Karnataka's 30 districts, less than 10% of adults living with HIV were on ART by 2010 . Across India, the 6-monthly rate of HIV testing among FSWs in 2011 was less than 10% in six out of 33 states [15,16]. Under existing baseline conditions in the districts modeled here, further prioritizing FSWs’ access could reduce the need for ART over the next 10 years. Increasing access for FSWs and clients would generally require more ART than maintaining the existing ART program, but will be more efficient than increasing access for everyone. If we were to change eligibility criteria, then doing so for FSWs first would be the most efficient approach. However, if baseline conditions are different – low condom use and poor ART coverage – then, as expected, changing eligibility criteria will have little impact. Under these conditions, the most efficient approach to expanding ART is to expand access to care, starting with FSWs. Overall, low condom use and poor ART coverage reduce the efficiency of ART expansion when the prevented fraction is measured. The exception is prioritizing FSWs’ access to care – a strategy that is more efficient (and ‘resource saving’) in settings without a targeted intervention and poor baseline ART coverage. However, it may not be feasible to prioritize FSWs’ access to care in settings without an established targeted intervention that has already addressed barriers to reaching and engaging FSWs in the context of HIV prevention [15,16,63]. Of note, regions with low condom use and poor ART coverage also have larger epidemics, and so for every person-year of ART, expanding ART prevents more HIV infections than in smaller epidemics with high condom use and better ART coverage.
Currently, HIV care and FSW interventions are implemented separately [13,15,16,27]. Established FSW interventions are well positioned to facilitate FSWs’ access to care, having already adopted such an approach with referral for 6-monthly HIV screening [15,16]. Our findings suggest that in regions with existing targeted interventions, even with medium-to-good ART coverage and high FSW HIV screening, FSW programs should sustain their efforts on condom use [15,16,64]; integrate FSWs’ linkage and retention in pre-ART care as part of their HIV prevention strategy; and expand access for clients to achieve the maximal preventive benefit of each additional person treated. Important next steps include operational research  to determine how best to integrate HIV prevention and care for FSWs and clients in settings like India with vertical delivery of prevention and ART services [13,15,16,27].
This study adds to a very small, but growing body of modeling literature to suggest that prioritized ART expansion to FSWs could be an effective and efficient strategy [7,66]. However, the feasibility of a prioritized strategy will depend on our ability to identify, reach, and engage FSWs and/or clients . As described above, existing FSW interventions in south India offer a unique opportunity to leverage their program coverage, and to examine the feasibility and individual-level impact of not only integrating prevention and HIV care, but also prioritizing ART expansion for FSWs.
To our knowledge, this is the first systematic examination of the epidemiological impact of ART under different baseline conditions, and the third to examine prioritized ART expansion for FSWs [7,66]. Strengths of this study include the data available for model parameterization and calibration. Nonetheless, there are limitations in the data, particularly with respect to group-specific ART coverage by CD4+ cell count, retention in pre-ART care, and treatment discontinuation. As a result, we assumed that linkage and retention in pre-ART care were similar across risk groups. However, if FSWs are diagnosed at lower CD4+ cell counts and/or less likely to be linked and retained in pre-ART care than the general population, it may influence the potential impact of prioritizing FSWs for expanded eligibility.
There are also limitations with the scenarios we modeled. The way in which we modeled maximal access to care (i.e., 80% annual HIV screening, 100% linkage and retention in pre-ART care) represents an idealized scenario. Although achieving this level of HIV screening may not be entirely realistic for the general population, it has already been achieved for FSWs in several regions – because frequent HIV screening has been integrated into existing FSW interventions . Efforts to increase linkage and retention in pre-ART care are underway in south India, via decentralized clinics for medication delivery and uptake and an outreach program to ’find’ individuals lost to care . As a result, linkage and retention in Karnataka in 2011 are higher than some other regions of the world [28,67]. As recommended above, existing FSW interventions offer an important opportunity to improve FSWs’ linkage and retention in pre-ART care, and efforts to integrate HIV care into existing FSW interventions are now underway in several African countries [65,68].
Finally, whereas findings from this study speak to the efficiency of various expansion strategies, they do not account for costs incurred, including costs of HIV screening. We also do not examine health benefits (life-years saved or disability-adjusted life years gained) in this study, and have not discussed discounting resource or health benefits. Important and planned next steps include a cost–effectiveness analysis of the different ART strategies across baseline conditions.
In summary, ART programs in concentrated HIV epidemics should consider their baseline intervention-specific (targeted condom use and existing access to HIV care) and local epidemic context when expanding ART for HIV prevention. In south India, maximizing and prioritizing access to HIV care among FSWs, followed by prioritizing access to FSWs and clients represents the most efficient way to expand ART for HIV prevention.
Members of Strategic Epi-ART Modelling Team are as follows: Stephen Moses (Department of Community Health Sciences, Centre for Global Public Health, University of Manitoba, Winnipeg, Canada), Shiva Halli (Department of Community Health Sciences, Centre for Global Public Health, University of Manitoba, Winnipeg, Canada), B.B. Rewari (National AIDS Control Organization, New Delhi, India), Taoufik Bakkali (UNAIDS India, New Delhi, India), Nalini Chandra (UNAIDS India, New Delhi, India), B.M. Ramesh (Karnataka Health Promotion Trust, Bangalore, India).
The authors would like to thank T. Raghavendra, Mahesh Doddamane, and Dr Gajanan Pise (Karnataka Health Promotion Trust) for providing ART-related program data from the FSW intervention, Dr Shajy Isac for providing district-specific census estimates, and Drs Ravi Prakash and Pradeep Banandur for assistance with data extraction. They thank the Karnataka State AIDS Prevention Society for their assistance. They thank Holly Prudden and Dr Kate Mitchell (London School of Hygiene and Tropical Medicine) for helpful discussions. They thank Ellen McRobie, Drs Jeff Eaton and Timothy Hallett from Imperial College London and the HIV Modelling Consortium for helpful discussion and incorporating this work within a broader context in the 2013 World Health Organization HIV Treatment Guidelines. S.M. is supported by a Canadian Institute of Health Research Fellowship and a Royal College of Physicians and Surgeons of Canada Detweiler Travelling Fellowship.
M.C.B., N.K.D., C.G., S.M. conceived of, motivated, and designed the study. S.M. designed, developed, and analyzed the model. M.C.B. and M.P. contributed to model design. E.M., S.M. contributed to data S.M. and E.M. synthesized the empirical data and parameterized the model. S.S., BBR, R.W., J.F.B., M.A., SMo, BMR, M.B., SH, C.G., TB, NC contributed data for model parameterization and calibration. M.C.B., C.G., S.M., M.P., P.V., E.M., M.B., J.F.B. contributed to research question and critical input into model analysis. S.M. wrote the article. All the authors contributed to interpretation of results and critically reviewed and edited the article.
Conflicts of interest
The authors have no conflicts of interest to declare. The study was funded by the Canadian Foundation for AIDS Research (Grant # 023–015), and UNAIDS.
1. Gupta S, Granich R, Suthar AB, Smyth C, Baggaley R, Sculier D, et al. Global policy review of antiretroviral therapy eligibility criteria for treatment and prevention of HIV and tuberculosis in adults, pregnant women, and serodiscordant couples. J Acquir Immune Defic Syndr. 2013; 62:e87–e97. 10.1097/QAI.1090b1013e31827e34992.
3. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011; 365:493–505.
5. Aral SO, Cates W. Coverage, context and targeted prevention: optimising our impact. Sex Transm Infect. 2013; 89:336–338.
6. Delva W, Eaton JW, Meng F, Fraser C, White RG, Vickerman P, et al. HIV treatment as prevention: optimising the impact of expanded HIV treatment programmes. PLoS Med. 2012; 9:e1001258
7. Kato M, Granich R, Duc Bui D, Tran HV, Nadol P, Jacka D, et al. The potential impact of expanding antiretroviral therapy and combination prevention in Vietnam: towards elimination of HIV transmission. J Acquir Immune Defic Syndr. 2013; 63:e142–e149.
8. Vickerman P, Foss AM, Pickles M, Deering K, Verma S, Eric D, et al. To what extent is the HIV epidemic in southern India driven by commercial sex? A modelling analysis. AIDS. 2010; 24:2563–2572.
9. Ramesh BA, Moses S, Washington R, Isac S, Mohapatra B, Mahagaonkar SB, et al. Determinants of HIV prevalence among female sex workers in four south Indian states: analysis of cross-sectional surveys in twenty-three districts. AIDS. 2008; 22:S35–S44.
10. Ramesh BM, Beattie TSH, Shajy I, Washington R, Jagannathan L, Reza-Paul S, et al. Changes in risk behaviours and prevalence of sexually transmitted infections following HIV preventive interventions among female sex workers in five districts in Karnataka state, south India. Sex Transm Infect. 2010; 86:17–24.
11. Reza-Paul S, Beattie T, Syed HUR, Venukumar KT, Venugopal MS, Fathima MP, et al. Declines in risk behaviour and sexually transmitted infection prevalence following a community-led HIV preventive intervention among female sex workers in Mysore, India. AIDS. 2008; 22:S91–S100.
12. Becker ML, Ramesh BM, Washington RG, Halli S, Blanchard JF, Moses S. Prevalence and determinants of HIV infection in south India: a heterogeneous, rural epidemic. AIDS. 2007; 21:739–747.
14. National AIDS Control Organization. India Country Progress Report. ; New Delhi: NACO; 2011 .
17. Boily M-C, Pickles M, Lowndes CM, Ramesh BM, Washington R, Moses S, et al. Positive impact of a large-scale HIV prevention program among female sex workers and clients in Karnataka state, India. AIDS. 2013; 27:1449–1460.
18. Moses S, Ramesh BA, Nagelkerke NJD, Khera A, Isac S, Bhattacharjee P, et al. Impact of an intensive HIV prevention programme for female sex workers on HIV prevalence among antenatal clinic attenders in Karnataka state, south India: an ecological analysis. AIDS. 2008; 22:S101–S108.
19. National AIDS Control Organization. National Behavioral Surveillance Survey (BSS) 2006: General Population. Ministry of Health and Family Welfare, Government of India; Delhi; 2007
20. Kumta S. Achieving high coverage of HIV prevention services for MARPS: Avahan's experience in six states in India. Sex Transm Infect. 2011; 87:A7–A8.
21. Lowndes CM, Alary M, Verma S, Demers E, Bradley J, Jayachandran AA, et al. Assessment of intervention outcome in the absence of baseline data: ’reconstruction’ of condom use time trends using retrospective analysis of survey data. Sex Transm Infect. 2010; 86:I49–I55.
22. Pickles M, Foss AM, Vickerman P, Deering K, Verma S, Demers E, et al. Interim modelling analysis to validate reported increases in condom use and assess HIV infections averted among female sex workers and clients in southern India following a targeted HIV prevention programme. Sex Transm Infect. 2010; 86:33–43.
23. Ramakrishnan L, Gautam A, Goswami P, Kallam S, Adhikary R, Mainkar MK, et al. Programme coverage, condom use and STI treatment among FSWs in a large-scale HIV prevention programme: results from cross-sectional surveys in 22 districts in southern India. Sex Transm Infect. 2010; 86:I62–I68.
28. Shastri S, Sathyanarayna S, Nagaraja SB, Kumar AM, Rewari B, Harries AD, Zachariah R. The journey to antiretroviral therapy in Karnataka, India: who was lost on the road?. J Int AIDS Soc. 2013; 16:18502
29. Banandur P, Rajaram SP, Mahagaonkar SB, Bradley J, Ramesh BM, Washington RG, et al. Heterogeneity of the HIV epidemic in the general population of Karnataka state, south India. BMC Public Health. 2011; 11:S13
30. Prasad R, Washington R, Shastri S, Rewari BB, Agarwal R, Munugan SK, et al. ART scale up in Karnataka: best practice document. Bangalore; 2011. www.khpt.org
31. Vijay S, Kumar P, Chauhan LS, Rao SVN, Vaidyanathan P. Treatment outcome and mortality at one and half year follow-up of HIV infected TB patients under TB control programme in a district of south India. PLoS One. 2011; 6:
32. Thompson L, Bhattacharjee P, Anthony J, Shetye M, Moses S, Blanchard J. A systematic approach to the design and scale-up of targeted interventions for HIV prevention among urban female sex workers. ; Bangalore, India; 2012. www.khpt.org
33. Rajaram S, Sangameshwar S, Jayachandran A, Bradley J, Alary M, Ramesh B, et al. HIV and STIs in Belgaum district, Karnataka, India. A general population survey. Bangalore; 2008. www.khpt.org./
34. Lowndes CM, Jayachandran AA, Banandur P, Ramesh BM, Washington R, Sangameshwar BM, et al. Polling booth surveys: a novel approach for reducing social desirability bias in HIV-related behavioural surveys in resource-poor settings. AIDS Behav. 2012; 16:1054–1062.
35. Buzdugan R, Halli S, Gaurav K, Satyanarayana R, Blanchard JF. Payana cohort of migrant and mobile female sex workers in northern Karnataka: baseline results. Karnataka Health Promotion Trust; Bangalore; 2009 .
36. Banandur P, Ramanaik S, Manhart LE, Buzdugan R, Mahapatra B, Isac S, et al. Understanding out-migration among female sex workers in south India. Sex Transm Dis. 2012; 39:776–783.
37. Becker ML, Mishra S, Satyanarayana R, Gurav K, Doshi M, Buzdugan R, et al. Rates and determinants of HIV-attributable mortality among rural female sex workers in northern Karnataka, India. Int J STD AIDS. 2012; 23:36–40.
38. Mishra S, Ramanaik S, Blanchard JF, Halli S, Moses S, Raghavendra T, et al. Characterizing sexual histories of women before formal sex-work in south India from a cross-sectional survey: implications for HIV/STI prevention. BMC Public Health. 2012; 12:829
39. Becker M, Ramanaik S, Halli S, Blanchard JF, Raghavendra T, Bhattacharjee P, et al. The intersection between sex work and reproductive health in northern Karnataka, India: identifying gaps and opportunities in the context of HIV prevention. AIDS Res Treat. 2012; 2012:842576–1842576.
40. Washington RW, Pradeep BS, Becker ML, Garady L, Yallapa A, Prakash R, et al. A longitudinal study on quality of life (QOL) of people living with HIV in Karnataka, south India [abstract]. In: 19th International AIDS Conference. Washington, DC; 2012.
41. Pillsbury B, Andina M, Maynard-Tucker G, Saha D, Sarkar K. Samastha project USAID/India: final evaluation of comprehensive HIV and AIDS program. ; Delhi; 2011 .
42. Fairall L, Bachmann MO, Lombard C, Timmerman V, Uebel K, Zwarenstein M, et al. Task shifting of antiretroviral treatment from doctors to primary-care nurses in South Africa (STRETCH): a pragmatic, parallel, cluster-randomised trial. Lancet. 2012; 380:889–898.
43. Kitahata MM, Gange SJ, Abraham AG, Merriman B, Saag MS, Justice AC, et al. Effect of early versus deferred antiretroviral therapy for HIV on survival. N Engl J Med. 2009; 360:1815–1826.
44. Lodi S, Phillips A, Touloumi G, Geskus R, Meyer L, Thiebaut R, et al. Time from human immunodeficiency virus seroconversion to reaching CD4+ cell count thresholds < 200, < 350, and < 500 cells/mm3: assessment of need following changes in treatment guidelines. Clin Infect Dis. 2011; 53:817–825.
45. Anglaret X, Minga A, Gabillard D, Ouassa T, Messou E, Morris B, et al. AIDS and non-AIDS morbidity and mortality across the spectrum of CD4 cell counts in HIV-infected adults before starting antiretroviral therapy in Cote d’Ivoire. Clin Infect Dis. 2012; 54:714–723.
46. Rajasekaran S, Jeyaseelan L, Raja K, Vijila S, Krithigaipriya KA, Kuralmozhi R. Increase in CD4 cell counts between 2 and 3.5 years after initiation of antiretroviral therapy and determinants of CD4 progression in India. J Postgrad Med. 2009; 55:261–266.
47. Mussini C, Cossarizza A, Sabin C, Babiker A, De Luca A, Bucher HC, et al. Decline of CD4(+) T-cell count before start of therapy and immunological response to treatment in antiretroviral-naive individuals. AIDS. 2011; 25:1041–1049.
48. Pantazis N, Morrison C, Amornkul PN, Lewden C, Salata RA, Minga A, et al. Differences in HIV natural history among African and non-African seroconverters in Europe and seroconverters in sub-Saharan Africa. PLoS One. 2012; 7:e32369
49. Curtis AJ, Marshall CS, Spelman T, Greig J, Elliot JH, Shanks L, et al. Incidence of WHO stage 3 and 4 conditions following initiation of anti-retroviral therapy in resource limited settings. PLoS One. 2012; 7:e52019
50. Etard JF, Ndiaye I, Thierry-Mieg M, Gueye NFN, Gueye PM, Laniece I, et al. Mortality and causes of death in adults receiving highly active antiretroviral therapy in Senegal: a 7-year cohort study. AIDS. 2006; 20:1181–1189.
51. Zwahlen M, Harris R, May M, Hogg R, Costagliola D, de Wolf F, et al. Mortality of HIV-infected patients starting potent antiretroviral therapy: comparison with the general population in nine industrialized countries. Int J Epidemiol. 2009; 38:1624–1633.
52. Mahy M, Lewden C, Brinkhof MWG, Dabis F, Tassie J-M, Souteyrand Y, et al. Derivation of parameters used in spectrum for eligibility for antiretroviral therapy and survival on antiretroviral therapy. Sex Transm Infect. 2010; 86:(Suppl 2):ii28–ii34.
53. Yiannoutsos CT, Johnson LF, Boulle A, Musick BS, Gsponer T, Balestre E, et al. Estimated mortality of adult HIV-infected patients starting treatment with combination antiretroviral therapy. Sex Transm Infect. 2012; 88:i33–i43.
54. Rodger AJ, Lodwick R, Schechter M, Deeks S, Amin J, Gilson R, et al. Mortality in well controlled HIV in the continuous antiretroviral therapy arms of the SMART and ESPRIT trials compared with the general population. AIDS. 2013; 27:973–979. 910.1097/QAD.1090b1013e32835cae32839c.
55. Johnson LF, Mossong J, Dorrington RE, Schomaker M, Hoffmann CJ, Keiser O, et al. Life expectancies of South african adults starting antiretroviral treatment: collaborative analysis of cohort studies. Plos Med. 2013; 10:e1001418–e11001418.
56. Cohen MS, Gay CL, Busch MP, Hecht FM. The detection of acute HIV infection. J Infect Dis. 2010; 202:S270–S277.
57. Boily M-C, Pickles M, Vickerman P, Buzdugan R, Isac S, Deering KN, et al. Using mathematical modelling to investigate the plausibility of attributing observed antenatal clinic declines to a female sex worker intervention in Karnataka state, India. AIDS. 2008; 22:(Supplement 5):149–164.
58. Pickles M, Boily MC, Vickerman P, Ramesh BM, Washington R, Deering K, et al. Time evolution of the fraction of new HIV infections due to primary infection among high risk groups in southern India [abstract]. In: 19th Meeting of the International Society for Sexually Transmitted Disease Research. Quebec City; 2011. A50–A51.
59. Garnett GP. An introduction to mathematical models in sexually transmitted disease epidemiology. Sex Transm Infect. 2002; 78:7–12.
60. Anderson R, May R. Infectious diseases of humans: dynamics and control. Oxford, UK:Oxford University Press; 1991 .
62. India Health Action Trust. HIV/AIDS situation and response in Uttar Pradesh: epidemiological appraisal using data triangulation. Bangalore; 2010 .
63. Beattie TSH, Bhattacharjee P, Suresh M, Isac S, Ramesh BM, Moses S. Personal, interpersonal and structural challenges to accessing HIV testing, treatment and care services among female sex workers, men who have sex with men and transgenders in Karnataka state, South India. J Epidemiol Community Health. 2012; 66:II42–II48.
65. Alary M, Lowndes CM, Van de Perre P, Behanzin L, Batona G, Guedou FA, et al. Scale-up of combination prevention and antiretroviral therapy for female sex workers in West Africa: time for action. AIDS. 2013; 27:1369–1374.
66. Nagelkerke N, Jha P, de Vlas S, Korenromp E, Moses S, Blanchard J, et al. Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission. Bull World Health Org. 2002; 80:89–96.
67. Clouse K, Pettifor AE, Maskew M, Bassett J, Van Rie A, Behets F, et al. Patient retention from HIV diagnosis through one year on antiretroviral therapy at a primary healthcare clinic in Johannesburg, South Africa. J Acquir Immune Defic Syndr. 2013; 62:E39–E46.
antiretroviral treatment; condom use; female sex worker; HIV prevention; HIV transmission; India; mathematical model
Supplemental Digital Content
© 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins
Highlight selected keywords in the article text.