Assessment of the impact and cost-effectiveness of the Avahan HIV intervention among high-risk groups in southern India

Funders: Support for this research has been provided by a grant from the Bill & Melinda Gates Foundation.

Collaborators: London School of Hygiene & Tropical Medicine, UK; Imperial College London; Centre Hospitalier Affilié Universitaire de Québec; University of Manitoba; Karnataka Health Promotion Trust; St. John’s Medical College and Hospital, Bangalore.

Data: Serial cross-sectional behavioural and biological data (IBBA surveys) have been collected from high risk groups and the general population in numerous districts across the southern Indian states of Andhra Pradesh, Karnataka, Maharashtra and Tamil Nadu. In addition, costs were collected prospectively over four years. For a selection of districts this included time allocation to different activities and the value of any donated goods.

Project background and summary
India, with an estimated 2.4 million individuals, has the second highest number of people in the world currently living with HIV/AIDS. The epidemic there is geographically heterogeneous and largely focussed in southern India, where it is believed to be mainly driven by commercial sex (heterosexual and homosexual) and high risk non-commercial male-to-male sex. Evidence suggests that to date, transmission within the general population has been limited, and remains mainly due to ‘first-wave’ transmission from male bridging groups to their low-risk female partners, with little onwards transmission within the general population.

The Avahan Programme, initiated in 2003 by the Bill & Melinda Gates Foundation, is the largest targeted HIV intervention in the world. It is the first serious attempt to implement and scale up targeted interventions for high-risk groups to a level sufficient to impact population-level HIV. Assessing the impact and cost-effectiveness of the Avahan intervention is thus crucial, not only to ensure that Avahan achieves its goals of reducing HIV transmission in India, but also to inform future large-scale interventions. Key questions are whether the intervention been implemented and scaled up, with sufficient quality, intensity and coverage of the target populations to change behaviour and consequently significantly impact on HIV/STI rates firstly among the target core and bridge groups and subsequently among the general population, and whether this was achieved in a cost-effective manner.

In the absence of an actual control group, the interpretation of time trends in HIV and STI prevalence for the purposes of evaluating an intervention is challenging, particularly in the short to medium term. This is due to the long latent period of HIV, and because only HIV prevalence, and not incidence, can currently be accurately measured. Furthermore, an HIV epidemic can be in an increasing, stable or declining phase. Increases in HIV prevalence can occur early on in an epidemic despite a successful intervention, since incidence may be decreased by the intervention but not sufficiently to result in a decline in prevalence in the short term. On the other hand, decreases in HIV prevalence can occur in later epidemic stages even in the absence of interventions, due to AIDS mortality, or other causes such as drastic changes in migration.

Thus an important component of the evaluation of Avahan is the use of transmission dynamics modelling, applied within a Bayesian framework, in order to be able to make stronger plausibility statements about intervention impact. To this end, a tailor-made model of HIV/HSV-2 and syphilis transmission has been developed and parameterised using district-level data from extensive high-quality serial behavioural and biological surveys, incorporating uncertainty from key parameters, and using novel methods to determine changes in behaviour in the absence of baseline data. This model is then fitted to serial prevalence data for the different STIs in high risk groups and the general population. Simulated control groups are used to assess what might have happened in the absence of the intervention, allowing calculation of the impact of the intervention.

To date mathematical models have been used to investigate the contribution of partnerships in the general population in Mysore and Belgaum to overall HIV prevalence in the population, to examine trends in ante-natal clinic sentinel surveillance HIV prevalence data across districts in India, and to test how likely or accurate any reported change in behaviour is in a situation where social desirability bias is likely to be important, as well as to generate interim estimates of the potential impact of the intervention on HIV prevalence and incidence among female sex workers and their clients given reported changes in behaviour in Mysore and Belgaum. These models will then be used to further explore drivers for heterogeneity across districts, as well as the potential impact of modifying impact strategies and the consequences of the changing structure of sex work in India.

The initial funding commitment for Avahan, made in 2003, was US $200 million for five years, with an additional US $58 million committed in 2006. From its inception, an economic evaluation has been an integral part of the programme. This was considered essential, both to provide national and local programme managers with cost estimates to sustain prevention services and to improve the information available on the costs and returns of investments in targeted HIV prevention strategies in concentrated epidemics more generally. There are very few studies on the costs of HIV prevention in Asia, and Avahan provides a unique opportunity to measure the costs of a large-scale programme. Cost data were collected from all sites, with a more detailed costing conducted in districts where behavioural data was also collected. Costs were also collected from the Bill and Melinda Gates Foundation head office in Delhi and the state level lead partners, in order to estimate the economic cost of the entire programme. These costs are then brought together with the impact estimation generated by the modelling to assess cost-effectiveness.

To date the cost data has been used to analyse the variation of costs at different scales of the programme. This found substantial economies as the programme scaled up. Initial work has also been conducted examining how costs vary by target group, setting and other contextual factors. Finally, in 2010, the first cost-effectiveness analysis was completed, for Mysore and Belgaum districts. This found the programme to be highly cost-effective. This work will now be extended to include three other sites, and an overall evaluation of the cost-effectiveness of the entire programme.

References related to project:
Pickles M, Foss A, Vickerman P, Deering K, Verma S, Demers E, et al. An interim analysis of HIV among high-risk groups in Southern India on changes in condom use due to the India AIDS initiative intervention. STI 2010, 86 Suppl 1:i33-i43.

Boily MC, 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 Suppl 5:S149-164.

Boily MC, Pickles M, Vickerman P, 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 Suppl 5:S149-64.

Chandrashekar S, Guinness L, Kumaranayake L, Reddy B, Govindraj Y, Vickerman P, Alary M (2010). “The effects of scale on the costs of targeted HIV prevention interventions among female and male sex workers, men who have sex with men and transgenders in India.” Sex Transm Infect 86 Suppl 1: i89-94.

Vickerman P, Foss AM, Pickles M, Deering K, Verma S, Demers E, Lowndes CM, Moses S, Alary M, Boily MC. To what extent is the HIV epidemic in South India driven by commercial sex? A modelling analysis. AIDS 2010; accepted for publication.