The contributions and future direction of Program Science in HIV/STI prevention

For HIV/STI program design, the necessary evidence involves an incisive appraisal of the social and epidemiological drivers and mediators of local epidemics. This includes understanding the places and drivers that might disproportionately place key populations (KPs) at higher risk of HIV/STI acquisition as well as characterizing population-level chains of transmission.

Innovations in Programmatic Mapping involves a systematic approach to generating key information about the size and distribution of KPs within a defined geographic area [26]. Other methods, including multiplier methods or capture-recapture techniques, provide overall size estimates but do not provide the granular information required for detailed program planning and implementation. For example, geographic mapping provide city-wide KP size estimates and also provide data on micro-level hotspot (places where KPs congregate to solicit sex/drug using partners) level KP size estimates, as well as generate information on the physical locations where KPs congregate and the characteristics of these locations, such as the typologies of sex work. The detailed population size data allow programs to set coverage goals and the location data enable programs to plan for outreach and concentrate resources in areas of greatest need. Programmatic mapping has been used by many countries in Asia and Africa [27‐30] and there is growing global recognition of the importance of mapping data [31]. David Wilson, the World Bank’s Global AIDS Program Director, recently wrote that “programmatic mapping are the foundation for high quality HIV programs” [32].

While programmatic mapping provides data on micro-level geographic concentration of risk, there has also been work to understand the macro-level spatial distribution of the epidemic at the province/state/district level as highlighted in work led by Tanser et al. and Abu-Raddad et al. [13, 16]. Tanser demonstrates that in regions where the HIV epidemic was traditionally felt to be a generalized epidemic, that in fact, there were important zones of high HIV transmission signifying the presence of concentrated sub-epidemics. Prioritizing finite resources by place (e.g. province or state) may be more efficient than universal distribution of resources across a country [33] to reduce HIV infections. Similarly, re-allocation of resources to better align service delivery with disease burden and disparities requires detailed mapping of health-states and services, including how individuals navigate health systems [34, 35].

Additional innovations have included approaches for characterizing HIV epidemics by understanding the causal pathway of HIV transmission at a population-level rather than focusing on HIV acquisition at an individual level. For example, condomless sex acts in the context of sex work may lead to a small number of HIV infections in the short-term, but contribute to a large number of HIV infections over time through onward transmission [36‐38]. Disentangling the causal pathways may require a more in depth understanding of the local context of sex partnerships, which in turn, leads to a better understanding of the sources of heterogeneity in risk of HIV transmission, and of acquisition. For example, the importance of transactional sex (sex in exchange for money/goods/resources wherein exchange was not explicitly negotiated prior to sex) leading to high proportion of HIV acquisition was recognized when a revised Modes of Transmission Model was parameterized to the local Nigerian context [36].

The uptake of some of these innovations into policy for resource allocation can be seen with an example from the Centers for Disease Control and Prevention (CDC). In 2013, the National STD Prevention Program in the United States was revised to incorporate a strategic planning component to its state funding allocation and provides a useful example of the application of Program Science in a northern hemisphere country context. The Division of STD Prevention at the Centers for Disease Control and Prevention in Atlanta is responsible for all of STI prevention in the United States. The funding requirements use a Program Science framework for resource allocation [22], using STI disease burden by subgroup, and subgroup population size, and thereby requiring programs/states to generate local knowledge about STI epidemiology through methods like programmatic mapping.

The Avahan India AIDS Initiative of the Bill and Melinda Gates Foundation was a large scale focused HIV and STI Prevention Program in South India for KPs. Avahan used programmatic mapping for strategic planning and specifically to determine where, when, and for whom interventions should be prioritized. Avahan is also a very nice example of using Program Science to determine what intervention mix is required and how to deliver these interventions in their programs [23, 39, 40].

Avahan clearly demonstrated the need to combine behavioural, biomedical and structural interventions to achieve the maximum impact in reducing HIV and STI rates. Biological and behavioural surveys conducted among female sex workers (FSWs) revealed a decline in HIV, syphilis, chlamydia and gonorrhea prevalence in most sex work sub-groups and most locations as a result of combination prevention interventions which included STI prevention and treatment [41‐44]. As the program matured, the “what” and the “how” also evolved. The program began to incorporate structural interventions aimed at reducing violence and improving community mobilization [23]. The inclusion of these interventions was driven by needs voiced by community members (members of KPs) as well as a program aim to further reduce HIV and STI rates. The process of designing and implementing these structural interventions centered on comprehensively engaging with policy makers, police, lawyers, media and sex work communities. With the incorporation of these interventions into the existing multi-pronged prevention programs, reductions in reported violence and improved individual and collective mobilization and empowerment were also seen [45, 46]. These changes also resulted in increases in the number of FSWs accessing government social programs and in some areas, improvements in condom use and service utilization [47]. The use of a Program Science framework allowed for a dynamic response; as the needs of the community changed, the program also evolved, using evidence to reshape and redesign the program and its’ implementation.

An important dimension of Avahan’s effectiveness is the integration of community knowledge in informing its intervention mix. Ashodaya Samithi [44, 48, 49], the first intervention site supported by Avahan, has developed community-centric processes and responses that allow communities to prioritize their issues, set the agenda for the way forward, and ensure community ownership of the intervention. This is achieved at multiple levels, initially through community engagement and involvement, and later through ownership of the intervention and capacity building that ensures sustainability of the intervention [23]. These levels of community involvement have been found to result in communities re-interpreting and translating intervention messaging at the local level to develop contextualized responses to public health challenges [24].

Improving program efficiency requires an approach to identify and define existing opportunity gaps. The Program Science Initiative in Kenya, through a Technical Support Unit (TSU) to the National AIDS and STI Control Programme (NASCOP), developed innovative field level tools to capture data on HIV/STI prevention program indicators. HIV prevention programs in Kenya follow a combination prevention approach with a focus on biomedical, behavioural and structural interventions. The tools developed and used by these programs were developed to collect data on all aspects of the program covering all three of these intervention focus areas. Kenya, as many other countries do, has several funders of KP programs. As such, implementers were using many different reporting formats used by the many different funders. The TSU, with support from NASCOP, worked with all funders and implementing partners through the National Key Population Technical Working Group to develop standard data collection tools. A basic 15 indicator reporting tool was developed and all implementing partners were mandated to report to NASCOP on a quarterly basis on all 15 indicators [50, 51]. This standard tool was useful to both simplify and harmonize data collection and reporting. The reports are compiled at the national level by TSU and NASCOP and county wise analysis is shared with the implementing partners, county governments and funders on a quarterly basis to: (1) examine data quality; (2) evaluate trends such as changes in HIV testing uptake over time and (3) assess program achievements as compared to national targets. Figure 3 illustrates the layers of data collected and highlights the differences in coverage across the counties in Kenya.