Use of dynamic network models to explore the role of social media use in HIV transmission and health promotion among gay men and other MSM

The majority of HIV transmission takes place between individuals via sexual contacts. At a population level, these partnerships can be viewed as connected into a dynamic sexual partnership network, whose structural and compositional characteristics influence, and are influenced by, the HIV epidemic. Although this is well understood, lack of data and a framework for utilising sampled network data and strategies for translating simulation findings into health promotion strategies has prevented the use of data-driven dynamic network mathematical models to inform the design of interventions in specific settings.

The way that individuals seek and form sexual partnerships is changing in many settings, and so plausibly are the characteristics of the sexual partnership network. A rapid rise in the use of online social media is occurring not only in high income settings, but also in sub-Saharan Africa. It is now both critical and feasible to develop a framework for data-driven dynamic partnership network models of HIV transmission, to understand the intersection of HIV acquisition and transmission risk with online socialising and partner-seeking, and to ensure that HIV prevention and engagement in care interventions can be most effectively designed and targeted.

A key group affected by HIV across the world is men who have sex with men (MSM), including in sub-Saharan Africa (SSA). Studies of MSM in SSA have found high HIV prevalence and incidence, alongside widespread experience of social stigma, violence, harassment and poor access to sexual health services. The incentives for MSM to socialise and seek partners online are high, and correspondingly, there is great potential to use online social media venues to provide information or referrals for in-person HIV prevention, testing and care services.

This project will use unique and current data on sexual partnerships, online and offline-initiated, HIV status and engagement in care, and social media use from representative samples of MSM from Nairobi and Johannesburg to develop dynamic network models of HIV transmission, explore the potential role of social media use in HIV transmission, and bring networks-based insight to improve targeting of health promotion and inform intervention development and future research to reduce HIV incidence.

This project aims to explore the feasibility and value of data-driven dynamic sexual partnership network models of HIV transmission applied to intervention design, particularly to investigate the role of social media in partnership formation and health promotion among gay men and other men who have sex with men (MSM) in sub-Saharan Africa.

This project will use population-based data on sexual partnerships and HIV epidemiology collected in 2017 from MSM in Nairobi (n=618) and Johannesburg (n=301). Sampled network data from a population representative sample can be used to infer whole network characteristics. For dynamic networks, models of tie formation and dissolution over a series of time steps can be built using separable-temporal exponential random graph models (STERGM). Target parameter values for network statistics can be set using the sampled partnership data, and Markov Chain Monte Carlo used to simulate networks whose characteristics reflect, though vary stochastically around, these target network statistics.

An individual stochastic model for HIV transmission can then be run alongside the dynamic simulated networks using current HIV epidemiological and behavioural characteristics estimated in from the data, supplemented by a literature review of HIV transmission among MSM. The model will be calibrated using Bayesian methods such that I will be able to generate a field of potential models that fit the data, including all the uncertainty in the inferred models. By varying the parameter values of the STERGM, the effect of different network characteristics on HIV transmission can be modelled.

The first 18 months of the project will be dedicated to developing the modelling framework, at which point I will discuss and confirm the specific health promotion questions to apply the model to informing in each setting, alongside local researchers, service providers and MSM community based organisations.

The overall aim of this project is to inform the targeting and development of HIV prevention and care interventions amongst gay men and other MSM in Nairobi and in Johannesburg, and to develop the methodological approaches to be applied amongst other populations and settings. My project aims to spent the first 18 months developing a framework for modelling of HIV transmission over a dynamic sexual partnerships network, and the second half of the fellowship applying this model to answering specific health promotion questions co-constructed alongside local researchers and policy-makers in each setting. A primary impact I hope to achieve includes improvements to the targeting of HIV prevention and care outreach activities via i) an appreciation of the role of the online environment in sexual partnership initiation and the implications for HIV transmission locally; ii) an understanding of which MSM are most critical to engage to reduce transmission, either through onwards transmission or prevention of acquisition; and iii) improved understanding of where MSM with different profiles, needs and potential roles in the HIV epidemic could potentially be engaged for health promotion, with particular consideration to online versus 'offline' environments.

I have planned and budgeted for 2-3 priority-setting and findings interpretation workshops with local researchers, service providers and MSM community-based organisations in each site. This is to ensure that the specific questions to which my modelling framework is addressed are those which are most important to local health promotion questions. In Nairobi, I plan to engage with and potentially to coordinate these workshops with the 'G10' forum for researcher and MSM community based organisation (CBO) engagement, and with a yearly conference held to discuss new research related to HIV/AIDS in Nairobi. I will work with Dr Joshua Kimani, the TRANSFORM Nairobi site PI from Partners for Health and Development in Africa and the SWOP-Kenya programme, which integrates HIV research, advocacy and community initiative with clinic services, and will invite a group of MSM CBO's who have been involved with TRANSFORM and with a previous project assessing HIV prevention interventions for MSM in Kenya. In Johannesburg, I will engage the TRANSFORM PI at Wits Reproductive Health Institute and with the Anova Health Clinic, who have been providing a large programme for MSM in the region, Health4Men. I will engage in existing initiatives to gather data and collate estimates related to MSM as part of monitoring South Africa's new Key Populations HIV strategy and their National Strategic Plan for HIV, TB and STIs 2017-2022.

There is a broader interest in understanding how social media can be used to inform HIV intervention planning and I will seek to engage in these discussions. I will explore opportunities to engage with international agencies (UNAIDS, WHO, CDC) through my existing relationship with the Measurement and Surveillance of HIV (MeSH Consortium) which aims to improve the use of routine data collection to inform HIV surveillance and target programming. I will seek opportunities to contribute to relevant guidance documents related to the use of social media and HIV surveillance and programming, as well as engaging in existing efforts to incorporate the role of key populations into models of HIV transmission in 'generalised' epidemic settings.