Sexual network structure, diffusion and prevention of HIV within populations: A case study using complete network data in Likoma Island (Malawi)

The sexual networks connecting members of a population have important consequences for the spread of sexually transmitted diseases including HIV. However, very few datasets currently exist that allow an investigation of the structure of sexual networks, particularly in sub-Saharan Africa where HIV epidemics have become generalized. This dissertation represents an attempt to fill this void. It builds on a unique sociocentric study of sexual partnerships conducted in small island of Lake Malawi, which provides the first map of a large sexual network through which HIV has diffused extensively. The four chapters comprising this dissertation highlight the role of network connectivity in determining the dynamics leading to the generalization of HIV epidemics. The first chapter describes the context and methods of the Likoma Network Study (LNS) and shows that the collection of sociocentric network data is feasible, even in a sub-Saharan context. Based on these data, the second chapter documents the existence of a large and robust sexual network linking a substantial fraction of the island\u27s young adult population. Contrary to claims that sexual networks are too sparse to sustain generalized epidemics through heterosexual contacts in remote areas of sub-Saharan Africa, the sexual network structure observed on Likoma Island appears compatible with a broad diffusion of pathogens to low-risk groups. This chapter also shows that the prevalence of HIV varies significantly across of the network, suggesting that network characteristics are an important determinant of the dynamics of HIV spread within a population. The third chapter investigates the role of migrants in diffusing HIV and other STIs into networks connecting the members of a local population. It shows that some STIs may circulate within rural populations independently of contacts with migrants and other outsiders. Finally, the fourth chapter systematically analyzes the quality of the network data. In particular, it emphasizes the potential impact of errors in link tracing on the estimates of network measures: such errors did not occur at random but were related to several important determinants of the disease diffusion process