3. Risk networks: the need for research in sexual networks

Mathematical models show that different patterns of sexual mixing have widely different implications for the spread of the HIV epidemic [5, 6, 7]. If people mix within relatively closed groups - homosexuals only with homosexuals, married people only with their partners or other married people, prostitutes only with a well defined group of individual clients - HIV may spread quickly within some of the groups but will have a limited impact on the population as a whole. But if there is much more mixing between groups, with injecting drug users having sex with prostitutes, whose clients have sex with their own wives, for instance, the disease may take off slowly but will insinuate itself into many more corners of society. Bridge populations, which form a link between otherwise unconnected groups, may be of particular importance for the dynamic of the epidemic by linking low and high risk behaviour populations.

Population surveys such as those recommended by WHO/GPA and UNAIDS have brought better knowledge of high risk groups and high risk behaviours. However, people are put at risk not just by their own behaviour but by that of others to whom they are linked in sexual networks. Policy makers should, from the shape of these networks, be able to identify useful points for intervention. It is up to social scientists to identify how sexual networks are spread across the society, by whom and why.

Thus, with the move from evaluation as the primary objective to that of more in-depth understanding of sexual networks, there is also a shift to an analytical framework that makes partnerships rather than individuals the primary unit of analysis [7]. Individual-based approaches explain behaviours by noting the characteristics of the individual: attitudes, knowledge, beliefs and education; while partnership-based approaches will try to explain behaviours by noting the characteristics of the relationship: its duration, mutual expectations or gender roles, for example.