O19.6 Professional injectors, injection networks, and HIV: cases studies from five cities in pakistan

Background There are large numbers of people who inject drugs (PWID) in Pakistan, with a high and growing HIV prevalence in many cities. Cross-sectional bio-behavioural surveys conducted over fifteen years have shown heterogeneous HIV epidemic trajectories in different cities with inconsistent associations between aggregate and individual measures of behaviour and epidemic trajectories. This study triangulates behavioural, biological, and network data to gain new insights into drivers of local epidemics among PWID. Methods A questionnaire was administered to 1,439 PWID in five cities in Pakistan in 2014 and dried blood spots were collected for HIV testing and sequencing. The questionnaire collected locations where participants injected recently and these were used to construct place-based injection networks. Demographic, behavioral, geographic, and network structure data were tested for associations with HIV and HIV phylogenetic analysis was conducted to identify transmission clusters. Results HIV prevalence varied between the cities, from 8.6% to 45.3%. Although aggregate individual behaviours were not strongly associated with HIV, there were differences between cities in injecting practices and network configurations. Injection network configurations differed between cities, ranging from networks with dense sub-regions to widely dispersed networks. A city that experienced an explosive HIV outbreak showed particularly dense geospatial networks and a large phylogenetic cluster (n=52) suggesting a link between geospatial concentration and rapid HIV transmission. In a city where the use of professional injectors was associated with HIV at the individual level, there were dense network subregions and HIV+ individuals were more likely to be more centrally located in networks (p Conclusion Contextual factors shape the injection practices and network configurations among PWID in Pakistan. Factors that influence HIV transmission dynamics are more complex than individual behaviours, and include professional injectors and geographic and network clustering. Disclosure No significant relationships.