Injecting in a park was another risk factor for HIV infection. It is well established that the size and density of social and injecting networks are key factors in the transmission of HIV among IDUs [12,13]. Indeed, a recent study in Pakistan pointed out large sharing networks as a key factor for explosive HIV prevalence growth among IDUs [14]. Assuming that IDUs who recently injected in parks did so in the past as well, the high prevalence of HIV found among those injecting in parks plausibly reflects their exposure to larger and more diverse needle sharing networks than those not injecting in parks.
Sexual risk behaviour, such as unprotected commercial sex, was not associated with HIV, indicating that the main route for HIV acquisition among IDUs is unsafe injections rather than sexual intercourse. Surprisingly, infection with gonorrhoea or chlamydia (NG/CT) was associated with lower likelihood of HIV infection, but our data could not identify any behaviours associated with both HIV infection and NG/CT.