Reductions in HIV incidence are likely to increase the importance of key population programmes for HIV control in sub-Saharan Africa.

Introduction: An efficient HIV response requires that resources be focussed on effective interventions for those most at risk of acquiring and transmitting infection. As HIV epidemics evolve the distribution of HIV across key and other populations will change. Here, the epidemiological concepts underpinning these changes are described and the importance of appropriate allocation of effective interventions is discussed.

Discussion: In many sub-Saharan African countries HIV epidemics have been categorized as "generalized," and HIV testing, treatment and prevention interventions have focussed on the "general" population. As HIV epidemics are better controlled the relative importance of "key" populations will increase, dominating the ongoing burden of disease and providing the potential for repeated outbreaks of HIV if interventions are relaxed. The basic reproductive number (R ) describes the potential for an infectious disease to spread at the boundary of invasion or elimination, whereas the effective reproduction number (R ) describes the current potential for spread. Heterogeneity in risk means that while R is temporarily below one and prevalence declining, the R can remain above one, preventing eventual elimination. Patterns of HIV acquisition are often used to guide interventions but inadequately capture the transmission dynamics of the virus and the most efficient approach to controlling HIV. Risks for HIV acquisition are not identical to risks for HIV transmission and will change depending on the epidemiological context. In addition to the challenges in measuring HIV transmission dynamics, there is a tension between using epidemiology to drive the HIV response and the social and political realities constraining how programmes and providers can practically and appropriately focus on key populations and maintain political support. In addition to being well focussed, interventions need to be effective and cost-effective, which requires a better understanding of packages of interventions rather than specific tools.

Conclusions: Continued control of HIV will increasingly rely on resources, programmes and interventions supporting key populations. Current epidemiological and programmatic approaches for key populations in sub-Saharan Africa are insufficient with a need for an improved understanding of local epidemiology and the effectiveness of interventions.