HIV has disproportionately affected southern Africa. This region, which comprises 2% of the
worlds population, is home to an estimated 34% of all people living with HIV, 29% of new HIV
infections globally in 2010, and 30% of AIDS-related deaths. A strengthened response to the
epidemic by countries in southern Africa in recent years has brought life-prolonging antiretroviral
therapy to the majority of those in need of treatment, and declines from peak levels of HIV
incidence over the past decade are a reason for optimism. But, in 2010, 770,000 new HIV
infections occurred. A better understanding of why the epidemic has spread so severely in this
region is required to inform strategies to reduce and eventually eliminate new HIV infections.
This thesis uses data analysis and mathematical modelling to understand the interaction between
behavioural and biological factors that may have contributed to the spread of HIV in
southern Africa, and the implications of these for controlling the epidemic. It focuses specifically
on two topics of recent attention for public health decision makers in southern Africa:
concurrent sexual partnerships and HIV treatment as prevention. Chapters explore the interaction
between high HIV infectiousness during primary HIV infection and concurrent sexual
partnerships, describe and evaluate a consensus indicator for concurrency, develop a method to
adjust for high levels of missing data in sexual behaviour surveys and examine trends in sexual
behaviours in a high HIV prevalence population in South Africa, create a mathematical model
to examine the potential impact of antiretroviral therapy on HIV incidence in hyperendemic
settings, and systematically compare the predictions of twelve different mathematical models
of the impact of HIV treatment as prevention in South Africa. Taken together, through these
topics we come to understand more broadly the complexity of the epidemiological context in
which HIV spreads in southern Africa.