Measuring the Utility of Surveillance Data For Monitoring the HIV/AIDS Epidemic In Sub-Saharan Africa

Since the early 1980s in sub-Saharan Africa (SSA), substantial human and financial resources have been dedicated to monitoring the HIV/AIDS epidemic. Throughout, surveillance data collected at antenatal care (ANC) clinics have been a key data source. ANC surveillance data are well-known to be biased when quantifying population HIV prevalence levels in SSA. Nonetheless, a routinely-accepted, although rarely-tested assumption has been that the data are representative of population-level HIV prevalence trends. More recently, HIV testing data from prevention of mother-to-child transmission (PMTCT) programmes have been proposed as a substitute for ANC surveillance, although these data can be subject to temporal biases too. The primary objective of this thesis is to add to the limited evidence regarding the representativeness of HIV testing data from pregnant women to monitor population-level HIV prevalence trends. Empirical analyses from repeated household-based population surveys and ANC surveillance were done for seven countries in SSA from 2000 to 2010 and among youth aged 15 to 24 years in Manicaland, Zimbabwe from 1985 to 2003. Also, a mathematical model was used to explore temporal bias in ANC surveillance trends in epidemics similar to those in Botswana, Côte d’Ivoire and rural Zimbabwe from 1985 to 2030. Finally, PMTCT programme data were assessed for their representativeness as compared to ANC surveillance data in Manicaland, Zimbabwe from 2006 to 2008. Results showed the representativeness of ANC surveillance data to vary by time period and setting, although trends among youth were more robust than those among adults aged 15 to 49 years across settings, and particularly so among men. Representativeness in the ART-era depends on coverage and scale-up, the setting, and the potential for changing fertility patterns among ART users. PMTCT data for surveillance purposes was of limited use in Manicaland, Zimbabwe from 2006 to 2008. In summary, caution is needed when using HIV testing data from pregnant women to monitor population HIV prevalence trends in SSA