Estimates of HIV epidemic trends in sub-Saharan Africa are generated by fitting a simple dynamic epidemic model to national data sources about HIV prevalence in a Bayesian framework, to provide inference about population-wide trends in HIV prevalence, new infections, treatment need and coverage, and other policy indicators. In this talk I will (1) discuss motivating observations for revisiting the interpretation of widely relied upon data sources for estimating HIV epidemic trends, (2) describe a new integrated demographic projection and HIV epidemic model for improving HIV epidemic trends and projections in sub-Saharan Africa, and (3) investigate the impact of accounting for biases on inferred HIV epidemic incidence trends in sub-Saharan Africa. Finally, I will discuss future opportunities, challenges, and considerations for combing next generation survey data and routine health facility data to generate more granular and real-time HIV estimates.