HIV Working Group

Ellie Caniglia, SD
Assistant Professor of Epidemiology, Department of Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania

“The Tsepamo Birth Outcomes Surveillance Study in Botswana – Challenges and Solutions when Conducting Rigorous Epidemiologic Research Using Observational Data”

ABSTRACT: Estimating causal effects to inform policy and clinical practice using observational data is challenging. In this talk, I will present some recent work from the Tsepamo Birth Outcomes Surveillance Study that highlights some of these challenges and proposed solutions. Tsepamo is an ongoing birth outcomes surveillance study in Botswana, which captures pregnancy and delivery information for more than 200,000 deliveries nationwide from 2014 to present. In the first study I will present, we describe a prolific bias in observational research to evaluate medication safety and effectiveness in pregnancy – immortal time bias – and how to overcome it using a (sequential) target trial approach. We apply this approach to answer the question of the safety of antibiotic initiation in pregnancy on preterm delivery. In the second study I will present, we discuss how a ‘quasi-experimental approach’ – difference in difference – may overcome some challenges with using observational data to estimate causal effects, but those challenges remain. In this study, we apply the difference-in-difference approach to examine the potential effect of the COVID-19 lockdown on adverse birth outcomes in Botswana.