My research program has three main axes. I conduct methodologic and substantive research in perinatal epidemiology and pharmacoepidemiology, and conduct methods research on statistical tools for causal inference.
Perinatal epidemiology presents challenges, because relatively little is known about the timing and measurement of maternal exposures and how they affect fetal and infant outcomes, e.g., growth. I have shown that growth references based on birth weight may give biased inference. My current research aims to develop new statistical tools to use fetal growth measures based on ultrasound measurements of fetal weight to capture the effects of exposures on fetal growth and of fetal growth on infant outcomes.
In pharmacoepidemiology, longitudinal studies allow us to understand effects of exposure over time. Marginal structural models are tools to estimate parameters in longitudinal studies that have been used successfully in pharmacoepidemiology to estimate effects of cumulative exposure to medications. However, many outstanding questions remain on optimal use of such models in practice. My research will develop answers to some of these questions, in the context of large studies of drug efficacy and safety.
Finally, traditional statistical tools are designed to measure associations, not causation. In recent decades, statistical tools designed for causal inference have been developed. These techniques allow us to estimate causal effects of exposure on outcome, under certain assumptions. My research program is oriented towards investigating both marginal structural models and targeted maximum likelihood, a technique for less-biased inference in marginal structural models under certain assumptions. I am investigating the theoretical and applied properties of both of these methods.
This combination of substantive and methodological research allows me to work at the interface of medical research and statistics, to help bring the best statistical methods into health research, and provide better answers to important health research questions.