Causal Inference from EHRs: Opportunities for Cancer Prevention

Join us for a seminar jointly sponsored by the Zhu Family Center for Global Cancer Prevention and the Department of Epidemiology.

There is growing interest in precision prevention and early detection strategies for cancer, in which interventions over time can be tailored to individuals’ unique and evolving characteristics. However, evaluating these dynamic (adaptive) strategies is challenging because a randomized trial may be infeasible and conventional methods may be inadequate. The intersection of large databases of electronic health records (EHRs) and causal methods designed to evaluate dynamic strategies provides the opportunity to inform decision-making about personalized strategies as well as the design of more efficient trials.

We (1) propose a framework to emulate target trials of dynamic cancer prevention strategies using nationwide databases of EHRs, (2) describe extensions to different cancer control questions, study designs, and data sources, and (3) discuss how this may enable a learning health system that quickly translates results from big data analyses to patient populations.