Dispatches from the field: EPA’s experience with very large literature reviews

For over four decades, protection from six major ambient air pollutants in the U.S. has depended on periodic reviews of the evidence for the effects of each of them on human health and on ecosystems. Those high stakes reviews are characterized by the very large scale of the literature considered, extremely high recall requirements, and scrutiny from a wide range of stakeholders. Since 2018, EPA has used machine-assisted screening, and developed its own software in 2022 with input from UNC. Dr. Dubois will share his team’s experience, highlight some surprising lessons learned, and propose some general questions for Systematic Review that arise from EPA’s particular needs, but apply to the entire field.