We will use difference-in-difference methods to study the effect of Medicaid expansion on diabetes diagnosis, treatment and health outcomes. We rely on electronic health records for a sample of over 10 million patients, in 5 expansion and 4 non-expansion states, whom we will follow from 2009 (providing a pre-treatment baseline) through 2019 (providing sufficient time, following the 2014 Medicaid expansion) to observe near- and medium-term outcomes. The large sample provides power to detect small changes in HbA1c levels and other outcomes, and will let us quantify the extent of adherence to diabetes and related medications.