Assessing a Medicaid Randomized Insurance Experiment Within Community Clinics

In 2008, the Oregon Medicaid program used a "lottery" (Oregon Experiment) to allocate limited insurance resources to low income adults who did not categorically qualify for traditional Medicaid coverage. Thousands were randomly selected to apply for Medicaid; many others were not selected. This "natural policy experiment" is the first population-level randomization of insurance coverage since Rand's 1971-1982 experiment. We will use data from safety net clinics that share a common electronic health record (EHR) operated by OCHIN, a non-profit organization in Portland, Oregon. As one of the nation's largest linked community health center networks, the OCHIN EHR database records nearly all visits to safety net clinics in Oregon with one patient chart linked across >100 community health centers. We will link Medicaid claims data and the OCHIN database to determine patients that had applied to participate in the Oregon Experiment, as well as to determine those that gained insurance via this experiment. We will test the hypothesis that public insurance coverage is associated with higher rates of receiving preventive care services and increased healthcare utilization among persons with a usual source of primary care. To that end, we will analyze preventive care and utilization patterns among the OCHIN clinic patients affected by Oregon's Medicaid Experiment. This study capitalizes on two unique opportunities: an EHR database linking >100 community health centers, and the quasi-experimental design of Oregon's randomized Medicaid experiment.

The specific aims related to the Oregon Experiment are:

1. To compare post-intervention rates of preventive services receipt among OCHIN patients randomized to the Oregon Experiment group (intervention group), compared to those not in the coverage group (comparison group); and 3) To compare post-intervention rates of ambulatory healthcare utilization among OCHIN patients randomized to the Oregon Experiment group (intervention group), compared to those not in the coverage group (comparison group).