Oregon Health Insurance Experiment - Background

Oregon's Heath Insurance Lottery

In 2008, Oregon implemented a limited expansion of its Medicaid program for low-income adults through a lottery, selecting names from a waiting list to fille a limited number of available spots. Those selected had the opportunity to apply for Medicaid and to enroll if they met eligibility requirements.

This lottery presented a unique opportunity to gauge the effects of Medicaid using the rigorous standard of a randomized controlled trial.

• In January 2008, Oregon opened a waiting list for its Medicaid program for low-income adults that had previously been closed to new enrollment. Approximately 90,000 people signed up. The state deemed random selection by lottery the fairest way within federal guidelines to allocate its limited number of openings.
• Between March and September 2008, the state drew approximately 30,000 names from this waiting list by lottery to fill the 10,000 available spots.
• This random selection allows researchers to gauge the many effects of health insurance itself, isolating it from the types of confounding factors that can plague observational studies.

Methods

Although the quasi-experimental approaches used in many studies can be an improvement over observational designs and often involve larger samples than are feasible with a randomized design, they cannot eliminate confounding factors as effectively as random assignment. The Oregon lottery thus presents an unprecedented opportunity to isolate the causal effect of insurance on the health care use, health outcomes, financial strain, and well-being of low-income adults.

• Although there are many studies comparing health or health care use between the insured and uninsured, inferring the impact of health insurance from such comparisons is difficult: the insured and the uninsured may differ in many ways – such as income, employment, or initial health – that may themselves affect the outcomes being studied. This makes it difficult to discern the effects of insurance itself.
• Random assignment of health insurance to some but not others avoids such confounding factors (since the treatment and control groups are divided by chance, eliminating systematic differences between them). Such an experiment has almost never before been performed in the Medicaid program, in part because there would be ethical concerns with deliberately withholding available health insurance for the purpose of research. The allocation of the limited number of spots available in Oregon’s insurance program, however, created a natural circumstance that allowed researchers to take advantage of such a randomized design.

The study compiles rich data from many sources to examine a wide range of potential effects of insurance, including both primary data from surveys and clinical measurements and administrative data such as hospital and emergency department records.

Statistical Analysis

The lottery can be used to assess both the effects of lottery selection and the effects of Medicaid coverage itself on a range of outcomes. Virtually all of the analyses were prespecified and publicly archived in detailed analysis plans. Prespecification was designed to minimize issues of data and specification mining and to provide a record of the full set of planned analyses.