The Oregon Health Insurance Experiment

A randomized clinical trial in Oregon provided health insurance to some low-income residents and not others. Results from the trial demonstrate better access to care, improved health, and reduced financial stressors. 

The state of in 2008 to select a fraction of uninsured low-income residents to enroll in Medicaid (Oregon Health Plan or OHP). Researchers familiar with the situation recognized this unique opportunity to conduct a randomized trial of the effect of insurance on this disadvantaged population. This experiment compliments evidence from the only other randomized controlled trial of health insurance – the RAND Health Insurance Experiment – which investigated the effects of cost-sharing among individuals with health insurance.

Individuals eligible for the lottery were Oregon residents (citizens or legal immigrants) aged 19-64,  uninsured for at least 6 months, with income below the federal poverty level, and assets under $2000. A total of 89,824 individuals signed up for the lottery which then provided access to apply for the 10,000 available OHP slots. Of those selected by lottery, only 30 percent successfully enrolled in OHP.

Information on outcomes were derived from hospital discharge data, credit records from TransUnion, mortality data from Oregon’s Center for Health Statistics, and mailed surveys from the investigators (with a response rate of 36-50 percent).

Researchers included nearly 29,000 individuals in both the treatment (lottery winners) group and the control group. Although the treatment group was slightly less likely to respond to the outcome survey, there was no difference among various baseline demographic characteristics between the treatment and control groups.

Data analysis compared the control group to both the treatment group (lottery winners) and a subset (lottery winners who successfully obtain Medicaid). Evidence from the experiment reveals that those who obtained health insurance were 2.1 percentage points (30 percent) more likely to have hospital admissions. Subjects with insurance received statistically greater hospital charges (an extra $1,000) and 7 percent more procedures while in the hospital. Outpatient measures revealed that obtaining insurance results in greater use of prescription drugs by 15 percent, a 35 percent increase in outpatient office visits, and an overall increase in annual health spending of approximately $226-$778 (compared to a baseline of $3,156).

Possessing Medicaid insurance allowed subjects to have access to better preventive care. Lottery winners more often were screened for diabetes and high cholesterol. Women were better able to obtain mammograms and cervical cancer screening.

Based on financial data, subjects in the treatment group were statistically less likely to have debt collections. These reduced financial strains were due purely to fewer collections for medical debt (as opposed to other consumer debts). However, there were no reductions in bankruptcies, liens, judgements, or delinquencies during the study period. Other proxies for financial strain – having out-of-pocket medical expenses, borrowing money or skipping other bills to pay for medicine, or refusing treatment because of medical debt – were all reduced by obtaining insurance.

The evidence to date did not reveal any difference in mortality during the 16-month study period (99.2 percent of participants were still alive). However, surveys clearly demonstrated subjective improvements in both physical and mental health for those in the treatment group.  Patients in the treatment group reported better access to care – having a usual site of care, a personal doctor, and getting all the needed medical care and drugs needed.

Commentary

The Oregon Health Insurance Experiment promises to transform the way health policy analysts look at health insurance in general and Medicaid in particular. As only the second in the health policy field, it will certainly become a seminal source for .

Data from the first year of this experiment, where some Oregonians received Medicaid while others did not, demonstrates improved access to care, increased health care utilization, and decreased financial stresses for patients. The data even reveal that patients who wound up on Medicaid were “happier” than those who did not.

Unfortunately, this paper – while ground-breaking – is written in a style unlikely to lead to easy dissemination of the facts. Hopefully, with continued reporting on the paper, the effects of Medicaid on the uninsured will be known by the lay public and decision makers.

The data also reveal several facts about the poor health of the low-income population. While almost 2/3rds of the sample was taking a prescription medication, fewer than half of the women received the above-mentioned screening tests. Nearly 7 percent were hospitalized during the study.

While these findings are useful, they may not be applicable to other states. The demographics – 4 percent Black and 12 percent Hispanic – do not mirror the low income population, which is only about 50 percent White.

Finkelstein, et al. “The Oregon Health Insurance Experiment: Evidence from the First Year.” NBER. (July 2011) Working Paper No. 17190.

by

Cedric Dark, MD, MPH