New data from the Oregon Health Insurance Experiment has thought leaders in the health policy world dueling over the statistics these past 24 hours. As background, we (Dr. Matthews) previously discussed the premise behind the :
“Oregon has recently opened its Medicaid program, the Oregon Health Plan Standard, by allowing the application of about 30,000 names drawn randomly from its waiting list. Such an expansion allows for an analysis akin to a randomized controlled trial of expanding public health insurance coverage for low-income adults. Such an analysis is of obvious interest with the adoption of the recent federal expansion of Medicaid due to the Patient Protection and Affordable Care Act. Oregon offers the first opportunity for a randomized evaluation of such an expansion.”
The initial data, released in July 2011, revealed several interesting findings. Consistent with the , people gaining Medicaid demonstrated increased utilization of health care services (hospital costs, procedures, prescription drug use, outpatient office visits, etc.). The RAND experiment showed that as the cost of health care gets less expensive for people with private insurance, the utilization of services goes up. Similarly, the Oregon experiment showed that as costs got cheaper for uninsured patients (by giving them Medicaid), utilization went up.
Table 5 from the newest data on the Oregon experiment, continues to show this significant effect on insurance and utilization. There were significantly more prescriptions (up 0.66), office visits (up 2.7 per year), and annual health care spending (up $1,171.63) for those with Medicaid compared with people who remained uninsured. Also, they were more likely to get Pap smears (up 14%), PSA tests for prostate cancer (up 19%), cholesterol screenings (up 15%), and mammograms (up 30%). Is this increase in utilization helpful? I guess that depends on if cervical, prostate, or breast cancer mortality and morbidity improves. The current study does not help us one bit with understanding those specific issues. What about cholesterol? (more about that later)
Both the original and new data from the Oregon experiment show financial improvements for uninsured patients moving onto Medicaid. Based on the original study, Medicaid provided statistically fewer debt collections, lower out-of-pockets costs, less skipping other bills to pay for medicine, and less refusal of treatment due to cost. At year 2, the new study shows more of the same: lower out-of-pocket costs, a near elimination of catastrophic expenditures (defined as >30% of household income), fewer skipped bills, and less medical debt. But isn’t that what insurance is supposed to do in the first place?
Martin Gaynor (@MartinSGaynor) May 2, 2013
In light of these data, it appears Medicaid is doing it’s job. So why are the policy wonks on both ends of the political spectrum engaged in intellectual jousting over this study?
Well, the controversy lies in the apparent lack of effectiveness of Medicaid to alter clinical outcomes. Blood pressure, no change. Cholesterol, no change in total cholesterol, HDL, or LDL, prevalence, or use of medications. Diabetes, no change in overall HbA1C levels or the proportion indicative of disease. Yet, the population of this study, based on the clinical values reported, isn’t that sick to begin with.
Aaron E. Carroll (@aaronecarroll) May 2, 2013
The average blood pressure is NORMAL. The total cholesterol is AT WORST, borderline. The LDL and HDL levels are downright EXCELLENT. Don’t believe me, check out the Mayo Clinic’s reference ranges. The average HbA1c is not indicative of diabetes, in fact only about 5% of the study group even met diagnostic criteria for diabetes. That’s less than in the general US population. Many people are slamming Medicaid in view of this study. Cooler heads, understanding that the study population isn’t as sick as it could be and that detecting minor changes in health could be difficult, should prevail.
Last night, I asked Avik Roy whether or not he thought the null findings on clinical outcomes meant Medicaid was equivalent to being uninsured. He had been going around a lot lately citing some (nonrandomized) trials indicating that Medicaid was worse than being uninsured. He must have been up all night furiously typing and pounding some Red Bulls because his response (which posted at 4:20am) and take on the study was pretty thoughtful.