Medicaid or Placebo: Don’t Worry, Be Happy

4690978242_e754286de1I really like the way Avik Roy went about explaining the Oregon Study, and for that matter, statistics to his readers. While he is right to try to break down the Oregon Medicaid study like a similar RCT for a drug (we really need to practice evidence-based health policy), he must remember that health insurance touches far more than a simple biochemical interaction within the human body. Health insurance largely influence whether or not someone even presents for care, regardless of whether or not that individual adheres to prescribed treatment.

When randomized controlled trials are designed for pharmaceuticals, they often select a group either already with a disease or at least at high risk in order to have the best chance to uncover a true effect. When the results of this Oregon study broke, I was listening to a presentation about a new drug and dissecting the pharmaceutical propaganda that was trying to convince me that drug X was “noninferior” to drug Y. I would much rather know if drug X is equal (or not equal) to its well-established competition than if it is possibly just as good. But you cannot expect the pharmaceutical industry to conduct that type of study.

At least with the Oregon experiement we have a true comparison of insured versus uninsured. If there is a true difference, and it’s big enough to be found based on that sample, we will soon know it. Avik points out some reasonable biases which should favor Medicaid  – better access to doctors, better Medicaid fees – in the Oregon experiment. Yet, there is still a lack of statistical benefit in health outcomes. [I was enthused that there was discussion of what a “p-value” is but disappointed that he brushed aside the the concept of statistical power. If you are going to talk about Type 1 error, you should at least give Type 2 error its due. But I digress.] Buried way down in Avik’s piece, and typically up high in articles written by those favoring Medicaid, is the study’s most important and enduring finding – improvements in mental health.  

  As a review, saw subjective improvements in mental health and physical health. The latest research fails to show objective physical health findings. However, the mental health findings are legitimate. Vastly fewer individuals screened positive for depression based on a common, validated survey (PHQ-8) once they obtained Medicaid coverage. In fact, while 30% of the control group screened positive for depression, the study group (Medicaid patients) experienced a slightly greater than 9% decline in depression. If Medicaid were a drug, as Avik likes to point out, it would be pretty safe and effective for treating depression with a number-needed-to-treat slightly above 10.

So while I will agree that the physical health impact of Medicaid appear negligible (at least at 2 years and in a population that does not appear as ill as most Medicaid recipients), the mental health impact is huge. That can’t be denied. If this were a drug study, they’d stop right there and get approval for an FDA indication to treat depression. The financial impact – which is really what health insurance is supposed to be all about anyway – is tremendous. In my view, another undeniable benefit to Medicaid.

What does this mean moving forward? Should states implement the Affordable Care Act’s Medicaid expansion? On the basis of this data, I would argue to proceed with caution understanding that physical health (at least for very limited measures of blood pressure, cholesterol, and diabetes) will not necessarily be immediately impacted. We have no clue how it impacted other disease processes. Did the improvement in cancer screening results in changes in morbidity and mortality from those diseases? Nevertheless, the psychological and financial benefits to America’s poor might make expanding Medicaid (or at least some kind of coverage) a worthwhile proposition.

by

Cedric Dark, MD, MPH