Per Avik Roy, the study we’ve been discussing is up online. Unfortunately, it’s behind a paywall, so most of you won’t be able to see it yet. Moreover, it’s complicated, and I don’t think it’s likely permissible for me to repost huge sections of it here. So I’m going to have to summarize, and you will have to trust, but verify.
As has been said over and over, this study used a large inpatient database to examine the association between primary payer status and outcomes. And, as has been said, it found that being uninsured or having Medicaid led to worse outcomes, including a higher risk of death.
I’ve read the whole thing now. If you came looking for me to tell you the methodology is fatally flawed, I will have to disappoint you. The authors were thorough, careful, and skilled. Their analysis is complex, well controlled, and uses established methodology. I do not doubt that their results are robust, nor do I think there is anything major that I would change.
Where Avik and I will likely agree is on the strict recitation of the findings:
Unadjusted mortality for Medicare (4.4%; odds ratio [OR], 3.51), Medicaid (3.7%; OR, 2.86), and Uninsured (3.2%; OR, 2.51) patient groups were higher compared to Private Insurance groups (1.3%, P < 0.001). Mortality was lowest for Private Insurance patients independent of operation. After controlling for age, gender, income, geographic region, operation, and 30 comorbid conditions, Medicaid payer status was associated with the longest length of stay and highest total costs (P < 0.001). Medicaid (P < 0.001) and Uninsured (P < 0.001) payer status independently conferred the highest adjusted risks of mortality.
Patients with private insurance do best. After controlling for important covariates, Medicaid had the longest length of stay and highest total costs. I could come up with explanations as to why this is, but – to be honest – it’s irrelevant to the larger point I want to make, so let’s leave that for the moment. The other key finding were that Medicaid and Uninsured patients were independently found to have higher mortality after adjusting for other factors. So far, I bet Avik and I won’t quibble.
But one key point, and again it is not huge, is that all the comparisons showing significance were done with private insurance as the reference. So it is totally correct to say that both Medicaid and Uninsured were significantly worse that private insurance. But there was no test comparing Medicaid with Uninsured. So this study did not show that Medicaid patients had a higher mortality than the uninsured. Yes, the odds ratio was higher for Medicaid than for the Uninsured, but the confidence intervals overlap with respect to mortality, so we can’t make clear distinctions between those two groups.
But, again, I don’t want to spend too much time on that point. I agree with the larger point that Medicaid patients and the Uninsured both fared much worse than those with private insurance. I accept the results.
Now on to likely disagreements.
First off let me address this:
Aaron also notes that Medicaid is voluntary: but this is weak support for the implication that Medicaid, in its current form, is the best we can do. In the Vietnam days, some conservatives used to tell liberals to “love [America] or leave it.” I don’t remember liberals being too happy about that. Nor does Medicaid’s voluntary nature mean, ipso facto, that it must be doing some good. Does the voluntary nature of Medicare overutilization mean that Medicare overutilization is a good thing? Most liberal health policy types that I know believe otherwise.
It’s not that I think Medicaid is voluntary. It’s that I don’t think Medicaid keeps you from seeing doctors you could otherwise see if you didn’t have it. As I said before, Medicaid patients are able to see physicians that accept Medicaid plus those that they can pay for out of pocket. I don’t believe this equals “doing some good”. I do think this equals increasing the available pool of physicians over what they would have without Medicaid. I think I know what a response might be to this argument, but I hate when people presuppose what the opposition must think, so I’m going to let others make their own arguments as to why this is wrong.
But the larger issue is to what this all means. I agree that Medicaid is associated with bad outcomes. I agree that Medicaid is flawed. I agree that Medicaid could be improved. But I don’t necessarily agree that Medicaid is the cause of the problem.
I wish I could repost huge tracts of the manuscript, because I think the authors do an excellent job of describing many of the reasons this association might exist. Almost none of them are causal. I will summarize as best I can:
- Elective operations were more common with Private Insurance and nonelective were more common in Medicaid and Uninsured patients, pointing to the fact that those populations don’t have the same choices up front. Elective surgeries usually are planned for and have better outcomes. Yes, this is controlled for, but still significant (and noted by the authors).
- Patients with private insurance may have better access to higher quality physicians or facilities. There is a good amount of evidence that surgeons’ experience matters. I don’t disagree with this. I do disagree that Medicaid worsens this over being Uninsured because of the argument I made earlier.
- Other differences may exist between these populations that are unaccounted for. The authors note:
Both Medicaid and Uninsured payer groups had the highest incidence of drug and alcohol abuse. In addition, Medicaid patients had the highest incidence of acquired immunodeficiency syndrome, depression, liver disease, neurologic disorders, and psychoses. Furthermore, Medicaid patients had the highest incidence of metastatic cancer, which likely reflects the combined influence of deficits in access to care, poor health maintenance, and delayed diagnosis resulting in the presentation of advanced disease stage within this population.
- Patients with Medicaid and the Uninsured use the system differently than those with private insurance. More of them get their primary care in the emergency department. They are more likely to have language barriers or health literacy issues. They are more likely to be malnourished and have other issues.
Again, though, we can’t know for sure.
So what should we do with this? If I read Avik’s argument correctly, he thinks Medicaid is fatally flawed and should be replaced with, perhaps, subsidies to buy private insurance.
Now – in theory – that’s not the worst idea in the world. In fact, it’s what the insurance exchanges in PPACA represent. I am going to take Avik at his word and go with this a little further.
In a previous post, Avik seemed to endorse giving the poor a lump sum payment and letting them try and get insurance. I don’t think that’s a good idea. With individual ratings, insurance for much of the Medicaid population would still be out of reach even with the subsidy he cites. But in his most recent post, Avik endorses the Swiss system. Now, that is much more interesting.
The Swiss system has a powerful mandate. They require everyone to buy private insurance. They also give out subsidies so that insurance is not more than 8%-10% of anyone’s income. All insurance companies are mandated to have a “basic” plan with regulated benefits. This basic plan also must be non-profit. All insurance is community rated. The Swiss system also takes a pretty heavy hand with practitioners. As Uwe Reinhardt notes:
On the surface, the Swiss health system may give the impression of a price-competitive, consumer-directed health care model. However, the heavy government regulation that pervades the entire system—including the health insurance sector—makes it a far cry from the vigorous, price-competitive health care market envisaged by the advocates of consumer-directed health plans in the United States. Some gestures to competition aside, the Swiss system so far has remained mainly a de facto cartel of insurers and health care practitioners who transact with one another in a tight web of government regulations.
I agree that the Swiss system costs less than ours, achieves comparable and sometimes better outcomes, and covers nearly 100% of their population. If this is the direction Avik thinks is a good idea, then we may agree more than I think.