For those who haven’t had their fill of the Medicaid reform discussion, the full results of the UVa surgical outcomes study have been published in the online edition of Annals of Surgery. (I ask everyone who has had their fill for forgiveness.) There are a couple of points that keep coming up in the comments and in responses from other bloggers, so I want to spend at least one post addressing them.
Austin Frakt writes that, contrary to my expressions of concern, he is quite open-minded to the possibility that outcomes with Medicaid are poorer than those of the uninsured (and especially those with private insurance). He remains reasonably skeptical that studies like the Virginia one adequately control for the poor social and health status of the Medicaid population:
There are undoubtedly studies that consider Medicaid vs. uninsured outcomes using the random variations provided by the natural experiment that is Medicaid. Characteristics of the program vary by state and year, making it a perfect set-up for such an analysis of this issue. This second I can’t point to a study. But I know where to look. One place to start would be to examine the literature cited by Stan Dorn on Ezra Klein’s blog at the Washington Post (tinyurl.com/StanDorn), Harold Pollack on The New Republic’s The Treatment blog (tinyurl.com/HPollack), and by J. Michael McWilliams on this blog (tinyurl.com/JMMcWill).
That’s it. That’s my position, and it always has been. If you read carefully you ought to notice that I didn’t actually condemn or praise Medicaid. I didn’t actually say how it should be reformed. I just listed the possibilities.
It is certainly important, in any study comparing Medicaid to other insurance populations, to control for the kinds of things, like prior health status, that negatively skew Medicaid outcomes. In medicine-speak, these factors are called comorbidities. As I have written before, one must be careful with comorbidity analyses not to eliminate the ways in which Medicaid actually contributes to poor health status.
Having said that, the authors of the Virginia study did control for 30 different comorbidity measures, using a widely used, highly validated methodology first worked out by Anne Elixhauser and colleagues in 1998. Using 1992 data from 438 California hospitals, Elixhauser et al. identified those factors that most significantly contributed to “substantial increases in length of [hospital] stay, hospital charges, and mortality.”
Notably, the Elixhauser comorbidities include several factors that many associate with the Medicaid population: AIDS, alcohol abuse, depression, drug abuse, liver disease (such as hepatitis), obesity, and psychoses.
(The others are: deficiency anemia, arthritis/collagen vascular disorder, chronic blood loss anemia, congestive heart failure, chronic pulmonary disease, coagulopathy, uncomplicated and complicated diabetes, high blood pressure, hypothyroidism, lymphoma, fluid and electrolyte disorder, malignant cancer, neurologic disorders (other than stroke), paralysis, peripheral vascular disease, pulmonary circulation disorder, renal failure, solid tumors (non-malignant), peptic ulcer disease (without bleeding), heart valve disease, and weight loss.)
The Virginia study authors also controlled for income, age, and gender.
To me, this is a responsible, comprehensive, and quantitatively validated list of comorbidities and adjustments. To those who disagree: what other adjustments would you have preferred to see? Which, if any, of these additional adjustments would have had a significant impact on the results of the study? What data can you cite to support their importance, and to support that they are independent of the effects of Medicaid itself or of welfare dependency? Austin has written extensively about the general concept of controlling for extenuating factors, so his thoughts on this topic could be illuminating.
Austin also writes that a definitive study to measure the benefits of Medicaid against those with other forms of insurance (or uninsurance) would need to be prospective—that is to say, we take X number of people, put half on Medicaid and half on something else, and see what happens. While he’s right that prospective studies are, all else being equal, better than retrospective ones, they can also be misleading if they aren’t stratified properly by health status, social factors, etc. Another disadvantage of prospective studies is that they usually sample a much smaller number of people than retrospective studies can. Even if we could design a good prospective Medicaid study, actually conducting one would be very difficult and take several decades to play out. I hope we can all agree to use the best available evidence in the meantime.
I would also point out to Austin that exactly the same arguments could be applied to PPACA: i.e., let’s do some well-designed, prospectively-controlled studies before adding 16 million people to the Medicaid rolls at a cost of $100 billion a year.
Aaron Carroll points out that the UVa study focuses solely on surgical outcomes. Aaron, as a pediatrician, wonders whether its results are applicable to non-surgical situations. I have addressed this issue in part, but I will take a look at pediatric outcomes in a separate post.
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.
A number of people ask: “Ok, let’s suppose for a minute that the UVa study is accurate, and that those on Medicaid do fare worse than the uninsured. Are you suggesting that we abolish Medicaid and do nothing about the problem of the uninsured?” No—I am suggesting that we transition to something akin to the Swiss model, whereby we offer graduated subsidies with which the poor can buy consumer-driven private insurance. Simply sending them the cash would be far more efficient than what we do now.
I would ask those who ask these philosophical questions if they are happy with Medicaid as it is, or if they have their own ideas for reform. Hopefully we can get past this idea that criticizing Medicaid equals seeking to abolish assistance for the needy.
The evidence of Medicaid’s problems is, in my view, overwhelming. I encourage those who believe otherwise, or are simply agnostic, to spend some time going through the data. Medicaid reform is an issue that should unite those who are concerned with the plight of the poor, and those who are concerned about America’s fiscal condition. These are the two things, after all, that health care reform was supposed to be about.