Monday, July 19, 2010

Re (2): The UVa Surgical Outcomes Study

Cross-posted from The Agenda on National Review Online.

I’m grateful for Austin Frakt’s continued interest in the problem of Medicaid’s poor outcomes. Here are his most recent thoughts:
Avik Roy responds to my post. He lists several more studies that find Medicaid patients have far worse outcomes than privately insured ones and the uninsured too. I’m not going to undertake a literature review. I don’t have time. So, I’m not really debating the merits of the studies Avik Roy cites or whether they are representative of the entire body of work in this area.

So, let’s presume they are credible and representative, then what is the implication? Should we make Medicaid more like private insurance or more like no insurance? Should we Federalize the program?

I believe that everyone should have access to affordable insurance that facilitates access to affordable, high-quality care…If Medicaid doesn’t fill that role for low-income individuals, some of whom are very sick and/or disabled, then it should be reformed. That probably means spending more money on it.

I’m not getting the sense that’s what Roy has in mind. He writes that “most people can afford to take on more responsibility for their own care, and indeed would be far better off doing so.” That sounds like he wants to make Medicaid more like no insurance.
Austin is a thoughtful guy, so I’m surprised to see him argue that throwing more money at a dysfunctional system is the best path forward. We already spend $436 billion a year on Medicaid, only to achieve the dismal results reflected in the Virginia study. Once PPACA is fully implemented, Medicaid may surpass Medicare as our most expensive health-care program. Medicaid is overwhelming state budgets, crowding out other essential government services. And I haven’t even mentioned our federal deficit. In the post-Obamacare era, increasing physician reimbursement for Medicaid patients to still-stingy Medicare levels will cost $200 billion a year. If Austin has that kind of change under his couch, he should let us know!

I instead favor, as a start, what Mitch Daniels has accomplished with the Medicaid program in Indiana (before PPACA destroys it): subsidized health savings accounts combined with consumer-driven health plans. And Indiana covers people at up to 200% of the poverty line, compared to Obamacare’s 133%. Instead of covering more people, I would more heavily subsidize those at or below the poverty line, in order to bring Medicaid’s low physician payments in line with those of the private sector. Ideally, we would move to a modified version of the Swiss model, in which everyone purchases consumer-driven plans in the individual market, with graduated subsidies for lower-income households.

Ultimately, the goal should be to minimize the number of people who require subsidized insurance. This requires comprehensive health reform aimed at reducing the cost of health care: de-linking employment from insurance; broadening the reach of consumer-driven care; creating a national insurance market; aggressive antitrust enforcement against providers; medical tourism; transparency; malpractice reform; and Medicare reform.

Austin also passes along the comments of some of his readers, who question the methodology of the Virginia study. Reader Jay worries that the Virginians didn’t control for prior health status. On the contrary, they controlled for 30 co-morbid conditions, along with age, gender, income, geographic region, and surgical procedure. They may not have directly controlled for smoking or alcohol consumption (we don’t yet know), but these issues would show up in the comorbidities and in the surgeries themselves. For example, lung resections are typically performed on smokers.

Reader Steve worries that they chose procedures that would be heavily tilted to academic centers, and that this biases the results. From my reading, the procedures they chose are among the most common high-morbidity surgical procedures performed today, which is what you would want an outcomes study to examine. A larger issue, which I discussed in my previous post, is that Medicaid patients appear to spend more time in low-volume surgical centers.

Steve also says:
I would bet almost anything that they do not look at all of the social factors that would contribute to worse outcomes for a Medicaid population. Docs don’t generally look for those. If you are uninsured, how long have you been uninsured? Are you working if you are uninsured? What is the functional capability of someone on Medicaid not working vs. someone uninsured who is working? Which group is more likely to have communication problems? Which group is more likely to give a better history? Which group is more likely to get family support? Which group is staff more likely to dislike?
Steve is missing the point. It’s not the job of these studies to eliminate every conceivable factor that leads Medicaid patients to underperform the uninsured. Rather, they are seeking to establish whether or not Medicaid patients underperform the uninsured, after controlling for basic factors like age, income, and comorbidities. Once you establish that Medicaid underperforms, you can then begin to propose hypotheses as to why this underperformance happens.

On this score, Steve has reversed the cause and the effect. Welfare dependency is what leads to problems like family breakdown, underemployment, and diminished functional capabilities. Dumping 16 million more Americans into our broken Medicaid system is not the answer, but rather its antonym. It is not compassionate, but cruel.


  1. "On this score, Steve has reversed the cause and the effect. Welfare dependency is what leads to problems like family breakdown, underemployment, and diminished functional capabilities."

    Is there data to support this belief? My experience with welfare patients, and some of my staff who were on welfare at one point, suggests a much more diverse group. Many were clearly non-functional people before they went on welfare.


  2. Hi Steve, it's fair to say it's a combination of both -- a good example of the non-functional type is the schizophrenic population. The question is, do the non-functional people who drop into Medicaid (as opposed to the people who become non-functional *because* of Medicaid) drive the data? I'm sure it has an incremental impact, but the problem of physician access is of far greater significance.

  3. I should also mention that the UVa study does control for many of these factors. Here is the complete list of co-morbidities that the study controlled for.

    You'll notice that the list includes AIDS, alcohol abuse, drug abuse, depression, psychoses, and liver disease (eg hepatitis).

    Alcohol Abuse
    Deficiency Anemia
    Arthritis/Collagen Vascular Disorder
    Chronic Blood Loss Anemia
    Congestive Heart Failure
    Chronic Pulmonary Disease
    Diabetes Mellitus (uncomplicated)
    Diabetes Mellitus (complicated)
    Drug Abuse
    Liver Disease
    Fluid and Electrolyte Disorder
    Metastatic Cancer
    Neurologic Disorder (not CVA)
    Peripheral Vascular Disease
    Pulmonary Circulation Disorder
    Renal Failure
    Solid Tumor (without metastasis)
    Peptic Ulcer Disease (non-bleeding)
    Valvular Disease
    Weight Loss