• In health care, the US is on a different planet

    This is a TIE-U post associated with Jonathan Oberlander’s Political Dynamics and Policy Dilemmas (UNC’s HPM 757, Fall 2011). For other posts in this series, see the course intro.

    This week’s reading included a 2009 paper by Henry Aaron and Paul Ginsburg titled Is Health Spending Excessive? If So, What Can We Do About It? It’s worth reading, and since an ungated pdf is available, I’m not going to unpack it. However, I want to share the contents of one table and two charts. (Scroll way down for the charts. Worth it!)

    Exhibit 1 of the paper asks and answers some good questions about the US health system. Here’s all that in Q&A format.

    Q: Does the U.S. spend more per capita than other countries?

    A: Yes, a lot more.

    Q: Does the U.S. spend more per capita than other countries after adjusting for income and other factors?

    A: Yes, a lot more.

    Q: Has per capita U.S. health care spending risen faster than that of other countries in recent years?

    A: More than some, less than others.

    Q: Has the excess of health spending growth over income growth been higher in the U.S. than elsewhere in recent years?

    A: In general, yes, but by varying amounts and not uniformly.

    Q: Does the U.S. spend a lot on low- or no-benefit care?

    A: Yes, but evidence on how much is poor.

    Q: Does the U.S. spend more on low- or no-benefit care than other countries?

    A: We don’t know, but our larger outlays mean that we could waste more.

    Q: Does the U.S. pay higher prices for health care services than other countries?

    A: Definitely.

    Q: Does the U.S. pay more for health care services, adjusting for quality, than other countries?

    A: Almost certainly, although measuring—and even defining—quality is difficult; in some dimensions, we seem to be doing very well (length-of-stay, cutting-edge procedures); in others, poorly (delivering recommended care, control of diabetes); we have few data from other countries.

    Q: Would cutting (growth of) health care spending raise welfare?

    A: Static: if one could target cuts, yes; if not, no. Dynamic: if cost limits improve targeting of research, yes; if not, could be harmful.

    Q: Will increases in health care spending impose stress on public budgets?

    A: Yes; were it not for projected increases in health care spending, no material long-term gap between revenues and expenditures under current policy would exist.

    Below are the two charts. They compare life expectancy and infant mortality of US states to OECD countries, both graphed against per capita health spending. Some things I already know about these types of charts, but I bet you they’ll come up in the comments anyway: They don’t prove causality. There are objections to infant mortality comparisons based on how they’re measured by country. These are just two (correlated) population measures. There are many others that possibly relate to health care quality and costs. They don’t control for lots of other relevant differences between the US and other countries (or among states, for that matter). There are lots of reasons we have high spending and less than stellar population health in the US. In fact, Aaron and Ginsburg discuss them. If it makes you feel any better, don’t apply any interpretation to the charts at all. Just gaze at them in stoic amusement, as illustrations of some random cuts through data.

    As for me, I still think they are nice illustrations of what many other measures suggest, which is that the US is in a very different (not good) place than other, wealthy nations when it comes to health care spending and population statistics related to health. It’s almost as if there is health and health care on planet Earth (or the wealthy parts thereof anyway) and then there is US-style health and health care, more costly and less effective on a population level. If you want to see vastly more about US health spending and quality see Aaron Carroll’s excellent series on those topics.

    • Unfortunately, the two graphs are tightly related since one of the greatest influences on life expectancy at birth is infant mortality.
      Including infant mortality in “life expectancy” distorts policy responses for both obvious and non-obvious reasons; an example of the latter is the current fashion for justifying increases in retirement and Medicare ages based on reductions in infant mortality.

    • I gaze in stoic amusement and notice that the gradient is steeper in the U.S. for both outcomes. Can’t prove a thing. Can’t even show the two slope estimates are statistically distinct. I just gaze and think, “Golly gee, do you suppose pre-natal care reduces infant mortality? Hmmm, I wonder if giving people targeted access to care when they know they are pregnant is somehow less effective than giving them continuing access to care that may include an episode of pregnancy.” Can’t prove a thing.

    • I am sincerely puzzled.

      Why the perpetual recourse to these two sets of data when they’re so massively compromised by so many confounding factors? If the statistical data is *really* that damning on every front – particularly when it comes to clinical efficacy relative to cost, there should be no shortage of data sets out there that substantiate the same claims in a much more convincing fashion than LEAB and IM, with wouldn’t require the same litany of caveats.

      It’s quite unlikely that there aren’t any data sets that could be used to compare clinical efficacy vs cost for particular medical conditions. The said data sets could be aggregated into a reasonable proxy for comparing the effectiveness of the things that doctors and hospitals actually do for patients relative to what they cost. That would make sense, if focusing on things that doctors and hospitals can actually do for patients is actually what’s motivating the analysis.

      In every publication/analysis that focuses on hopelessly confounded and massively over-aggregated data sets, it’s far from clear that that’s the case. If it were – a data set that compares how much was spent on equally dead patients without worrying about how spending influenced survival probably wouldn’t be serving as the intellectual foundation that we were building the edifice of our health policy upon.

      Stoicly praying for the day when we’re looking at data sets which arthroscopic MCL repair and amputation aren’t aggregated into equivalent outcomes in a cost/benefit plot…

    • Which are to states that do well? ND and Utah?