• The meaning of health spending skewness

    The skewed distribution of health spending means something important. I’ll get to that. But first, to the data! I showed the following in a prior post. It’s for the entire population (2004 data).

    Some may have thought that the elderly were a big source of skewness. Not so. In a recent NEJM paper, Linda Blumberg presented in tabular form the distribution for the nonelderly (2011 data). I graphed it:

    I know this graph is not in exactly the same form as the prior one. Study both for a moment or two until you get what they’re showing. Here’s the key: the blue bars in the second graph are the averages within the indicated percentile ranges. The orange bars are the lowest spending within each range. It’s the orange bars and their values (at the top of each bar) that are most comparable to the values in the prior figure (the ones in parentheses below the horizontal axis), though keep in mind that this is a 2004 vs. 2011 comparison and the percentile ranges are not the same in both graphs (some are, some aren’t). I don’t have the raw data, so this is the best I can do. Still, it is enough. Try to stay with me.

    It appears to me that the degree of skewness for the entire population is not that different from the nonelderly population. What’s the point?

    The point I want to make about this is not that health spending is skewed. We already knew that. The point is what it means. It means that for the vast majority of us, how we use health care is not how we should think about health care. We are not representative of where the money is. Unless you’ve had a major surgery or struggle with a costly, chronic illness, you’re not where the action is.

    Take this further. Most of us should stop thinking about we would or do respond to the incentives of the health system, should stop thinking about how we would or do behave with “more skin in the game,” should stop thinking about how much we value the health care we use or don’t. Why? Because most of us are not that relevant to the nation’s struggle with health care costs. Thinking about ourselves is thinking about the wrong part of the distribution.

    I confess, I do not have the intuition to think like a very sick individual who requires costly care to stay alive or to maintain some crucial functionality. I do personally know some people, even young people, who have required such care. I’ve spoken to them. But I am not them. They are the ones who we should be talking to in trying to understand how to address our health care cost problem. We should ask them how they would respond to incentives. We should try, as best we can, to walk a mile in their shoes. I bet that walk would be a lot harder than most of us think.

    Comments closed
    • Thanks for this, Austin. I note in the first graph the almost-perfect application of the 80/20 rule 🙂

      One reaction I have to these graphs was that we, the majority of the population that hasn’t had to deal with catastrophic medical bills, would do well to have More “skin in the the game”.

      This is contrary to what you say: “Most of us … should stop thinking about how we would or do behave with “more skin in the game,” …”

      The reason I say so is that as our medical costs are less variable, and the situations in which we need to make drastic and quick large-ticket decisions are fewer, we will probably function more rationally than the top 20% in your graph. The ‘skin in the game’ makes a more rational and more meaningful difference for us in deciding how we appropriately allocate our spending.

      For example, we are deciding which primary care doctor to visit, which method of paying for prescription drugs (mail order?), etc. which are well within the realm of rational and every-day thinking. The skin-in-the-game makes more sense here to me, as it does when we spend our own money on all of our other predictable annual household costs.

      We don’t (thankfully) usually have to make $40,000-ticket, quick decisions, where surely it is more difficult to act rationally, skin-in-the-game or not.

      • Would have to be a tremendous amount of skin! A large amount of spending is well above even a pretty high deductible. Maybe you’re thinking about huge copayments too. Then, one should also consider Nyman’s theory of health insurance. Large copayments would discourage a lot of economically efficient care, even as it discourages a little inefficient care.

    • Health care decision are not always rational. People faced with fatal illnesses hope to win the treatment “lottery.” Their expectations fed by providers and suppliers, lead them to treatments offering a 10% chance of long term improvement with a median two to four month benefit–no matter the cost. Consider the new agents for cancer and their results. Most have offer no chance of cure and virtually no chance of long term remissions, but they are something to try and are making drug companies billions.

      Having skin in that game might help people be more rational but do we as a system have the ability to discern the better approaches from the worse? (There are a few that really work) And are we willing to accept explicit rationing by wealth.


    • There seems to be an assumption that the same people are in the same spending categories every year. Is this true?

      From my experience health spending is irregular, but the time scales of the irregularity vary among groups. Young, healthy people spend very little over years, but then are in an accident, spend $20,000 to $50,000 on diagnostics, surgery, rehab, etc., and the next year are back in the bottom 50%. Back pain patients can go years between episodes, but as their problems escalate, the interval starts dropping. Even cancer patients typically have a lot of expenses in the peri-diagnostic phase, then spending drops to a lower level until a recurrence.

      Has anyone studied the duration of membership in the various spending percentiles? It seems this would have implications on the conclusions.

    • Most of the skin in the game ideas seem aimed at the low consuming 50%. For those in the upper 50% expenditures are going to be determined by geography, networks and physicians preference as much or more than consumer direct choice. The chronically ill are going to go over their deductibles every year anyway, so they are going to be less affected by skin in the game.

      Lastly, we have seen a dramatic growth in the last ten years in deductibles of $1000 or more. It has not slowed growth below general inflation.


    • I wonder what the skewness of life time health care spending is.

    • BTW as a data point, the Amish who do not carry health insurance sometimes go to Mexico for care:

      It is not uncommon for Amish to travel great distances in seeking medical treatment, to out-of-state clinics or even to Mexico.

      So the deductible may needs to be huge but savings can be found on high cost procedures and that could bring in competition. A life time spending overage is what most people would need coverage for as most Americans can amortize even a $300k bill that comes 10 years before they are retired and retired could save for such a bill if they new that was needed. Insurance companies and Governments are a hassle to deal with and create great inefficiencies.

      For another idea, here is Dean Baker on free trade in health care:

      This paper outlines a mechanism for taking advantage of these potential gains from trade: a globalization of the Medicare and Medicaid programs. Since most of the beneficiaries of Medicare are retirees, as are a substantial portion of the beneficiaries of Medicaid, they need not live near a workplace. Many beneficiaries have family or other ties to other countries. The globalization mechanism proposed in this paper would allow beneficiaries of these programs to have a voucher that would allow them to move to other countries and buy into their health care systems, with the government and the beneficiaries splitting the gains. To provide an inducement for other countries to participate, they would receive a premium (e.g. 10 percent) above their costs to ensure that they benefit from this process as well.

    • Regarding the high-spending percentile, vigintile, decile, and quintile: can we make any generalizations about their maladies? How do they generate such high expenses?