• Physician services spending: Medicare vs. private insurers

    Before I get into it, with this post I am not claiming that health spending on physician services is the only or even the most important reason for overall health spending growth. What I’m really doing is documenting some figures I’ll be using in a talk. Since I went to the trouble of making a figure, and copying another, I might as well practice getting my thoughts straight about them here. You get to see the rough draft. (Lucky you. Of course,  you can avert your eyes if you wish.)

    We all know that Medicare has a physician fee schedule, the basic structure of which governs how private insurers pay physicians as well (which is not the same thing as saying private plans pay physicians the same rates as Medicare; they don’t). It’s the overall growth in Medicare physician fees that is the subject of much, and frequent, debate: the “doc fix.”

    About that, last month Uwe Reinhardt posted some intriguing graphs that illustrate what a mockery the doc fix issue and the system it aims to “fix” have made of the control on costs they’re supposed to exert. From those, here’s the graph I’ve selected for my talk.

    Reinhardt explains,

    The lowest line in the first chart (in red) shows Medicare’s fee increases from 2000 to 2009. […]

    [T]hat line is so much lower than the green line, which represents growth in the M.E.I. [Medicare economic index], the index that measures the annual increase in the cost of operating a medical practice. […]

    Indeed, even if one assumed that the M.E.I. should be adjusted downward a bit because physicians should be able over time to increase the productivity with which practice inputs are used, basing fees on the productivity adjusted M.E.I. (the blue line) would still have raised Medicare’s physician fees substantially more than the actual fee increases.

    Finally, however, the top line (in black) shows that, in spite of Medicare’s miserly fee updates [the red line], total Medicare spending on physician services per Medicare beneficiary actually has grown by fully 60 percent from 2000 to 2009, at an average annual compound rate of 5.4 percent.

    (By email, Reinhardt wrote me that he believes the figures in the chart are not adjusted for overall inflation.)

    What’s driving that huge increase in physician services spending? Not fee levels (the red line), but utilization. Medicare’s inability to control utilization is, by now, a familiar bugaboo of its payment system, the very reason for proposals of bundled payments, ACOs, and the like.

    Since private plans piggyback on the structure, though not payment levels, of the Medicare physician payment system, I wondered how the rate of increase in spending by them compared. Unfortunately, after much searching, I could find no figure comparable to the one produced by Reinhardt, above. So, I made my own, using National Health Expenditure data for private plan physician spending and OECD/Census data for number of people insured by such plans. Here’s the resulting figure:

    The black line is cumulative (nominal) growth in private plan spending on physician services per covered life. The red line, for comparison, is growth in (nominal) per capita GDP. Physician spending by private plans has far outpaced overall economic growth in the last decade.

    Comparing the figures, private plan spending growth is above that of Medicare. So, perhaps Medicare’s controls on fees do have an effect, though clearly a modest one.

    UPDATE: As pointed out in the comments, my use of the term “volume” above is not quite right. I should have written “utilization.” Read the comments for details.

    UPDATE 2: The post and second figure have been redone because I learned that the Medicare figures are not adjusted for overall inflation.

    • Good start. I wish we had more data. We need the equivalent for Reinhardt’s red line for private insurance fees. I have never found this anywhere. I suspect that the ability of private insurers to dump costly patients is a factor here, but data will be hard to find. I wonder what happened between 2006-2008? I also wonder if hospital integration into networks affects the elderly much more than younger patients? If it is easier to get additional consultation and testing, that seems more likely to happen in the elderly.


    • Austin, unless you are using “volume” loosely, it is not true that this is the primary reason total expenditure growth has so vastly outpaced fee schedule increases. There are two other factors: upcoding and the introduction and increased use of higher cost codes to address new technologies. It’s greater utilization of the highest cost codes that is the greatest problem (MRI vs x-ray, newer types of surgery, etc.), not volume across the board. It’s been shown many times that the US does not have more utilization of services in general than other nations, and if anything we have become lower utilizers of some things in recent years (inpatient hospital visits come to mind). I know you’re aware of this.

    • One complaint: The figures are not 0-based. If you’re doing a “cumulative percent change”, the initial starting point should be at height 100, not at height 0. (I guess I would label such a graph as “Relative spending, 2000 = 100” or something like that.) The numbers are alarming enough without exaggerating them.

      • @Dylan T – What is the %change between the underlying year 2000 figure and itself? I’d say zero. You seem to want it to be 100%. Maybe it is the axis label you object to. What should it read so zero is the right answer in 2000? Remember that I want to compare with Reinhardt’s figure. I couldn’t as easily do so if I shifted everything by 100%.

    • I of course agree that the %change from 2000 to itself 0. But that’s a measure that leads to misleading graphs, and the more natural measure is “% of 2000 level”.

      I don’t really care what the axes are labelled; I just want the graphs to be shifted up so that the bottom of the graph is the natural 0 for this data, which would be 0 spending on physician services, etc, which with the way your and Reinhardt’s axes are labelled would be at -100%. Alternately, you could do it all on a log scale, and then I’d have no complaints. Can’t you regraph his data?

    • Thanks to Jonathan for hitting the nail on the head — i.e. upcoding.
      The Japanese have the answer, if the goal is cost control. Their fee schedule is by and large as flat as a pancake. Both X-rays and MRI’s are reimbursed at $100, not $70 for one and $1200 for the other.

      Controlling US medical spending will require someone who has no respect for the costs of providers. When Walmart tells toothpaste manufacturers that the selling price should be $1.45, Walmart will not listen to complaints that resarch costs have gone up, labor costs have gone up, etc etc.

      Virtually every drug in common usage could be sold for $30 a month. Virtually every diagnostic test could be performed for $200 or less. Most busy hospitals could provide a day of care for $1500.
      Doing all these will mean lower profits and fewer stock millionaires.
      We’ll survive.

      Bob Hertz
      Director, The Health Care Crusade

    • New to this, realize it’s an old, probably dead thread. Gee, not sure that WallMart is a model we should strive for here unless you want my employees to go without health insurance- and you would be misinformed if you think that all Drs offices provide health insurance to their employees as a staple. Don’t know how to incorporate Chinese sweatshops into my office practice business model either. The upcoding concerns are spot on, and I don’t think EMR is in that realm anything less than a Pandora’s box. After all, apart from enhanced Medicare reimbursement for adopting these platforms, they’ve routinely been sold and engineered with upcoding in mind, so expect that curve to go more vertical as more and more providers adopt EMR.