• Health care spending since 1980

    I’m getting e-mails and comments asking about health care spending since 1980. Specifically, why did the U.S. pull away from the OECD pack beginning then?

    The divergence since 1980 is not that puzzling and we’ve covered it here before (and recently!). But maybe we didn’t make it explicit enough.

    Aaron’s chart from a recent post shows outpatient care taking off is the biggest factor.

    That outpatient spending is a big factor in excess U.S. health spending (above that predicted by GDP) is one conclusion documented in Aaron’s series on health care spending. As for what it is about outpatient care that is responsible for the rapid increase, Uwe Reinhardt and colleagues told us fairly convincingly that it’s the prices, stupid (true for other settings too).

    We could keep asking why. Why are prices rising faster in the U.S. than elsewhere? At this point I think it gets complicated and there are likely multiple factors relating to policy, provider market power (e.g. hospital consolidation), technology, etc. Even though we may not know all the answers, we certainly know some. What’s happened since the 1980s is not a complete mystery.

    It’s also important to keep in mind that the first graph above is health care spending divided by GDP. So, changes in GDP affect the lines too. (On that, more here and here.)

    • This is just a bit of microeconomic musing based on my personal experience, but I wonder if the shift to HMOs isn’t the biggest culprit.

      When I was a child, my parents’ insurance had a deductible, then covered 80% after the deductible, so a visit to the doctor had a not insignificant out of pocket cost. In y adult life, however, I have always had HMO coverage, with fairly co-pays from $10-25.

      Now, if I think about it, I am aware that premiums have been going up, but my premium is paid on my behalf by my employer, so I don’t feel it as acutely as I do the out of pocket cost; and I also know I pay more attention to this than the average consumer (I am one of the “intelectial eleet”), so many people are probably less aware of this connection than I am.

      Because of the low co-pays, I am much more willing to see the doctor for things Mom would have smeared some Neosporin on, wrapped in gauze, and told me to be tough and tell her if it got infected.

      So, do low out of pocket costs lead to overutilization, which in turn leads to higher total costs when the total costs are largely hidden from the consumer?

    • Wow, I should really get in the habit of proofreading before hitting “send.”

    • I’ve been thinking about the claim that cross country differences in expenditures per capita are largely driven by price differences (the point that my friend Uwe Reinhardt makes in the ‘Its the Prices Stupid’ article. My reading of this work is that the price-variation explanation is premature: all that this literature does is to point out that _certain_ inputs (physicians per capita, hospital days per capita, beds per capital, MRI) are available to a similar degree in the US relative to other OECD countries. It interprets this as a statement about the Qs being the same and the Ps (prices) being different.

      Four observations on this line of reasoning:

      1. What folks are calling a difference in prices may be a difference in unmeasured utilization. Folks are comparing office visits per capita in the US to Canada (5.8 v. 6.4 in Exhibit 4). But this is a problematic comparison because (a) it doesn’t measure the intensity of each office visit– what was prescribed within the visit or what outpatient care was recommended (b) that 20% of the US population receives little ambulatory care by virtue of being uninsured. Even a small correction for (b) would make the US have more office visits relative to Canada.

      2. A a big source of cost-growth comes from the use of outpatient hospital services (use of laprascopic surgery) and skilled nursing home facilities. There is nothing in the paper that speaks to these margins, and it’s probably one reason for why we have fewer hospital days per capita.

      3. Exhibit 6 notes that the US has 3 times as many MRI scanners per capita and over 1.5 times as many CT scanners relative to Canada. This comparison doesn’t even measure the more frequent use of these scanners, which are a big source of cost-growth in the US. It’s true that we have fewer scanners per capita than Austria and Japan– but it’s the frequency of use that matters, not the presence of a scanner.

      4. Finally, note that Exhibit 6 shows that the US does almost 500% more angioplasties relative to Canada and close to 800% more than the UK. Angioplasty is a big source of cost-growth in the US. Angioplasty for stable coronary disease (which is what a lot of these are) was not shown to confer any survival or quality of life benefit in the COURAGE trials.

      All in all, I don’t think we know if Qs matter more than Ps for the cross-country comparions–and certainly not with the certainty with which it is being stated. Moreover, given that no one is trying to make the US look like Canada, the within US differences in expenditures (which the Dartmouth work shows is driven by Q much more than P) are the key thing to focus on.

      As always, a pleasure to read your blog.


    • This seems like good news to me because outpatient spending seems to be the sort of medical care that is most likely to be controllable by patients. Reducing insurance through higher deductibles and co-pays could create a market with prices for outpatient care more than say emergency care or critical care. Most of it is non emergency and the patients needing outpatient care are not generally in too bad a shape to shop for price. Also it seems the sort of care where payment plans might work. So it sounds like good news for health savings account and more consumer involvement.

      Also to Amitabh Chandra,
      Just a single data point to add to your point # 1 my doctor is from Germany and he said that he left Germany for the USA because he was not allowed to spend what he thought was enough time with each patient in Germany. I haven’t checked it out so it could be wrong but you could be on to something.

    • Before blaming outpatient cost growth for the bulk of American exceptionalism since 1980, please take into account that a large majority of the outpatient care would formerly have been delivered in an outpatient setting. So, as outpatient costs grew massively, inpatient should have seen a roughly equivalent decline. But that is not what happened. Instead, inpatient costs increased as well, even though bed days per thousand have declined.

      So, really you need a more detailed analysis before you say that outpatient rather than inpatient is the primary culprit. What I would say instead is that facility costs are the primary culprit. International comparisons show that mostly its the prices, not the volume.

    • Amitabh is right in pointing out that the crude “output” measures contained in the OECD Data Base may hide variations in resource intensity per unit of crude output not apparent from the available data.

      Our paper “It’s the Prices Stupid” was mainl a reaction to the widely held belief that cross national variations in the level of purchasing-power adjusted health spending per capita reflect mainly different degrees of rationing heath care. ( We in the US, of course, do not ration health care, even for the uninsured I suppose. At least that’s what I constantly hear.)

      While the evidence for our thesis is admittedly circumstantial, I do find it persuasive. First, there is the fact, first noted by Pauly, that many other naitons have more of certain inputs — notably doctors, hospital beds and nurses, than we do. Second, we know that prices for standard products and procedures — e.g., drugs or MRIs — ar esubstantially higher in the US than elsewhere. I would include in that even normal deliveries. Thus it is not far fetched to assume that prices even for identical physician visits in the US would be higher than those abroad.

      Unfortunately, even if the procedures packaged into a crude output measure such as a “physician visit” in country A are more plentiful than those packaged into a visit in country B, we do not know what actual benefits the extra procedures yield.

      We all know that ideally we would like to measure prices as money spent per unit of properly measured outcome, and not per bundle of real inputs. Perhaps one day we’ll have it. Amitabh is young and smart enough to perhaps get that done.

      In the meantime, there is much room for argument, even on the intra-national variations to which Amitabh alludes — the Wennberg variations.

    • Interesting discussion. I’ve only been in the health policy field for about 2 years now, but the cost drivers of healthcare (why prices are so high here) seem pretty obvious, and are at least hinted at in the Paul Starr, “social transformation” book. They are, in no particular order:

      1) Technology–we have the fanciest, most expensive stuff, and the marginal utility of much of it is questionable at best. Nonetheless, everyone demands access to it, and substantial public and private dollars are spent pursuing it.

      2) Insurance–Employer-based insurance effectively masks the true cost of care from the patient, and creates an extra level of bureaucratic interference. The dual effect is to drive more demand, and make administration more complicated/costly.

      3) Culture–this is pretty vague, but our lifestyles (diabetes, cancer, heart disease, etc.), and our expectations (everyone wants to live forever, nobody wants to pay for it) all drive up the need for, and the cost of, those fancy new technologies.

      An obvious solution seems to be increasing the number of physicians and other providers (particulary nurses, Physician’s assistants, etc.) to meet demand, and unwinding the Employer-sponsored insurance paradigm.

      Unfortunately, the fourth factor in driving up prices is also the lead barrier to addressing 1-3: Our dysfunctional (corrupt?) political institutions and an (a)pathetic body politic.

      • @Steve B. – Part of the question is, why since the 80s? All the stuff you point to existed in the 70s too. What changed in about 1982?

        Increasing physician/nurse supply may not be the answer. Physician induced demand increases costs too. Barriers to use (waiting times) can decrease them (though there are some compensating factors in both cases).

        There may not be a simple, knowable answer here. If there were someone would have said it (or did I miss it?).

    • Thanks, Austin.

      Coming from a background more rooted in political science than economics or medicine, my gut reaction is that the politics of the country changed. Dr. Reinhardt alludes to this in his Economix peice on physician fee schedules, with the implementation of more elaborate methodologies (jealously gaurded by the AMA(!)) for calculating physician reimbursement. I wager an otherwise ill-informed guess that the physician-induced demand you mention is abetted by these methodologies, and that FFS does drive overutilization (I swear I’ve seen data on this somewhere).

      There were also significant changes to tax law around that time that seem to correlate with the observed trend of employees unwittingly substituting wages for health insurance. “Winner-take-all politics,” by Jacob Hacker and Paul Pierson, touches on this factor in service to their larger argument about inequality.

      The early 80s were also a boom time for technology, with the mass proliferation of PCs into homes and workplaces being the most obvious example. While this tech has clearly increased productivity over time, there are always trade-offs, one of them being decreased productivity as businesses and workers acclimate (Edward Tenner’s “technology bites back” documents these effects superbly). I’d risk some intellectual street cred to advance the notion that the broader trend of exponential technological growth (Moore’s law, for example) really began to take off around this time, and that if we were to (somehow) plot technological advancement (patent applications, perhaps?) in the medical field, we would observe some degree of this trend (how about the introduction of Ritalin, etc. to a broader marketplace?).

      The 90s and 00s saw this accelerate, particularly with the removal of old-timey restrictions on direct advertising to patients by pharmaceutical companies and the proliferation of drugs for a whole host of new problems like Erectile Dysfunction and “restless leg syndrom.” Again a political decision.

      Assuming my hypothesis is remotely correct, the real cost issue lies in the fact that many of these technologies (particulary the drugs) are largely untested in terms of marginal cost/utility over existing technologies, and our politicians are loathe to fund any such research on the grounds that its “killing granny.” (Fingers crossed PCORI gets off the ground).

      Anyway, this is obviously just meaningless theorizing–I have no data on hand to confirm anthing I’ve just said–but thanks for providing a platform to daydream about this stuff.

      • @Steve B – It’s all good hypothesis generating musing. That’s valuable. I just crave something more researched and evidence-based. It may not be out there. If it is, I’m hoping someone will point me to it. (Or if they have already and I seem to have ignored it, do it again. I’ll read it.)

    • I hope I can engage either Uwe or Amitabh on a question I have posed to Austin both on and offline

      It concerns the price vs volume dilemma, similar in spirit to what is discussed above. However, rather than speak in terms of US vs the OECD and “who is better,” my desire is to focus on our quandary domestically, mainly, what is the primary cost driver in the US?

      Not to oversimplify the two opposing paradigms, but it appears to that there are two schools of thought on the subject: some folks weigh in with an “its the prices” position, and others will focus primarily on the overuse of technology, we “do too much stuff” angle.

      Two focus on one, exclusive of the other, of course is foolhardy. But there does seem to be a parting of the ways on which of the two badder actors contribute to the inflationary trends in health care.

      Here is my question:
      In a two by two table, you have high and low utilization on the “x” axis, and on the “y” axis you have high and low prices. Which quadrant is the biggest problem in our country?

      I understand that the answer is, a) not cut and dry, and b) not complete as far as existing evidence, but will graciously settle for best informed guesses.

      Thanks, and I hope I can engage one or both of you to elaborate just a bit. I find this issue a real challenge.


      • @Brad F – Of course it is “both,” but my understanding is that volume and intensity are major drivers of cost in every country, the US included. However, a big difference between the US and other nations is cost (though volume may play a role there too). Happy for others to weigh in on this.