• Health care productivity

    This is a TIE-U post associated with Jonathan Kolstad’s The Economics of Health Care and Policy (Penn’s HCMG 903-001, Spring 2012). For other posts in this series, see the course intro.

    This week I’ll return to a paper by David Cutler, about which I’ve blogged before: Where Are The Health Care Entrepreneurs? The Failure of Organizational Innovation in Health Care [1]. In it, Cutler wonders why health care productivity is so low. Here’s his figure:

    According to the figure, productivity growth in health, education, and social services was negative between 1995 and 2005, whereas the average across industries was 2.4%.  Here’s an excerpt from another, similar figure for the period 1990-2010 from a paper by  Robert Kocher and Nikhil Sahni:

    (More about the Kocher and Sahni paper here.)

    Cutler’s paper explains why productivity and productivity growth are low in health care and what can be done about it. He hypothesizes why we see so little innovation in health care and suggests ways to promote it. It’s a familiar set of problems (asymmetric information, inability of plans and providers to capture long-term returns on short-term investments, plan turnover, third-party payment, etc.) and solutions (bundling, provider integration, pay-for-performance, etc.). Along the way, he points to more granular evidence of low productivity: under-, over-use of care.

    First under-use:

    Diabetes is a chronic disease, requiring regular dietary and (often) pharmacological intervention, and testing for possible complications. There are consensus guidelines for how frequently these should occur. […]

    Adherence to guideline recommendations is low. Only 43 percent of diabetics in the United States receive recommended therapy. The issue is not just lack of insurance. [… O]ther countries [with] universal coverage [have] an average success rate (46 percent) [that] is no better.

    Diabetes is not unique. Only one-third of people with high blood pressure have their cholesterol under control (Cutler et al., 2008), and only one-quarter of those with high cholesterol are under control (Hyre et al., 2007). Outcomes for patients with conditions such as depression are even worse. Again, this appears similar in all countries. Unlike excessive use of care with low value, poor chronic disease care management is a feature of all developed country medical systems.

    Next over-use:

    Almost all elderly men have cancer of the prostate. In many cases, however, the cancer grows slowly, and the person will die of something else before the cancer becomes fatal – or even clinically meaningful. Thus, ‘watchful waiting’ is a common strategy. In some cases, the cancer will grow rapidly and should be treated. However, it is not always clear whether a patient has a rapidly growing cancer or not. […]

    Some clinical evidence has examined the effectiveness of [the] different [treatment] strategies. The results suggest that the therapies are approximately equally efficacious in men aged 65 and older, the most common group diagnosed with localized prostate cancer. In particular, there is no evidence that the newer and very expensive radiation therapies have better outcomes. There is some evidence of adverse side effects with surgery – impotence and incontinence are common outcomes – making watchful waiting even more appropriate for many men.

    Still, rates of invasive treatment remain high. Only 42 percent of elderly men with prostate cancer receive watchful waiting. One-third receive a radical prostatectomy, 15 percent receive brachytherapy, 1 percent receive external beam radiation therapy, and 5 percent receive intensity modulated radiation therapy. A final 4 percent of patients receive a combination of intensive treatment – which has not even been explored in the literature. […]

    Patient preferences are not a major part of the variation in treatment. Sommers et al. (2008) show that patients differ in their preferences for side effects and risks of metastatis, but these preferences do not predict the therapy a patient receives. Rather, patients get referred to a particular type of specialist, and this specialist then recommends the  therapy that they judge best. Thus, patients who see only a urologist most frequently undergo a radical prostatectomy, while patients seen by a radiation oncologist undergo some form of radiation.

    Though we can identify reasons for low health care productivity (as above) , there is still an underlying puzzle. Why has neither market nor government found a way to wring more of the quite obvious inefficiencies from the system? Why does productivity in the health sector remain so low?

    One caveat to all of the above, neither of the productivity figures illustrated account for gains attributed to changes in quality of health care. It is not the case that drugs and procedures are the same in 2000 or 2005 as they were in 1990 or 1995. Those who are consuming appropriate, necessary care are probably getting a better dose of health care than they would have a decade ago, with better health outcomes to show for it. So the productivity losses are exaggerated. By how much? I don’t know.


    1. Cutler, D.M. “Where are the Health Care Entrepreneurs? The Failure of Organizational Innovation in Health Care” NBER WP# 16030.


    • Dear Austin,

      I agree that healthcare productivity is below the “average” of 2.4%. But to simply report this mean value obscures the fact that there’s a lot of variation around that mean, and that many industries are well above that average value, and that many other industries are well below it. Just looking at the graph, it appears that the really big gains in productivity are in businesses that have the potential for substantial automation: durable and non-durable goods, information, agriculture, and trade. But if we look at other businesses that require human capital (e.g. the arts, professions, even finance) there is a much smaller gain in productivity. And healthcare doesn’t even have the weakest gains — those are in construction and mining!

      I’ve noticed in general that whenever a mean is reported, there’s a often a subconscious tendency to interpret it as saying that everything has that value, when of course it just reports an overall central tendency. Nobody intends to do this, but it’s a difficult psychological trap to avoid. Our brains are wired to think in terms of single point estimates, and it’s difficult to keep in mind the concept of variation about the mean, although that can often be considerable.

      None of this means that we shouldn’t be trying to improve productivity in healthcare. Of course we should always be trying to make things better. But it does suggest that the productivity bottlenecks that healthcare experiences are not unique, nor are they even exceptionally bad.

      • I provided a lot more than a mean. But, more to the point, I’m writing briefly about the content of the work of others. I extracted some graphs that present the data they reported. Perhaps there is more detail on productivity figures in their papers, but I don’t recall (I read them long ago). You can certainly follow the links!

    • @ afrakt. This is a very great point that you make. Critics are so overwhelmed at the lack of efficiency in health care distribution that they overlook it’s reason for inception; better care. While technological advancements in health care are being achieved, provided care IS improving. The only downfall here is whether or not the beneficiary has the financial flexibility to afford this type of service. This is where the lack of efficiency is spawned. I believe that better provided care will now be accessible to the masses, bridging the gap of inefficiency. If more people can receive better care, there in turn will be a lower rate of sicker individuals seeking care, making room for those who need care. Am I wrong in assuming this? I had written an article responding to the nearly $4.1 billion recovered by the HHS fraud prevention and enforcement program in the 2011 fiscal year. Individuals and companies defrauding seniors and taxpayers were trying to seek payment for services they were not entitled to. This being said, even with the advances we are making in technology, ( hence improvement in care) can we really expect physicians to “go the extra mile?” Productivity is, and should be, a culmination of quality, first, then efficiency. If our current administration is cracking down on fraudulent practices, one can only assume that physicians and companies would be more likely to comply for fear of penalties or prosecution. With that in mind, shouldn’t productivity increase? Just a thought….

    • Thank goodness for low productivity in health care!

      That is what is propping up the American economy, Per Michael Mandel and Michael Spence, among others, health care has provided the majority of new jobs for the past ten to twenty years!

      As men have lost jobs due to rising productivity in factories, at least their wives have found jobs in unproductive hospitals. That is propping up the middle class.

      Go online any day to a job website like Indeed.com.

      In many if not most geographic areas, You will see fast food jobs paying nothing……and health care jobs with very good pay…..and very little else!

      The low productivity of health care means that our hospitals have more employees per bed than virtually any other country. It means that each doctor’s office employs extra people for billing and records. and so on!

      One could in theory make health care productive by automation, offshoring, medical tourism, and vicious price competition. Huge hospitals that are the major employer in many cities could be closed down in two years if we stopped overpaying them for intensive care and for outpatient procedures.

      The focused hospitals and ambulatory surgery centers that survived would be far more productive…….

      and unemployment in the US would surge upwards.

      I guess what I am saying is that the cure for low productivity may be worse than the disease.

      Another way to phrase this is that increases in productivity can have huge human costs. American agriculture had marvelous increases in productivity in the mid 20th century, especially in the South.
      The result was to some extent the Great Depression and the large black ghettoes of the North as unproductive farm workers were forced to move. (see Arnold Kling and others)

      As the old saying goes, be careful what you wish for………

      Bob Hertz, The Health Care Crusade

    • “So the productivity losses are exaggerated. By how much? I don’t know.”

      It may not come up in the book but I can’t help but I can’t help but wonder what the disaggregated data would look like, and whether there wouldn’t be areas where productivity has remained essentially flat for dedades, like catheterization, vs procudures which are dramatically less invasive and require far less in-hospital secondary care that they used to (ACL repairs, etc).

      I also can’t help but wonder how one contends with the measurement problems arise when evaluating the cost-performance of clinical interventions that effectively address conditions that were previously untreatable, technologies that are so superior to their antecedents that they hardly belong in the same category (LASIK vs glasses, etc), or drugs like treat infections/diseases that were previously lethal.

      I have a superficial familiarity with Baumol’s central hypothesis as applied to the arts and higher ed, and was underwhelmed, as the central argument seemed to hinge upon measuring top-line costs over-time, without much thought as to what the underlying composition of the costs was.

      When my Dad went to college it was dramatically cheaper in real terms, but there was no such thing as an IT department, the faculty-to-admin ratio was dramatically higher, living conditions were comparatively spartan, and state-subsidies as a percentage of tuition were higher, legacy costs were lower, etc.

      To recapitulate your point, today’s universities aren’t like the universities of 50 years ago. It’ll be interesting to learn more about how Baumol addresses these problems arising from the changing composition of the measured phenomena over time in a full-length book.