• Physician behavior and demand inducement

    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.

    What’s the central issue in health economics? Tom McGuire’s answer in the first sentence of his Physician Agency (ungated pdf) [1] is that it is physician behavior. Though McGuire covers many related topics, one of the main areas of exploration in his paper is physician (or supplier) induced demand (SID).

    Historically, within health economics SID has been a contentious and controversial topic. Perhaps that’s because it rejects the canonical independence of supply and demand schedules and because it has been associated with possibly pejorative notions of wasteful over-provision of marginally beneficial (or nonbeneficial) care. I blogged on these aspects of SID yesterday.

    The consternation over SID may be less common outside of health economics. McGuire writes that,

    The literature outside of economics is without soul searching about whether physicians influence demand. It is nearly universally considered obvious that of course they do. The concern in this literature is usually with identifying the factors, such as socioeconomic status of patients, that lead physicians to direct patients to different courses of therapy.

    Within health economics the evidence cited by McGuire at the time of his paper (year 2000) was somewhat mixed, but he concludes that,

    Adding up the evidence, on obstetricians doing more C-sections, surgeons doing more bypass operations, physicians referring more frequently to their own labs, and other studies, makes a convincing case that doctors can influence quantity and sometimes do so for their own purposes.

    What’s meant by “for their own purposes”? Many point to financial incentives. McGuire quotes Mark Pauly [2],

    “Other things equal, physicians would rather tell the truth, but they would be willing to surrender some accuracy for some amount of money income.” Once that tradeoff is admitted, it is hard to avoid the conclusion that the physician will be inducing some demand.

    Though I don’t think it is necessary to imply physicians are lying to their patients, recent evidence also supports the very conclusion McGuire reached, that physicians induce some demand, consistent with their financial interests. You can click back to my post about a randomized experiment that reaches that conclusion and links to other work. More recently, Christel van Dijk and colleagues found that in the Netherlands,

    introduction of fee-for-service for socially insured consumers led to a higher increase in physician-initiated utilisation. […] Differences in the trend in physician-initiated utilisation point to an effect of supplier-induced demand.

    (Their paper also includes an up-to-date literature review of SID.)

    However, physicians might induce demand even in the absence of financial incentives to do so. This can occur when patients are not fully informed and defer to their physicians for treatment decisions. John Wennberg describes the stereotypical physician-patient model in Tracking Medicine (highly recommended, by the way),

    The patient, as a layman, does not know what he or she truly needs; it is the physician who knows the nature of the patient’s illness and can select the right treatment. For these reasons, many social scientists thought it was rational for patients to do something they would not dream of doing in most markets—that is, to delegate decision making to the seller of services, the physician, who by virtue of his special knowledge and skill, could act as their “rational agent” in health care purchasing decisions.

    This, as Wennberg goes on to illustrate, is completely wrong. Evidence shows that physicians cannot act as perfect agents for patients. This comports with common sense. The nature and brevity of typical physician-patient interactions suggests that it is unlikely the physician can divine the quantity and type of care the patient would elect if fully informed.

    In large part, concern over physician behavior and what it means for Medicare spending is what gives rise to the doc fix debate. There a key issue is that, despite relatively modest increases in Medicare fees for doctors, per beneficiary physician spending by the program has gone up dramatically, as illustrated in the following figure by Uwe Reinhardt. (More about it here.)

    It’s not a big leap to hypothesize that something about physician behavior is altering utilization patterns and driving spending up at a rate well beyond that of unit prices. A counter-hypothesis is that patients are to blame. Though (natural, un-induced) patient demand may play a role, in this population it is not credible to me that it is the only or even the most important factor.

    Naturally, problems with Medicare spending motivate the search for payment systems that provide incentives for more efficient provision of care, as articulated recently on the Health Affairs blog by Michael Chernew, Darius Lakdawalla, and Dana Goldman. This is not unique to public programs. Private insurers are also seeking ways to shape provider behavior through innovations in contracting and financial incentives.

    Whether or not you agree with McGuire that physician behavior is the central issue in health economics, it is central to health policy and health insurance.


    [1] McGuire, T.G. (2000) “Physician Agency,” Chapter 9 in Handbook of Health Economics, Eds. A.J. Culyer and J.P. Newhouse, Amsterdam: North-Holland.

    [2] Pauly, M.V. (1980), Doctors and Their Workshops: Economic Models of Physician Behavior (University of Chicago Press, Chicago).


    • Interesting stuff from Austin with the implication that FFS needs to be at the very least changed to counter these issues with physician behavior. I have three questions though:

      1. In my experience, ER’s are the biggest single contributor to unnecessary testing, overutilization, etc. (not aware of any peer-reviewed literature outside of the radiology journals). Their physicians and other providers are overwhelmingly salaried yet indulge in this. Liability is the obvious culprit and even if that’s totally incorrect (my attempt to avoid the morass of tort reform discussions!), this has obvious implications about the efficacy of any payment reform model-based solution given that there are seemingly multiple stable equilibria that result in overutilization.

      2. Fee for service is very very simple. That might seem trite but I think has big implications as when we switch from the model that most developed nations (no matter the health provisioning system) use, we invite a lot of complexity into a system with strong incentives to discriminate against the sickest members of society. I can foresee a situation where increasing regulations attempting to counter the skewed incentives eventually fails or is gamed…

      3. Patient opposition to a system where their provider is incentivized to deny care has always been extremely strong. The political incentives to dismantle such a system are compelling. Even with the strong research supporting its need, I think we have a responsibility to consider the political economy here of this move…

    • Just realized I didn’t put the above in question form. Anyway, would like to see what people think in any case about the 3 points…

      • 1) A lot of ERs have a productivity clause in their salary for their docs. There is often subtle, and not so subtle, pressure to use more studies that reimburse better for the hospital.

        2&3) As Austin has noted, whatever reforms we embrace must be politically acceptable. There is major fear of explicit rationing, especially by the government. Those fears can be, and have been, used to derail reform. Just one of the hurdles.


    • I would broaden this beyond physician behavior. For the last 15 years, at least, hospitals have worked with a revenue centered model. They figure out where they can make the most money, then emphasize that kind of care. Hence, all of the total joint programs. They often induce demand just by making it very easy for patients to have the procedure done, or offering hotel-like accommodations. I remember the guy who sent out a van to pick up his cataract patients to bring them to his facility.


      • Fair points that I agree with about hospital responsibilities being ignored.

        In relation to the ER argument though, I am not clear the impact is that simple. Productivity clauses from hospital contracts do typically prioritize throughput (which would increase overall costs) and thus, revenue, but not the type of tests done. Since when these docs order imaging, they can quickly dispose of a patient in a legally safe manner, this indirectly helps them financially but also on every other potential metric (e.g., clinical satisfaction as patients see something done, workload reduction, wait time reductions, liability safety, etc.). Undoing the relative contributions is difficult.

        The fears about 2) and 3) are to some extent justified though, right? An interesting question is whether it makes more sense to use the argument that these fears are groundless or that we should argue that the trade-offs are worth it…

        • I would prefer that we directly confront the issue of rationing, but that is just my management style. Not sure it is good politics.


    • The rising per beneficiary physician costs comment appears to focus on either the doctor or the patient demanding more treatment. I am curious whether there is also an effect of pricing for new services. It would be plausible that as each new treatment arrives, it gets priced while it is fairly new and that the price remains relatively sticky. That seems like it could (help) explain rising costs per physician based on a preference for new treatments independent of a preference for more treatments.

    • If the notion of increasing patient demand is “not credible”, then pharmaceutical firms are wasting a LOT of money on advertizing! I don’t see how you can ignore the fact that new technology and treatment improvements drive patient demand. A hundred years ago the doctor could often do nothing but hold your hand while you suffered and died. Now they can even treat your restless leg syndrome.

    • Will the increase in the number of patients per physician if ACA is fully implemented in 2014 reduce the incentive to do more procedures per patient?

      Physician investments in labs and imaging services creates financial incentive to refer there, but there is evidence from social psychology that making those investments may strengthen physicians’ belief in the value of those services.

    • -If you actually want to understand the extent to which “Supplier Induced Demand” is driving spending, then the Reinhardt chart is less useful than no chart since it isn’t adjusted for inflation, and assumes that a (1) constant composition of services being delivered/paid for over the interval being measured.

      The number of dollars being spent on physician per beneficiary is increasing over time. Hmmm – how much of that is driven by inflation (easy adjustment that should have been incorporated into the chart)? How much of the change is driven by changes in technology, changes in what defines the standard of care? If physicians are charging more for the same services – how much of that is driven by provider aggregation? When it’s all said and done, how many of the additional dollars per beneficiary can SID actually account for?