• Supplier-induced demand vs. supply-sensitive care

    Last week Thom Walsh explained some differences between supplier-induced demand and supply-induced demand (aka, supply sensitive care), terms often confused with one another. In this post I cover what I think is the crucial distinction.

    The first few pages of Victor Fuchs’ The Supply of Surgeons and the Demand for Operations (The Journal of Human Resources, 1978) includes a nice, graphical treatment of the concept of supplier-induced demand (SID). It’s ungated, so there’s no barrier to you reading it for yourself. Below I reproduce the main points anyway. Then I’ll contrast SID with supply-sensitive care.

    SID is a deviation from standard economic theory. Normally, we think of supply and demand as independently determined. However, SID theory is that physicians (the suppliers) can induce a shift in demand by patients. (Fuchs calls it the “demand-shifting hypothesis.”) The distinction between standard and SID theory is easily seen in the following two charts from the paper, labeled Figure 1A and Figure 1B.

    Figure 1A illustrates the standard theory: a shift in supply (from S1 to S2 or vice versa) has no effect on the demand schedule (D). The market equilibrium changes, as indicated by the different points of intersection between D and S1 or S2, but in both cases it’s the same demand curve that’s relevant. Consumers’ quantity-price demand relationship has not changed. As this figure illustrates, all care is supply sensitive to some extent (greater supply leads to greater quantity, Q). The degree to which it is depends on the slope of the demand curve (or the elasticity of demand). Some services are more supply sensitive than others for this reason.

    Figure 1B illustrates SID theory. If the supply schedule shifts from S1 to S2, suppliers induce a corresponding shift in the demand curve from D1 to D2. Somehow suppliers have caused consumers to demand a larger quantity at every price. (Though Fuchs doesn’t mention it, another way physicians might induce demand is to change the slope of the demand curve.)

    Figure 2, below, provides another way to consider SID. MC is the marginal cost of care, assumed constant for all quantities. MB is the schedule of marginal benefit. As quantity increases, marginal benefit decreases. The next unit of care is not as valuable as the previous one.

    One question is, who chooses the actual quantity of care delivered, Q? Fuchs points out that even if the provider chooses Q, that is not itself evidence of SID. The physician, acting as an agent on behalf of the patient, may select the same Q that a fully informed patient would himself. The more interesting case is when the Q selected by the physician is not what a fully informed patient would choose.

    If the physician/population is relatively high in an area (for reasons unrelated to demand), they may push quantity to the right of Q in order to keep prices and incomes from falling drastically [as in the shift from D1 to D2 in Figure 1B]. If there are relatively few physicians in an area, and if they cannot or do not raise price to an equilibrium level, they may push quantity to the left of Q in order to avoid excessive work. This latter situation, sometimes characterized as “excess demand,” has been offered as an explanation for the observed correlation between supply and utilization. It would be described in Figure 1A by a price that is below the intersection of S1 and demand. A shift of supply to the right results in higher utilization because it takes care of some of the excess demand.

    Note that the presence of demand-shifting should not be equated with “unnecessary care.” If “necessary care” is defined as Q in Figure 2, demand-shifting to the left implies that some patients are not getting the care they should, and does not imply that any patients are getting unnecessary care. Moreover, necessary care may be defined differently than the quantity that maximizes the patient’s utility (i.e., Q). If, for instance, it is defined as the quantity that maximizes the patient’s health regardless of cost, the optimum would clearly be to the right of Q, and such demand-shifting would not necessarily imply “unnecessary care.”

    The severing of SID from “necessary” and “unnecessary” care is important. Though most of us could point to care that fits those descriptions (even if we disagree on which specific types of care do), SID need not correspond. It’s my hypothesis that sensitivity to the possibility of insulting physicians with an implication of provision of unnecessary care (or withholding necessary care) has led many authors to back away from clear language about SID. That’s a shame.

    The term “supply-sensitive care” (SSC) has been suggested to me as a less pejorative alternative to SID. But I don’t think it’s synonymous. (I covered supply-sensitive care last week.) To see this, recall that SSC is care delivered at a volume that responds to availability of provider supply and in ways that cannot be explained by other factors like the health of the patient population. That’s consistent with Figure 1A, for which SID does not apply. SID is different. It’s a shift in demand induced by suppliers (Figure 1B). It’s not merely a shift in supply that leads to greater utilization.

    Both SSC and SID might be associated with over- or under-provision of care, something that is implied by the geographic variations literature. However, that observation does not itself define what the “right” rate of care should be.

    Why might SID occur? I’ll address that question tomorrow.

    @afrakt

    Share
    Comments closed
     
    • Two questions:

      1) Conceptually,”elasticity” of demand is operational when considering SSC (as opposed to SID)

      2) You are addressing the “why” tomorrow, but the issue I am interested in is impact. Papers have minimized SID effect when seeking causes for care variation–not as potent as we once believed, ie, contributes “10% of variation effect, etc”

      Brad

      • 1) Elasticity pertains to the slope of the demand curve. Formally it’s the (% change in Q) / (% change in P). Thus, it’s always relevant. Just look at the charts. Imagine how things look with a demand curve of a different slope. Draw it if need be.

        2) I’d have to do far more work than I have to see if there is anything like that info in the lit. Don’t hold your breath. However, I will ask around to see if there is a short cut. 🙂

        • Elasticity from perspective of patient (Q), if SID present…

          Better way of saying, is yes, elasticity present, but patient does not impact, ie, they are “inelastic.”

          Correct? And that assumes 100% SID.

    • Does it really matter if S shifts alone (your SSC example) or S and D shift together (your SID example)? What really matters is whether the market adjustment occurs because of rent-seeking or proft-seeking activities. The former is bad because the size of the economic pie shrinks for all (inefficiency–benefits of the additional medicine less than costs) whereas the latter is good because the economic pie expands (efficiency=benefits outweigh costs).

      • Rex,

        McGuire has a nice paragraph discussing why SID matters in his Physician Agency paper I will cite tomorrow. See 2nd full paragraph on page 519.

        Fundamentally, if a “true” demand curve exists, shouldn’t we want the market to function according to it? Under SID, it doesn’t. Under SSC, it does. As I wrote, the quantity response to SSC may be perfectly reasonable, rational, and good. Though, in practice it may not be in all cases. There is such a thing as too little supply, after all.

        • You are misinterpreting my comment. It’s the motivation (rent vs. profit seeking) behind the actions of health care providers that matters for efficiency and not whether they shift supply or demand.

          • OK. Your opening question didn’t give that impression. You seemed to be asking if shifting S vs. shifting S+D should matter. I now think you’re more in the ballpark of the idea that it is OK for S and/or D to shift for some reasons but not others. You seem to be saying we should not care about SID if it is motivated by profit seeking, right? I find that hard to swallow. If the shifted D is not the demand curve that would obtain by fully informed patients, then we have a market failure (information asymmetry). Why should I be happy about that?

            UPDATED: I had a bad typo in my 4th sentence. Fixed!

            • I am allowing for the real-world situation where informed health care providers may correct markets that would otherwise fail because of uninformed consumers who may undervalue medical care.

        • Regarding Austin’s comment on “too little supply.” This is indeed a concern, particularly when it relates to the cost of producing care. The Minimal Efficient Scale (or MES) of hospitals is an older line of research but still relevant. Try this for starters – http://www.ftc.gov/be/workpapers/wp203.pdf

          MES concerns the cost of production. SSC concerns expenditures. I’ll be posting about costs and spending on The-141 tomorrow.

          Thom

    • A busy surgeon looks at a lump and says, “it is probably a lipoma. We can just watch it”. A non-busy surgeon says, “It is probably a lipoma, but it might be something else. Why dont we just take it out to be sure”. It just is not that hard to induce demand, if you want to do so. There are relatively few bright lines in medicine. There is a fair amount of leeway on what to do or not do for much of procedural medicine.

      What is not clear to me, as alluded to by Brad, is how big a factor it is, and how much is done explicitly to raise income, how much is just practice style (many institutions are more aggressive than others) and how much is done just to keep busy.

      Last of all, this is all complicated, I believe, by the observation that the physicians who are most entrepreneurial will usually be the ones most interested in maximizing income. All those new surgicenters are being started with an eye towards making money by most of the docs starting them.

      Steve

      • I have it on good authority that nobody knows the extent to which SID explains geographic variation. Maybe if I read deeply into some papers and thought long and hard I could figure out a back-of-the-envelope guesstimate of some bounds. Don’t hold your breath. What I will do is keep watching the literature. Maybe something will turn up.

        • Not really something I see covered much, but there are also issues, as alluded to above I think, when demand is unable to move supply. Wait times for an appointment with a developmental pediatrician in our state are on the order of months. It has been that way for years.

          Steve

    • I’m not sure the distinction between supplier-induced demand (SID) and supply-sensitive care (SSC) really adds much to previously existing discussions. But there are as-yet under-tested hypotheses about these issues.

      The Dartmouth shop has repeatedly demonstrated a correlation between the density of service availability (such as physicians or hospital beds) per capita and the utilization of those services per capita. Since the 70’s, Jack Wennberg described the phenomena as small area variations (SAV). (I had not seen the Glover cite; thank you.)

      As Austin mentioned in an earlier post [http://theincidentaleconomist.com/wordpress/supply-sensitive-care/], some of their more recent work has distinguished between those types of health care with large (from those with small or no) correlations between the densities of suppliers and utilization rates. The 1998 Dartmouth Atlas (and perhaps those more recently; I just referenced some old notes) tabulated the ratio of hospitalizations with hospital referral areas to average US hospitalization rates for various common conditions. Conditions where hospital capacity had no effect, such as hospitalizations for hip fracture, were considered to be supply insensitive. Conditions with significant correlations between capacity and utilization, such as hospitalization for congestive heart failure, were considered to be supply sensitive medical conditions. While aggregating up to All Medical Discharges doesn’t break any rules, it misses the dynamics of what’s happening.

      While there’s a good deal of literature that attempts to identify how much of our rising expenditures may be attributed to SID or to find any rationality for SAV, all of the literature that I know about presumes that healthcare’s only “product or service” is better health. In this framework, any expenditure that can’t be expected to improve a patient’s health gets demoted as “unnecessary.”

      But as Austin points out, more care is not necessarily all bad. Indeed, I’ve suggested that some of the rising health care expenditures can be justified by an income-related increased demand for reduced uncertainty, hope, and amenities in health care. [http://onlinelibrary.wiley.com/doi/10.1002/hec.1765/abstract] [and mentioned in TIE at http://theincidentaleconomist.com/wordpress/most-important-chart-in-health-policy/%5D

      To my mind, the interesting question is how much of the care above and beyond that-which-can-be-correlated-with-improved health can be correlated with satisfying a demand for risk aversion, hope, and amenities in health care. Such questions were not included in the study by Amber Barnato and colleagues [Are Regional Variations in End-of-Life Care Intensity Explained by Patient Preferences?]

      Moreover, there’s an easy explanation for the unequal distribution of health care services. If I am a newly minted physician trying to decide where to locate, and I have a choice between Minneapolis where I believe stoicism creates an inelastic demand and Miami where I believe an additional demand for hope, reduced uncertainty and amenities in health care creates a greater own-price demand elasticity, my rational choice, ceteris paribus, is going to be Miami.

    • I don’t think that supply-sensitive care is intended to supplant the concept of SID, or to soften it in any way. It is fundamentally a different process.

      My understanding of SID is that it implies that the physician induces the patient to undergo (in effect, purchase) a procedure the patient otherwise would not have demanded. The usual motive for this inducement is greed, but in the physician’s case there may also be other motives. Mark Chassin called it the “enthusiasm hypothesis.”

      http://www.ncbi.nlm.nih.gov/pubmed?term=Chassin%20enthusiasm

      We know that doctors, particularly surgeons, are more or less enthusiastic about the procedures they can perform, and more or less likely to recommend them for their patients. This is a pretty good explanation for the variation in rates of procedures like angioplasty, knee replacement, back surgery, mastectomy, prostatectomy, gall bladder surgery etc. Doctors in different communities have different opinions about the value of surgical treatments and are more or less likely to recommend them. Is that supplier induced demand? Seems like it to me, but I’m not an economist.

      What’s interesting here is the supply of surgeons doesn’t correlate very well with rate of surgery, except at the grossest level: when you measure the rate of all surgeries it correlates with the supply of all surgeons. But the rate of knee surgery does not correlate to the supply of orthopedists and the rate of gall bladder surgery doesn’t correlate strongly with the supply of GI or general surgeons.

      The mechanism for supply sensitive care is different. It doesn’t have much to do with decisions around surgical procedures, and it doesn’t have much to do with the individual physician’s enthusiasm for this or that procedure. It involves discretionary decisions that are affected by the supply of resources. For example, the decision about whether or not to put a diabetic patient in the ICU or treat him in the nursing home. This is fundamentally different from the elective decision around having knee replacement surgery.

      But you’ll probably get into all of this tomorrow!

    • Robert, Shannon,

      You might like to see my post from last week on differences between supplier induced demand and supply induced demand, referenced in Austin’s opening paragraph –

      http://the-141.blogspot.com/2012/03/supply-induced-or-supplier-induced.html

      There is actually quite a long history of economic research on induced demand in health care. Not surprising, it has it’s own controversies. See Health Economics 4th Ed, by C. Phelps. pgs. 225-238.

      Thom

    • The “Why might SID occur?” will definitely provide us with the pertinent reasons for geographic variations and other disparities in utilization. I look forward to reading your take on this.

    • I’m interested in the ultization of surgical supplies related to physician preference. Could the term “supplier-induced demand” be used to describe other areas, or is it specific to “over” prescription of medical care?

      I see that surgeons rely heavily on product vendors to provide information and education about new surgical supplies. Where vendors are able to build a relationships with surgeons, supply use increases. In this case the supplier is the product vendor, and the demand comes from surgeons.

      Is there term to discribe this?

      • Elizabeth,

        Supplier induced demand is not confined to healthcare. See car-repair and plumbers, or search ‘induced demand’ in econ text books. Also consider TV and print ads by pharma companies – “Ask your doctor about our new medication.”

        I’m not familiar with the literature on medical device manufacturers. Try using the term ‘induced demand’ in your literature searches.

        Good luck.

        Thom