Unnecessary Care Explained
The biennial Dartmouth Atlas of Health Care is out and the findings are as eye-popping this year as they have been in the past. Among chronically ill patients in the last two years of life:
- New Jersey patients spent almost three times as many days in the hospital as patients in Utah.
- Patients in Manhattan had 3½ times as many hospital days as patients in Bend, Oregon.
- Among teaching hospitals, the variation in the amount spent was more than four to one.
So what impact did this wide variation in care have on the health of patients? Not a whit.
- There is no evidence that extra care and extra spending produce better outcomes, and some evidence that they produce worse outcomes.
- Further, variations in care correlate with variations in supply: the more hospital beds, the more bed days; the more CT scanners, the more scans; the more cardiologists, the more cardiac care, etc. [See Associated Press article ]
Is this the whole story? I'll shelve that question for another day. For the moment, what do we make of all this?
Before anyone launches into another bash-the-American-health-care-system tirade, be forewarned. Researchers have found wide variations in medical practice from doctor to doctor, town to town, region to region – all over the world. The type of health care system doesn't seem to matter very much.
So what's going on? I'll start with the conventional theory. Then add four of my own.
1. The Venal-Mercenary-Doctor Theory. The conventional theory is "supply induced demand." Providers convince patients to want whatever health resources are available. Why would they do that? Fee-for-service payment is often blamed. Providers make more money if they can convince patients to demand unnecessary care.
Anyone who has spent much time around doctors will recognize this theory as mainly poppycock.
2. The Doctor-as-Altruistic-Agent-of-the-Patient Theory. In case anyone hasn't noticed, we have made health care free (or almost free) to patients all over the developed world. Now if you were a physician advising patients and the perceived cost of virtually every resource was zero, what advice would you give? It's a no-brainer: Use every resource in sight as long as there is some marginal benefit, no matter how small. With willing patients, the only thing restraining usage will be limits on supply and the occasional (managed care) limit on demand.
3. The Patients-Acting-as-Their-Own-Doctors Theory. Who needs doctors to induce demand? Any curious soul armed with a computer can find all kinds of ways to spend (other people's) health care dollars for his or her own benefit. In fact, 113 million people have searched the Internet for health information. [link]
4. The Misguided-Donor/Misguided-Legislator Theory. Why do resources differ from place to place? It usually comes down to arbitrary decisions by donors and politicians (far removed from real patients). Suppose a donor pays for an MRI scanner and also pays for the technical staff and the radiologists. Treating these costs as fixed, the marginal cost of a scan is almost zero. So it makes sense to run the scanner around the clock. If a second scanner becomes available, it makes sense to run that one around the clock as well. Ditto for the third. And the fourth. You can do an awful lot of scanning before the marginal benefit gets all the way to zero.
5. The Medicine-As-Art-Rather-Than-Science Theory. Imagine three ways of approaching your job. First, as an artist – relying on experience and judgment but also wafting wherever the spirit moves you, with no bad consequences. Second, as a business manager – being forced to compare profit and loss on every decision with bankruptcy always a threat. Third, as a scientist – being constantly under the scrutiny of your colleagues, with professional reputation hanging in the balance.
Our system (and, indeed, the health systems of all developed countries) gives doctors the freedom to choose approach number one.
Years ago, a RAND study found widespread variation of medical practice and concluded that one-third of all care was unnecessary. Yet as explained in Patient Power and again in Lives at Risk [hot link], even among RAND's expert panels there were wide differences of opinion. Absent economic and scientific pressure, conformity of treatment modes is unlikely to ever occur in medicine.
Bottom Line. I believe theories 2, 3, 4 and 5 explain 80% to 90% of everything we observe, while the conventional explanation (theory 1) explains a small part of the remainder. Wide variation in the use of health care resources is not a surprise. It is the natural and inevitable consequence of a system in which normal market forces have been systematically suppressed and nobody ever faces a real price for anything.
Have a great day.
John
Bravo…once again.
Great Health Alert on Dartmouth claims.
Have you seen the Doyle paper “Returns to Local Area Health Spending”? I’ve attached a copy. He presents evidence that higher local spending does lead to better outcomes. Which I find interesting as I’ve always been somewhat skeptical of using highly aggregated results to pronounce on the intensity or adequacy of care given the issues of measurement error and self-selection bias.
I love this epistle, because it is the clearest illustration yet of what I call “Faith Based Health Policy Analysis.” In fact, you courteously flag your analysis as faith based in your section Bottom Line:
“I believe theories 2, 3, 4 and 5 explain 80% to 90% of everything we observe, while the conventional explanation (theory 1) explains a small part of the remainder.” (Underlining added.)
You dismiss the venal doctor theory (1) out of hand, without citing any evidence on why we should dismiss it out of hand. I find that sycophantic to the medical profession.
Next, you subscribe to the “Doctor-as-Altruistic-Agent-of-the-Patient Theory.” It could explain, of course, why doctors would recommend treatments that are free at point of service to patients up to the point at which the expected marginal benefit from added resource use approaches zero. That much my own belief (and a belief it is) in economic theory finds intuitively appealing. But how can that theory explain the Wennberg variations with which you open your epistle? Since presumably health care is free at point of service in all regions under Medicare (although actually there is a fair amount of cost sharing in Medicare, even for hospital episodes), why do altruistic doctor-agents prescribe so many fewer hospital admissions and days in Utah than in, say, heavy resource-using Texas or New Jersey? Even within New Jersey the use of hospital care during the last two years of a Medicare patient’s life varies by a factor of up to 3 across NJ hospital market areas. (As the Chair of the recently concluded New Jersey Commission on Rationalizing Health Care Resources, I was told by numerous hospital executives and by the Commissioner of Health, himself a doctor who has practiced in NJ for decades, that Theory 1 actually furnishes a major explanation for these variations. They even informally identified types of doctors more likely to be driven by economic motives in their treatment decisions. I am not sure whether you are prepared to have your insight as an economist override those of actual practitioners in the field; perhaps you are.)
As to the “Patients-Acting-as-their-own-Doctors Theory,” I am not sure what to make of your views on it. Isn’t that what the CDHC people want patients to do? Is the problem, then, that these patients don’t pay enough out of pocket? I recall your telling me and the audience in DC a few years ago that you were NOT a champion of high deductibles, but merely a champion of better information for consumers. (I had asserted that you were a champion of high deductibles). Are you now in favor of high deductibles, but were not then? And do you really think many Americans will have themselves admitted to a hospital for painful surgery just because it is free—because it tingles so nicely? Finally, the impact of cost sharing on patients’ decision surely must vary by income class. We learned that much in graduate school. Do you sincerely think that people in your and my income class would be deterred from using whatever health care we fancy just because we face, say, a $5,000 deductible—peanuts to you and me?
Next, your “Misguided-Donor/Misguided-Legislator Theory.” You open up that section with “Why do resources differ from place to place? [sic].” Are donors in Utah genetically different from donors in Texas and NJ, which can explain regional differences in resource use? Do federal legislators have different Medicare rules for Utah than they do for Texas and NJ, and that can explain differences in regional resource use? Am I missing something here, or might I be staring at a classic non-sequitur?
Finally, in connection with 5. The Medicine-As-Art-Rather-Than-Science Theory, does it explain regional differences in resource use?
My own reaction to the Wennberg variations is that they most likely represent the confluence of all of the theories you list, in different regional mixtures whose nature we do not yet understand because we have never devoted research to it. I find it odd that, as an economist, you would axiomatically, as an act of pure faith, rule out Theory 1. Doing more operations and MRI scans is highly profitable to the providers of these services. And yet, as an economist, you believe that these profits play only a minor role in treatment decisions?
You may wish to get a copy of the book Overtreated: Why Too Much Medicine is Making Us Sicker and Poorer. It is by Shannon Brownlee and it is a provocative read. The author highlights the Dartmouth folks' work in the opening chapter and profiles the lone doctor who basically started this field of study from scratch. I tend to come down more on the fragmented nature of medicine and think the best overview of this fragmentation was in the recent CED reports discussed here: http://www.ced.org/ What I would find most useful is how Medicare and Medicaid policy planners should use the Dartmouth insights as far as developing policy. One person at the GAO summed it up beautifully when looking at the difference between expenditures of the Mayo Clinic and UCLA: "Why should the best medical care in the world cost twice as much as the best medical care in the world?" Good question. I look forward to the work by the NCPA on policy options for Medicaid that use the lessons unearthed by Dartmouth's researchers.
As usual, your remarks are right on target! I am sending them to all the cardiac and thoracic surgeons I know, which is a large number.
Keep up the inspiring and informative messages. It is a blessing to have your talent and experience working on all of these important issues!
Superb analysis — congrats.
Please clue in our wonderful Presidential candidates!
In 1975, I was an Economics major, a senior at Wellesley College doing my honors thesis. I won’t bore you with details but the gist goes like this.
I asked the question: “what determines the cost of health care in any demographic hospital community?” I researched the country and looked at hundreds of variables. The answer….
The number of doctors on staff at the hospitals was the sole variable involved in determining the cost of care in that hospital community.
I went on to suggest the following: 1) the doctor (at that time) was both the demander and supplier of care (therefore Keynesian principles did not apply); 2) the hospital was the “capital playground” for doctors. They had no personal investment on the line, but they got to “play” and charge for it; 3) in order to contain cost we needed to better define the product “health care” and in order to do that we needed a relationship clinical data base.
So in 1975, I proposed the computerization of the medical record (the clinical record) as the means to controlling cost of care.
Prophetic??? I am not so sure, but now you know why I am so involved in these economic and political issues. I wrote the initial study when I was young and naive.
Since then and since I have earned my medical degree, I have the same modified conclusion. When you look at the Dartmouth study and all the studies asking why demographic communities are SO different I would add the following: 1) the communities are different. There are states in the north where LOS in a hospital are different in the winter because of transportation problems. Communities are different because they are genetically different. California rates of diabetes are bound to be greater than Maine because we have a Mexican population (you know that data). Communities are culturally different and there are different expectations. 2) Physician training is different. Doctors tend to stay put or to live where they are trained and the fact is that some communities have stronger cardiac programs and that changes treatment standards.
So this is my 2 cents.
PS: I was not at Wellesley when Hillary was there but my Economics professor says the following with respect to Hillary: “History has never been so beautifully re-written.”
I think you are letting us off too easy. Doctors do play a role in inducing demand and the variance explained by physician behavior is likely to be similar to the effect size explained by the other models. However, most physicians don’t do it out of raw greed. We are trained to be complete and comprehensive in our investigation and so medicine is practiced in an inefficient manner. Furthermore, like all people, we are attracted to shiny new toys, even when the tried and true still work. Plus the reimbursement is better with the new toys.
Until recently, there was little discussion of quality, especially in routine care; and even today there is no teaching about value (quality at a reasonable dollar cost) in medical school, residency, or CME.
Finally, don’t be fooled into thinking that doctors don’t do things for higher reimbursement. The difference is that most doctors aren’t businessmen, in the sense that they aren’t looking for maximal profits. But we do have a general sense of how much we think we should be paid, and will induce consumption until that payment level is reached.
This issue of value needs to be explored further. If patients and doctors had a better sense of value and concordance on what value means, then the issues of regional variation in costs (and quality) would improve. This is the part about market economics that seems to get lost. Without transparency over value and its metric (price), there really is no market economics. Your archetypal example of markets for Lasik surgery is right on par. We need better methods of communicating quality, value, and price at the doctor-patient interaction level. But none of the stakeholders (insurers, governments, employers, etc.) seem very interested in empowering that process.
Aanand Naik, MD