What explains regional variation in health care spending?

It doesn’t seem to be demand side factors, but rather what doctors believe, including false beliefs.  That is scary.  There is a new NBER paper by David Cutler, Jonathan Skinner, Ariel Dora Stern, and David Wennberg and the abstract is this:

There is considerable controversy about the causes of regional variations in healthcare expenditures. We use vignettes from patient and physician surveys, linked to Medicare expenditures at the level of the Hospital Referral Region, to test whether patient demand-side factors, or physician supply-side factors, explains regional variations in Medicare spending. We find patient demand is relatively unimportant in explaining variations. Physician organizational factors (such as peer effects) matter, but the single most important factor is physician beliefs about treatment: 36 percent of end-of-life spending, and 17 percent of U.S. health care spending, are associated with physician beliefs unsupported by clinical evidence.

There is an earlier ungated version here (pdf).


More unsurprising news on healthcare....

Seriously, wasn't this stuff well-known 5-10 years ago? Perhaps it was just suspected....

I might have believed the surface narrative more back then. Are we to seriously believe that 17% of medical spending (3 some odd percent of GDP) is simply because doctors haven't read Cutler et. al.?

"because doctors haven’t read Cutler et. al.?"

They often haven't even read the relevant cost-effectiveness literature in their own specialty area.

There is very little cost effectiveness research for most specialties. Who would pay for it? The drug companies? LOL. It pretty much needs to be govt financed, and half of our govt opposes the research. (Guess which half.)

I think Cutler et al are a bit too flip about the unsupported by clinical beliefs bit. There is an awful lot of gray area in medicine. Going through charts afterward you can draw bright lines if you want, and researchers do, but there is a fair amount of latitude in practice styles.


If no one even has incentive to pay for it, why would anyone hew to the findings?

The problem is that patients have little incentive to shop on cost-effectiveness. Insurers would like to create a competitive advantage, but it's illegal for them to do so -- insurers aren't even allowed to suggest a cost-effective alternative because then they'd be "practicing medicine" and of course you need the secret handshake to do that or the state sends men with guns after you.

Which half?

So, ban reimbursement of treatments not supported by clinical evidence and reduce US Health Expenditures 17% saving, what, 3% GDP. That's huge. What's the argument against?


Freedom. HSAs. Skin in the game. Medicaid in Oregon. Never again. If Cutler found that demand-side effects don't matter much in spending variability, then that's only because he's wrong.

Wait, my cribsheet said we're also supposed to mention waiting months for surgeries. Did they kick me off the distribution list again!?!

Are you two doctors? Economists?

You paper over demand-side effects by breaking the link to financial incentives. Lo and Behold the supply-side will then dominate the equation. What exactly do you think is a gotcha about something TallDave says right here and has been saying for years?

You can go find the Rand study where HSAs were essentially the only thing that worked and people like. Managed care "worked" to cut costs but people don't like it. You can claim that it isn't scalable (speculation) or say that it doesn't meet your definition of fairness, but you can't seriously ridicule HSAs.

The subject's leading researchers find close to 40% of people who elect qualifying high deductible health plans don't put a dime into HSAs.

you're damn right they can be ridiculed, as can anyone who has some silly-ass view that they're some kind of magic bullet

I will probably lose my job if that happens so I will make many, many more annoying posts here.

A good place to start looking for that answer would be by asking the people who oppose these efforts when insurance companies utilize them.

Then we can move into the theological aspects of the debate.

Because there is no clinical evidence until doctors do something.

1. until doctors do something ... for which there is no clinical evidence.
2. unless they are doing it as part of of, you know, a clinical study or something.

Seriously though - I've heard this weak argument. If the policy is 'Let doctors, in their judgment, do a bunch of random things to everybody out of free vacations to Hawaii, social influences and hunches and for no evidence-based reasons - one gigantic moveable feast of a rolling random-clinical-trial experiment in which we all get to be guinea pigs whether we like it or not and Uncle Sugar foots the bill - and then we'll deep-dive the big-data and through epidemiological regression analysis find out what works and what doesn't and then ...

Ban what doesn't work. Right? We've apparently been doing that for decades, so, time to ban it now, neh?

The NIH Budget is $30 Billion. 3% of GDP is $500 Billion., If Cutler's only 7% right (I agree, leaving out fraud is a big hole in the study), we'd save enough to double research, and we could spend that on medical effectiveness research and producing new clinical evidence so we can practice ... evidence-based, cost-effective medicine. What a pipedream!

As I've been saying, oh for years now, while you were bankrupting the government with Medicare paying for ineffective stuff and the NIH developing new expensive marginal treatments, why weren't you already doing effectiveness research?

And guess who does most of the effectiveness research? Doctors publishing their case studies.

Because doctors are god-like beings who can't possibly have their unlimited discretion shackled without risking the loss of the mandate of heaven.

Because doctors are all god-like beings who can't possibly have their unlimited discretion shackled lest we lose the mandate of heaven.

I would think many of these beliefs are of the form, "I suspect that the fact that this isn't supported by the clinical evidence won't save my ass or the hospital's ass in court if we don't do it."

Except that in Texas, where they tried that by limiting payouts for malpractice suits, it didn't have any effect. In fact Medicare spending in TX rose faster than other parts of the country. Now, malpractice suits have gone down substantially, but it hasn't done patients or payors any good.

No. Capping only the non-economic damage portion of an award at 'only' $250,000 does not have a significant effect on reducing unnecessary defensive medicine.

'No clinical evidence' is relative when it's a battle of hired-gun paid-experts on the stand, one of whom is a PhD who said you should have done procedure-X that a lot of other doctors do, and the jury's looking at some poor guy who lost his legs from diabetes.

'I tried everything' and a paper-trail to prove it is decent liability protection when it's ordinary folks and not experts who are the finders of 'fact'.

Banning non-clinically-supported practices would be a great summary judgment bar to legal liability.

Let's say lawsuits are the cause of defensive medicine and this is primarily driving spending. We know this payout cap in TX resulted in dramatically fewer lawsuits (not a coincidence--malpractice insurance premiums went way down). We can conclude that this policy has dramatically reduced the incentive for physicians to practice defensive medicine and reduced their exposure to risk, but health care spending didn't go down. Perhaps the doctors simply continued to practice defensive medicine because they didn't understand why their insurance premiums kept dropping. However, the fact that health care is still dominated by a fee for service model that incentivizes volume makes me think that the payment model has more to do with health care spending than the whole defensive medicine canard.

Except that in Texas, where they tried that by limiting payouts for malpractice suits, it didn’t have any effect.

In May 2005, the American Medical Association removed Texas from its list of states in crisis.[12] Texas is so far the only state to be removed from this list.

The reform bill’s most significant achievements have been increased access to health care and an unanticipated positive economic impact on the Texas economy. By the end of 2013, 10 years and three months after the effective date of HB4, the number of licensed physicians in the state will almost have doubled. It is anticipated that Texas will have somewhere close to 60,000 doctors to care for its citizens, almost twice as many as it had in 2003. The number of physicians in Texas is now growing at twice the rate of the state’s population—a statistic that helps prove the success of HB4’s reforms in increasing access to health care. ...
As a result, the malpractice insurance premiums paid by Texas doctors have fallen by more than 60 percent on average. Consequently, most Texas doctors are paying less than half of what they were paying 10 years ago."


That looks like a pretty significant effect. Malpractice reform significantly increased the supply of physicians and reduced their insurance costs.

A group of academics who published a study in a peer-reviewed journal found no evidence of an effect on health care spending: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1635882.

The authors' response to those who cite an influx of physicians is that the data on "new physician applications and licenses do not account for doctors who left the state or retired, creating vacancies for their jobs; physicians who don't treat patients but do research or administrative work; and physician growth compared with other states. When those factors are weighed, Texas saw the number of direct patient care doctors grow more slowly after tort reform than it did before, the study says. Afterward, Texas did slightly worse than other states in attracting doctors, the study says." So, the growth in physicians practicing was actually slower in TX than other states.

By the way, the Heritage study was authored by the former Texas lawmaker who authored the tort reform law. Yeah, same guy. The researchers who produced the peer-reviewed study consisted of two Republicans, a Democrat and one foreign national.

Post-postmodern conflicting evidence means I can "believe" whatever I want even though it's fairly obvious which source is more reliable.

Your post is misleading. The study you linked to is specifically about Medicare spending. Which does not reflect the broader healthcare picture. Medicare spending is a function of politics fmore than it is a function of a normal market place.

"We study how Medicare spending changed after Texas adopted comprehensive tort reform in 2003, including a strict damages cap."

Indeed, I'm at a loss how anyone would expect that Medicare spending would go down due to tort reform. The incentives aren't even close to being aligned.

Variations in regional health care spending are analyzed via Medicare because that is the most comprehensive data set available, and the most comparable because the benefit is the same across regions. Regional variations are what this whole discussion is about. The very author of the legislation in TX and the Heritage study, Joe Nixon, published an op-ed that explicitly states that the goal of tort reform is to reduce health care costs: http://www.mysanantonio.com/news/article/Medical-liability-reform-crucial-to-health-care-843293.php

It turns out tort reform did reduce doctors' insurance premiums, but it didn't reduce costs. It also didn't increase the supply of practicing physicians. So let's move on.

Not the effect being discussed. More physicians paying less for malpractice insurance is obvious. But costs of procedures and consumption of medicine both went up. Not being sued just makes it better to be a physician, but it doesn't bring down costs for consumers.

Yes,. the number of MDs. in Texas expanded after the law change.

But the growth rates actually fell. the law was changed in
In the five years 1998 to 2002 the number of Texas MDs grew at a 3.4% average annual rate. In the five years 2003 to 2008 the average annual growth rate slowed to 2.2%.

So how effective was the new law?

A whole paper on regional variation in Medicare spending and not one mention of fraud. South Florida has the highest Medicare spending per enrollee of any region in the country. South Florida has the greatest amount of fraud. The Bronx has significantly higher Medicare spending per enrollee than Manhattan. The Bronx is a hotbed of Medicare fraud. When roughly 25% of spending is fraud and there is significant regional variation in fraud, the explanation for regional variation is, not suprisingly, fraud. That economists -- the people who think that you cannot tell whether a coin is fair even when it lands on heads 25 times in a row -- would miss the most obvious explanation when that explain is criminal activity does not surprise me.

+1, not considering that fraud might be a significant issue seems like a pretty big miss.

No mention of fraud in the whole article. I would encourage people to look at the highest cost per enrollee regions against the regions with the most Medicare fraud. The overlap is stunning.

If fraud is a set percentage of total billings for every area, this is what you would expect. It's like the xkcd with heat maps for radically different things, but they all looked similar because they all corresponded with population density.

Do some areas have proportionally higher frauds? Sure. How much of the variance in Medicare spending do they explain? Well, it's hard to make it more than a couple of percent. For most of the country, fraud isn't that much more likely between, say, Des Moines and Grand Rapids, but such comparisons can have dramatically different Medicare costs.

But the Republicans think it would be wrong wrong wrong for Obama to establish a government body tasked with trying to determine what treatments are effective.

There's no need. Just learn from the equivalent bodies in other countries.

Agreed. If there's one thing my government can do, it's contract out work. We just won't tell anybody what NICE stands for and pay them a couple billion a year. It would be the highest ROI move Medicare could make, assuming they were allowed to use the findings to alter their coverage decisions.

That is because there is already such a group. It's called doctors.

If you want to help them, by all means be my guest.

What you are talking about is a cost-effectiveness panel concerned with saving the government from bankrupting itself.

Just a little evidence, baby. It is first a clinical effectiveness panel and second a cost effectiveness panel. If it saves the Medicare from ill-advised physician decisions that are neither backed by evidence nor in the best interests of the program, yes.

If a giant insurance company is going to decide which treatments are effective, at least let it be one I can sue.

That seems to be rational rationale for evidence-based rationing in Medicare.

Correcting physician views of the cost effectiveness of different procedures ought to be easier than changing the public's views.

"...36 percent of end-of-life spending, and 17 percent of U.S. health care spending, are associated with physician beliefs unsupported by clinical evidence."

And the responsible docs all live in fraudulent Florida [KLO], or ... self-delusional Massachusetts!

Surprise surprise. Another bunch of economists re-inventing the wheel. "Structured vignettes" are discrete choice experiments - the econometrics was pioneered by McFadden (1974) which got him the nobel prize. Plus you have to "calibrate" stated preferences (in this case we're interested in the doctors) before you compare with revealed preferences due to different variance scale factors in the data. Again, known for 20+ years.

I'm a practicing pulmonary/critical care doctor, and I just read the ungated article. While the discussion in this article relates to more defined problems than I usually confront, it resonates. Just today, I had an end-of-life discussion with a patient with a rare and terminal condition. I recommended no advanced care, that is, intubation and mechanical ventilation. Another patient with a huge burden of pre-existent disease, and a recent pneumonia, I recommended waiting until the more familiar pulmonologist came back on service, but I think withdrawal of care is appropriate. Yet, another patient who I have cared for over many years is struggling under a huge burden of cardiopulmonary disease, but he's always pulled through before, so I recommended continued care.
The article points out that physician's recommendations carry great weight. I can anecdotally confirm this. I think that as one becomes better in one's field, one accumulates success stories. Confirmation bias plays a large role though, and the patients who suffer and die are under-represented in a doctor's recommendations. This is certainly true in my case. Some patients recognize that their life is over, and it's easy to make them comfortable and let them die, but others want to fight to the last dying breath. If you have many tools, like I do, you'll always look to use these to eke out every benefit and advantage for these patients. In the latter cases, it's always easier to try some additional desperate maneuver than to tell the patient that they are dying. That is, of course, why these patients come to us physicians.
I additionally don't think that professional society recommendations are the last word in patient care. These recommendations are not fine grained. They don't apply to special situations, and most of them can merely state that a given practice is unproven as opposed to be proven to provide no benefit. Cardiology is less this way than my field, as cardiology has tremendous funding and some things in that field can be proven to be detrimental. Additionally, many practices that are not beneficial upon initial study prove beneficial later. As an analogy, if aviation had to prove every innovation to be beneficial, many things such as LORAN, and small modifications to hydraulics and airframes might have failed muster because the incidences of the failure modes they sought to remedy are so rare.
Finally, here in Minnesota, a low litigation state, I have witnessed little medicare fraud, but not none. Here, I don't think it's a large part of expenditures. I don't fear lawsuits, but instead family complaints that I was too aggressive. I want to do the best for my patients. I thought that I and my partners were listening to patients wishes, but maybe we should be more attuned to desires for minimal care. Too many times the desire for less aggressive care comes from the family, and not the patient. Ethically, we have to listen to the patient, no matter how drugged out or delirious they are. We struggle with the huge expense to keep people alive in their last year, but also have the occasional success story to prove that it's not all hopeless. Maybe we should give into nihilism. It's not my constitutive propensity.
Sorry for the long comment. This really hit home. I wish I had a good answer to these problems.

I am waiting for a mandated second opinion by IBM's Watson or even an initial interview with Watson as a precondition for extraordinary end of life spending..

Would people be more willing to hear from a machine about their chances, rather than a doctor who wants to sugar coat the bad news and holds out too much hope for the effectiveness of futile care? And, it takes the pressure off the doc.

Watson recommends apricot pits, megadoses of vitamin C, and removal of all of your mercury amalgam tooth fillings.


David, thanks for the thoughtful post.

All healthcare is marginal. More expensive healthcare is almost always more marginal.

Do you really need that MRI for a headache? Hip replacement at 82?

Somewhat paradoxically, the more a system spends the more likely it will be spending money on things that have very small or even negative benefit.

But "we" don't want individuals decisions to be affected by financial concerns. If the Federal government is going bankrupt and their decisions are affected by financial concerns, that is okay.

Lynn Payer's classic "Medicine and Culture" is great on this. Her research focused on treatment variations from doctor beliefs in France, Germany, U.K., and U.S. Fascinating stories of "liver panic" in France, heart problems from romanic-thinking Germans, and stoic, suffering of Brits making do, plus U.S. surgeon cowboys. Culture influences medical views of doctors and medical expectations of patients.

Makes sense similar cultural variations would influence U.S. medical beliefs of doctors and patients. It's a big country.

2 points, with evidence. First, without meaning to be glib, the "answer" varies across medical practices, AND there is a real need to define "clinical evidence, depending on the setting. Second, there are people at work on figuring out how to improve doctors and systems as to cancer, which of course implicates end of life.

Change and cultures of treatment do vary. For example, the pulmonologist practices in an area where, to my knowledge, not too much is truly new for persons with advanced disease. For oncologists, however, much is new and changed, for some people with some tumors. Precision treatment of tumors now exists for some tumors arising from some specific and now known mutations. Therefore, treatment "miracles" do happen even with advanced disease, but in fact are not miracles and instead are examples of precision medicine being put to service to treat a specific mutation with a specific drug, producing years of high quality life or even remissions. This WSJ story presents an example in a real person. http://online.wsj.com/article/SB10001424127887323300004578557473861805376.html. The story also illustrates that many (most?) day to day to doctors have no clue about what's new in cancer treatments, but doctors at research institutes and true centers of excellence (e.g MD Anderson, Mayo, Sloan- Kettering) do tend to know the answers and/or know they can and should turn to a specialist colleague who does know the answer. For a broader example of what's new, read about Adcetris, a new genomic drug producing complete remissions in well over over 80% of patients with certain blood and lymph cancers. See http://investor.seattlegenetics.com/phoenix.zhtml?c=124860&p=irol-newsArticle&ID=1765797&highlight=. Now, when phase 1 trials show such stunning results, is that "enough" clinical evidence or do you have to wait for phase 2 or 3 studies ? The FDA increasingly seems to accept and act on "breakthrough" drugs, moved in part by the stunning reality that 575,000 americans die of cancer, every year, and 1.6 million new cancers are diagnosed each year.

As to doing better, NIH has funded some really smart and creative folks at a multi-disciplinary ASU center for law and science to recommend ways to change medical practices and systems to deliver precision medicine and thereby avoid litigation and liability. Doing so should save lives and money. See http://www.law.asu.edu/News/CollegeofLawNews.aspx?NewsId=4233. Multi-disciplinary thinking is critical.

Also see the following brief but pointed note on NCI's new effort to change how we think about and define cancer. The author is physician and long term cancer survivor who is actively involved in speaking and writing for both doctors and patients on ways to improve their communication and thereby make better choices in the face of uncertainty. http://wendyharpham.typepad.com/healthy_survivorship/2013/08/one-way-to-address-the-issue-is-to-change-the-language-used-to-describe-lesions-found-through-screening-said-dr-laura-j-es.html .

Plainly we will not get there when CMS and others refuse to pay a decent amount for doctors to communicate.

By "clinical evidence" for every medical procedure I have 2 questions:
1. Are we talking about the huge amounts of clinical evidence supporting something as mundane as the use of fluoride?
2. Who has a set of knee-pads handy?


Consider the following two alternatives....

1. You are 80 years old with a terminal disease - your life will end in the next year regardless of what is done, but you have $1,000,000 in assets and if you chose to spend them on extending your life you will live the full 12 months - if not you will leave an estate that will allow several of you grandchildren to attend college and get a great education - something that cannot happen otherwise. What does the discussion with your doctor sound like - what are your concerns - what do you think you will do?

2. Now consider the same circumstances where you are covered my Medicare - and yes you will have to draw down some of your net worth to get care - but would your choices - decision be much different? any different? - if so, Why?

One of the things I am most sad about is how easy it seems to be for people to spend other peoples [tax payers] money without even thinking they should say thank you!

Medicare is not other peoples' money. If a guy has a million in assets, likely he has paid into Medicare -- how much? Couple million, over the course of his lucrative life? Medicare was not sold as welfare, it was sold as a savings plan. Along with SS, of course, the stupidest savings plans ever. I'd gripe at your hypothetical guy if he volunteered for it, but of course, none of us have a choice.

Take it down a rung, we have a society that deifies "science" while not having the tools to evaluate whether something is science or not. Consequently, we tend to vest our trust in anyone with a diploma on the wall saying he's smarter than we are, if he has a license it's all the better. Our doctors are our priests. So of course what they believe in is going to shape the way we consume medical care, because we follow their lead. When the majority of doctors speaking in public told us a pregnancy begins at implantation instead of conception (not a scientific distinction, a semantic one based on belief systems) there were huge changes in reproductive health care consumption. When the majority of doctors moved from the old belief that the soul determined life to the new belief that brainwaves did (not indicated, but determined), we had huge changes in how end of life situations were addressed. There are more belief system changes in the works, and that will work similar changes in the way we consume health care. And those changes will move in waves regionally, with hold outs in some places and folks well ahead of the curve in other places. There's even little offshoots that branch out and flourish or die, like doctors who now cater to the belief that conventional doctoring is not very useful, bringing in alternative medical practices to their offices to compete with their non-doctor competitors (and also because their belief systems have changed). In those regions were more people see alternative providers, doctors start to go alternative, then the patients of doctors get exposed to things like nutritional medicine, etc. An example is the number of doctors now prescribing Vitamin D testing and supplements, now people consider that regular medicine where even ten years ago it would have been wacky tree-hugging stuff.

Comments for this post are closed