“Unwarranted practice variation”: an essential e-patient awareness topic

Headline and body edited Oct 6, 2013: the original post talked about “practice variation,” but that was bad wording. The problem is unwarranted practice variation: variation that, when studied, is not warranted by actual differences between cases.

This is the first of the follow-up posts I hope to write from participating last week in the Salzburg Global Seminar titled “The Greatest Untapped Resource in Healthcare? Informing and Involving Patients in Decisions about Their Medical Care.”

One of our purposes on this site is to help people develop e-patient skills, so they can be more effectively engaged in their care. One aspect is shared decision making, which we wrote about in September. A related topic, from August, is understanding the challenges of pathology and diagnosis. Both posts teach about being better informed partners for our healthcare professionals.

I’ve recently learned of an another topic, which I’m sure many of you know: unwarranted practice variation. This is a big subject; I’ll have several posts about it. It’s complex, the evidence about it is overwhelming, and its cost is truly enormous. I’m no expert at it yet, but I also know it’s important, so let’s get started. Corrections welcome. Here it is, in  a nutshell:

  • Very large parts of healthcare are delivered inconsistently from area to area.
  • In other words, the care you get depends on where you live.
    • That’s right; very often, care decisions aren’t based on some objective standard of care. The same patient in a different local area might or might not get a prescription for treatment. Very often.
    • Which one is right? Is one overtreated, or is the other  undertreated?
  • This isn’t a matter of economics: it’s a matter of local medical practice. It cuts across all economic levels.
    • That’s why it’s not called discrimination, it’s called practice variation.
  • The people involved – the doctors – mostly don’t know they’re doing it.
  • Bottom line: depending on where you live, you may be getting care you don’t need – hospitalizations and even surgery.
    • Since both of those carry risks of infection and even death, e-patients need to be aware so they can make informed, empowered choices.

Pardon the heavy italics; I’m trying to convey concisely the things you’ll see when/if you start studying this. Examples:

  • For decades, tonsillectomies were performed in some regions 3-4x more often than in others. (Even between neighboring towns.)
    • Here is a seven page paper from the United Kingdom showing a threefold variation in how many kids got tonsillectomies. It’s from 1938, and Dartmouth researchers found the same in the US in the 1970s and 80s.
    • The end of the report carries the nasty impact: in one year the nation had sixty deaths from tonsillitis, and over 500 deaths from tonsillectomies – most of them children. Unnecessarily dead children because of this issue.
  • The same has often been true with hysterectomies. And gall bladder surgery. And coronary bypass grafts. And many other things.
    • A current non-US example: a post on the NPR blog, written at the end of last week’s seminar by participant Chris Weaver (@cdweaver) of Kaiser Health News: UK citizens in Oxford are 16 times more likely to get a particular type of hip replacement than similar people in London.
  • For any given condition, your odds of being hospitalized are often proportional to how many hospital beds are in your area.
    • Yes, that’s true after controlling for demographics, severity of illness, everything.
  • At the end of life, your odds of dying in an ICU are proportional to how many ICU beds your region has.
    • I’m not making this up; this is well-vetted, carefully-culled data, controlled for confounding variables. For any given illness, your mother is less likely to die at home – even if she requests it – simply depending on how many ICU beds your local hospitals have.
    • There are tons of data to support this.  It’s been validated and cross-checked every which-way from Sunday, for years and years.

Yes, to a large extent, recommendations for some types of surgery and hospitalization are driven by local superstition and the mere availability of empty beds (or a particular type of specialist).

This is generally not medical plundering. Doctors generally do not know they’re doing this. (I imagine some do, but this is not a matter of rooting out greed – there’s a bigger issue of widespread denial about how things work.)

This is by far the hardest healthcare issue to comprehend I’ve ever seen. Neither the problem itself nor its intractability – its resistance to change –  make any sense to me. Most of the people involved can’t even believe it’s happening – even though they’re doing it, and the evidence is clear!

When that happens, it’s a sure sign we’ve been overlooking something big. And our efforts to argue for change are doomed until we understand the actual situation.


I see two major impacts.

  • Cost of unnecessary hospitalizations.
    • Cost to society
    • Cost to the patient and family for the care
    • Lost income
  • Risk of harm, including infection and death.
    • And the cost of those complications.

E-patient Takeaways:

Smart people have been trying to change this for decades, and it hasn’t changed. While they work on it, the matter is in our hands. In my view empowered, engaged, educated patients need to

  • Realize this happens
  • Educate ourselves about the region we live in
    • Information is available about which areas are high-utilization. More on this in upcoming posts.
  • Get to work at spreading the word.

As I said, this is all part of a larger issue, SDM – shared decision making, which is a bigtime participatory medicine topic. We’ll have much to say about this in the coming weeks. The first posts will discuss more about practice variation.


Background / homework: I’ve started reading the new (2010) book Tracking Medicine, by Jack Wennberg of Dartmouth, which chronicles his 30+ years working on this. You can buy it on Kindle if you want and start reading immediately. (You don’t need a Kindle; there are free Kindle apps for almost everything including smartphones.) Or if you’re like me, you’ll want to scribble all over the pages of a paper copy.


Posted in: e-pts resources | practice variation | shared decision making | understanding statistics | Why PM




25 Responses to ““Unwarranted practice variation”: an essential e-patient awareness topic”

  1. Annie Stith, e-Patient says:

    Hey, Dave!

    Wow. I mean, I’ve heard about this before, but… WOW.

    I was wondering when I was sent to a surgeon quite a while back now if he was “prone” to surgery just by *being* a surgeon, but this is much more vast a problem.

    I imagine that here in St. Louis, with our multiple medical schools and all those students needing hands-on experience — and *plenty* of hospitals in the area — maybe for or because of the schools — it’s just plain scary stuff.

    Added to that, the MRSA in our hospitals is just awful. One had their ER shut down and they had to rebuild off a different entrance. That’s not the one I was in for that gallbladder surgery, either, where I ended up with MRSA, or the one Dad (my almost-in-law) got it from during a bout of pneumonia. Three separate hospitals.

    Can’t wait til I get paid and I can get that book!


  2. Mike Pierce says:

    Dave – Great article and nice to see all the well researched supporting facts. It’s an incredible revelation and like you say, it isn’t necessarily driven by misappropriation but often a result of the confluence of experience, cultural norms and information symmetry to name a few key drivers.

    It seems that these geographical anomalies could be put into check by getting patients and care givers help from outside medical expertise with access the the latest in EBM care and SDM – so as to network knowledge and experience across geographies. It’s hard to see the forest from the trees when you are in the middle of it.

    Thanks for the great insight!

  3. Gangadhar says:


    Thanks for the great post. In my opinion, this is essential reading for all empowered patients. Here is another aspect of practice variation:

    I recently came across this:


    The part that attracted my attention was that Vermont’s law, which prohibited pharma companies from using a doctor’s prescription history to market drugs, was over ruled by a court. I felt that the prescription history of a doctor is a very useful piece of information for empowered patients.

    Now why do I bring this up when we are talking about practice variation? Imagine if you could pull up the drug prescription history of a particular doctor and you could see if they prescribe more drugs or less drugs on an average (compared to other doctors in your city, state etc). This is one form of practice variation. Now this information will go a long way in helping the patients to decide if they should really take that drug being prescribed or go for a second opinion.

    • What a great idea, Gangadhar! What if patients (and policy people) could see, transparently, each physician’s prescribing patterns?

      • Claire says:

        Individual prescription patterns are only of value if the individual practice population is known — only then can you assess appropriate behavior.

        The other difficalut thing about following pure scientific practice guidelines is that medicine treats people not widgets — no two people are identical or respond identically. Their psycho-social state impacts both disease course and treatment response, which is why a holistic approach is essential. This is not to refute the intrinsic value of science but to temper its authority a little and emphasize that basing everything on the known science is not always the best choice.

        Every procedure/treatment has been invented for a very good reason and is 100% appropriate for the right person/ailment — it is honestly figuring out if you are that person that is tricky.

        • Hi Claire – yes, Tracking Medicine documents the years of studies that went into establishing that practice variation is just as robust after controlling for population differences. The same analysis, and same results, were reported in the 1938 paper on tonsillectomies in the UK.

          I take your point, but since the averages vary by hundreds of percent from area to area (the book cites one case where variation is 36fold, I think), what we’ll surely find is that individual docs vary by those numbers, and probably more.

          If our goal is to work together to improve healthcare, it seems clear to me that physicians and consumers alike will benefit from opening our eyes and trying to understand reality. With that comes responsibility for understanding the issue you cite: the raw number itself needs to be taken in context. That’s part of understanding the data in the first place.

  4. Go to http://www.citizensforpatientsafety.org and view the trailer of the new training documentary regarding shared decison making… The faces of Medical Error from Tears to Transparency the Story of Michael Skolnik. This is an award winning series with the Lewis Blackman film being the first of the series.

    • Hi Patty – what a pleasure to have you here. (For those who don’t know, Patty Skolnik is one of the great leaders in the patient safety movement, having suffered an enormous loss herself, as shown in the trailer.)

      I watched the trailer last week and as someone who’s new to SDM, I didn’t get the connection. In last week’s seminar I learned that an important shift is underway, from informed consent to informed choice. As I understand it, in the legal doctrine of informed consent, the doctor decides what’s recommended, and hands you a document to sign.

      In practice I’m told we have abundant evidence (as exemplified in Patty’s trailer) that there’s no way for an innocent patient or parent to know what’s not being said, and there’s every reason not to assume you’re being told what all the options are. That’s where informed choice is different: we have a right to know what all the treatment options are, and especially to know the risks of each, so we can choose.

      In Patty’s trailer it’s clear that they didn’t know the risks, much less alternatives, and paid a very dear price when things when wrong. The lesson is that we have to ask, and we ought to do everything in our power to learn about all the options.

      We’ll get into more depth about SDM and informed choice (vs consent) later in this series. For now, Patty’s trailer gives a realistic and painful picture of the cost of the breakdowns in today’s system.

  5. Hi Dave,

    Great post! Practice variation has many implications; one of which is also patient safety as you’ve so nicely highlighted. I agree with you that in most cases doctors don’t know they’re doing this; instead it’s the payment incentives that dictate much of this.

    Patient engagement is the ultimate goal and to do that we need more hospital quality outcome data (useful information from which decisions are made) to stimulate that engagement.

    Thanks for your passion!

  6. Kerry OConnell says:

    These are indeed very interesting times. The availability of beds has increased significantly since the reccession kicked in as many people can no longer afford procedures. It woould appear that you have a tug of war between doctors filling beds and patients resisting needed care for economic reasons. Once the newly insured get into the system next year the beds will fill up fast and if your studies are right there should be a huge spike in medical harm.

    • I don’t know, but I wouldn’t necessarily leap to that conclusion… I don’t think the idea is that there’s a direct almost-mechanical link, I think the point was that a large amount of practice volume that does correlate with # of beds and doesn’t correlate with any other observed variable.

      But we shall see.

  7. Chris Hall says:

    As shocking as this post is, Dave, I can’t say that it surprises me. Human beings are shaped by their own experiences, and will rely on those experiences in the absence of contradictory data. We go with what we know… hopefully until proven otherwise.

    Its the bringing everything together part to create an easily identifiable standard of care that really interests me. :)

    • Yeah, Chris, the more I learned about it, the more stunned I was – especially the robustness of the effect, i.e. how it persists even after controlling for other variables.

      But after the stun I was left with an odd sense of being empowered by it, because there’s always power in a new realization about reality. If we don’t know what’s going on, we have zero ability to deal with it effectively.

      So, next up is to learn more, understand what’s behind the numbers (to fortify our knowledge), and then look at what we can do to create worthy change.

  8. […] Practice variation – the first of several follow-up posts about what we learned. This post introduces the new book Tracking Medicine, which recaps the 30-year work of Jack Wennberg and colleagues at Dartmouth, analyzing vast amounts of Medicare data to understand how we actually spend our money. As the post explains, we think medicine is practiced scientifically from a common body of knowledge, so we’d think it would be pretty consistent, but the data show it’s common to have threefold, fourfold, or even greater variation in what doctors recommend, and it pretty much varies by geography, with no rhyme or reason. If your doctors recommends surgery, wouldn’t you want to know that most doctors elsewhere wouldn’t? […]

  9. lionel says:

    “For any given condition, your odds of being hospitalized are often proportional to how many hospital beds are in your area”,
    then, of course
    “your odds of dying in an ICU are proportional to how many ICU beds your region has.”

    After all, when people die, they have to die *somewhere*. I am convinced that the statistical gruntwork was done, and well done, but please, when advocating patient empowerment and opendata and such, provide meaningful links to studies or raw data, including methods of correcting bias, etc.

    “There are tons of data to support this. It’s been validated and cross-checked every which-way from Sunday, for years and years.” is *not* helpful.

    • I couldn’t agree more, Lionel. That’s why I cited Wennberg’s book, in a section explicitly labeled “Background/homework,” and included a direct link to go get it on Amazon. :–)

      • I see Lionel’s point. If there is tons and tons of data then there are tons and tons of articles published about that data. And therefore there should be some links to pubmed references to help us learn more and decide which way we want to look at this issue. If the only link is to a book who is sitting behind a paywall we haven’t help much and are, in fact, promoting the paywalls instead of the concept of Open Access and Open Science.

        As we move along and this blog gains coverage this issue will return to haunt us. Everything we write must be based on openly accessible references and everything we publish should be Openly Accessible.

        • KC says:

          For folks interested in more empirical evidence on practice variation, you might try looking up the Dartmouth Atlas. It’s a project going back more than 30 years trying to figure out why Medicare spending per patient is so radically different in different areas, and whether those differences mean anything for patient care.

          Like Dave, they cite practice variation as a factor, and even include some thoughts on whether patients are generally better off if their health care provider decides to admit them to a hospital (very expensive) or attempt to manage their condition via outpatient visits, nursing homes, or home health orgs (not cheap, but less expensive than long term admissions and possibly less traumatic to the patient).

          Here’s a summary of some of their work:

          • Hi KC! Yes, that’s exactly what I’m leading up to – the Dartmouth Atlas is the major outcome of Wennberg’s work. Thanks for teeing up my next drive, coming up this weekend. :)

            Good to meet you!

  10. […] the evidence shows that a lot of medical practice is simply not done scientifically. (See the December post for […]

  11. […] this was three months ago, en route to a seminar on SDM (shared decision making); my first post after the seminar was shortly after. The whole subject has bent my thinking about healthcare so severely that it’s […]

  12. […] this was three months ago, en route to a seminar on SDM (shared decision making); my first post after the seminar was shortly after. The whole subject has bent my thinking about healthcare so severely that it’s […]

  13. […] to be engaged in decisions, especially before they understand the reality of physicians’ unwarranted variation in practice. Unwarranted variation has been shown in decades of research to cause immense unnecessary surgery, […]

  14. […] Unfortunately that led to unwarranted practice variation (see my 2010 post on it. […]

Leave a Reply