Posted by: notdeaddinosaur | June 21, 2010

Atul Gawande and Complexity: the Flaw in his Argument

In a fascinating but commonplace confluence of circumstances, I’ve just started reading Atul Gawande’s The Checklist Manifesto at the same time as Gawande gave the commencement address at Stanford medical school (and Buckeye Surgeon blogged about it). Reading the latter felt like deja vu in places, as Gawande plagiarized himself making his points about complexity; specifically the increasing complexity of medical care in the 21st century:

The truth is that the volume and complexity of the knowledge that we need to master has grown exponentially beyond our capacity as individuals….

Half a century ago, medicine was neither costly nor effective. Since then, however, science has combatted our ignorance. It has enumerated and identified, according to the international disease-classification system, more than 13,600 diagnoses—13,600 different ways our bodies can fail. And for each one we’ve discovered beneficial remedies—remedies that can reduce suffering, extend lives, and sometimes stop a disease altogether. But those remedies now include more than six thousand drugs and four thousand medical and surgical procedures….

It should be no wonder that you have not mastered the understanding of them all. No one ever will. That’s why we as doctors and scientists have become ever more finely specialized. If I can’t handle 13,600 diagnoses, well, maybe there are fifty that I can handle—or just one that I might focus on in my research. The result, however, is that we find ourselves to be specialists, worried almost exclusively about our particular niche, and not the larger question of whether we as a group are making the whole system of care better for people.

Wow! Over 13,000 diagnoses in ICD-9-CM! Actually, while there may be more than 13,000 entries in ICD-9, there are huge swathes of them that aren’t really medical diagnoses at all. I like to say that ICD-9 represents over 13,600 reasons to see a doctor. V70.0, periodic health examination of the well adult (V20.2 for a child), shouldn’t count as one of the “different ways our bodies can fail.” Nor should V76.12 (screening mammogram) or V72.31 (routine gynecological exam).

But I digress. I will grant that scientific advances of the last century have resulted in a much expanded understanding of the function and failings of the human organism, such that its complete knowledge is indeed beyond the mastery of any one individual. And I agree that this fact necessitates specialization — even super-sub-specialization — for mastery of the most arcane sub-specialized details. But Gawande is still wrong.

Here’s why: even though he quotes average numbers of diagnoses per patient (usually double-digits, mainly for hospitalized ones, though he also includes similar numbers for outpatients) that imply every case is mind-numbingly complicated, NOT EVERY PATIENT IS THAT COMPLEX. In point of fact the simple, straightforward medical cases vastly outnumber the ridiculously complicated ones. Offhand, I would say the ratio is at least nine to one, based on the oft-quoted statistic that 10% of the population uses 90% of the medical resources.

Gawande’s figures on outpatients with dozens of diagnoses in their electronic medical records are also misleading. Typically they include every diagnosis the patient ever had. Pink eye three years ago does not materially increase the complexity of dealing with your right lower quadrant abdominal pain now. Even ongoing chronic conditions like hypertension or psoriasis don’t meaningfully impact management of your penile discharge.

It’s easier to lose sight of this fact when you spend most of your time in the hospital. Inpatients are sicker than outpatients. (Where’s Captain Obvious when you need him?) When every patient you see requires ten specialists just to get him through til dawn, the complexity is indeed overwhelming, and it can seem appropriate to generalize one’s observations to the practice of medicine as a whole. But this “denominator effect” leads to the incorrect conclusion that specialization is the only answer.

There are plenty of patients (90-95% of what walks in my door, per AAFP figures) with medical needs that do NOT require Gawande’s or anyone else’s super-dee-duper-sub-specialized skills. There really is such a thing as a simple sore throat, a straightforward case of hypertension, and uncomplicated diabetes. And it really is possible for a single doctor, a well-trained family physician — me! — to manage these patients.

An enormous part of my job is to identify those patients with problems too complex for me to handle, or who require treatment beyond my scope of practice. Even then, my job is to correctly identify which of the myriad of super-sub-specialists is the best choice for a given patient; frequently a very complex decision in and of itself!

So to the two or three graduating Stanford seniors (statistically) going into primary care, don’t worry. Gawande may be a rock star, but his reports of the death of the generalist physician are greatly exaggerated.


  1. The champion of the Dartmouth Atlas and the scourge of McAllen, Texas urging graduating medical students to all become super specialists (as if any additional encouragement was necessary). I hope the next paragraph of his speech had the essence of your comments. Atul, do you know how much money I save Medicare and the whole system daily? So which is it, Atul, do you want to save money or do you want to cut/scope/stent/image every last patient for every minor complaint? Or do you expect all the fancy Stanford specialists to sit idle and only cut/scope/stent/image the minority of patients that actually need it rather than the universe of patients that they could get away doing it to.

    Our local cardiology mega group bought a Cardiac PE. Scanner. Do you know what the indication for cardiac PET scan is here? 1. Patient has a heart. 2. Straight Medicare insurance. (Because no prior auth required, payed every time).

  2. Actually I thought the book made some good points if you consider the context in which Gawande works (mostly hospital with very sick folks, who really ARE mind-numbingly complex). I don’t think he recognizes that most patients/cases aren’t that complicated because that’s not how things are in his world. And it was kind of interesting to learn the source of all these checklists we keep getting hit over the head with.

  3. One quibble: you cannot plagiarize yourself!

  4. Actually, JPB, you CAN plagiarize yourself — if you do not own the copyright to your own work. For example: I was approached by a professional organization to write a book about my area of expertise, but the book contract specified that the organization would hold the copyright to the work, and that I would not be able to use the work myself in the future (at least, not w/o the organization’s written permission). Since the bulk of my self-employment involves training in this specific area, that limitation would have prevented me from using the training materials I’ve developed over the years, and which would have been the guts of the book.

  5. Oh, you certainly can plagiarize yourself — if you take big chunks of a previously-published article and then try and plunk that text in the body of a brand new article that is supposed to be going to a prestigious, peer-reviewed journal that only accepts original submissions, you’re toast.

    Have I caught people doing it? Yes.

  6. Every field would say that you cannot re-publish results and conclusions as if they were entirely new. That is known as duplicate publication. But some fields do allow you to copy verbatim some of your own previous published text.

    The British Medical Journal allows new submissions to be 10% identical to one of your previously published articles. In reality, that 10% means 0%, given that articles generally only present one or two studies. But if you wrote a non-empirical paper, you can copy 10% of your own previous publication.

    The American Psychological Association (APA) allows you to publish up to 400 words of your own text in background and methodology sections, without quotation. The major journals in Operations Engineering allow you to copy entire sections verbatim as long as you say, “This section is taken from [your previous paper]”

    The WAME is lenient with regards to copying your previous text for Methods sections or other sections in which you really can’t say it as clearly another way. Beyond that, it’s debated.

    This only applies to ethical considerations though. Publications may have copyright policies that are stricter than the ethical guidelines.

  7. In response to the above comments, you are actually discussing copyright violations rather than plagiarism. Please check your dictionaries!

  8. @JPB You’re right! 🙂

  9. @JPB: It started as a discussion of copyright violation and not plagiarism, but rethoryke and I discussed plagiarism in addition to copyright violation.

  10. Yeah, but nobody’s discussing the fact that Atul Gawande lives in an ivory tower of complexity, which makes his approach to medical care unwieldy and hideously expensive in many outpatient settings. *pout*

  11. NotDeadDinosaur: One of the Laws of the Internet is that online discussions (even in blog comments) have a life & trajectory of their own and it’s hopeless to try to direct them in the direction you think they should go. So just enjoy the roller-coaster! 🙂

  12. Right ON, Dinosauer! Joust with the Knight Gawande and show that being a whole person knight is good for the kingdom. The parts oriented knights should report to the whole person knights for further instruction about the big picture.
    We need more big picture generalists to lead patients to the right situation at the right time, and to lead medical staffs to realign with patient needs. A tiny focus in the wrong place can be costly and disastrous. Go Dinosauer!

  13. Someone else sounding the death knell. This time the cause of death is non-rural “rural medicine.”

  14. perhaps not germane but definitely interesting.
    John Fogarty of Creedence Clearwater once got sued for singing his own songs in concert and lost.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: