Missing the Forest For the Granularity

Nortin Hadler

European health care systems are already awash in “big data.” The United States is rushing to catch up, although clumsily thanks to the need to corral a century’s worth of heterogeneity. To avoid confounding the chaos further, the United States is postponing the adoption of the ICD-10 classification system. Hence, it will be some time before American “big data” can be put to the task of defining accuracy, costs and effectiveness of individual tests and treatments with the exquisite analytics that are already being employed in Europe. From my perspective as a clinician and clinical educator, of all the many failings of the American “health care” system, the ability to massage “big data” in this fashion is least pressing. I am no Luddite – but I am cautious if not skeptical when “big data” intrudes into the patient-doctor relationship.

The driver for all this is the notion that “health care” can be brought to heel with a “systems approach.”

This was first advocated by Lucien Leape in the context of patient safety and reiterated in “To Err is Human,” the influential document published by the National Academies Press in 2000. This is an approach that borrows heavily from the work of W. Edwards Deming and later Bill Smith. Deming (1900-1993) was an engineer who earned a PhD in physics at Yale. The aftermath of World War II found him on General Douglas MacArthur’s staff offering lessons in statistical process control to Japanese business leaders. He continued to do so as a consultant for much of his later life and is considered the genius behind the Japanese industrial resurgence. The principal underlying Deming’s approach is that focusing on quality increases productivity and thereby reduces cost; focusing on cost does the opposite. Bill Smith was also an engineer who honed this approach for Motorola Corporation with a methodology he introduced in 1987. The principal of Smith’s “six sigma” approach is that all aspects of production, even output, could be reduced to quantifiable data allowing the manufacturer to have complete control of the process. Such control allows for collective effort and teamwork to achieve the quality goals. These landmark achievements in industrial engineering have been widely adopted in industry having been championed by giants such as Jack Welch of GE. No doubt they can result in improvement in the quality and profitability of myriad products from jet engines to cell phones. Every product is the same, every product well designed and built, and every product profitable.

If patients were widgets, if care givers were production workers, and if caring conformed to “six sigma” principals, even slight deviations from standards of care would be as easy to recognize, those responsible could be singled out for improvement, and remedies would be obvious. This is the mantra of the quality agenda that drives health policy. In a corollary exercise in linguistic determinism, health economists and hospital administrators are wont to speak of patients as “units of care”, physicians as “providers”, and clinical demand as “throughput.”  But manufacturing cell phones or producing fast food has little in common with managing the care of patients in all their variability and with all their unpredictability. No clinical metric conforms to a six sigma standard; we in medicine are barely comfortable defining “normal” with a 95% confidence interval and we are fully aware that “outliers” can be clinically normal and those in the normal range can be abnormal. Forcing the clinic into the six sigma paradigm is worse than an exercise in futility; it is an exercise in iatrogenicity. To wit:

It was the end of a long day in clinic a couple of weeks ago. I found myself sitting next to a colleague, a young, talented and well-read academic orthopedic surgeon. Each of us was staring at a computer screen displaying our “Physician’s Dashboard”, the gateway to the granularity-generating Electronic Medical Record (EMR) purveyed by EPIC. EPIC is a privately-held, exceedingly profitable company that dominates the multi-billion dollar EMR marketplace. Both of us are novitiates with the interface; UNC Hospitals only recently “went live” with EPIC at great cost in terms of personnel time and in terms of hundreds of millions of dollars. I was grumbling as I fought with the program to create a clinically useful medical record for the sake of the patients I had been seeing that day. My surgical colleague urged me to create templates and “smart sets”; he found EPIC easily manageable thanks to these short-cuts to record keeping which took advantage of any and all aspects of patient care that were stereotypical.

First, I took him to task on behalf of my patients. I explained that for a rheumatologist like me, and all others who are labeled “cognitive” specialists, the care of the patient revolves around the “granularity” of the narrative. I needed to listen actively to my patient to understand the particular idioms that populate any narrative of distress. And I need to communicate the approach the patient and I would take to sorting out the illness and formulating management to all involved in this patient’s care. It is an exercise that takes advantage of individual differences and idiosyncrasies rather than pummeling narratives into common denominators. I never assume the average, nor do I ever presume the outcome. Templates and “smart sets” are anathema.

Next, I took him to task on behalf of his patients. Most “interventionalist” physicians and surgeons think in terms of the average, the usual, and presume the outcome. It’s the old adage about having a hammer. However, this is a mindset that is a disservice to patients seeking elective orthopedic solutions. One of the greatest advances in clinical medicine in my 45 years as a physician is not a product of laboratory science or translational research; it’s a product of life-course (“social”) epidemiology. We are now advantaged by a rich literature addressing the experience of “health”. It turns out that many a person in the community is coping effectively with the same physical predicaments that others find so overwhelming as to feel the need to seek medical advice. That’s true of heartache, headache, heartburn, and so much more including the experience of regional musculoskeletal pain, which are the most frequent complaints voiced in the primary care setting and the reason for most orthopedic consultations. I coined the term “regional musculoskeletal disorders” 30 years ago to denote localized musculoskeletal discomfort and impairment experienced by working-age adults who were otherwise well and who suffered no traumatic precipitant. Regional musculoskeletal disorders are intermittent predicaments of life. If one goes a year without an important episode of low back pain, that’s abnormal. For knee or neck/shoulder/arm regional disorders, we should not be surprised by an episode during the course of 3 years. It was clear even 30 years ago that a central, often the central clinical issue was not the region that hurts, but what is it about this episode that proved so overwhelming that this person chose to be a patient?

This perspective has revolutionized our understanding of the illness that presents as back pain. However, I chose knee pain as the object lesson for my colleague. We’ve long known that elderly women with regional knee pain who remain people cannot be distinguished from similar women who choose to be patients with knee pain by the quality of the pain, the consequent incapacity, or by features discerned on physical or radiologic examinations. Affective challenges are far more likely the distinguishing features. There is a prominent case report that is relevant to this insight. It is tucked into the August 13, 2012 issue of the New Yorker, in “Big Med,” an article written by Atul Gawande. I have had numerous interchanges with Atul going as far back as his earliest article in the New Yorker in 1998, an article titled “The Pain Perplex.” Atul is a distinguished endocrine surgeon, a brilliant student of health policy as it relates to surgical practice, and an outstanding commentator. In “Big Med” he calls for standardization of medical practice, particularly surgical practice, in a fashion he likens to the line chefs in the Cheesecake Factory restaurant chain. He offers an example of the value of such standardization: the way he chose the orthopedist to operate on his recently widowed mother’s painful knee. No doubt, the procedure was performed in a technically exemplary fashion. Need I say more?

That brings us to the “forest” that is missed in any exercise that demands one parse clinical predicaments to create the granularity required to harvest “big data.” Nothing about this granularity can obviate the need for the person who chooses to be a patient to be advantaged by a discussion with the physician in the context of an empathic therapeutic relationship. “Big data” does not recognize or serve this relationship. It may facilitate invoicing, it certainly creates a new, self-serving industry, but it is a hazard for patient care.

42 replies »

  1. Thanks for wonderful article post. One of the best blog post I read in recent past. This post have great content. I can stand up and say superb! There is sufficient reason why you got more than thousand visitors to your article. 🙂 Certification Training

  2. Personally, I agree with the observerr above who laid the blame at the feet of Taylorism 2.0 Right-on, Bobby Gvegas!
    I can hardly pick on Deming, or even “To Err Is Human.” Deming said that 85% of mistakes in the auto industry were SYSTEMS mistakes. PERSONAL mistakes were far less – the mistakes when someone was not doing something right.
    After 15 years, it has turned itself inside-out. Barry says “. I hope they’re willing to use checklists and consistent protocols where appropriate like inserting a central line.” Errors in inserting central lines come from not enough experience, and not enough time. The endpoint of the central line is what matters – a functioning sterile access to the central venous system at the location anatomically desired. The protocols, however, put the cart before the horse very frequently. A second person with a checklist observing the procedure is entirely fine – but the systems have neglected to allocate the time for that second person to be there and not be doing something else. All the rules of the masters wind up with the summary, “OK we got 4 minutes to drop this line.” THAT is on the checklist for disaster.

  3. “I like your Christ, I do not like your Christians. Your Christians are so unlike your Christ.” MK Gandhi

  4. “One has to spend time to think and learn and experiment, which doesn’t fit the M.O. of most current doctoring.”
    Thanks for the back-of-the-hand, there, Rajiv. May I offer that audacity that medical information is supposed to supply something to the physician, not demand something from them.
    I am a physician, not an alimentary tract. My duty is not to perform machine-language differential rationality to a datastream.
    Perhaps the differential refinement of the Data Mass and its further peristalsis towards an infinite endpoint serves some reason to finer minds than mine – if only I might think, learn and experiment on my own, perhaps I might understand.
    But, as a bear of very little brain, I’ll give it a try to add my little iota of contribution to Lord Data. I am skeptical, though, as I am familiar with the products of peristalsis in other situations.

  5. Yes, and jolly good show for you! I was wondering who placed in the Becker’s Hospital Review in 2014. By the way – who is Becker, what are the grounds for his audacity to review hospitals, and who votes on HOPEA and VVPMS? How does one measure the Virginia Mason Product – and what sort of Product is it?
    The senseless quest for the Ideal Management Method is as perpetual as it is witless. It is of the People Magazine mentality, looking for the next boffo starlet in Hollywood – and what’s wrong with that Lindsay Lohan, anyhoo?
    The labeling of American Healthcare as the Major Business Problem of America virtually doomed it to extinction – one might note that there are very few opportunities in non-governmental manufacturing elsewhere, as it has been fairly well driven into the Rust Belt by the best minds of forward-thinking, move-ahead, Six Lean Sigma cowboys. If businesses could be redeemed by a sufficient plethora of gibberish, ours is second to none.

  6. This article fails to include examples of health care systems or hospitals that have already implemented lean quality practices like Virginia Mason Medical Center in Seattle. (From their web site)

    In 2002, Virginia Mason embarked on an ambitious, system-wide program to change the way it delivers health care and in the process improve patient safety and quality. It did so by adopting the basic tenets of the Toyota Production System (TPS), calling it the Virginia Mason Production system, or VMPS.

    Virginia Mason is ranked as one of the best hospitals in Washington state for the fifth consecutive year by U.S. News & World Report, which has also designated it as a national high performer for cancer care; gastroenterology and gastrointestinal surgery; nephrology; orthopedics; and urology.

    Virginia Mason was today named a recipient of Healthgrades Outstanding Patient Experience Award for the second consecutive year. This distinction places it among the top 10 percent of hospitals in the nation for patient experience.

    Virginia Mason is included on the Becker’s Hospital Review list of the “100 Great Hospitals in America” for 2014. The industry publication today released its compilation of some of the most prominent, forward-thinking and focused health care facilities in the nation.

  7. That’s a great Deming primer. Thanks.

    I expect you agree that there is enormous potential for better health based on much richer data than we’ve had in the past, just used much more wisely than insisting that each individual match population “norms”.

    In the quantified self community, many are using new technologies (as well as pen and paper) to gain knowledge about their own situations and use that to improve their health and wellbeing (however they themselves define their goals). Sometimes people collect a great deal of data, with very fine granularity. It’s not at all easy, but sometimes people are able to make great improvements because of their data. Unfortunately, the professional medical community is not prepared to make use of such data. One has to spend time to think and learn and experiment, which doesn’t fit the M.O. of most current doctoring.

  8. Six Sigma has largely become a business. I’m surprised the phrase hasn’t been patented. I laugh when I see people touting their “Lean Startup®” or “Just Culture[TM]” business lines. I love the “just culture” concept (mutual respect, mutual accountability, systems thinking), but do I owe someone a residual every time I say or write that phrase? 😉

    I am always reminded of the late Alan Watts’ jibe regarding “a religion ABOUT Jesus rather than the religion OF Jesus.”

  9. Deming didn’t like terms like TQM and he probably wouldn’t have liked terms like Lean or Six Sigma either. That said, Deming’s management mindset is a huge part of the Toyota philosophy and Lean (although far too many Lean people have learned tools and not that underlying philosophy, hence the trouble). Deming’s statistical methods are part of Six Sigma, but Deming would have been pretty appalled by many of the GE management practices that are associated with Six Sigma (requiring an ROI, belt certifications, firing the bottom 10%, etc.).

  10. Although Deming deserves credit for the “Six Sigma” paradigm, he also deserves credit for the Deming/Shewhart Cycle and for realizing, early on, the provisos Rajiv is emphasizing above.

    My old friend, and leading industrial psychologist, Brooks Carder, has taught me that about Deming, and lots more: http://asqhdandl.org/uploads/3/3/3/8/3338526/leadership_and_profound_knowledge_final_b_opt.pdf

    It is the “quality agenda” policies that take Six Sigma as the standard to be applied to the care of patients. It is this agenda that I am targeting. I am not targeting Deming.

  11. I agree whole-heartedly with your point, but criticizing Demming and a “systems approach” is off-target. We should be criticizing *superficial* applications of Demming and systems, while encouraging deep understanding of the ideas.

    For example, Demming taught that blaming workers for errors and defects was almost always the wrong thing to do. The problem almost always lies instead with poor processes, poor tools, bad management, bad design, etc. With this in mind we’d do well to stop blaming doctors for non-hand-washing and stop blaming people (aka patients) for non-adherence.

    Similarly a deep appreciation for complex systems (including appreciating how complex a system our health is) leads us away from simplistic ideas of controls and norms. In “Thinking in Systems”, Donella Meadows writes, “Systems thinking leads to another conclusion, however, waiting, shining, obvious, as soon as we stop being blinded by the illusion of control. … We can’t impose our will on a system. We can listen to what the system tells us, and discover how its properties and our values can work together to bring forth something much better than could ever be produced by our will alone.”

  12. Not to detract from the main point of the article, but the opening paragraph is a little misleading when you consider that the ICD10 code set used in Europe bears very little resemblance to the ICD10 code set that will eventually be implemented here in the US. They really are not even the same thing considering that the US will be using over 68,000 codes for diagnosis plus another 76,000 procedure codes. The international code set is 14,400.

  13. Anything that Nortin Hadler writes is worth attending to. Again he nails it with this blog. I am proud to call myself among the founding “Hadlerians”

  14. @Mark Graban -“Lean is conceptually simple, but I wouldn’t underestimate how difficult it can be… ”

    Nor do I, having worked on Lean initiatives in clinics when I was with the REC. All I said was it’s quite simple conceptually. In psychosocially toxic work cultures, it will likely prove problematic. As I observed on my blog:

    “My recent posts have ruminated on what I see as the underappreciated necessity for focusing on the “psychosocial health” of the healthcare workforce as much as focusing on policy reform (e.g., P4P, ACOs, PCMH), and process QI tactics (e.g., Lean/PDSA, 6 Sigma, Agile), including the clinical QI Health IT-borne “predictive analytics” fruits of “”Evidence Based Medicine” (EBM) and “Comparative Effectiveness Research” (CER). Evidence of psychosocially dysfunctional healthcare organizational cultures is not difficult to find (a bit of a sad irony, actually). From the patient safety-inimical “Bully Culture” down to the “merely” enervating emotionally toxic, I place it squarely within Dr. Toussaint’s “8th Waste” (misused talent).”

  15. Would this work?

    Study web sites that you like and which have about as many bits of information as an average patient chart over a course of years. Be sure they can handle video (for echoes, e.g), audio, text and have good search engines. Be sure the site example you like can add, correct and subtract data and time stamp it. Amazon is a great site. Have your programmer copy this chosen templet and put each patient on his own web site. Encrypt those portions the patient wants.
    Publish the public key and give the private key to the patient, who may give it to his insurer and his doctor and biller.
    Let google do the searching. You don’t care if all the patients have a different format. And you don’t care if everyone uses the same format.

    You have to use a discharge summary supplied by the payer.

    If anyone wants to study big data in your patients they have to do the programming. You can give them metadata so they can access de-identified patients’ charts. Patients would have to give keys for anyone to study what they encrypted.

  16. I’m just curious but are hospital CEO’s typically physicians? From what I’ve seen MD’s are at the top of administration more often than not.

  17. Hospital administrators will speak any language that allows them to optimize payer mix and maximize reimbursement. It’s their mantra and repeated robotically. Speaking “Lean” or Six Sigma or anything else is to them like learning a few phrases of a foreign language. It makes them seem oh-so-cool (mostly to themselves), but it doesn’t mean anything because it doesn’t really help them pursue the only two things they care about. See first sentence.

  18. Barry what you are talking about is not the problem with EHR’s. Physicians themselves have sought these type of protocols.

    However, what happens when one has a good protocol for his particular needs and is told he has to use another protocol that doesn’t work?

    Think of it this way. There are accidents in the operating room transferring things and maintaining sterility. Let us say that some genius comes up with a protocol to handle how things are passed around and what should be held in what hand which would lead to less complexity. It may work for many but what happens when you have a left handed surgeon?

  19. Lean is conceptually simple, but I wouldn’t underestimate how difficult it can be… teaching the concepts is just the starting point. Changing a culture and learning new management habits is another thing.

    I’m not trying to hijack the discussion here, but if Lean were really that simple, more hospitals would have transformed themselves by now.

    Lean Thinking and Just Culture are VERY compatible, conceptually and practically speaking. We don’t want this to be Taylorism 2.0 where experts tell people what to do and people are just pressured to work faster. That’s not Lean.

  20. Lean is conceptually simple, and can be rather quickly taught to any high school graduate staffer. “Six Sigma” is old wine in a new bottle where service industries are concerned. Priesthood-led sophistry.

    Absent a Just Culture, any of it runs the risk of simply devolving into productivity treadmill Taylorism 2.0. See Simon Head’s book “Mindless.”

  21. What you describe is just good management, no lean, six, or sigma required.

    But, with CMS, insurers, and hospital corps running the show, all with different incentives and goals, good management is a pipe dream.

  22. Agreed. Many (most?) of the hospitals out there who have learned to “talk Lean” are not really walking the walk, unfortunately.

  23. Agree. It is easy to chant lean six sigma. It is another thing to live up to it or, heaven forbid, actually hire up to it.

  24. “If patients were widgets, if care givers were production workers…” and if we were in a Lean management environment:

    1) care givers would have enough time to do their work
    2) supplies and medications would never be missing
    3) care givers would have the right level of helpful support from managers
    4) processes would be mistake proofed (rather than relying on individuals to be careful) and quality/safety would be far better
    5) care givers would have their ideas listened to and everybody would be able to participate in improving the care that’s provided and the work that’s done

    I’d encourage people to understand Lean (and maybe even Six Sigma) rather than just bashing it as some sort of knuckle-dragging manufacturing thing… figuring out HOW to apply these helpful principles to healthcare rather than dismissing it out of hand.

    It’s hard for anybody to defend the current levels of quality and safety that are being achieved through traditional healthcare management methods. We have to try something different.

  25. Life improved for millions of years without any measurement or administrative supervision. Improvement might be natural without all the self appointed overlords trying to siphon wealth and power out of the system.

  26. “Interoperability” is a misnomer anyway. Google “Interoperababble update” for my blog post on it (THCB continues to block my URL).

    “As a patient, I think interoperability would be a good thing because it would allow any provider to access my records and history if they have compatible technology. It’s unfortunate that EPIC, the industry leader, doesn’t see it that way.”

    Yeah. But transparency is the inverse of margin. Efficient Markets hypothesis 101. Opacity + barriers to entry = profit.

  27. I don’t know much about Big Data in this context. However, as a patient, I hope doctors don’t consider every case so unique that they’re not willing to follow standard protocols like a time out before a surgical procedure to ensure that they have the right patient in the room and will operate on the right body part and that every team member is present. I hope they’re willing to use checklists and consistent protocols where appropriate like inserting a central line. I hope that when something goes wrong, the team will undertake a root cause analysis to understand the cause and, if necessary, implement appropriate changes. As Paul Levy would say, hospital management should be “hard on the problem and soft on the people.”

    As for EPIC, my understanding is that it has chosen to keep its technology platform closed and proprietary rather than make it interoperable with non-EPIC systems. As a patient, I think interoperability would be a good thing because it would allow any provider to access my records and history if they have compatible technology. It’s unfortunate that EPIC, the industry leader, doesn’t see it that way.

  28. The UNC and the Duke EPIC systems do not communicate across 8 miles of geographic distance.

    The inmates are running ….

  29. “Nothing supplants compulsive carefulness on the part of the ordering and the administering practitioners”

    And that mental attitude, of compulsiveness and carefulness, is almost impossible when dealing with MU clicking, CPT bullets, and endless scrolling and screen changing.

  30. “If patients were widgets, if care givers were production workers, and if caring conformed to “six sigma” principals [sic], even slight deviations from standards of care would be as easy to recognize, those responsible could be singled out for improvement, and remedies would be obvious.”

    A great passage.

    My favorite EHR story is that two major health systems in St. Louis both bought the same product. They each made so many modificiations that the two systems now cannot communicate with each other, even though they were the same out of the box. Great thinking. Wonderful benefit for the patients who get referred from one to the other.

  31. Just one person taking responsibility for a comprehensive look at the patient anywhere in the chart is a blessing these days. It’s like no one takes responsibility anymore. But I guess that is “team based care” for you.

  32. Remarkably little information has been made public as to the efficiency and efficacy of EPIC. However, there is some related to EPIC and to other forms of CPOE. Caveat emptor. This is very far from error free. Nothing supplants compulsive carefulness on the part of the ordering and the administering practitioners.

  33. The only hope for finding real information in Epic is to sift through and hope you can find an actual transcribed dictation from a primary physician or consultant who knows the patient. The rest is cut and pasted repetition of quasi facts that often are simply wrong.

    I’m fortunate enough to be an anesthesiologist who does big, complicated cases, and usually only 3, at most four a day. I have the luxury of going through the Epic records, reading old operative reports and old consult notes, and correcting a lot of what is on the lists as the patient’s medical and surgical history. That way, if I happen to take care of that patient again, I know that at least what is there has been personally verified as far as possible. The errors (both of omission and commission) I find are astounding.

    The computerized order entry has merit in terms of legibility and avoiding gross dosage errors. If you tend to write the same orders a lot, creating your own template is quite efficient. Otherwise, the EHR is a travesty in the sense of garbage in, garbage out; the front-line clinicians are simply too rushed to deal with it.

  34. “The driver for all this is the notion that “health care” can be brought to heel with a “systems approach.”

    I’ve been shouting the fallacy of this premise at national meetings and have had little luck against the managerial tyrannosaurus with the most predictable (and predictably dull) quip “if you can’t measure you can’t improve. Measure, measure, measure!”

    I’m glad I’m not alone.

  35. “f patients were widgets, if care givers were production workers, and if caring conformed to “six sigma” principals [sic], even slight deviations from standards of care would be as easy to recognize, those responsible could be singled out for improvement, and remedies would be obvious.”


    Good post. I will be citing it.