Medical Errors, Or Not

Nortin HadlerIn a recent post, the renowned neurologist, Martin Samuels, paid homage to the degree to which uncertainties create more than just anxious clinicians, they can lead to clinical errors. That post was followed by another  by Paul Levy, a former CEO of a Boston hospital, arguing that the errors can be diminished and the anxieties assuaged if institutions adhered to an efficient, salutary systems approach. Both Dr. Samuels and Mr. Levy anchor their perspective in the 1999 report of Institute of Medicine Report, “To Err is Human”, which purported to expose an alarming frequency of fatal iatrogenic errors. However, Dr. Samuels reads the Report as a documentation of the price we pay for imperfect knowledge; Mr. Levy as the price we pay for an imperfect organization of health care delivery. These two posts engendered numerous comments and several subsequent posts unfurling one banner or the other.

I crossed paths with Dr Samuels a long time ago when we were both speakers at a CME course held by the American Geriatrics Society and the American College of Physicians. I still remember his talk for its content and for its clinical perspective. His post on THCB is similarly worthy for championing the role of the physician in confronting the challenge of doing well by one patient at a time. Mr. Levy and his fellow travelers are convinced they can create settings and algorithms that compensate for the idiosyncrasies of clinical care. I will argue that there is nobility in Dr. Samuels’ quest for clinical excellence. I will further argue that Mr Levy is misled by systems theories that are more appropriate for rendering manufacturing industries profitable than for rendering patient care effective.

Errors in process, obvious errors such as giving the wrong dose or leaving a sponge in the abdomen, all of which are incontrovertibly errors, are not the preponderance of preventable medical errors in general and preventable fatal medical errors in particular. Most of the fatal errors that led to the horrifying statistics that elicits national outrage were judgment calls, defined as errors either by peers reviewing medical records or by virtue of the voluntary reporting of errors by doctors and hospitals. Not long after the National Academies published “To Err is Human. Building a safer health system”, a study from the US Veteran’s Administration demonstrated that the preventability of hospital deaths due to medical errors was very much “in the eye of the reviewer.”[1] A panel of 14 board-certified, trained internists undertook multiple structured reviews of the records of 111 hospital deaths, accumulating 383 reviews. They were measuring whether the deaths were “preventable by better care.” About a quarter of the deaths were rated as possibly preventable by optimal care. However, a tiny minority would have left the hospital alive had optimal care been provided. The reviewing clinicians estimated that only 0.5% of patients who died would have lived 3 months or more in good cognitive health if care had been optimal representing 1 patient in 10,000 admissions to the study hospitals. In-hospital deaths are largely the fate of very ill patients suffering from diseases in terminal phases, or suffering from multiple confounding conditions simultaneously, or the frail elderly. Many of these patients die in intensive care units where much is happening to them quickly, often under the pall of desperation, and where errors are usually apparent in retrospect rather than in the heat of the moment.

The American way in dying in hospital demands a national discourse. This is the setting where most “errors” occur, errors of commission in the desperate attempt to “save” the lives of the elderly, the frail, and the terminally ill. The message that merits wide debate is whether this is an appropriate way to die in America. My colleagues estimate that at least 40% of the patients in our many intensive care unit beds at any given time have terminal illnesses. The best we can do with “optimal care” is to prolong their dying. The other 60% may benefit from all the hustle, bustle and hassle of intensivist medicine, perhaps in part because their biology is more forgiving of occasional “suboptimal” care that eludes the checklists and other systems safeguards.

America has made a tremendous investment in intensive care units. We have many times the ICU beds per capita as any other resource advantaged country, 25 per 100,000 people as compared to 5 per 100,000 in the United Kingdom. Not surprisingly, when we build them we also build the demand, so-called demand elasticity.[2] The indications for admission in America result in a very different case-mix than anywhere else. We need ICU beds for patients with acute or potentially reversible conditions, but do we need them for the frail elderly or the terminally ill? Maybe the error is not so much in their medical treatments as in the lack of appreciation of their humanity.

The specter of avoidable complications, including avoidable deaths, has permeated notions of health care reform for 25 years. Blame has been spread thickly and widely under the rubric “human error”. Solutions have come in a torrent of regulations aimed at improving human performance. Some have proved ineffective if not counterproductive, such as restricting trainee hours which has fragmented care and toned down intellectual rigor. In the torrent of regulations are attempts to supplant human performance with computerized algorithms – with consequences that were intended and unintended. The emphasis on the “error” in human error is readily defensible even though the notion that “error” is an epidemic is indefensible. Even more indefensible is the fashion in which emphasizing “error” has come to deemphasize the “human” of human error.


Most hospital administrators, thanks largely to the minions they hire and pay as consultants, burden and brandish Lean, six sigma, and similar rallying cries in facing the regulatory avalanche in the wake of “To Err is Human.” In a report a year later[1], the same Committee detailed the changes in the system of health care that were necessary to effect a 50% reduction in these errors within 5 years. The report called for Congress to establish a Health Care Quality Innovation Fund of some $1billion to “produce a public-domain portfolio of programs, tools, and technologies…and to help communicate the need for rapid and significant change throughout the health system.” This funding had to wait for the “stimulus” legislation of the early days of the Obama administration. But lesser projects were funded by many. Despite a decade of these efforts, the Office of Inspector General for the Department of Health and Human Services said in 2010 that adverse events contributed to the deaths of 180,000 Medicare insured patients the previous year and 10-times as many suffered from a non-fatal adverse event.[2] By 2013, we were learning that preventable medical errors had climbed to the #3 cause of preventable deaths in America.[3] Something is very rotten in…our hospitals or in our calculations or both. But the press, the public, some policy wonks and some researchers were convinced only of the former. This is a dialectic that precludes alternative explanations, including explanations that could promote patient safety.

Hence, the “quality” zealots hold sway to this day. But maybe they continue to miss the forest for the blighted trees. First off, let’s examine their definitions of avoidable errors more carefully. Some errors are clearly a reflection of the system of care gone awry: operating on the wrong patient or the wrong knee, forgetting to remove surgical instruments from the abdomen, switching blood samples, administering medicines to the wrong patient, faulty equipment, and many more errors of this nature should be avoidable by improving the delivery system. In fact, thanks to the efforts of many in health policy and patient advocacy, great progress has been made. There are checklists in the operating room, redundant labeling systems, and much more that should have greatly reduced the possibility of human errors of this nature with technological solutions. But progress has been slow. Between 2005 and 2011, the adverse-event rates in American hospitals did not decline for surgical patients or patients with pneumonia, only for patients with cardiac conditions.[1] More specifically, the implementation of surgical safety checklists into the hospitals of Ontario, Canada was not associated with a reduction in operative mortality or complications.[2] Lucian Leape is one of the most influential of those advocating for improving the quality of care and a principal voice in the National Academy’s “To Err is Human” polemic for a “systems” approach. He is inclined to dismiss these disappointing observations on the basis that they did not exclude the possibility of various forms of non-compliance.[3] Perhaps he’s correct. However, adjusting for the magnitude of the procedures and for the severity of the illnesses of the patients, operative mortality and 30-day post-operative mortality in acute care hospitals was less than 1% with or without a checklist. Furthermore, all surgical complications, fatal or non-fatal, afflicted less than 4%. These numbers are not trivial, nor are they alarming as they are likely to reflect the degree of desperation that drove the decision to operate in the first place. Routine procedures in well patients, such as hernia repairs or breast biopsies, do not carry anywhere near this risk.


So, I suggest all embroiled in this debate return to Dr. Samuel’s essay. All of us who strive to be trustworthy at the bedside share Dr. Samuel’s uncertainties, anxieties, and drive to provide the best care we can. For the sake of our patients, we need the collegiality that promotes peer review and that is the essence of continuing medical education (https://www.youtube.com/watch?v=iknU6XpYG5Q#t=363  ). I only wish the ever growing corps of hospital administrators realized as much. It’s what W. Edwards Deming called “profound knowledge”, which is as important to his legacy [1] as his pioneering insights regarding the relationship of quality to productivity in industrial systems.

[1] http://asqhdandl.org/uploads/3/3/3/8/3338526/leadership_and_profound_knowledge_final_b_opt.pdf

[1] Hayward RA, Hofer TP. Estimating hospital deaths due to medical errors. Journal of the American Medical Association 2001; 286:415-20.

[2] Gooch RA, Kahn JM. ICU bed supply, utilization, and health care spending. An example of demand elasticity. Journal of the American Medical Association 2014; 311:567-8.

[3] National Research Council. Crossing the Quality Chasm: A new health system for the 21st Century. National Academies Press, 2001.

[4] https://oig.hhs.gov/oei/reports/oei-06-09-00090.pdf

[5] James, JT A New, Evidence-based Estimate of Patient Harms Associated with Hospital Care. Journal of Patient Safety 2013; 9: 122–8 doi: 10.1097/PTS.0b013e3182948a69

[6] Wang Y, Eldridge N, Metersky ML, et al. National trends in patient safety for four common conditions, 2005-2011. New England Journal of Medicine. 2014; 370:341-51.

[7] Urbach DR. Govindarajan A, Saskin R, Wilton AS, Baxter NN. Introduction of surgical safety checklists in Ontario, Canada. New England Journal of Medicine 2014; 370:1029-38.

[8] Leape L. The checklist conundrum. New England Journal of Medicine 2014; 370:1063-4.

[9] http://asqhdandl.org/uploads/3/3/3/8/3338526/leadership_and_profound_knowledge_final_b_opt.pdf

Nortin M. Hadler, MD is a graduate of Yale College and Harvard Medical School. He joined the faculty of the University of North Carolina in 1973 and has been a professor of medicine and microbioogy/immunology since 1985. His assaults on medicalization and overtreatment appear in many editorials and commentaries and 5 recent monographs: The Last Well Person (MQUP 2004) and UNC Press’ Worried Sick (2008), Stabbed in the Back(2009), Rethinking Aging (2011) and Citizen Patient (2013).

Spread the love

Categories: Uncategorized

Tagged as:

15 replies »

  1. My earliest exposure to the risks patients face came from reading a letter of reference written by Physician A who really wanted Physician B to take his substandard practice elsewhere…but wasn’t “allowed” to say anything negative due to collegial protocol. The letter contained subtle wording that would, according to Physician A, signal Physician B’s incompetence (without actually expressing it). In other words, they were kicking the can down the road, arguing that there was nothing they could do to fix the underlying problem.

    It is reductive to focus only on the fates of extremely ill patients when it is well-documented that errors in the practice of medicine impact a significant cross-section of patients from the OR to the ED to the PCP’s office. The problems that put those patients at risk are much greater than bad apples and bad software. If we want physicians, and the systems they work in, to identify and fix the underlying problems that cause real errors and do real harm, then saying that the mediocre results from the first wave of technical solutions prove it’s a fool’s errand, and that cognitive processes can’t be improved, is once again kicking the can down the road.

  2. PaulLevy said ” “Learn from and (over time) eliminate variation arising from the professionals, while retain variation arising from patients.” Is there really anyone reading this that finds such a concept anathema? If so, please address that matter directly.

    Rocky Bilhatz said: “My own experience has often been that external regulations from a larger central authority (government) often impair this type of leadership’s true flexibility.”

    I think Rocky’s comment is excellent, but I would go a step beyond what he said.. There has been so much control flowing upward to government and insurance and think tank social engineers that it appears inevitable that the QI movement will be seriously messed up by top down dictates, and the mess created will likely be worse than the positives from locally controlled implementation.

    • Don’t give up hope. There are lots of places that are making good progress on quality and safety improvement through locally directed efforts. Our philosophy was always, “We’ll do what we are required to do under the regulations, but we won’t let that distract us from doing the things we know make a difference.”

  3. It seems everyone wants to have it both ways. It is true that Dr. Hadler’s dismissal of QI is specious, because of the completeness of his dismissal. But is it not also true that the sure gains of QI do not mean that the science of determining a medical error is not shakier than an earthquake in San Francisco?

    It is true that standardization and judgment can co-exist. But intellectual honesty requires us to acknowledge the two vectors are not in the same direction. They are not a false dichotomy, but a tension. It does no good denying this tension. Simply saying: we want you to follow rules but use your judgment, is having one’s cake and eating it.

    Perhaps the greatest travesty is the conflation of the serious error with the marginal one, that is the one determined after the fact by a retrospectoscope more powerful than the Hubble telescope. This does two things. It makes one argue – that is the nature of hindsight. It also detracts from solving the serious errors. Such egalitarianism ultimately hurts the most avoidable, whilst doing little to hinder the least avoidable.

    • There is nothing wrong with completely dismissing a wholly bad idea. The notion that a bureaucratic system of unaccountable outside observers can deliver a meaningful QI program has no redeeming features.

    • Re: “Simply saying: we want you to follow rules but use your judgment, is having one’s cake and eating it.” Well, no. It is how you manage the very tension you correctly set forth.

  4. I agree that reducing inappropriate variations in clinical practice is a good idea. Giving antibiotics on schedule is a good example. No reason to vary from sound pharmacologic principles.

    But I agree strongly with the author on many points. Most in-hospital deaths occur for patients who are at the end of life. Many are in ICUs, or receiving heroic procedures to stave off the inevitable–TAVR procedures on extremely aged patients are prime examples. The likelihood of some type of error, significant or not, escalates with the number of medications and interventions done, and many of these take place in the last weeks to months of life. Too often, the last month of life has no quality whatsoever, but no one is willing or authorized to deescalate care. So of course there is an association between errors and deaths.

    Protocols are fine for concrete tasks like central line insertions. But in humans, multiple medical problems may result in protocols that directly conflict with each other. Liberal IV fluid administration may be appropriate for the patient with asymmetric septal hypertrophy under anesthesia, but completely inappropriate if the same patient has renal failure. If only the practice of medicine were as simple or predictable as many administrators and bureaucrats think it should be.

    • I remain concerned that you choose to use this kind of perjorative “many administrators and bureaucrats” statement. I’ve visited hundreds of hospitals around the world and have met very few administrators and bureaucrats who think the practice of medicine is simple or predictable. Why not accept the good that can occur with clinicians and administrators work toward the common purpose of reducing preventable harm? Why do you feel a need to dismiss those activities as less relevant?

      • The success of a quality program should not be judged solely on the basis of the measures it chooses to implement. For one thing, Campbell’s Law is operative in healthcare as it is in other social contexts. More importantly, there’s more to quality than the strictly measurable, and improvements in certain measures can (and do) affect quality in other, unmeasured areas. The present system does not hold itself accountable for the totality of its effects, but merely (and barely) for the goals it has set for itself. In such a scheme, declaring success is relatively easy.

  5. I’m sure I’ve made my own share of errors, but to say that I’ve been “misled by systems theories that are more appropriate for rendering manufacturing industries profitable than for rendering patient care effective,” simply ignores the evidence from places that have calmly and thoughtfully applied what is sensible from other fields to the area of clinical process improvement. I have seen it from personal experience at Beth israel Deaconess Medical Center, where our clinical leaders (Note: not administrators or bureaucrats) engaged in systemic process improvement. Simply put, they applied the principles of the scientific method to clinical process improvement. This led to a reduction in harm caused to patients in a variety of settings on the floors and units of our hospital. The same has occurred in many other places throughout the US and elsewhere in the world.

    It seems to me that Dr. Hadler doth protest too much to make his point about other aspects of the current health care environment which (I agree) result in doctors and nurses being asked to do things that do impinge on their good intentions, expertise, and training–and, by the way, also can interfere in a healthy patient-doctor relationship.

    Why he seems intent on denying the improvements that have taken place is a mystery to me. Why not acknowledge the good that can come from reducing unwarranted variation in clinical practice? Read my column again and note the conclusions set forth by Brent James. We are not suggesting that doctors should become automatons and become unthinking rule followers. Far from it. In Brent’s words: “Learn from and (over time) eliminate variation arising from the professionals, while retain variation arising from patients.” Is there really anyone reading this that finds such a concept anathema? If so, please address that matter directly.

    And, a final point. Our work at BIDMC was not based on some misreading of–or imperative from–the IOM studies from years ago. It was based on our Chief Of Medicine and other chiefs deciding that working together to try to prevent preventable harm was, simply, a good idea.

    • I found the blanket pejorative dissing of process QI professionals as “quality zealots” off-putting, serving nothing more than weakening an otherwise good and necessary argument. You cite Dr. Brent James. I trained in QI under him at IHC back in the 90’s. I also have to wonder how Dr. John Toussaint (also an MD) of ThedaCare, whose Lean work I have long admired and cited, would respond to such a transparent Straw Man proffer.

    • Regarding “Learn from and eliminate variation arising from the professionals…” in many ways, this speaks to Dr. Samuels’s article, cited at the start of this post.

      Do not forget that growth comes from the outliers, not the conformers. Many substantial advances in Medicine were initially at odds with academic wisdom. Seeking to eliminate variance, in this manner, would be downright harmful.

      Also unmeasurable in this discussion is the “tone” of patient safety organizers, the willingness (or admittedly, in a minority of cases, unwillingness) of ground workers to improve upon quality, and the inflexibility of policy establishers to adapt to the feedback provided from these ground workers.

      The latter point is arguably one of the greatest current challenges to collaboration, as the process of QI should actually be a very dynamic and flexible one. For example, say that a QI improvement policy is recommended from the top and implemented downward. Intended and/or unintended consequences result, causing it to be a complete flop. This policy must be immediately eliminated (not added to), hence, the importance of flexibility.

      Of course, the further removed the QI central authority is from the ground workers, the harder it will be to maintain this dynamic component.

      Your work at BIDMC (as described in your last paragraph) appears to have recognized the importance of this local environment leadership, and I commend you for that. My own experience has often been that external regulations from a larger central authority (government) often impair this type of leadership’s true flexibility.

      Kind regards,
      Rocky Bilhartz

      • Don’t give up hope. There are lots of places that are making good progress on quality and safety improvement through locally directed efforts. Our philosophy was always, “We’ll do what we are required to do under the regulations, but we won’t let that distract us from doing the things we know make a difference.”

Leave a Reply

Your email address will not be published. Required fields are marked *