Vienna General Hospital, 1846 – a young pregnant woman arrives at the hospital, very apprehensive. Surprisingly, her concerns are not what one might expect - such as the pain of childbirth, worry about her baby, or even how long the delivery might be. Instead, she is literally begging on her knees not to be admitted to the ward named First Obstetrical Clinic; word on the street is that women who deliver there often end up dying. She pleads to be admitted to Second Obstetrical Clinic. Some women who anticipated being assigned to First Clinic after figuring out the hospital’s admitting rules had even resorted to giving birth on the street rather than chance admission and delivery in that ward. Indeed, this open secret is substantiated by the measured mortality rate of First Clinic, which is consistently 3-4 times higher than Second Clinic. Little wonder then that one of the doctors overseeing First Clinic distraughtly wrote, “It made me so miserable that life seemed worthless.”
That doctor was Ignaz Semmelweis, and in 1846, he was as yet a young physician, freshly minted from medical training. Determined to understand the reasons for the higher death rate, Dr. Semmelweis undertook a rigorous and exhaustive study of the problem. He meticulously studied and systematically eliminated every potential difference between the two clinics – exploring factors as diverse as climate, patient overcrowding, and even religious practice. Finally, the breakthrough came when he postulated that the higher mortality must be from puerperal fever (childbed fever) caused by “cadaverous particles” on the hands of the First Clinic medical staff who performed autopsies, since those in Second Clinic did not. His finding was exceptional even more so because it came years before Pasteur and Lister established the germ theory of disease. At the time of his discovery in April 1847, the mortality rate in First Clinic stood at 18.3%. In mid-May 1847, he instituted an antiseptic hand-washing policy. By June, the mortality rate had dropped to 2.2%, which was comparable to Second Clinic. In the following year, expanding his techniques even achieved months of 0% mortality. Needless to say, these were astounding results, measured not just as percentages on paper, but in the lives of each of the women who were saved by his efforts.
It was in my first year of medical school that I had learned about and been deeply inspired by Dr. Semmelweis. Having just graduated with a Bioengineering degree the year prior, my engineering mindset naturally embraced his thoughtful and rigorous problem-solving approach, which struck me as a quintessential example of practical clinical research. Unfortunately, delving deeper into his story also opened my eyes to a rigid closed-mindedness within the medical profession to adopt new concepts. What is less known is that despite Dr. Semmelweis’ immense contribution in saving the lives of his patients, he paid a heavy price for his commitment to doing the right thing. Tragically, he was shunned by the Vienna medical community of the time, who could not accept the fact that their own hands were responsible for their patient’s infections. Their defensiveness may have stemmed from feeling faulted even while they were hard at work serving these very patients. As a consequence, they simply rejected Semmelweis’ hand-washing dictum. These resistant physicians were unable to free themselves to adopt a new way of thinking, a phenomenon now fittingly named the Semmelweis Reflex.
Semmelweis Reflex is the reflex-like tendency to reject new evidence or new knowledge because it contradicts established norms, beliefs or paradigms.
Dr. Semmelweis fought an increasingly isolating battle to make his work accepted, even though the objective results of his work should have spoken for itself. His conviction and passion were misunderstood. He was eventually forcibly committed to a mental asylum, and in the ultimate tragic irony, the man who devoted his career to eliminating his patient’s infections died from infected wounds himself after a severe beating at the institution. In my own reflections over the years, I have always been struck by how an individual doctor - using keen perception and committed resolve - could make such a profound contribution. Thankfully, the climate today for doing the right thing in medicine is better compared to when Dr. Semmelweis practiced (although even now, over a century and a half later, hospitals still struggle to achieve 100% hand-washing rates!).
In June 2003, I was entering residency as a fresh-faced surgical intern. Serendipitously, a book titled “Moneyball” was published that same month. Aside from my love affair with medicine, I am also a huge sports fan (in fact the combination of engineering, sports, and medicine was what led me to choose orthopedic surgery as my specialty), so this was reading I devoured immediately. Moneyball chronicles the early days of analytics being used in baseball to drive better decision making, standing in contrast to the heuristics-based approach that coaches and scouts had used for generations. Heuristics are rules of thumb that allow us to make quick approximations in complex situations, especially where incomplete information is available. Billy Beane, then general manager of the Oakland A’s, embraced this sabermetrics approach over conventional methods, employing more objective tools and predictive analytics to create value and achieve a competitive edge for his organization. Like Dr. Semmelweis, it was by being open to a completely new methodology that he was able to achieve measurably better outcomes. The results sparked nothing short of a revolution that spread from baseball quickly throughout the sports world, where analytics are now ubiquitous. However, at the time, I could not help but be struck by how baseball experts also exhibited the Semmelweis reflex.
I had chosen the Bronx for my surgical residency to learn the skills of serving high risk patients and communities with complex and varied pathology. Observing my attendings artfully problem-solve case after case, I imbibed the heuristics being employed to make these expert judgements. Doctors synthesize their vast knowledge into heuristics for practical application in clinical decision-making. I could see why heuristics were helpful, especially when the quality of information available on each patient is highly variable and the time available to make decisions is limited. Doctors, like any other experts, almost never use just a single heuristic, and our training is focused on developing our minds to layer multiple heuristics and artfully weave them into the sophisticated decisions required of us. With progressively deeper specialization and greater experience, pattern recognition accelerates and physicians even start to take pride in these abilities, bragging in the same way that a baseball scout would about the right ‘gut’ or intuitive judgement on a player. Thankfully, heuristics generally tend to work and lead to appropriate decisions for patients most of the time. When exceptions arise, we treat them as just that - exceptions. But it’s only when we encounter these ‘exceptions’ that we realize that an over-reliance on heuristics in medicine can be limiting at best, and dangerous at worst. The seeds had been planted as I started to wonder whether human ‘gut’ intuition was constraining optimal medical decision-making in the same way that Moneyball had described.
While these ideas were germinating in my head, a lot else was happening in my life - I graduated from residency, completed a fellowship in minimally-invasive surgery, got married, was recruited to found an Orthopaedic Department, and then became Vice Chair of the Board at a ~200 physician multi-specialty group. Through all this, I relished direct patient care above all else. The negative corollary to being so personally invested in my patients’ outcomes was how hard I took even the slightest complication (which my wife would thoughtfully support me through). I remember one particularly illustrative case. A fit, muscular gentleman came to see me after experiencing ongoing knee pain from years of soccer and long-distance running. He looked much younger than his chronological age, and yet the number of miles he had logged on his knees had taken their toll. An old soccer injury had resulted in prior open reconstructive surgery of his right knee which was complicated by a postoperative infection. X-rays showed clear evidence of bone-on-bone arthritis in both his knees, so he was eager to know whether both knees could be replaced simultaneously. After discussing the technique of the operation, the expected recovery timeline, and how his rehab therapy would progress, he ultimately decided to proceed with surgery on his left knee first. This was ostensibly the lower risk option, so he surmised a faster recovery and sooner return for his right knee replacement. Arrangements were made, including a full medical evaluation by his primary care doctor, for which he received “clearance” to proceed with surgery.
On the morning of the surgery, operating to the calming sound of classical music, the procedure went exactly per plan. Like the well-executed orchestral rendition of Debussy’s Claire de Lune playing in the background, our team worked harmoniously to achieve technical perfection in the operating room (since our story started in Vienna, I could just as easily have referenced one of many Viennese composers - Mozart, Beethoven, or Strauss – but Debussy is one of my favorites for an early morning in the OR!). Pleased with how well the surgery had gone, I saw my patient walking on his new knee and in very good spirits later that day. However, when I evaluated him the following morning, his vital signs showed an elevated heart rate, and he was experiencing some mild shortness of breath. I immediately ordered a STAT spiral CT scan. When it confirmed my suspicion of a pulmonary embolism (blood clot in the lung), we were able to take prompt action, averting its progression to a fatal complication. Even so, this left me with lingering questions. What would have happened if we had gone ahead with the riskier simultaneous bilateral surgery? With today’s outpatient surgery capabilities, could he have been offered a potentially fatal option, unbeknownst to the surgeon and patient? What was the role and meaning of the medical “clearance” he received - shouldn’t that have identified a potential issue? Could there be a way to predict, and thereby potentially prevent, such a complication from occurring?
Over-reliance on heuristics in medicine can be limiting at best, and dangerous at worst.
There is a generally accepted idea in the medical community that a certain number of postoperative complications are par for the course; in fact, a famous surgical adage states - “the only surgeon who doesn’t have complications is the surgeon that doesn’t operate”. While I dedicated much time and energy to attending surgical conferences that focused on the newest surgical techniques, in tracking my own patients, complications tended to involve factors seemingly out of a surgeon’s control. Technical precision in surgery is important, yet it is only one dimension of what is necessary for an optimal patient outcome. What is the point of perfection in the operating room and delivering the best knee flexion if it could still result in a complication, such as heart attack, stroke or pulmonary embolism, for the patient? Semmelweis had discovered this in First Clinic - delivering a healthy baby was not an optimal outcome if the mother risked dying from an infection; women were begging for physicians to look at the totality of the problem.
Many surgeons assume that handing off “clearance” for surgery to the patient’s primary care doctor or to an anesthesia pre-op nurse addresses the issue of surgical risk. Not only do these fail to protect against litigation, but these approaches don’t address procedure-specific risks and comprehensive episode-level issues. Ultimately, only surgeons (with their patients) own the totality of the outcome. In fact, at the recent American Academy of Orthopedic Surgeons Annual Meeting, an entire session (titled “Is Medical Clearance Enough”) was dedicated to highlighting the deficiencies of this “clearance” approach. Indeed, newer episode-of-care reimbursement models have incorporated this holistic approach in their very design, as patient outcomes fundamentally depend on it.
Enormous progress has been made in the discovery and implementation of important medical, surgical and anesthetic measures that have materially reduced complications. For instance, consider the pulmonary embolism (PE) that my patient developed postoperatively. PE is one of two blood clotting conditions that come under the category of VTE (venous thromboembolism), the other being deep vein thrombosis (DVT). It was only as recently as the 1980s that knee and hip replacement patients were experiencing postoperative VTE rates ranging from 41% to as high as 85%. Since one third of patients develop long term complications after VTE, known as post-thrombotic syndrome, and 10-30% die within a month after their VTE diagnosis, this translated to unacceptable morbidity and mortality after knee and hip replacement surgery. Once this was understood, surgeons moved towards adopting standardized regimens for postoperative VTE prophylaxis for all patients, sharply reducing postoperative VTE rates to about 1-2%. This undoubtedly made a tremendous difference for patients. Once the low hanging fruit had been plucked, a prevailing sense amongst us surgeons has set in that most of what can be done for this problem is already being done; the rest were “acceptable” complications, par for the course. To move the needle further now requires a degree of consistency and nuance that pushes at the boundaries of human cognitive abilities. It is no longer as simple as implementing operational solutions, such as hand-washing or standardized VTE prophylaxis. High volume surgical facilities with a large number of surgeons are doubling down on heuristics with standardized guidelines, such as simplistic pre-op testing protocols, fixed anesthesia modalities, or routine antibiotic agents for all patients. While standardization is better than an ad hoc approach, and it can simplify and potentially enhance system-level efficiencies, it will never in and of itself lead to the most optimal outcome for each and every patient.
In difficult endeavors, the last mile is often disproportionately harder than the journey preceding.
Before long, I was hard at work systematically studying all the factors affecting each type of surgical complication. In those early days, I would go in to patient visits with pen-and-paper and record detailed notes by hand on spreadsheets to collect and weigh all risk factors objectively. Thousands of iterations and a few years later, I started seeing the difference that this data-driven and detail-oriented approach was making. I had to resist the urge to give up collecting data when a case seemed ‘obvious’ and be disciplined to avoid defaulting to my ‘gut’ intuition.
As it so happened, my patient came back sometime later to have his other (right) knee replaced. I would now use what I had gleaned from my studies to personalize his care. A typical plan of attack to mitigate the blood clot risk would have been to put him on stronger blood thinners after surgery. However, surgeons know that while this may reduce his VTE risk, the potential cost of this simplistic solution could be increased postoperative bleeding and hence higher risk of infection. Layering of multiple heuristics such as these can quickly become complex and subjective, particularly with incomplete data about all the risk factors. Decision making in such a context screams for a systematic data-driven methodology (a la Moneyball) to achieve the optimal balance of the various dimensions. It was the insights from this methodology that helped my patient avoid a complication, despite the right knee actually being the riskier of the two operations; it was the prior left knee operation using the old heuristics-based approach that had suffered the complication. These types of wins started to feel like my Semmelweis moment. Was there a way to do this consistently for every patient?
To move the needle further now requires a degree of consistency & nuance that pushes at the boundaries of human cognitive abilities.
The explosion of scientific knowledge has now pushed us against the limits of human cognition. Heuristics-based mitigation strategies can be useful in avoiding mistakes and get us in the ballpark, but are not effective in consistently achieving optimal outcomes. Dr. Daniel Kahneman’s Nobel Prize-winning behavioral model, Prospect Theory, explains why expert decision-making fails especially in the context of risk and uncertainty - a context that perfectly applies to surgical decision-making. Since the human mind processes things in terms of “expected utility” rather than absolute outcomes, heuristics bias us in a context of limited information to make decisions that may result in suboptimal or variable outcomes, despite our depth of knowledge and best intentions. We surgeons are not immune from the many well-established human cognitive biases and limitations - focalism, availability bias, base-rate neglect, probability neglect and recency bias, to name just a few. As Dr. Kahneman identified, the best decisions occur when we can delay intuition and judgement, and instead gather as much objective data as possible on the various dimensions underpinning a particular decision. This formed the conceptual basis for the methodology I developed - a hybrid creation melding the predictive analytics style of baseball’s sabermetrics revolution with decision science concepts from prospect theory and the clinical problem-solving approach of Semmelweis. Crystallizing in the form of Ready Surgery, today the methodology is continuously enhanced with the best new evidence by harnessing artificial intelligence. As long as we don’t fall prey to the Semmelweis reflex, the science of Surgical Risk Intelligence can get us through the last mile of making surgery safer.
Just a month ago, I seized an opportunity to make a personal pilgrimage of sorts to Budapest, home of Dr. Ignaz Semmelweis. It was a humble museum, on the grounds of his former residence on the Pest side of the Danube River, and on the day I visited, I was the only one there. I studied Dr. Semmelweis’ surgical instruments in the curiously studious way only a fellow surgeon probably would. Yet, the significance of the moment was not lost on me. For me personally, from the time I first came upon his story as a medical student, it has always served as a north star for my own approach to medicine. Not only was he an intelligent and thoughtful scientist, but he was committed and relentless in solving an important problem; when his convictions were tested, he stood up for doing the right thing for his patients. As I reflected alone in the museum courtyard after my visit, it was clear to me that while much has changed in medicine, and will continue to change in the years ahead, the imperative and quest of doctors to make a difference is truly timeless.
The author, Dr. Jaideep Iyengar, is an award-winning orthopedic surgeon who was first to perform robot-assisted joint surgery in Silicon Valley. He is also an engineer, published researcher, and the scientist behind Ready Surgery, an AI-Enhanced Surgical Risk Intelligence Platform.