I feel like most people would agree with the statement that American healthcare is inefficient. Because of this, our lab began using a big data approach and data mining to see where we might be able to make effective improvements. We wanted to identify patterns in department and hospital workflows with the expectation that we would find clear targets for optimization.
Often, however, the things we thought were wasted moments – like delays in access to advanced imaging like MRIs or delivering medications - weren’t a waste at all. They were essential to assessment and planning that may have prevented downstream complications. A physician taking a few extra minutes before administering clot-busting drugs to a stroke patient wasn’t being slow but instead were gathering information to avoid causing dangerous bleeding. The apparent delay in giving fluids to a patient with a critical infection? That may have represented crucial time spent determining whether aggressive fluids would overwhelm a failing heart.
In evaluating the data, we began to see hospitals not as assembly lines to be perfected, but as complex, adaptive systems. They are dynamic networks that must constantly balance efficiency with adaptability. This dual nature creates a fundamental tension, especially in the emergency department, where any single-model approach to improvement will almost certainly fail. This tension plays out in ways both subtle and profound, shaping everything from national policy to individual patient experiences.
Embarrassingly, I experienced this conflict firsthand while trying to arrange fertility testing. At a healthcare system optimized for high-volume, predictable primary care (let’s call it Two Medical), my request created chaos. I faced wrong appointments, inappropriate testing facilities, and staff unable to navigate anything outside their standard workflows. This ludicrous situation ended in a broken sample container and a need for urgent dry-cleaning. Sigh.
The system had been streamlined for efficiency but, in doing so, had lost its ability to handle edge cases. Frustrated, I restarted my testing at a major academic medical center. Explaining what I needed, a clerk at the front desk figured out a solution in real time. After finding the correct department, a nurse then was able to locate an order and necessary supplies to the problem. Sure, it was probably less efficient with regards to the organization’s flowchart, but the staff was able to handle the unexpected.
This difference isn’t about competence. It’s about whether a system has optimized away its ability to adapt, and the resulting repercussions.
Two Methods to the Madness
My father worked in industrial design for many years. After every doctor’s office visit, he always asks, “How can healthcare be so inefficient?” He loves McDonald’s, Panera, and Toyota and feels that maintaining a consistent experience is the highest honor a system can attain. But viewing hospitals as replicable experiences isn’t helpful. Instead, hospitals should be seen as a living, adaptive network, and doing so, reveals why they resist traditional optimization. Unlike a factory producing the same widget repeatedly, a hospital runs two fundamentally different systems under one roof.
The first system—which includes scheduled surgeries, elective procedures, and routine follow-ups—does look remarkably like a factory. A diagnosis is obtained weeks in advance. Resources can be planned, protocols standardized, and workflows optimized. When LEAN methodology, the efficiency principles Toyota pioneered for manufacturing, is applied here, it works. Studies consistently show LEAN improves on-time starts for elective surgery, reduces operating room turnover times, and increases utilization. One study found LEAN reduced perioperative time by over 12 minutes in spine surgeries. Another showed first-case on-time starts improving from 23.5% to nearly 70%. Predictability enables optimization.
But the emergency department operates in an entirely different manner. This is a world of managed chaos. Patients arrive with unknown problems that require diagnostic assessment, and all the intervention they need can be determined. No one can predict how many strokes will walk through the door, or whether tonight’s chest pain cases will be panic attacks or heart attacks. Accordingly, the system must constantly adapt to whatever arrives. But that does not stop most patients from asking why it takes so long for them to be seen.
When researchers apply LEAN to emergency departments, the results tell a different story. Success comes almost exclusively in “fast-track” areas treating low-acuity, predictable cases—essentially carving out a tiny factory within the chaos. But attempts to optimize the core ED show mixed and inconsistent results. Systematic reviews note “remarkable difficulty in analyzing and reporting implementation” and “heterogeneous methodologies.” The improvements that do stick often depend heavily on local conditions, suggesting they aren’t universal principles but temporary workarounds.
COVID-19 demonstrated why this adaptability is paramount. Emergency departments suddenly needed wholesale transformation to create separate respiratory zones, convert wards into isolation areas, and see people outdoors or in tents. The system survived not because it was efficient, but because it retained enough capacity to reorganize when everything changed.
Efficiency Can Kill
The pressure to resolve this issue in favor of efficiency is immense, driven by demands for cost containment and rising patient expectations. After all, research has shown that the single biggest predictor of patient satisfaction is how quickly they are seen. But this relentless pursuit of speed has a dark side. Safety researcher Erik Hollnagel formalized this as the Efficiency-Thoroughness Trade-Off (ETTO) principle: healthcare workers constantly choose between being efficient and being thorough, and “it is not usually possible to maximize both.” When organizations overemphasize efficiency, safety suffers.
More troubling, this trade-off can be lethal. A study of Italian hospitals in Lombardy found a direct and measurable relationship between efficiency and mortality. Analyzing ward-level data, researchers made a chilling discovery: more efficient hospitals have higher mortality rates.
This relationship held, even after controlling for patient characteristics and hospital procedures, with the effect particularly pronounced in general medicine and surgery. The data also revealed an exception that proves the rule: oncology wards that were more efficient actually had lower mortality rates. Cancer treatment, with its highly protocolized chemotherapy regimens and predictable side effects, behaves like a factory process and can indeed be optimized. But for the variable, assessment-first work of emergency and general medicine, efficiency comes at the cost of lives.
Our own data mining work supports this. Algorithms flag certain clinical time as “inefficient,” but when we investigated further, that time is often essential. The physician who seems to be “delaying” treatment is actually gathering information needed to choose the right one. The workarounds that look inefficient are often just healthcare workers being thorough.
The Inflection Point
The theory of dynamic systems warns that optimizing for a single variable can push a system past a critical threshold, triggering a “regime shift”, that is, a fundamental transformation to a degraded state that is difficult to reverse. Worryingly, nobody has identified where that threshold is in a hospital. One study looking at hospital IT investment found a U-shaped relationship suggesting diminishing returns beyond a certain point. But that’s about technology, not operational design. No one has quantified the efficiency-adaptability inflection point for all the hospital’s operations.
The COVID pandemic suggests we have already come dangerously close to that line. The adaptations required were so dramatic precisely because so much organizational slack had been eliminated in the name of efficiency. Emergency departments that had once absorbed unexpected surges by flexing staff and repurposing space instead needed to undergo wholesale reorganization, scrambling for basic supplies.
How to Choose the Right Goal
Perhaps the reason healthcare isn’t as efficient as it could be is that we’ve been asking the wrong question. We’ve assumed efficiency is the goal, and that any system can be made more efficient if we just apply the right methodology: LEAN, Six Sigma, or now AI-driven optimization.
But our research suggests something altogether different: hospitals must maintain some of that inefficiency because it is actually adaptive capacity in reserve. The “extra” time physicians spend gathering information, the “redundant” staff who can be redeployed during crises, the “underutilized” spaces that can be converted when needs change aren’t bugs to be engineered away. They are features that allow the system to handle what cannot be predicted and to move smoothly through the complexities of the healthcare system. As I was reminded during a recent resident simulation case, “Slow is smooth, and smooth is fast.”
“How do we make emergency departments more efficient?” isn’t what we should be asking. Instead, we should focus on which parts of the hospital are factories to be optimized, and which are adaptive systems that must maintain flexibility. Because they are vital for both productivity and safety. The next breakthrough in healthcare operations will be one that understands that some systems need to remain “inefficient” for patients to stay alive.