Three weeks out of residency, and the morning schedule looked like any other. Eight patients. The first was a forty-four-year-old with low back pain — onset after lifting at work, six days of symptoms, no prior episodes, no red flags. The evaluation took twenty-three minutes. The hypothesis was generated in ten. Lumbar segmental dysfunction with a directional preference into extension. The treatment plan wrote itself: mechanical diagnosis, graded loading, functional restoration. The new graduate documented efficiently, moved to the next room, and was on time for the third patient of the morning.
This was the residency’s most visible product. Twelve months earlier, the same evaluation would have taken forty-five minutes. The same hypothesis would have emerged, but through a slower, more effortful process — each differential weighed deliberately rather than recognized. The difference was not additional knowledge. It was knowledge reorganized. The clinical reasoning had been compressed, patterned, and made efficient enough to serve a real-world caseload. Routine expertise.
The second patient arrived at ten-fifteen. Fifty-one years old, bilateral lower extremity pain, intermittent numbness, eight months of symptoms that had not followed any identifiable mechanical pattern. Two prior courses of physical therapy — one focused on core stabilization, one on neural mobilization — each produced modest short-term improvement that did not hold. The medical record showed a recent MRI read as “mild multilevel degenerative changes, no significant stenosis.” The patient described pain that moved, sleep that had deteriorated over the past year, and a work schedule that had shifted to nights three months ago.
The new graduate spent thirty-eight minutes on the evaluation. Hypothesis generation was slower — not because of uncertainty about what to test, but because the findings did not organize into a clean, mechanical pattern. There was some directional preference, but it did not explain the bilateral distribution. The neurological screening was normal, but the symptom behavior suggested something the mechanical framework was not capturing.
Here is where the residency’s less visible product appeared.
The new graduate did not file the discrepancy away. Did not round the findings into the nearest familiar pattern. Did not default to the approach that had worked on the morning’s first patient. Instead, the graduate treated the most defensible findings — the directional preference, the movement impairments that were clearly present — and added a note in the medical record: “Recommend close monitoring. Consider fellowship consultation if response is atypical.”
That note. That single sentence. That was the product.
The residency had not taught the new graduate what to do with this patient. That was never the residency’s purpose. What the residency had built — across twelve months of structured mentoring, case-based reasoning, and progressive clinical exposure — was something harder to see and harder to measure than clinical knowledge. It had built the capacity to notice when a case does not fit the pattern.
Hatano and Inagaki identified this decades ago in their foundational work on expertise development: the distinction between routine expertise and adaptive expertise lies not in the volume of knowledge but in how that knowledge is organized and deployed (Hatano & Inagaki, 1986). Routine expertise produces efficient, reliable performance on familiar problems. It is the forty-four-year-old with textbook low back pain — recognized, treated, documented in twenty-three minutes. Adaptive expertise produces the capacity to recognize when the familiar framework is insufficient — to feel the friction between the pattern and the presentation, and to preserve that uncertainty rather than suppress it.
The fourth capacity that routine expertise develops — and the most important one — is this emerging metacognitive awareness. The capacity to recognize when a case does not fit the pattern. To feel the friction. To preserve uncertainty rather than suppress it. That is the seed of adaptive expertise (Mylopoulos & Regehr, 2009).
Most clinical education systems do not measure this capacity. They measure knowledge. They measure procedural skill. They measure hours completed and examinations passed. The residency assessment in one ABPTRFE-accredited program measures something different — it crosses entrustment levels with complexity tiers, creating a developmental picture that can capture not only what the clinician can do independently but also where the clinician’s reasoning currently ends. The moment the new graduate wrote “consider fellowship consultation if response is atypical” was the moment the assessment architecture was designed to see.
The current healthcare system lacks a structural mechanism to honor this kind of clinical judgment. There is no billing code for noticing. No reimbursement differential for the clinician who recognizes the limits of their own reasoning and routes the patient toward someone with deeper expertise. Medicare reimbursement, adjusted for inflation, is roughly half of what it was thirty years ago (MedPAC, 2023). In that economic reality, the incentive runs the opposite direction — toward treating every patient with the tools you have, regardless of whether those tools match the complexity of the case.
The residency cannot fix the reimbursement structure. But it can instill a reasoning habit that the reimbursement structure has no mechanism to instill on its own. Bohle Carbonell and colleagues, in a systematic review of adaptive expertise, found that what distinguishes adaptive from routine experts is not just different knowledge but different knowledge representation — different cognitive flexibility, different analogical reasoning, different capacity to generate novel solutions rather than applying stored ones (Bohle Carbonell et al., 2014). The learning environments that foster this give individuals responsibility for developing their own solution strategies under supportive supervision. That is a precise description of what a well-designed residency does.
The data on what residency produces are consistent with this. Residency graduates report that training had a major positive influence on clinical reasoning — ninety-four percent for logical reasoning, ninety-five percent for thorough examination (Smith, Tichenor, & Schroeder, 1999). Briggs and colleagues found that residency- and fellowship-trained clinicians spend more time on teaching, mentoring, and research than non-trained peers — training shifts practice patterns, not just knowledge (Briggs et al., 2023). Fewer than 12% of DPT students apply for residency upon graduation (Osborne et al., 2019). The gap between what the training produces and how many clinicians receive it is the structural problem the profession has not solved.
And there is a nuance the data insist on. Rodeghero and colleagues, in a retrospective cohort of over 25,000 patients, found that fellowship training improved clinical outcomes — functional status and treatment efficiency — but residency training alone did not yield the same gains (Rodeghero et al., 2015). The residency builds the foundation. The noticing — the capacity to recognize when a case exceeds the clinician’s current reasoning architecture — is the seed. The fellowship is what grows it into the adaptive expertise that produces measurably different clinical results.
The system has built these training pathways. It has not built them at scale. It has not built the economic incentives that would make them accessible to more than a fraction of the workforce. And it has not built the clinical infrastructure that would allow the kind of complexity matching the entrustment framework makes possible.
The new graduate saw six more patients that afternoon. Four were straightforward — the kind of clinical work the residency had made efficient and reliable. Two required slower reasoning, longer evaluation, and careful documentation of what did not quite fit.
At the end of the day, the patient from ten to fifteen was still on the schedule for Thursday. The note in the medical record — “Recommend close monitoring. Consider fellowship consultation if response is atypical” — sat in the chart like a quiet signal. Not a diagnosis. Not a referral. A noticing.
The residency had not taught the new graduate what that patient needed. It had taught the graduate not to pretend otherwise. That capacity — the willingness to sit with what you do not yet know, to preserve the discrepancy rather than erase it — was the product. No knowledge added, but a reasoning habit installed.
