It is 8:30 on a Tuesday morning in an outpatient orthopedic clinic. Two new patient slots have opened. Two referrals came in overnight, both flagged as routine. The scheduling coordinator checks who has an opening and places them.
The first patient is a forty-one-year-old software engineer with eight weeks of right knee pain that started after a weekend hike. Pain is localized to the medial aspect, worse with stairs and downhill walking. No prior injuries. No systemic symptoms. The MRI shows a small medial meniscus tear, possibly degenerative. The orthopedist wrote: Try physical therapy, consider arthroscopy if conservative management fails. The problem is straightforward. Mechanical, regional, classic presentation.
The second patient is a fifty-eight-year-old who arrives with right shoulder pain and left lower back pain that began simultaneously eight months ago, three weeks after starting a new chemotherapy regimen. The pain is constant, non-mechanical, with inconsistent findings on examination. This patient has altered sleep architecture from cancer treatment, takes six different medications with known musculoskeletal side effects, and carries a persistent fear that the pain signals a return of disease. Two prior episodes of physical therapy — both technically competent, appropriately dosed, well-structured — produced no meaningful change. The referring oncologist wrote: patient very anxious about recurrence, wants PT as part of comprehensive management.
The clinic has two openings. A fellowship-trained clinician with eight years of experience is available at 9:00. A clinician who graduated eighteen months ago — six months of independent practice — is also available. The coordinator assigns the software engineer to the fellowship-trained clinician and the oncology patient to the new graduate. The assignment was made on the basis of who was free, not who was prepared.
This is not a scheduling error. This is the system working exactly as designed.
The mechanism is efficient at matching logistics. It is indifferent to matching clinical capacity. And this particular Tuesday morning is not unusual. It is a portrait of what happens every morning, in every outpatient orthopedic clinic, across the country.
Consider what the schedule actually contains. The software engineer presents a problem that entry-level training was built to manage. The diagnosis is clear, the examination findings are consistent, the evidence-based protocol is appropriate, and the pattern recognition required is exactly what foundational education develops. A new graduate should see this patient. The match serves both well — the clinician builds reliable expertise with cases appropriate to their training level, and the patient receives competent, efficient care.
The oncology patient presents something entirely different. The pain is not mechanical. The prior treatment failures signal that a standard approach will not work. The medical complexity, the psychosocial burden, the fear of recurrence — these are the dominant drivers of the pain experience, not the pathomechanics of the shoulder and spine. This case requires a clinician whose reasoning framework can hold all of it simultaneously. A fellowship-trained clinician would have conducted a different assessment — not more intensive, but differently directed. They would have recognized that the pain and disability were proportional to the medical trauma and the fear, not to the anatomy. They would have adjusted the intervention accordingly, coordinated with oncology and psychology, and established a prognosis based on the person’s trajectory rather than the joint’s imaging.
This is not a more complex technique. It is a different reasoning. And the system has no mechanism to distinguish between them.
Response rates to first-line conservative interventions are 60–75% for most musculoskeletal diagnostic categories, according to systematic reviews including Chou and colleagues’ analysis. That means a quarter to forty percent do not respond as expected. Not all of them are complex — some reflect poor treatment selection, inadequate dosage, and insufficient adherence. But a meaningful proportion reflects genuine complexity: the treatment was appropriate for the diagnosis, but the diagnosis did not capture the full clinical problem.
The convergence is not slowing down. The population is aging — the proportion of Americans aged sixty-five and older has risen from 12.4% in 2000 to over 17% currently, with projections estimating 21% by 2030. Comorbidity prevalence is climbing — federal surveillance data document multimorbidity increasing steadily across all adult age groups, with more than three-quarters of adults now reporting one or more chronic conditions, and young adult prevalence rising by seven percentage points in a single decade. Each chronic condition interacts with the musculoskeletal problem. Diabetes affects tissue healing. Cardiovascular disease limits exercise tolerance. Depression and anxiety alter pain perception. The patient with a single chronic condition may be managed using a standard approach. The patient with three or four is increasingly unlikely. And that patient is becoming the norm.
The caseload is not getting simpler. It is getting more layered. A system that treats all clinicians as interchangeable was always flawed. As patient complexity grows, the flaw is no longer tolerable.
Yet every morning, the schedule is built around availability. A referral arrives. The coordinator checks who has an open slot that works for the patient’s geography, insurance, and work schedule. The patient is assigned to that clinician. This is not negligence. This is standard practice. It is how nearly every clinic operates, because the system has developed no alternative.
When complex patients are assigned based on availability rather than a match, consequences accumulate. The clinician applies the routine approach, and it does not work. The patient does not improve. Both the clinician and the patient interpret the failure as personal — either the clinician was not good enough, or the condition is simply not amenable to conservative care. Neither explanation captures what actually happened: a mismatch between case complexity and clinician capacity. The patient gets referred for unnecessary imaging. The clinician adds it to a growing list of cases that did not work, and begins to wonder.
We know what happens to that pattern over a career.
The two patients from Tuesday morning had different outcomes because the system had no way to match them with clinicians whose training matched their needs. The software engineer did fine—would have done so with either clinician. The oncology patient completed another course of technically competent physical therapy that addressed none of the actual drivers of the pain experience. The documentation read: patient non-responsive to conservative management. The system moved on. The patient did not.
A different system would have started that Tuesday morning with a different question. Not who has an open slot, but what does this case require? Not a more expensive question. Not a slower one. A four-minute screening at intake — prior treatment failure, comorbidity burden, psychosocial risk factors, medication profile, sleep quality — would have changed what happened next. The oncology patient would have been flagged. The match would have been made. The fellowship-trained clinician would have seen the patient who needed that level of reasoning, and the new graduate would have seen the patient whose case would build their confidence and their skills.
Same building. Same clinicians. Same Tuesday. Different systems around them.
Stay up to date on The Development Gap and subscribe to us on Susbtack!
