The patient is fifty-six years old, with right-sided low back pain that has not responded to two courses of physical therapy over the past fourteen months. The first course — eight visits at a clinic near the patient’s workplace — followed a standard protocol: lumbar stabilization exercises, general conditioning, ergonomic advice. The patient reported modest improvement at discharge. The pain returned within six weeks. The second course — ten visits at a different clinic, different therapist — identified the same mechanical presentation, prescribed a similar approach with the addition of manual therapy, and produced the same arc: initial improvement, early return of symptoms.
The referring physician’s note for the third referral reads: chronic low back pain, failed conservative management x2, patient requesting one more attempt before considering surgical consultation.
The patient arrives at a third clinic. The therapist reviewing the intake paperwork sees a familiar pattern — mechanical low back pain, two prior episodes, incomplete response. The examination confirms what the previous therapists found: limited lumbar flexion, segmental hypomobility, and mild motor control deficits. The findings support a clear clinical framework. The evidence base for managing this presentation is well established. The therapist — competent, efficient, two years out of a residency program — initiates a plan that is clinically defensible and consistent with current best practice.
What the therapist does not do — because nothing in the training, the referral, or the system prompted it — is treat the two prior failed episodes as data.
The patient’s sleep has deteriorated significantly over the past year. A statin was added eight months ago, and the patient reports diffuse muscle aching that predates the current episode but has been attributed to aging. The patient’s spouse recently entered long-term care, and the patient is managing the logistics of that transition alone while working full-time. None of this is on the intake form. None of it was asked about during the first two therapy sessions. None of it is visible from within the clinical framework the therapist is using — a framework that is entirely appropriate for the mechanical presentation the examination identified and entirely insufficient for the life in which that presentation is embedded.
The patient does not improve.
There is a reasonable question here, and it is not about this therapist. Three clinicians in fourteen months identified the same findings, applied evidence-based treatment, and produced the same result. None of them did anything wrong. Each was competent. Each followed the evidence. Each did precisely what their training prepared them to do.
The question is what happens when doing precisely what training prepared a clinician to do is not enough.
Evidence-based physical therapy primarily relies on studies of patients who are not complex. Randomized controlled trials — the foundation of clinical practice guidelines — typically exclude patients with significant comorbidities, psychiatric diagnoses, pending litigation, and multilevel pathology. Rothwell’s analysis found that the populations studied often bear little resemblance to those treated in routine practice. The evidence that guides clinical reasoning derives from populations selected for homogeneity.
This is not a criticism of the research. Controlling variables isolates treatment effects. But it means the evidence-based approach works best for patients resembling the populations studied. The more a patient deviates from that population, the less directly applicable the evidence becomes. The patient with low back pain, a statin-induced myalgia, chronic insomnia, and caregiving stress was not in the trial. The guideline was not designed for that patient. Applying it is not evidence-based practice. It is an extrapolation.
And it is the most common clinical scenario in outpatient orthopedic practice.
The CDC’s most recent data on chronic conditions show that 93% of adults over 65 have at least one chronic condition, and 78.4% of midlife adults do the same. Among younger adults — the cohort many clinicians assume to be straightforward — chronic disease prevalence has increased by seven percentage points over a single decade. The proportion of adults taking five or more medications has more than doubled over the past two decades, from 8.2% to 17.1%, with the figure among adults over 65 rising from 23.5% to 44.1% (NHANES data). Statins cause myalgia and tendinopathy. Gabapentinoids cause cognitive dulling. Fluoroquinolones cause tendon damage. The medication list is not a sidebar in the chart. It is a diagnostic artifact.
The patient sitting in a waiting room in 2026 is not the patient who sat there in 1996. The musculoskeletal impairment may be the same. The person carrying it is not. And the three competent therapists who treated the mechanical findings and observed that the patient did not improve were not failing. They were operating within a framework designed for a different patient — the uncomplicated one, the one in the trial, the one who no longer represents the majority of the caseload.
Two months later, the same patient — now with a fourth referral, this one to a fellowship-trained clinician at a specialty clinic forty minutes from home — is seen for an initial evaluation. The evaluation takes longer. The questions are different.
The fellowship-trained clinician recognizes the mechanical findings and sets them aside — not because they are wrong, but because two competent therapists have already treated them and the patient is not better. That fact reorganizes the clinical picture. The sleep history becomes relevant. The medication profile becomes relevant. The caregiver burden becomes relevant—not as background noise but as part of the clinical presentation, interacting with the pain experience in ways the standard framework does not account for. Sleep deprivation alters pain processing and impairs tissue recovery (Finan et al., 2013). Psychosocial stress changes the neurobiological environment in which rehabilitation occurs (Lumley et al., 2011). A statin-induced myalgia may be amplifying mechanical symptoms in ways no amount of stabilization exercise will resolve.
The clinician constructs a management approach that addresses the mechanical findings and the contextual complexity simultaneously. Coordinates with the prescribing physician about the statin. Incorporates sleep hygiene as a treatment variable, not a wellness suggestion. Adjusts exercise dosing to account for the global muscle aching the patient has been attributing to age. Builds the treatment plan around the patient’s actual life — including the forty-five-minute drive to this clinic, the caregiving schedule, and the energy budget available for rehabilitation on top of everything else.
Over twelve weeks, the patient improves — not perfectly, not dramatically, but meaningfully and for the first time in over a year.
The question is not what the fourth clinician knew that the first three did not. Fellowship-trained clinicians do not possess secret techniques unavailable to others. The difference was not in the hands. It was in the questions. It was in the capacity to look at two failed episodes and treat that fact as clinical data rather than as a footnote in the referral.
The first three therapists each assessed the patient using a framework that processes the examination findings. The fourth assessed the patient through a framework that also accounts for what the examination cannot reveal — the contextual factors that modify how the impairment behaves in a particular life.
The system that assigned this patient to three clinicians over fourteen months offered no mechanism for identifying which clinical encounter the patient actually needed. No complexity screening at intake. No triage pathway that treated prior non-response as a routing signal. No infrastructure for getting the patient to a clinician whose reasoning framework matched the actual demands of the case, before fourteen months, three courses of therapy, and the accumulating demoralization of a patient being told, implicitly, that the problem is their failure to respond.
The patient with right-sided low back pain is not an unusual case. The patient is the caseload. The question is whether the system will be built to recognize that — or whether the next patient, the one arriving tomorrow with a familiar referral diagnosis and an unfamiliar life, will take the same fourteen-month path to someone who finally asks different questions.
Stay up to date on The Development Gap and subscribe to us on Susbtack!
