Medicine tells a clean story about diagnosis. A set of symptoms appears. A clinician evaluates them. Tests are ordered. Patterns emerge. A name is assigned. Uncertainty resolves. That version is comforting because it suggests order: cause, method, resolution. It also collapses under sustained complexity. For some patients, that collapse is not theoretical. It is lived over years.
Our son started entering hospitals at 8 months old.
Not once or twice, but in cycles that became part of the rhythm of childhood. Weeks out of every year spent inpatient with respiratory illness that did not behave predictably. Recurrent pneumonia. Severe lung issues that repeatedly defaulted into an asthma framework because that was the closest available explanation. Hospital stays became familiar in a way that never felt normal: IV poles, oxygen monitors, and escalation conversations becoming part of childhood vocabulary. For years, he was treated under a pulmonary diagnosis that made sense locally, even as the broader pattern kept resisting containment.
Over time, the system expanded its explanations. Immune dysfunction was added. Then questioned. Then redefined. At one point, a specialist identified what looked like B-cell dysfunction, his body producing B-cells that never fully matured, leaving him unable to mount an effective immune response. That interpretation made sense of pieces of the picture, but not the trajectory.
Real physician voices, twice a week
Free, and one click to unsubscribe.
He was eventually diagnosed with common variable immunodeficiency at age 11. That diagnosis brought structure. It also brought a new set of decisions: IVIG therapy, risks, side effects, uncertainty about long-term benefit. None of it was simple. All of it was negotiated. This is what diagnosis looks like when it is not stable yet. Not clarity, but a series of constrained choices made under shifting interpretations of incomplete data.
As he got older, the system did what it is designed to do when uncertainty persists: It escalated. White blood cell counts dropped. Platelets dropped. Cancer was seriously considered. Bone marrow biopsies were done. Extensive lab workups followed. He was referred across specialties and across institutions because no single system had a complete explanation for what was happening.
Each specialty did its job well. Pulmonology managed lungs. Immunology managed immune dysfunction. Hematology evaluated malignancy risk. Dermatology tracked evolving autoimmune manifestations. But none of them were seeing the whole structure. And no system was responsible for holding it together. So it became mine. Not formally. Functionally.
I tracked labs across institutions. I followed consult notes across specialties. I carried forward interpretations that would otherwise have been lost between systems that did not naturally communicate with each other. It meant keeping track of lab trends across institutions that did not share systems. It meant remembering what one specialist had ruled out months earlier when a different department started from scratch. It meant recognizing patterns no single visit could show.
The clinicians were thorough in their domains. But what they saw individually did not resolve into a coherent whole. And over time, a different pattern emerges in cases like this: not lack of care, not lack of intelligence, but lack of integration. And no mechanism inside the system that consistently corrects for it.
At one point, while monitoring yet another round of bloodwork after tapering a medication suspected of contributing to the abnormal counts, a nurse said something that changed the trajectory of the entire case.
“I just feel like there is something more going on here. Have you ever considered genetic testing?”
That sentence was not dramatic. But structurally, it was an expansion event. Because until that moment, genetic testing had never been part of the working diagnostic space. It had not been proposed. It had not been discussed. It was not part of the active search space being used by the system.
We had to create access to it. Insurance would not cover it. The cost was prohibitive. So I did what families in these systems often end up doing when standard pathways stop moving: I started searching for alternate structures that could expand eligibility rather than assume it.
That search led to a study involving a geneticist at a university in California and a clinician at a university in Utah. There were strict inclusion criteria. Extensive medical records had to be assembled and submitted. An intake process determined whether he would qualify for testing. Because of the complexity and longevity of his medical history, he did. He was accepted. The testing that followed was, at the time, among the most comprehensive available. And through that process, he was diagnosed with an extremely rare autoimmune condition that had only been identified in 2013.
He received that diagnosis in 2022, at age 15, nearly a decade after the condition itself had been defined in medical literature, and more than a decade into a system that had been trying, accurately and unsuccessfully, to explain what was happening.
Seen from the outside, this can look like a long arc that finally resolves. But that is not what it feels like from inside it. It feels like years spent moving through partial explanations that are locally correct but globally incomplete. Asthma was not wrong. CVID was not wrong. Cancer workups were not wrong. Each diagnosis explained a portion of what we were seeing. None of them explained the trajectory we were living.
And that is the point. Every clinician involved was operating inside a bounded hypothesis space, shaped by training, specialty, and the need to narrow uncertainty in order to act. Within those boundaries, each decision was reasonable. A hospital admission is a safety response. A biopsy rules out catastrophic risk. A referral escalates appropriately. These are not failures. They are local optimizations. But local optimization does not guarantee global convergence. Over time, the system becomes very good at refining the explanations it already has and less capable of recognizing when the space of possible explanations needs to expand.
In theory, health care systems integrate. In practice, that integration is fragmented across time and institutions. Each specialty sees correctly, but partially. Pulmonology builds one model. Immunology builds another. Hematology evaluates another layer entirely. Genetics often sits outside the working frame until explicitly activated. No single actor is responsible for synthesis. So synthesis shifts. And in long, complex cases, it shifts to the family. Not symbolically. Functionally.
Tracking contradictions across systems. Maintaining continuity across fragmented interpretations. Holding together a coherent picture that no single institution is structured to maintain. That is not advocacy in the usual sense. It is systems work being done outside the system.
At some point, diagnosis stops being about finding the right answer. It becomes about expanding the space in which answers are even allowed to exist. And when that space finally expands, through a nurse’s question, through a study, through persistence, the system does not suddenly become simple. It becomes navigable.
Seen this way, diagnosis is not a revelation of truth. It is a negotiated convergence across fragmented systems operating under constraint. Each specialty contributes partial truth. Each test refines part of the picture. Each intervention reduces uncertainty in one dimension while sometimes increasing it in another. Over time, a diagnosis emerges when enough of these partial systems converge on something stable enough to act on. Not complete. Not final. Stable.
Rare and complex disease does not break medicine. It reveals how it actually works under pressure. Most patients are not experiencing a different system. They are experiencing the same system with more visible seams. And in that environment, what looks like failure is often structural.
The system is not malfunctioning. It is performing as designed, just not designed for cases that fall outside its initial assumptions: within bounded search spaces, through fragmented expertise, relying on local optimization, and depending, quietly and consistently, on patients and families to bridge the gaps between those fragments.
Which leads to a harder conclusion: Diagnosis is never just a medical act. It is an architectural one, and in complex disease, the architecture often includes the patient or family as an invisible structural beam.
Sarah Whaley is a patient advocate.
