Why Some Diseases Remain Hard to Diagnose

A diagnosis is a match between a pattern of symptoms and a category that medicine has already defined. Diagnosis fails in two ways: when the pattern is not recognized, and when no category fits. Both failures are common, and they persist despite extraordinary advances in imaging and testing.

Rare is common

Any single rare disease is rare. Collectively, more than seven thousand of them affect roughly one in ten people. A general physician may see one case of a given rare disease in a career, and the symptoms often mimic common conditions. Studies of patients with rare diseases consistently find average diagnostic delays of several years and multiple misdiagnoses along the way. Genome sequencing has shortened that path for children with suspected genetic conditions, but access remains uneven.

When the category does not exist

A second group of conditions is defined by symptoms rather than by a test: a pattern of fatigue, pain, cognitive difficulty or autonomic dysfunction with no marker that distinguishes it. Myalgic encephalomyelitis, fibromyalgia and many cases of long COVID fall here. Without a test, diagnosis depends on excluding alternatives, which is slow, and on the physician believing the patient, which is not guaranteed. Research increasingly suggests that some of these labels cover several distinct diseases, which would explain why treatments help some patients and not others.

The structure of medicine

Medicine is organized by organ system. A patient whose disease crosses systems, as many autoimmune and genetic conditions do, is passed between specialists who each see part of the picture. Appointment times measured in minutes leave little room for the long history that complex cases require. The problem is less a lack of knowledge than a lack of time and coordination to apply it.

What might help

Multidisciplinary clinics for undiagnosed disease, in which specialists review a case together, have shown good results and remain scarce. Software that scans records for patterns suggestive of rare conditions is in trial. Wider access to genome sequencing would shorten the search for many. None of these is new; all of them are slow to spread because they are expensive and because the system that would pay for them is not the one that bears the cost of delay.