The Thirty Percent

What studying one of the rarest illnesses in child psychiatry taught me about diagnostic humility

For almost two decades at the National Institute of Mental Health, I helped run the largest study of childhood-onset schizophrenia in the world. This is a rare and devastating diagnosis. Families traveled to Bethesda from every corner of the country because a doctor somewhere had already told them their child had schizophrenia, and had started treatment. Most child psychiatrists never encounter a single case of the disease. We had chosen to study it partly because an early form of schizophrenia might offer unique insights, and partly because it was a study that could only be done at a place like the NIH.

But that is not why I am writing this essay.

What we found still gives me pause even after more than twenty years. In almost thirty percent of the cases, doctors had made the wrong diagnosis. Said another way, close to a third of the children admitted with a suspected diagnosis of childhood-onset schizophrenia, children already being medicated for it, children who had passed rigorous screening including a full day of evaluation by our own team, did not have it.

It would be tempting, and easy, to say the referring clinicians had been careless or inexperienced. That would be far from the truth.

The problem was not clinical skill. Before coming to the NIH, these children had been seen by multiple psychiatrists and therapists and had multiple inpatient admissions. Our team scrutinized their records, and each child sat for an in-person interview lasting about eight hours. After this assessment, our team had to agree that the child was likely to have schizophrenia before admitting him or her to our ward.

What enabled us to eventually make a correct diagnosis was not a better test. There is no blood draw, no scan, no rating scale that works as a gold standard. What we had was time, and the willingness to observe carefully, without any insurance constraints, long enough to see the child underneath the diagnosis. When appropriate, and under close supervision, we took them off the medications they arrived on. Some were taking dozens of tablets a day, a cocktail of antipsychotics, mood stabilizers, antidepressants and more. Then we watched. In almost a third of cases, what had been labeled the psychosis of schizophrenia either simply resolved or turned out to come from something else: depression, severe anxiety, trauma, OCD, or behavioral problems. These children did not have schizophrenia, whatever their charts insisted.

Distinguishing among the many causes of early psychosis is one of the hardest things to do in psychiatry, particularly in children. But the main problem was that certainty and accuracy had quietly come apart, and from the inside there was no way to feel the difference. A wrong diagnosis feels exactly like a right one.

These numbers expose a flaw in how we handle child psychiatric evaluation; about how a diagnosis hardens into fact. More than seventy percent of U.S. counties do not have a single child psychiatrist. Where one is available, a typical outpatient visit runs fifteen minutes during which that psychiatrist must evaluate the child, talk with the parents, gather collateral information, write a prescription, and complete a note. When a child is in crisis, they end up in an emergency room, and if admitted, are held four or five days at most, “stabilized,” and sent home with no follow-up. It is no surprise that labels and medications get added and never taken off. No one has the luxury of time that we had at the NIH.

So a label and a treatment begin to prop each other up. Once a child is given a diagnosis of schizophrenia and started on an antipsychotic, everything afterward is read through that lens. Sedation reads as flat affect. A side effect reads as a negative symptom. The medication meant to treat the illness begins, quietly, to generate evidence for it. The diagnosis stops being a hypothesis under test and becomes the floor everyone is standing on. Removing the treatment, along with daily careful observation, was often the only way to see the child underneath it.

One boy I will not forget came to us on three medications and a diagnosis of schizophrenia. Off all of them, over a month of watching, what emerged was a frightened child with severe obsessive-compulsive disorder and a home he did not feel safe returning to. The voices everyone had been treating as psychosis were his own, narrating his fear.

None of this is unique to psychiatry. Every specialty has its version of the confident label that outruns the evidence and then grows harder to take back with each year of treatment built on top. What child psychiatry offers is something unusual: for years, our most trustworthy diagnostic instrument was also the crudest and most honest one available. Stop, wait, and look.

And yet psychiatry is also different, in a way that makes all of this harder. Most of medicine has an anchor outside the clinician’s judgment. For example, when a gastroenterologist suspects ulcerative colitis, a biopsy can settle it under the microscope. These tests may not be infallible, but they support the clinician in an objective and scientifically proven way. Psychiatry, unfortunately, has almost nothing of the kind yet. There is no diagnostic scan, biopsy, or blood test that can confirm or even support a diagnosis. We still heavily rely on clinical judgment. Time thus becomes the closest thing psychiatry has to a confirmatory test.

The children I saw at NIMH changed how I practice. I pause before I name things. When a family wants an answer on the first visit, and they nearly always do, I have learned that the most rigorous thing I can offer is often not a diagnosis but a question. Not “here is what your child has,” but “let us find out, and let us design this so that we can still be wrong.” That is not indecision. It is a clinical stance with consequences: I will not commit a child to a label, and to the medications and the self-understanding that travel with it, before the evidence has earned it.

This often cuts against how we are trained. Medicine rewards closure. A quick diagnosis is often taken as a sign of confidence, knowledge, and experience. But the thirty percent were not a failure of effort. They were a failure of a system that rushed to certainty. The remedy is building the discipline to be wrong out loud, and structuring the encounter so the diagnosis can still move.

The practical form of this is not that hard, if it can become routine. Watchful waiting, wherever the stakes allow it, should be the norm, with a bias toward the least committing explanation rather than the most frightening one. It means treating only the diagnosis you are most sure of, and having the nerve to pause it and see what remains and what evolves.

The children who did not have schizophrenia were, in a way, the fortunate ones, because somebody eventually paused long enough to find out. To this day, I cannot stop thinking about the ones nobody paused for, still carrying a diagnosis that was assigned with confidence and never revisited.

We should be more willing to take it back.

Nitin Gogtay, MD, is a psychiatrist scientist. He currently serves as the Vice President of Research and Deputy Medical Director at the American Psychiatric Association. He is the author of ‘The Vanishing Children’ (HarperCollins, January 2027) and writes about the human story beneath psychiatric diagnosis in his Substack ‘Beneath the Diagnosis’. Views expressed are his own and do not represent either the APA or the NIH.

Photo by Vitaly Gariev on Unsplash