Adult ADHD and the Overdiagnosis Debate, Explained by the Evidence

Ask two clinicians whether adult ADHD is overdiagnosed and you may get two confident, opposite answers, yet both can be reading the same evidence honestly. The reason is that the argument rarely turns on whether the condition exists, which the research supports, but on where its edge should be drawn and what the word "overdiagnosis" is meant to capture. Once those two things are pinned down, most of the heat in the debate turns out to be about definitions, not facts.

Key points#

• Overdiagnosis is a population-level idea: a label that is correct but unlikely to help the person, once side effects, cost, and stigma are weighed. It is not the same as a wrong diagnosis.
• Adult prevalence estimates range from about 2.5 percent to nearly 7 percent, and that spread comes from a definitional choice, not a measurement mistake.
• Successive revisions to diagnostic criteria enlarged the group near the threshold, which is exactly where disagreement concentrates.
• Overdiagnosis and underdiagnosis can be true at the same time in different parts of one continuous trait.

First, four words that are not the same#

Public debate stalls because a single term is asked to carry several unrelated problems. It helps if you keep them apart. Misdiagnosis means the label is simply wrong, applied to someone who has something else or nothing at all. Underdiagnosis means real cases are being missed. Overtreatment means intervention outpaces need even when the label fits. Overdiagnosis is the subtle one: a diagnosis that is technically accurate but unlikely to improve the life of the person receiving it, and possibly a net negative once medication effects, financial cost, and stigma are counted.

That last idea is inherently a population-level claim, not a verdict on any single patient. It says that somewhere along the mild end of a spectrum, adding a diagnosis stops producing benefit and starts producing harm. Deciding where that point sits is a value judgment as much as a clinical one, which is why reasonable people land in different places.

Why the prevalence numbers refuse to agree#

If ADHD were a sharp category, counting it would be straightforward. Because it sits on a continuum, the count depends entirely on where the line is drawn, and researchers do not all draw it the same way.

A 2021 systematic review and meta-analysis in the Journal of Global Health, pooling 40 studies across 30 countries, made the split explicit. "Persistent" adult ADHD, which requires documented childhood onset, came in near 2.6 percent worldwide. "Symptomatic" adult ADHD, counting anyone who currently meets symptom criteria regardless of childhood history, came in around 6.8 percent. These are overlapping but genuinely different populations, and choosing between the two definitions shifts the global headcount by hundreds of millions. The same review found prevalence declining with age, so a sample of college students and a sample of middle-aged adults will disagree before anyone even reaches the philosophy of diagnosis.

National surveillance behaves the same way. Data from the CDC's National Center for Health Statistics, collected in late 2023 and published in the Morbidity and Mortality Weekly Report in October 2024, estimated that 6.0 percent of US adults, about 15.5 million people, carried a current ADHD diagnosis, and roughly 55.9 percent of them were diagnosed at age 18 or older. The report is unusually frank about its own limits: those diagnoses were self-reported and never checked against medical records, so recall and reporting effects are built into the figure. A large number resting on self-report is a place to start a conversation, not a place to end one.

How the criteria themselves moved the line#

Diagnostic criteria are not natural landmarks. They are decisions made by committees, and small edits ripple straight through the prevalence figures. A 2013 review in Neuropsychiatry catalogued how the DSM-5 revised ADHD's definition, and the direction of travel was consistent. The age-of-onset requirement moved from before age 7 to before age 12. The symptom count required for anyone 17 or older dropped from six to five. The cross-situational rule loosened from requiring impairment in two or more settings to requiring only symptoms in two or more settings, and the impairment bar softened from "clinically significant" to "reduces the quality of" functioning.

Each of these changes was defensible on its own terms, often meant to capture adults and older adolescents that an earlier, child-centered definition had left out. But every one of them also enlarges the group sitting near the threshold, and that group is precisely where the label does the least clear-cut good. Broaden the entry point and you will find more people just inside it, which is neither proof of a hidden epidemic nor proof of runaway labeling. It is simply what happens when a line is redrawn on a continuous trait.

Context bends the ruler#

Even with fixed criteria, ADHD is judged from behavior measured against expectations, and expectations are not constant. They vary by classroom, workplace, culture, and whoever happens to be the comparison group.

The cleanest illustration is the relative-age effect. Within a single school grade, the youngest children are consistently more likely to be diagnosed than their oldest classmates, a pattern that points to developmental timing rather than neurology alone. A child who is eleven months younger than the peers he is ranked against can look inattentive simply for being younger. In adults the reference point becomes a demanding job, a strained relationship, or a self-report questionnaire, all of which flex far more than a laboratory measurement would. When the ruler bends with circumstance, the count bends with it.

The most systematic look at the evidence#

The most thorough attempt to test the overdiagnosis claim is a 2021 scoping review in JAMA Network Open. Its authors screened more than 12,000 records and analyzed 334 studies, and they concluded there was convincing evidence of overdiagnosis, organized around a set of recognizable signals: a large reservoir of undiagnosed but potentially diagnosable people, diagnoses climbing over decades, newly captured cases clustering at the milder end, treatment expanding in step, and, for those milder cases, harms that can outweigh benefits.

One caution matters. That review examined children and adolescents, not adults, so its numbers cannot be lifted wholesale into adult care. What does carry over is the framework. It gives you a way to ask, of any given expansion in diagnosis, whether it is uncovering real and previously ignored suffering or mostly relabeling the ordinary spread of human attention.

Both things can be true at once#

The unsatisfying but honest resolution is that overdiagnosis and underdiagnosis are not rivals; they can coexist in different regions of the same distribution. Adults whose symptoms were dismissed for years, including many women and older patients who never matched an earlier stereotype, can be genuinely underserved. At the same time, the mild margin of the same trait can absorb people for whom a formal diagnosis adds little and may subtract something. Both statements hold, because they describe different segments of one continuous characteristic rather than contradicting each other.

Read plainly, then, the evidence does not collapse into a slogan. It shows a real condition, defined through criteria that were assembled and revised for defensible reasons, then applied in settings that stretch and compress the numbers. Holding all of those moving parts in view is what separates arguing about a headline from understanding what the headline is actually counting.