A mental health diagnosis is frequently received as a discovery about what someone has. Understanding how the categories were constructed makes it clear that this is not quite what is happening.
How the categories are made
Diagnostic manuals define conditions by lists of symptoms, with thresholds for how many must be present and for how long.
Those lists are agreed by committees of clinicians and researchers reviewing evidence, and they are revised across editions.
Which means categories appear, disappear, merge and split between editions, and the changes reflect evolving consensus rather than new discoveries about underlying biology.
This is not a criticism. It is how the system openly works, and the manuals say so.
The heterogeneity problem
Where a diagnosis requires a subset of symptoms from a longer list, two people with the same diagnosis may share almost none of them.
For some categories the number of distinct symptom combinations meeting criteria runs into the hundreds or thousands.
Which means the label groups together people whose experiences differ substantially, and it limits what the category can predict about any individual.
Comorbidity
People frequently meet criteria for several conditions simultaneously, and this is the norm rather than the exception.
Which suggests the categories are not carving up distinct entities cleanly, since genuinely separate diseases would not co-occur at those rates.
The alternative interpretations are that shared underlying processes produce multiple presentations, or that the boundaries are drawn in the wrong places.
Both are actively argued.
The research response
Major research funders have moved toward frameworks organised around dimensions of functioning — reward processing, threat response, cognitive control — rather than around diagnostic categories.
The reasoning is that biological research organised around categories that may not correspond to biological reality was making limited progress.
This is a research framework rather than a clinical one, and it has not replaced diagnosis in practice.
What diagnosis is actually for
Stated plainly, since the criticisms above can be read as dismissal.
It provides a shared language between clinicians, which allows treatment knowledge to accumulate and transfer.
It guides treatment selection, since evidence for interventions is organised by category.
It provides access to services, benefits and accommodations that require a documented condition.
And it frequently provides relief to the person, who now has a name for something that felt inexplicable.
All four are real benefits and none requires the category to be a natural kind.
Where labels cause harm
When they become an identity that constrains expectation, which research on diagnostic labelling has examined.
When they attract stigma, which varies substantially by condition and is measurable.
And when they close off inquiry, so that new symptoms are attributed to the existing diagnosis rather than investigated.
That last is a documented phenomenon in medicine generally and affects people with psychiatric diagnoses disproportionately.
The threshold question
Where normal distress ends and disorder begins is a judgement, generally made on the basis of functional impairment and duration.
Critics argue thresholds have loosened across editions, medicalising ordinary experience.
Defenders argue that broader criteria reach people who were previously untreated and suffering.
Both concerns are legitimate and the balance differs by condition.
What to do with this
Treat a diagnosis as a useful working description that opens access to treatment, rather than as a fact about your nature.
It can change, and changing it does not mean anyone was wrong.
And any of this should be discussed with the clinician who made it, who will generally be more willing to explain the reasoning than people expect.
Second opinions
More common in psychiatry than people assume and entirely reasonable to request.
Because diagnosis rests on clinical judgement applied to reported experience, two clinicians can reach different conclusions from the same presentation.
Inter-rater reliability studies — measuring how often clinicians agree — have found substantial variation between categories, with some achieving good agreement and others considerably less.
Which is published, known within the field, and rarely communicated to patients.
Diagnostic overshadowing
The specific harm worth naming.
Once a psychiatric diagnosis is recorded, subsequent physical symptoms are more likely to be attributed to it rather than investigated.
Research has documented worse physical health outcomes and delayed diagnosis of physical conditions in people with severe mental illness, and this is one of the identified mechanisms.
Which is a reason to be persistent about physical symptoms and to say explicitly that you want them investigated, uncomfortable as that is.
Records
Diagnoses persist in medical records and can affect insurance, employment in some fields, and how future clinicians read a presentation.
Which is worth knowing before agreeing to one, and asking what will be recorded is a legitimate question.
Asking useful questions
What is the diagnosis based on, what else was considered, and what would change it are all reasonable to ask.
So is what treatment follows from it and what the expected course is.
Clinicians generally answer these readily and are rarely asked.