Where I Stand

I'm Autistic, and I don't use the word "autism." "Autism" is a noun the institutions built: a condition, a disorder, something a person has. Autistic is what I am. The difference sounds small until you notice that one of those words lives in diagnostic manuals and billing systems, and the other one lives in people.

I hold the social model of disability. The human nervous system is wider than the manuals admit. Treating Autistic people as having a disordered brain is a choice institutions made, and it is worth asking who that choice benefits. That is a longer argument than this one. This one is about what happens after they make it: each institution writes its own definition into a record, and the person being defined gets the least say in any of them.

It's also important to point out that for a lot of Autistic people, being identified is a relief. Sometimes it's the first time anything about their life has made sense. Both of those things are true at once.

So here's what each definition actually says, and what each one writes down. I quote where the institutions use their word.

Three Definitions, One Count

When the CDC counts Autistic children, it accepts any one of three things in a child's records: a clinician's written diagnosis, a school special education eligibility under the "autism" category, or a diagnostic billing code.

In its most recent report, covering 2022, the CDC reviewed the records of 8,613 eight-year-olds who met that definition. Only 34.6% had all three. Nearly a third had no clinician's diagnostic statement on file at all. Most of the children in the country's headline number are Autistic according to some systems and not others.

The same report found identified prevalence ranging from 1 in 103 children in Laredo, Texas, to 1 in 19 in California. The CDC's own explanation is differences in identification practices and access to services. California has spent years training pediatricians to screen early, and its children get counted.

The Manual

The DSM-5's criteria open with the phrase "persistent deficits." Everything after that is a list of ways a person differs from a neurotypical baseline, scored by how much support they're judged to need, on a scale of one to three.

The World Health Organization's ICD-11, which took effect in 2022, describes the same people differently. It has no support levels. It sorts Autistic people by intellectual functioning and spoken or signed language instead. Two manuals, two ways of slicing the same population.

ICD-11's diagnostic guidance also contains a sentence I'd like every parent to read. It says some Autistic people "are able to function adequately in many contexts through exceptional effort, such that their deficits may not be apparent to others," and that the diagnosis still applies. The manual is conceding, in its own words, that whether anyone sees you as Autistic can depend on how hard you're working to hide it.

In the US, the code that ends up on a claim is F84.0. It sits in Chapter 5 of ICD-10-CM, "Mental, Behavioral and Neurodevelopmental disorders," alongside schizophrenia. The code's name is "Autistic disorder." They took the adjective there, too.

And the definition moves. In 2013, DSM-5 folded Asperger's and PDD-NOS into a single "autism spectrum disorder." A 2014 meta-analysis of 14 studies found the new criteria cut diagnoses by 31% overall, and by 70% for people previously diagnosed with PDD-NOS. The individual studies ranged from a 7% decrease to 68%, which tells you how little anyone knew about who would still count.

The Doctor

Getting diagnosed depends heavily on which doctor you see, and whether that doctor has ever been trained to see it.

Most physicians haven't. Missing it is the default. The specialists doing this work have more of the picture, and still have holes. PDA is one of them: if it isn't in the manual they were trained on, it often isn't in the room.

Across the CDC's monitoring sites, the share of Autistic eight-year-olds with a documented diagnostic test in their records ranged from 24.7% in New Jersey to 93.5% in Puerto Rico. At least five different assessment tools were in common use. Boys were identified at 3.4 times the rate of girls.

The DSM-5 field trials found that two clinicians evaluating the same child separately agreed often enough for the diagnosis to rate "very good." That was trained evaluators, same child, trial conditions. It is not your pediatrician. The variation is in who gets into the room, whether the person in the room has been trained, and which test is on the table.

The people who don't get in early often get something else instead. Autistic women in particular are frequently diagnosed with borderline personality disorder first. In a small 2025 study from Brighton and Sussex Medical School, participants described feeling "powerless to challenge" that diagnosis, and said it delayed anyone recognizing they were Autistic. A wrong definition, written by someone else, that stayed in the file after the right one arrived.

The School

Schools have their own definition, and it's written into federal law. Under IDEA, "autism" means a developmental disability "significantly affecting verbal and nonverbal communication and social interaction" that "adversely affects a child's educational performance."

That last clause does a lot. State guidance from Connecticut and Ohio spells out what it means in practice: a clinical diagnosis is neither required nor sufficient. A child can be diagnosed by a doctor and still not be Autistic at school, because their grades are fine.

Put that next to the ICD-11 sentence about exceptional effort. The child who is working hardest to hold it together is the one the school's definition is least likely to see.

Whatever the school does decide goes into an education record, and under FERPA, the school can send that record to the next school your child attends without asking you.

The Government

To the Social Security Administration, an Autistic child qualifies for disability benefits only with documented deficits plus "marked" limitations, roughly two standard deviations below average on standardized testing, so being Autistic there means scoring far enough below the mean.

To Australia's immigration system, it's a cost. Visa applicants must meet a health requirement, and a family can be refused if one member's projected health and community services costs exceed a threshold, currently A$86,000 over ten years. Autistic children's special education costs have counted against their families. In a February 2024 decision, a tribunal upheld the refusal of a family whose thirteen-year-old was projected to need special education and disability support, finding it would "result in a significant cost to the Australian community." Other tribunals have granted waivers. A 2024 reform now exempts children born and living in Australia, but everyone else is still assessed.

To the US Department of Health and Human Services, on April 16, 2025, it was this. Announcing the CDC's 1-in-31 figure, the Secretary of Health and Human Services, Robert F. Kennedy Jr., said: "These are kids who will never pay taxes. They'll never hold a job. They'll never play baseball. They'll never write a poem. They'll never go out on a date." I am going to leave that sitting there.

Five days later, the director of the National Institutes of Health told the agency's advisers that NIH would be "developing national disease registries, including a new one for autism." After the backlash, an HHS official told CBS News: "We are not creating an autism registry. The real-world data platform will link existing datasets to support research into causes of autism." On May 7, NIH and the Centers for Medicare & Medicaid Services announced that platform. It links insurance claims, electronic health records, and data from consumer wearables, and it starts with Medicare and Medicaid enrollees who have an "autism" diagnosis. Those two programs cover roughly 36% of Americans.

Same data, new word. Even the government's name for its own project depended on who you asked and when.

NIH has since funded $51 million in projects under its "Autism Data Science Initiative." One of its four priority areas is "etiology and causation." In Part III I wrote that cause is a prelude. That hasn't changed here. What's new is the input. Every one of these definitions runs on the medical record, and the code a pediatrician entered when your child was four is exactly what they read.

The PDA Twist

Even among specialists, this is a common hole. PDA has no official definer at all. It isn't in the DSM-5 or the ICD-11, so there's no code for it. A claim still needs a code, so a demand-avoidant kid often gets filed under one that exists, commonly oppositional defiant disorder (ODD).

The DSM-5's text on ODD notes that it's more prevalent in families where "harsh, inconsistent or neglectful child-rearing practices are common." When no institution has a definition for your child, the record borrows one that implicates the parent.

PDA is still debated in the United States, but if you know the term at all, you already understand your child better than the coding system does.

The Individual

Ask Autistic people, and you get a different answer again, starting with the grammar. In a 2023 US survey, 87% of Autistic adults preferred identity-first language: "Autistic person." Professionals were the group most likely to prefer "person with autism," which is what clinical style guides prescribe. The people and the professionals can't agree on the sentence.

None of that reaches the record. No medical system has a field for how a person describes themselves. Most diagnoses happen in childhood, with a parent's consent. If the definition in the file is wrong, the provider who holds the record decides whether to change it.

As I mention in Part I, there is one opening under HIPAA. If a provider denies your request to amend a record, you can file a statement of disagreement. It's attached to the file and has to accompany that information whenever it's disclosed. As far as I can find, it's the only place in the medical system where an Autistic person's own description of themselves is legally guaranteed to travel with the record. It exists only after someone else has already said no.

And the women misdiagnosed with BPD? When the Autistic identification finally came, they called it "life changing." It let them stop trying to fix themselves. Being recognized can be the most validating thing that ever happens to a person, and the person still has no say in what the definition says.

The Noun

A manual that calls it a deficit. A doctor whose answer depends on the zip code. A school that only sees it when the grades slip. A disability office that measures it in standard deviations. An immigration department that prices it. A health secretary who described it as a list of severe disabilities. Every one of those definitions is in a file somewhere, attached to a child who was probably too young to have an opinion about any of them. The one definition that never makes it into the file is the one written by the person.