Health information to help you understand symptoms. It isn't a diagnosis.
Check symptoms

Research deep dives

Early identification: screening, and what happens next

Last reviewed November 17, 2026. OwieClear editorial team. Waiting for clinician review. Not medical advice. Talk with your doctor about your own care.

Autism is identified by watching development rather than by a test. Screening tools flag children who need a closer look; they do not diagnose, and they miss some children entirely.

The two most useful things to know are that screening is recommended at specific ages, and that support can begin on a developmental concern alone, long before a diagnostic appointment is available.

Related pages: Autism

The bottom line

What we know
  • Autism screening is recommended at 18 and 24 months in routine care.
  • A diagnosis can be reliable from around age 2 in many children.
  • Girls, and children from several racial and ethnic groups, are diagnosed later on average.
  • Early intervention services do not require a formal diagnosis to begin.
What is still open
  • How well current screening tools perform in girls and in older children.
  • How much earlier intervention changes long-term outcomes, where effect sizes are moderate and often overstated.

The story, step by step

  1. 1. Watching development ()
  2. 2. Regression ()
  3. 3. Who gets missed ()
  4. 4. Before the diagnosis ()
  5. 5. Honest expectations ()
What the evidence labels mean
GuidelineAn official recommendation from an expert group after weighing all the evidence.
Review of many studiesCombines the results of many studies (a systematic review or meta-analysis).
Randomized trialPeople were randomly given a treatment or a comparison, the best way to test cause and effect.
Large observational studyFollows many people over time. Shows strong links, but not always cause and effect.
Small or early studyA small study, case series, or first results. Needs to be confirmed.
Lab or animal studyDone in cells or animals. May not apply to people.
Part 1 of 5

Milestones, and what clinicians track

Review of many studies

What happens

Surveillance at every visit, screening at set ages.

What the research shows

  • Developmental surveillance happens at every routine check: whether a baby smiles socially, responds to their name, babbles, points, follows a point, brings things to show, and later whether they join in pretend play.
  • Formal autism screening is recommended at 18 and 24 months, usually with a parent questionnaire such as the M-CHAT-R, followed by structured follow-up questions when something is flagged.
  • The two screens matter because development changes between them, and some children screen clear at 18 months and not at 24.
  • A positive screen is not a diagnosis. It means a referral for a fuller assessment, which may conclude autism, another developmental difference, or nothing.
  • Signs that warrant asking regardless of age: not responding to their name by 12 months, no pointing or gesturing by 12 months, no single words by 16 months, no two-word phrases by 24 months, limited eye contact or shared attention, strong reactions to sensory input, and any loss of skills at any age.
  • Parent concern is one of the better predictors. Clinicians are advised to take it seriously rather than to reassure and wait, and 'let's watch and see' without a plan to review is not surveillance.

Why it matters

If a concern is raised and the response is simply reassurance, asking for a formal screening questionnaire is reasonable and specific.

What's still uncertain

  • How well screening tools perform across different populations, where sensitivity varies.

Read more

Next: Regression

Part 2 of 5

What losing skills means

Review of many studies

What happens

A common experience that the standard explanation does not cover well.

What the research shows

  • A substantial minority of autistic children, commonly estimated between a fifth and a third, show a loss of skills, usually language or social engagement, typically between 12 and 24 months.
  • This is a recognized developmental pattern in autism, not evidence that something happened to the child. Differences in brain development are present from before birth and become apparent on their own timetable.
  • The distinction matters enormously, because regression is the experience that drives parents to look for a recent cause, and the age at which it occurs coincides with the routine vaccination schedule. Large studies covering millions of children have found no link.
  • Documents stating flatly that autism cannot appear later can leave parents who watched a regression feeling that their experience is being denied. Both things are true: the regression is real, and it was not caused by something external.
  • Regression at any age should still be assessed, because a small number of other conditions, including certain genetic and metabolic disorders and epilepsy, can cause loss of skills and need identifying.
  • Loss of skills in an older child or adult who was previously managing is more likely to be autistic burnout, which is different again and is covered on its own page.

Why it matters

This is the single most common place where an accurate statement, honestly meant, lands badly on parents. Saying both parts is better than saying one.

What's still uncertain

  • Why some children regress and others do not, which remains unexplained.

Read more

Next: Who gets missed

Part 3 of 5

Late diagnosis, and why

Review of many studies

What happens

The patterns in who is identified and when.

What the research shows

  • Girls and women are diagnosed later and more often missed. Diagnostic criteria and screening tools were built largely on observations of boys, and the presentation is frequently different: more masking, more socially motivated, interests that look conventional in subject but are unusual in intensity.
  • Black and Hispanic children in the United States have historically been diagnosed later than white children, and more often misdiagnosed with behavioural or intellectual conditions first. The gap has narrowed and has not closed.
  • Children without intellectual disability, and those who speak early, are frequently identified only at school age or later, when social demands outpace coping.
  • Many adults are now diagnosed after recognizing themselves through a child's assessment, which is a common and valid route.
  • Access matters as much as recognition. Waiting lists for assessment run to a year or more in many places, and the wait is longer for families with less money and less access.
  • A later diagnosis is not a failure and is not too late. Adults consistently describe it as reframing a life rather than changing one.

Why it matters

Knowing the pattern is practical: if a girl or an adult is being told they cannot be autistic because they make eye contact or have friends, that reasoning is out of date.

What's still uncertain

  • How much of the remaining diagnostic disparity is access versus recognition.

Read more

Next: Before the diagnosis

Part 4 of 5

Starting support early

Review of many studies

What happens

What can happen while waiting.

What the research shows

  • In the United States, early intervention services for children under three run through IDEA Part C and are available on the basis of developmental concern. A formal autism diagnosis is not required.
  • From age three, support moves to the school system, again based on need rather than on a specific diagnostic label.
  • This is the most useful practical fact in this area and the one families most often learn late. Referrals can be made by a parent directly in most states, without waiting for a paediatrician.
  • Speech and language therapy, occupational therapy and developmental support can all begin while an assessment is pending.
  • What early support should look like: following the child's lead and interests, building communication in whatever form it takes including gestures and AAC, coaching parents, and adapting the environment. Our support principles page covers the debates about specific approaches.
  • Beware anything promising recovery or a cure. Chelation, bleach-based products sold as miracle mineral solution, hyperbaric oxygen and severely restrictive diets have all been marketed for autism. They do not work, several are actively dangerous, and children have been seriously harmed by them.
  • Augmentative communication does not delay speech. The opposite is true in the evidence, and this misconception costs children years of communication.

Why it matters

Waiting for a diagnosis before starting support is the most common avoidable delay in this whole pathway.

What's still uncertain

  • How much benefit comes from starting earlier within the usual range, which is harder to establish than it sounds.

Read more

Next: Honest expectations

Part 5 of 5

What early intervention does and doesn't do

Review of many studies

What happens

Keeping the claims proportionate.

What the research shows

  • Early support improves communication, daily living skills and family wellbeing, and the evidence for that is reasonable.
  • It does not make a child non-autistic, and claims of transformative outcomes outrun the trials, several of which are small and few of which follow children for long.
  • Autistic adults have been clear that interventions aimed at producing typical-looking behaviour, particularly suppressing stimming and enforcing eye contact, carry a cost that was not measured in older studies and is associated with masking and burnout later.
  • The approaches with least controversy focus on communication, regulation, participation and the environment rather than on appearing typical.
  • Goals are worth examining directly: is this to help the child communicate and be comfortable, or to make them look less autistic? The first is support, the second is something else.
  • Progress is not linear, and plateaus are normal rather than evidence that support failed.
  • Families do better with a diagnosis explained as a description of how a child's brain works than as a list of deficits, and that framing is itself associated with better outcomes.

Why it matters

Both overclaiming and dismissing early support do harm. The honest version is that it helps meaningfully and does not transform.

What's still uncertain

  • Long-term outcomes of early intervention, where follow-up beyond a few years is scarce.

Read more

More plain-language reading

See something wrong or unclear on this page? Suggest a correction. OwieClear is for education only and isn't medical advice.