Autism in women and minorities: why diagnosis takes decades

Autism diagnosis can take decades for women, girls and minoritised people because the system has historically been better at spotting a narrow, male, white, childhood presentation. Many people are missed because they mask, internalise distress, have strong language skills, are labelled anxious or difficult, or are judged through gender, race, class and cultural stereotypes. Late diagnosis is not proof that the autism is mild. It often means the person had to compensate for longer.

Why the old autism picture missed people

The public stereotype of autism was built around boys: visible social difference, obvious restricted interests, disruptive behaviour, language delay, and a family or school system that pushed for assessment early. That picture describes some autistic people well. It never described everyone.

Girls, women, gender-diverse people and people from minoritised communities may be missed for different reasons. A girl who copies friends, reads novels about relationships, studies people carefully and collapses after school may look "shy" rather than autistic. A Black boy may be judged as defiant rather than overloaded. A South Asian girl may be seen as quiet and compliant. A migrant family may face language barriers or fear stigma. A high-achieving adult may be told that their degree or job rules autism out, even when the cost is burnout.

Lai and colleagues argued in 2015 that sex and gender shape autism presentation, detection and research, and that the field needed to move beyond simply comparing boys and girls after diagnosis.⁶ That point is still practical in clinic: if the assessment only asks whether you looked like the old stereotype, it may miss the actual pattern.

NICE adult guidance is broader than the stereotype. It says comprehensive assessment should consider communication, social interaction, repetitive behaviours, sensory sensitivities, daily functioning, mental health, physical health, risk and coexisting conditions, using developmental history where possible.¹ It also says an AQ-10 score of 6 or above, or clinical suspicion based on history, should lead to comprehensive assessment. A screening tool can help. It should not replace clinical judgement.

Masking, internalising and the cost of looking fine

Masking is not just "trying to fit in". It can mean rehearsing facial expressions, forcing eye contact, copying social scripts, suppressing stims, monitoring tone, hiding confusion, and studying other people like a survival manual. Hull and colleagues' 2017 qualitative study of 92 autistic adults described camouflaging as effortful, identity-blurring and linked to delayed recognition.⁷

For many women and girls, the visible problem is not disruptive behaviour. It is anxiety, perfectionism, eating difficulties, exhaustion, shutdowns, self-criticism, chronic people-pleasing or burnout. Bargiela, Steward and Mandy interviewed 14 late-diagnosed women and described social imitation, missed signs, vulnerability and misinterpretation through non-autistic expectations.⁸ Small qualitative studies do not prove a universal female autism type, but they explain why a person can be autistic and still appear socially skilled in short bursts.

Lockwood Estrin and colleagues' 2024 narrative review on improving diagnosis for girls and women argued that assessments need broader behavioural examples, active consideration of camouflaging, and care that co-occurring mental health conditions do not overshadow autism.¹¹ This matters because a lifetime of being misread can produce real anxiety and depression. Those conditions deserve treatment, but they do not automatically explain the underlying neurodevelopmental pattern.

How race, class, language and gender identity affect diagnosis

Minority diagnosis delays are not only about biology. They are about who is believed, who is referred, who can wait, who can pay, who has a school that documents need, who has a GP who understands adult autism, and whose behaviour is interpreted as distress rather than disobedience.

The Roman-Urrestarazu study used England's National Pupil Database and included 7,047,238 pupils aged 2 to 21 in state-funded education. It identified 119,821 pupils with recorded autism and found prevalence differences by ethnicity, free-school-meal status and English as an additional language.¹⁰ The authors cautioned that differences could reflect diagnostic bias, detection, referral patterns or true variation. The important clinical message is humility: records show who gets identified, not simply who is autistic.

Gender-diverse people also deserve explicit mention. The 2026 BMJ Swedish birth cohort study included 2,756,779 people born from 1985 to 2020 and found the cumulative male-to-female ratio for autism incidence was 1.2 by age 20 in the final follow-up year, because many females were diagnosed later in adolescence.¹² The study does not erase sex differences, but it does make delayed recognition hard to ignore.

What a better assessment should look for

A good assessment should not depend on whether you perform autism in a recognisable way during one appointment. It should ask for a developmental pattern across childhood and adulthood, including sensory sensitivities, routines, transitions, social confusion, friendship history, interests, burnout, shutdowns, family observations, school reports, workplace adjustments and coexisting ADHD, dyslexia, dyspraxia, anxiety, trauma or eating difficulties.

For adults, the NHS says an autism assessment may include questionnaires, discussions with other people, childhood history, current support needs and reports from a GP or school if available.³ In England, the NHS also notes that people can choose which NHS service carries out the assessment, and may be able to find a shorter wait through Right to Choose.³

Current waiting lists are long, so preparation matters. Use the wait to gather evidence rather than trying to become more convincing through distress. The health library and insights section can help you organise sleep, stress, burnout and sensory patterns. For appointment preparation, keep the evidence concrete.

How to prepare if you have been missed

Do not rely on a list of traits. Build a timeline. Start with early childhood: play, sensory issues, food, clothes, sound, friendships, school reports, routines, transitions, shutdowns and special interests. Then adolescence: social rules, bullying, exhaustion, eating, anxiety, self-image and masking. Then adulthood: work, relationships, burnout, parenting, alcohol, sleep, health appointments and recovery time after social demand.

Ask one trusted person for examples if that is safe. If family evidence is unavailable or unsafe, use school reports, old diaries, therapy notes, workplace records, partner observations or your own timeline. Lack of a parent informant should not automatically end the assessment, especially for adults whose family relationships are complicated.

The core point is simple: diagnostic systems are human systems. They carry the assumptions of the research, services and cultures that built them. If you have been missed for decades, the task is not to prove you suffered enough. It is to document the lifelong pattern clearly enough that the system finally has to look.

This article is educational and does not constitute medical advice, diagnosis, or a treatment recommendation. Medication uses described as “off-label” are not licensed for that purpose in the UK and should only be considered under qualified clinical supervision. Always speak to your GP, pharmacist, or a registered specialist before starting, stopping, or changing any treatment. If you have severe or alarm symptoms - unintentional weight loss, blood in your stool, difficulty swallowing, persistent vomiting, a fever, or severe pain - seek urgent medical care.