Autism around the world

Autism is defined by clinical criteria, not a laboratory test, so a prevalence figure is a count of people who meet a definition, found by some method.¹ The main methods are registers and insurance records that hold existing diagnoses, review of health and school records, surveys that ask parents, and screening a population and assessing the people who screen positive.

The method of counting is a major source of the differences between estimates. A global systematic review of 99 estimates from 71 studies found a median prevalence of about 1 in 100, with single estimates from 1.09 to 436 per 10,000.¹ It attributes the spread to methodological differences and to complex interactions among community awareness, service capacity, help-seeking and sociodemographic factors, rather than to any single cause.¹

Countries also count in different places. Brazil added a question on a prior medical diagnosis of autism to its 2022 census, and the preliminary figures released in 2025 gave 1.2% of the population, about 2.42 million people, reporting a diagnosis.² The share was 2.6% among boys aged 5 to 9, 1.6% among adolescents and 0.3% among adults aged 30 and older; the authors read the lower rate among older adults as past under-identification, and caution that the figure cannot be compared directly with US record-based counts.²

No study found as of October 2026 has counted autism the same way across many countries at once, so a national figure cannot be read as that country's rate of autism.¹ A count from one country's records, an estimate from another country's screening and a third country's self-reported census do not measure the same thing.

The Global Burden of Disease study covers 375 diseases and injuries; for its 12 mental disorders, autism among them, it fits a Bayesian meta-regression to published epidemiological data and reports rates by country, sex, age and year.³ Its 2023 round is a modeled estimate, not a measurement. It estimated 1.17 billion cases of those 12 mental disorders worldwide in 2023, and reports disorder-specific numbers in its appendix; the autism-specific estimate is not quoted here.³ The 2021 round's autism figure, about 1 in 127, is set out in Prevalence.

Most children with a developmental disability live in countries that produce little of the research. An estimated 52.9 million children under 5 had a developmental disability in 2016, a broad category that includes sensory impairment, intellectual disability and autism; 95% of them live in low- and middle-income countries, and most lack access to care.⁴ Autism research itself is heavily skewed toward western high-income countries, and culturally appropriate screening and diagnostic tools are lacking in most low- and middle-income settings, where most of the global autistic population lives.⁵

A 2026 systematic review and meta-analysis pooled seven cross-sectional studies from Egypt, Kenya, Uganda and Nigeria, covering 71,341 people; their reported prevalence ranged from 0.54% to 23.8%.⁶ Pooled, clinically confirmed autism was about 1% (95% confidence interval 0.0 to 1.0%), while screen-positive autism was about 4% (95% confidence interval 0.0 to 16.0%).⁶ The two figures come from different sets of studies grouped by how they counted, not from the same children screened and then confirmed, and the review's own comparison between the two groups was not statistically significant; the screen-positive pool was carried by a single study of toddlers screened with a checklist. The authors read the gap between the two figures as the gap between identifying risk and confirming a diagnosis, and call for more diagnostic services and training for primary-care staff.⁶ Their conclusion is that autism prevalence across these African studies "appears comparable to those reported in high-income countries".⁶

In one rural area of Kenya, the Kilifi Health and Demographic Surveillance System, researchers screened 11,223 children aged 6 to 9 and confirmed a neurodevelopmental disorder in 522 of them.⁷ Children who screened positive, and a randomly selected share of those who screened negative, were invited for clinical assessment, and the prevalence figures were adjusted for non-attendance between the two stages and for the screening tool's accuracy. Autism was found in about 1.6% (15.6 per 1,000 children; 95% confidence interval 12.6 to 19.1), against attention-deficit/hyperactivity disorder at about 5% and cognitive impairment at about 2.7%.⁷ The study also reports associations between neurodevelopmental disorders and preventable environmental factors: adverse events around birth, febrile infections and being landless were each linked to a higher likelihood, and malnutrition was more common among the children who had one, in a cross-sectional design that cannot show cause.⁷ This is one site, in one country; it does not stand for a continent.

How autistic traits are read depends on where they are read. A four-level framework sets out how culture and context shape the expression, recognition, interpretation and reporting of autistic traits, and notes that culturally appropriate screening and diagnostic tools are missing in most low- and middle-income settings.⁵ What counts as typical behavior, how families understand child development, and how much mental-health literacy a community has all sit between an autistic person and a diagnosis.⁵

The measured differences between countries are large. On the ASK-Q knowledge questionnaire, adapted for each country and given to 6,830 people in 13 countries, the gap between the highest-scoring country (Canada) and the lowest (Lebanon) was 17 points, and countries with higher economies scored higher.⁸ In one survey, South Korean adults said they wanted greater social distance from autistic people than US adults did, and part of the gap tracked "cultural tightness", a stronger emphasis on following social norms; the US adults, by contrast, more often raised concerns tied to heredity.⁹ In a survey of 277 French adults, knowledge and familiarity each went with lower stated stigma, but only close personal familiarity went with lower implicit bias.¹⁰

Across low- and middle-income countries, reviews of care for young children keep finding the same barriers: missing detection and diagnostic tools, low awareness and experienced stigma that reduce demand for care, and specialist service models that cannot scale.¹¹ What is being tried is task-sharing, adapted tools and community involvement.¹¹

The World Health Organization's Caregiver Skills Training program is for families of children with developmental disorders or delay. It was designed to be delivered by non-specialists in low-resource settings and adapted to local language and culture, and it was undergoing field-testing in more than 30 countries as of 2019.⁴ The program teaches caregivers skills to support their child's communication and daily living, and it does not require a formal diagnosis to join.¹²

Caregiver skills training as a class of programs has the strongest evidence. A systematic review and meta-analysis of 75 randomized trials, covering 4,746 individuals with neurodevelopmental disorders and their caregivers, found improvements in children's development (g = 0.30) and in their reported problem behavior (g = 0.41), no significant change in adaptive behavior, and larger improvements in caregivers' skills and knowledge (g = 0.72), wellbeing (g = 0.52) and family relationships (g = 0.76).¹³ The child effects are the smaller ones; the trials cover neurodevelopmental disorders generally, not autism alone, and they measure the child's behavior and the caregiver's wellbeing, not the autistic person's own life later on.

For the WHO program itself the evidence is small so far. Its pilot randomized trial, in Italy with 86 children with autism and their parents, found a large effect on the parents' skills supporting joint engagement and, as secondary outcomes, significant effects on parenting stress, self-efficacy and the children's use of gestures, but no statistically significant change in the children's autism severity or joint engagement.¹⁴ A small before-and-after study in urban India, with 22 caregivers, reported improved caregiver skills and reduced stress, and improvements in child communication and social interaction.¹⁵ Caregivers in rural Ethiopia found the program relevant but said the idea of caregiver-child play was unfamiliar to them and that a lack of toys made some exercises hard, with transport and homework time as practical problems.¹⁶ A small controlled pilot of a modified program in mainland China (15 caregiver-child pairs per group) found significant changes from before to after in the program group but mostly no statistically significant difference between the two groups.¹⁷ A single-arm pilot with 25 Eritrean and Ethiopian mothers in the United States found gains in parents' knowledge, skills, self-efficacy, depression and empowerment and in children's communication, sociability and sensory or cognitive awareness.¹⁸ A fully powered cluster randomized trial of the program, the SPARK trial, began recruiting in Kenya and Ethiopia in April 2024.¹⁹ A scoping review of 17 studies across Asia, Africa, Europe and North America found adaptations in language, content and delivery, with task-sharing to non-specialists as a facilitator and logistics, stigma and resources as barriers.¹²

These programs are offered to the caregiver, and what they measure is mostly the caregiver's skills and wellbeing and the child's behavior as rated by adults. The program is recent and its trials enrolled preschool children, so the children in them cannot yet be adults. No study found as of October 2026 has measured how autistic adults who were children in a caregiver skills training program fare in their own lives, which is the outcome that matters most to the person themselves.

Autistic writers who live outside the countries that produce most of the research describe how autism is seen where they are. Katie, a blogger in the Philippines, writes in Alex Ratcliffe's collection of accounts from around the world: "Autism isn't really spoken of in Asia, especially not here in the Philippines. Down's syndrome is, but not autism."^c1 She adds that the people who talk about autism where she lives are mostly parents, and that "there are programs for these kids. But nothing as yet for the adults."^c1

Where awareness exists, it does not always bring access. Harry, writing from Canada and diagnosed at 70, says that "the term 'autism' was unheard of when I was born", and that support for autistic adults "just doesn't exist around here."^c2 Pete Wharmby, writing in the UK, puts the point plainly: "Awareness, after all, does not necessarily equal action."^c3