What we’re looking at
Autism spectrum disorder (ASD) represents one of the most intensively researched neurodevelopmental conditions of the past two decades, yet fundamental questions about prevalence, diagnosis timing, treatment effectiveness, and co-occurring conditions remain only partially answered. Parents, educators, and clinicians face a bewildering landscape of interventions—from behavioral therapies to medications to physical exercise—often without clear guidance on what works, for whom, and when.
This evidence review synthesizes eight major meta-analyses published between 2018 and 2025, collectively analyzing data from hundreds of individual studies and over half a million participants worldwide. We examine five core questions: How common is autism globally? When are people typically diagnosed? What other conditions frequently co-occur? Which interventions show the strongest evidence? And crucially, where do significant gaps in our knowledge remain?
The evidence — what studies actually found
Global prevalence and diagnosis timing
A comprehensive 2022 systematic review analyzed 74 studies encompassing over 30 million participants across five continents [PMID 35804408]. The overall global prevalence of ASD was estimated at 0.6% (ranging from 0.4% to 1.0% depending on methodology). Significant geographic variation emerged: Asia showed 0.4% prevalence, Europe 0.5%, the Americas 1.0%, Africa 1.0%, and Australia the highest at 1.7%. These differences likely reflect both genuine variation and disparities in diagnostic infrastructure and awareness.
Regarding when diagnosis occurs, a 2021 meta-analysis of 56 studies covering 40 countries found the current mean age at diagnosis is approximately 60 months—just over five years old [PMID 33213190]. When restricting analysis to studies of children aged 10 and under, the mean dropped to 43 months (about 3.6 years). This represents the diagnostic landscape from 2012-2019, suggesting modest progress compared to earlier decades, but still falls short of the goal of diagnosis by age two that developmental screening guidelines often target.
Co-occurring mental health conditions
Two major meta-analyses examined psychiatric comorbidities in autism populations. The first, published in 2019, synthesized 96 studies and revealed strikingly high rates of co-occurring mental health diagnoses [PMID 31447415]. Attention-deficit hyperactivity disorder (ADHD) showed a pooled prevalence of 28%, anxiety disorders 20%, sleep-wake disorders 13%, disruptive and impulse-control disorders 12%, depressive disorders 11%, obsessive-compulsive disorder 9%, bipolar disorders 5%, and schizophrenia spectrum disorders 4%. Nearly all these rates substantially exceeded general population estimates.
A 2023 review expanded this analysis beyond psychiatric conditions to include physical and developmental comorbidities, examining 340 publications with roughly 590,000 participants [PMID 37913872]. The most frequent co-occurring conditions included developmental coordination disorder, sleep-wake problems, gastrointestinal issues, ADHD, anxiety, overweight/obesity, feeding and eating disorders, elimination disorders, disruptive behavior, and somatic symptom disorders. Importantly, prevalence estimates differed significantly between population-based registry studies and clinical sample studies, with the latter showing higher rates—likely reflecting referral bias, as individuals seeking clinical services typically have more complex presentations.
Cognitive and behavioral profiles
A 2018 meta-analysis of executive function (the mental processes that enable planning, focus, task-switching, and impulse control) included 235 studies with over 14,000 participants [PMID 28439105]. Results demonstrated a moderate overall executive dysfunction in ASD (effect size g=0.48), with similar impairment across specific executive function subdomains including working memory, planning, inhibition, and cognitive flexibility. Contrary to some theories, the data did not support the idea that autism involves selective impairment of certain executive functions while sparing others—instead, broad executive challenges appeared relatively stable across childhood and adolescence.
Emotion regulation difficulties represent another well-documented feature. A 2024 meta-analysis of 55 studies found children and adolescents with ASD demonstrated significantly greater emotion dysregulation compared to neurotypical peers [PMID 38401510]. They employed more maladaptive emotion regulation strategies (like suppression or rumination) and fewer adaptive strategies (like cognitive reappraisal or problem-solving). Importantly, greater emotion dysregulation correlated with more severe autism symptoms, poorer social skills, and higher levels of both internalizing symptoms (anxiety, depression) and externalizing problems (aggression, defiance).
Intervention effectiveness
Two meta-analyses examined specific intervention approaches. For pharmacological treatment, a 2025 Cochrane review analyzed 17 randomized controlled trials of atypical antipsychotic medications, primarily in children [PMID 40396498]. Risperidone and aripiprazole showed modest reductions in irritability compared to placebo (mean differences of -7.89 and -6.26 points respectively on standardized scales), though evidence quality was rated as low due to methodological concerns. Lurasidone showed little to no benefit. Critically, weight gain occurred at more than twice the rate with active medication versus placebo, and extrapyramidal side effects (movement problems) occurred at 2.36 times the placebo rate. The review found very limited data on adults with autism.
Physical exercise therapy (PET) emerged as a potentially promising non-pharmacological intervention. A 2024 meta-analysis of 28 randomized controlled trials with 1,081 participants found improvements in motor performance (large effect size), restricted repetitive behaviors (moderate effect), and social dysfunction (moderate effect) [PMID 38986177]. However, only three studies met high-quality standards, and the overall risk of bias was substantial, warranting cautious interpretation.
The emotion regulation meta-analysis also examined intervention studies, finding that non-pharmacological interventions (primarily cognitive-behavioral and skills-based approaches) showed significant improvements in both emotion regulation capacity and reductions in emotion dysregulation [PMID 38401510]. This suggests targeted psychological interventions may address a core difficulty for many individuals with autism.
How it might work — biological mechanisms
While these meta-analyses focused primarily on observable outcomes rather than underlying biology, several mechanistic themes emerge from the pattern of findings.
Executive function and emotion regulation connections
The co-occurrence of executive dysfunction and emotion dysregulation makes neurobiological sense: both rely heavily on prefrontal cortex function and connectivity with limbic regions like the amygdala. Neuroimaging studies (not included in these meta-analyses but referenced in their discussions) have documented atypical development of these brain networks in autism. When executive functions like cognitive flexibility and inhibitory control are impaired, regulating emotional responses becomes substantially harder—you need intact executive systems to implement adaptive emotion regulation strategies like reframing a situation or distracting yourself.
The executive function meta-analysis noted that deficits appeared broadly distributed rather than selectively affecting specific processes [PMID 28439105]. This pattern suggests autism may involve widespread differences in how brain networks mature and communicate, rather than damage to discrete brain regions—consistent with current theories emphasizing altered neural connectivity.
The social-emotional cascade
The emotion regulation research revealed a crucial finding: greater emotion dysregulation predicted worse social skills [PMID 38401510]. This suggests a potential cascade where difficulties reading social cues combine with challenges managing emotional responses, creating a compounding effect on social functioning. When you struggle to interpret others’ expressions and also have difficulty modulating your own emotional reactions, social interactions become exponentially more challenging.
This helps explain why social skills and emotion regulation often improve together during intervention—they’re functionally intertwined, not separate domains.
Physical exercise mechanisms
How might physical exercise improve autism symptoms? Proposed mechanisms include increased brain-derived neurotrophic factor (BDNF, which supports neuron growth and plasticity), reduced inflammation, improved executive function through enhanced prefrontal cortex activity, and structured social interaction during group activities [PMID 38986177]. The relatively large effect on motor performance and moderate effects on repetitive behaviors and social function suggest exercise may benefit autism through multiple pathways simultaneously.
Medication mechanisms and trade-offs
Atypical antipsychotics like risperidone and aripiprazole primarily work by modulating dopamine and serotonin neurotransmitter systems [PMID 40396498]. While this can reduce irritability and aggression, these medications don’t address core autism features like social communication difficulties. The substantial side effects—particularly weight gain and movement problems—reflect these drugs’ broad effects on brain systems regulating metabolism and motor control. The risk-benefit calculation becomes complex: modest symptom improvement against potentially serious adverse effects.
Where the evidence is strong (and weak)
Strong evidence
Several findings rest on robust foundations. The global prevalence estimate of approximately 0.6% comes from 74 studies and over 30 million participants [PMID 35804408]. The high rates of co-occurring mental health conditions—particularly ADHD, anxiety, and sleep problems—emerged consistently across nearly 100 studies [PMID 31447415]. The presence of broad executive dysfunction in autism is supported by 235 studies and over 14,000 participants [PMID 28439105].
For interventions, the evidence that risperidone and aripiprazole can reduce irritability in children with autism is reasonably strong, based on multiple randomized trials, though effect sizes are modest and side effects significant [PMID 40396498].
Moderate evidence
Emotion dysregulation in autism is well-documented across 55 studies, and non-pharmacological interventions show promise for improving emotion regulation [PMID 38401510]. However, the specific intervention components most critical for success remain unclear—studies varied widely in approach, from cognitive-behavioral therapy to mindfulness to parent training.
Physical exercise shows potential benefits, but only three of 28 included studies met high-quality standards [PMID 38986177], limiting confidence in effect size estimates.
Weak or absent evidence
Critical gaps appear throughout. Evidence in adults with autism is remarkably sparse—the antipsychotic medication review found only one study of 31 adults versus 16 studies with 996 children [PMID 40396498]. We lack good data on optimal diagnosis timing strategies, long-term intervention outcomes, treatment effectiveness for different autism subgroups, and comparative effectiveness of different intervention approaches.
The diagnosis timing meta-analysis found numerous factors potentially influencing age at diagnosis (type of ASD diagnosis, presence of additional conditions, gender), but results were often “conflicting or inconclusive” [PMID 33213190]. We don’t yet have clear evidence on what strategies most effectively lower diagnosis age.
Geographic differences in prevalence remain poorly understood—is Australia’s 1.7% rate versus Asia’s 0.4% primarily due to diagnostic practices, awareness, actual biological differences, or environmental factors [PMID 35804408]? The available data can’t definitively answer this.
Practical takeaways
For parents and caregivers
If you’re concerned about possible autism in your child, pursue evaluation promptly—the current average diagnosis age of five years suggests many children aren’t identified as early as possible [PMID 33213190]. Early diagnosis enables earlier intervention, which generally yields better outcomes.
Be prepared for co-occurring conditions: if your child has autism, there’s a 28% chance they also meet criteria for ADHD, 20% for an anxiety disorder, and 13% for sleep problems [PMID 31447415]. These aren’t separate bad luck—they’re intrinsically connected to autism neurobiology and deserve attention alongside core autism features.
Emotion regulation difficulties are common, not a personal failing. Targeted interventions addressing emotion regulation skills show meaningful benefits [PMID 38401510]. Consider asking providers about evidence-based approaches to building these skills.
If considering medication, understand the trade-offs. Drugs like risperidone may reduce severe irritability and aggression, but improvements are often modest and side effects like significant weight gain are common [PMID 40396498]. Medication works best as part of comprehensive treatment, not as a standalone solution.
Physical exercise and structured movement activities may offer benefits beyond general health—emerging evidence suggests improvements in motor skills, repetitive behaviors, and social function [PMID 38986177].
For clinicians
Comprehensive assessment should routinely screen for common comorbidities, particularly ADHD, anxiety, sleep disorders, and gastrointestinal problems [PMID 37913872]. These co-occurring conditions substantially impact quality of life and treatment planning.
Executive function assessment provides valuable information about cognitive strengths and weaknesses, though no single executive function test achieves adequate diagnostic sensitivity [PMID 28439105]. Use executive function measures to guide intervention rather than diagnosis.
Emotion regulation should be part of standard evaluation and a target for intervention [PMID 38401510]. It predicts both internalizing and externalizing symptoms and impacts social functioning.
When prescribing atypical antipsychotics, carefully monitor weight and metabolic parameters. The evidence supports use for severe irritability in children when behavioral interventions prove insufficient, but benefits are moderate and side effects substantial [PMID 40396498].
For researchers and policymakers
Adult autism remains severely understudied. Nearly all intervention trials focus on children, leaving clinicians with minimal evidence when treating adults [PMID 40396498].
Methodological quality needs improvement. Most physical exercise studies had high risk of bias [PMID 38986177], and heterogeneity across studies remains enormous even after accounting for known moderators [PMID 31447415]. Standardized measurement approaches and transparent reporting of population characteristics would substantially improve evidence synthesis.
We need better understanding of what lowers age at diagnosis. While average diagnosis age has decreased modestly, 43-60 months still exceeds the goal of diagnosis by 24 months [PMID 33213190].
Limitations of this review
This synthesis relies entirely on published meta-analyses, inheriting their limitations. Publication bias affects all meta-analyses—studies with null or negative findings are less likely to be published, potentially inflating apparent effect sizes.
The included meta-analyses varied in quality. Some used rigorous Cochrane methodology [PMID 40396498], while others had less stringent inclusion criteria. Risk of bias in underlying primary studies was often substantial, particularly for intervention research.
Heterogeneity was enormous across most meta-analyses, even after subgroup analyses. The autism population is highly diverse—ranging from minimally verbal individuals with intellectual disability to verbally fluent individuals with average or above-average IQ. Intervention effects may vary dramatically across this spectrum, but most meta-analyses lacked sufficient data for fine-grained subgroup analysis.
Geographic representation was uneven. Most studies came from North America and Europe, limiting generalizability to other world regions. Cultural factors may influence both symptom presentation and intervention effectiveness.
Most data came from children, with adults severely underrepresented. We can’t assume findings in children apply equally to adults.
No meta-analysis included covered important topics like specific behavioral interventions (Applied Behavior Analysis, social skills training, etc.), communication interventions, educational approaches, or most dietary and biomedical treatments. This review cannot address those intervention categories.
Bottom line
Autism spectrum disorder affects approximately 0.6% of the global population, with substantial geographic variation in documented prevalence. Despite advances in awareness and screening, average diagnosis occurs around age five—later than optimal for maximizing early intervention benefits.
The autism phenotype extends beyond social communication and restricted interests to encompass broad executive dysfunction, emotion dysregulation, and high rates of co-occurring mental health conditions. ADHD, anxiety, sleep problems, and depressive disorders each affect substantial minorities of the autism population at rates well above general population baselines. These comorbidities aren’t coincidental—they appear intrinsically linked to autism neurobiology and significantly impact quality of life.
For intervention, modest evidence supports atypical antipsychotic medications for reducing severe irritability in children, though side effects are concerning and benefits limited. Emerging evidence suggests physical exercise and targeted emotion regulation interventions may offer benefits, but more high-quality research is needed. Critically, evidence in adults with autism remains extremely limited across all intervention types.
The path forward requires earlier identification, comprehensive assessment of co-occurring conditions, individualized treatment planning that addresses both core features and comorbidities, and urgent attention to the evidence gap in adult autism. While substantial progress has occurred in understanding autism epidemiology and phenomenology, major questions about optimal intervention remain only partially answered.
FAQ: Autism Diagnosis, Prevalence, and Treatment
What percentage of people worldwide have autism spectrum disorder?
According to a comprehensive 2022 systematic review analyzing over 30 million participants across five continents (PMID 35804408), the global prevalence of autism spectrum disorder is approximately 0.6%. However, this figure varies significantly by region: Asia shows 0.4% prevalence, Europe 0.5%, the Americas 1.0%, Africa 1.0%, and Australia the highest at 1.7%. These geographic differences likely reflect both genuine variation in autism rates and disparities in diagnostic infrastructure, clinical awareness, and screening practices across different countries and healthcare systems.
At what age is autism typically diagnosed?
A 2021 meta-analysis covering 56 studies from 40 countries (PMID 33213190) found that the current mean age at autism diagnosis is approximately 60 months—just over five years old. When analyzing only studies of children aged 10 and under, the mean diagnosis age dropped to 43 months (about 3.6 years). While this represents modest progress compared to earlier decades, it still falls considerably short of the developmental screening goal of diagnosing autism by age two, which is when early intervention can be most effective.
Why does early diagnosis of autism matter for treatment outcomes?
Early diagnosis enables early intervention, which research consistently shows can dramatically improve developmental outcomes for children with autism. The critical window for brain plasticity—when neural pathways are most adaptable—occurs in the first few years of life. Behavioral interventions started during this period can significantly enhance communication skills, social interaction, and adaptive behaviors. Despite current guidelines recommending diagnosis by age two, the global average diagnosis age remains around five years, meaning many children miss the optimal intervention window during their most formative developmental period.
What other conditions commonly occur alongside autism?
Meta-analytic evidence from 96 studies (PMID 31447415) reveals remarkably high rates of co-occurring conditions in autism populations. Attention-deficit hyperactivity disorder (ADHD) affects approximately 28% of individuals with autism, anxiety disorders 20%, sleep-wake disorders 13%, disruptive and impulse-control disorders 12%, and depressive disorders 11%. Additional comorbidities include obsessive-compulsive disorder (9%), bipolar disorders (5%), and schizophrenia spectrum disorders (4%). These rates substantially exceed what’s observed in the general population, highlighting the complex clinical picture many individuals with autism experience and the importance of comprehensive assessment beyond autism symptoms alone.
How do autism prevalence rates differ between boys and girls?
Research consistently demonstrates that autism is diagnosed more frequently in males than females, though the exact ratio varies by study methodology and population. The geographic variation in overall prevalence (ranging from 0.4% to 1.7% across different continents) partly reflects diagnostic practices that may overlook autism in girls, who often present with different symptom profiles than boys. Girls with autism may exhibit better social camouflaging, less obvious repetitive behaviors, and different special interest patterns, potentially leading to underdiagnosis or later diagnosis. This sex difference in diagnosis rates represents an important area where diagnostic criteria and clinical awareness continue to evolve.
Studies referenced
-
PMID 38401510 — Meta-analysis of 55 studies examining emotion regulation and dysregulation in children/adolescents with autism, finding significant challenges and evidence that non-pharmacological interventions can improve emotion regulation skills.
-
PMID 35804408 — Systematic review of 74 studies with over 30 million participants estimating global autism prevalence at 0.6%, with significant variation across continents ranging from 0.4% (Asia) to 1.7% (Australia).
-
PMID 37913872 — Comprehensive review of 340 publications examining co-occurring conditions in autism, finding high rates of developmental coordination disorder, sleep problems, gastrointestinal issues, ADHD, anxiety, and other conditions.
-
PMID 28439105 — Meta-analysis of 235 studies with 14,081 participants demonstrating moderate executive dysfunction in autism that is relatively stable across development, though individual measures showed limited diagnostic sensitivity.
-
PMID 33213190 — Review of 56 studies from 40 countries finding current mean age at autism diagnosis is approximately 60 months overall and 43 months in studies restricted to children aged 10 and under.
-
PMID 38986177 — Meta-analysis of 28 randomized controlled trials examining physical exercise therapy in autism, finding improvements in motor performance, repetitive behaviors, and social function, though study quality was generally low.
-
PMID 31447415 — Systematic review of 96 studies examining mental health comorbidities in autism, finding elevated rates of ADHD (28%), anxiety disorders (20%), depression (11%), and other psychiatric conditions compared to general population.
-
PMID 40396498 — Cochrane review of 17 randomized controlled trials examining atypical antipsychotic medications in autism, finding risperidone and aripiprazole reduce irritability modestly in children but with concerning side effects including weight gain and movement problems.