Multiple large-scale studies confirm that early behavioural and developmental interventions begun before age 4 lead to significantly better long-term outcomes in communication, adaptive behaviour, and cognitive development. The Early Start Denver Model (ESDM) — a naturalistic, play-based therapy — showed in randomised controlled trials that children who received 20+ hours/week of ESDM from 18–30 months showed significant gains in IQ, language, and adaptive behaviour compared to community control groups.

A 2020 meta-analysis published in JAMA Pediatrics covering 6,240 participants across 34 studies found that early comprehensive behavioural intervention produced moderate-to-large effect sizes for language and adaptive behaviour outcomes. The key finding: earlier start age consistently predicted better outcomes, regardless of intervention type.

Practical takeaway for parents: If you suspect autism, request a developmental screening immediately — do not wait for a formal diagnosis. Early support can begin before diagnosis is confirmed.

Autism is highly heritable — twin studies show concordance rates of 60–90% in identical twins compared to 0–30% in fraternal twins, indicating a strong genetic component. However, no single "autism gene" exists. Research has identified over 100 genes associated with increased autism risk, most contributing small individual effects.

Large genomic studies (including the SPARK cohort with 50,000+ participants) have identified de novo (new) mutations — genetic changes not inherited from parents — in approximately 10–15% of autism cases. Copy Number Variations (CNVs) — deletions or duplications of chromosomal segments — account for another 5–10%.

Environmental factors also play a role, though no single environmental cause has been established. Advanced parental age, prenatal exposure to certain medications (notably valproate), and very premature birth are associated with modestly increased risk. Critically: vaccines do not cause autism. This has been conclusively established by studies involving millions of children across multiple countries.

Neuroimaging research has identified several consistent brain differences in autistic individuals, though these are averages across populations and vary widely between individuals. Key findings include:

• Atypical connectivity: Differences in long-range neural connectivity — some networks show reduced connectivity (underconnectivity), others show increased local connectivity. This may relate to differences in sensory processing and executive function.
• Amygdala differences: The amygdala — involved in emotional processing — shows atypical development in many autistic individuals, potentially contributing to different emotional responses and social processing.
• Mirror neuron system: Some research suggests differences in mirror neuron activity, though the significance and interpretation of these findings remains debated.
• Cerebellum: Structural and functional differences in the cerebellum may contribute to motor coordination differences and some cognitive features.

Importantly, brain differences do not equate to deficits. Many autistic individuals show exceptional abilities in pattern recognition, attention to detail, and systematic thinking that are directly related to their neurology.

Autism is diagnosed approximately 4 times more often in males than females — but growing evidence suggests this ratio reflects diagnostic bias rather than true prevalence differences. Research on the "female autism phenotype" and "masking" (camouflaging autistic traits to fit social expectations) shows that many autistic girls and women go undiagnosed for years or decades.

A 2020 study in Autism (the journal) found that autistic women received their diagnosis on average 2.5 years later than autistic men, often after years of misdiagnosis (most commonly anxiety disorder, borderline personality disorder, or eating disorders). The social camouflaging common in autistic females — mimicking others' social behaviour, scripting conversations, masking sensory distress — leads clinicians to miss core features.

The diagnostic criteria themselves were largely developed from studies of male subjects, meaning standard assessment tools may underidentify autism in females. New assessment approaches and clinician training focusing on the female phenotype are being developed.

The majority of autistic individuals have at least one co-occurring condition. Understanding these is critical for proper support and treatment: