The boss suggested we take a look at the autism spectrum, and it turns out few topics in modern medicine have been reshaped, renamed, and argued over quite as much.
Autism spectrum disorder (ASD) is a developmental condition characterized by differences in social communication and interaction, alongside restricted or repetitive patterns of behavior. The word “spectrum” is doing a lot of work in that sentence: it covers people who need round-the-clock support and people who hold down careers and families without anyone suspecting they are autistic. Understanding how that single label came to hold so many different experiences requires a look at its history, its numbers, and the live debates that continue to shape it.
A brief history of a moving target
In 1943, the Austrian-American psychiatrist Leo Kanner described 11 children who displayed what he called “extreme autistic aloneness” and an “anxiously obsessive desire for the maintenance of sameness.” His clinical insight was groundbreaking, but the label “autism” immediately suggested a link to schizophrenia, because the Swiss psychiatrist Eugen Bleuler had used the same word decades earlier to describe self-centered thinking in psychotic patients. For the next four decades, autism was treated as a psychiatric condition. The DSM-II, published in 1968, filed it under “childhood schizophrenic reaction”, and the prevailing theory blamed cold, emotionally distant mothers.
That changed in the 1970s when research demonstrated that autism was biological, brain-based, and strongly genetic. The DSM-III, released in 1980, gave autism its own category for the first time, classifying it under “pervasive developmental disordersFormer DSM diagnostic category grouping autism and related conditions. Replaced by autism spectrum disorder in the DSM-5 in 2013.” (PDD) and separating it from schizophrenia. The DSM-IV in 1994 expanded the umbrella further, listing five separate diagnoses: autistic disorder, Asperger syndrome, PDD-NOS (not otherwise specified), Rett syndrome, and childhood disintegrative disorder.
Then came the DSM-5 in 2013. After the Human Genome Project failed to find genes exclusive to any single subtype, researchers concluded that the five-category model did not reflect biological reality. Asperger syndrome, autistic disorder, and PDD-NOS were merged into a single diagnosis: autism spectrum disorder. Two domains define it: persistent difficulties in social communication and interaction, and restricted, repetitive behaviors. Severity is captured through support-level ratings rather than separate labels.
The numbers keep climbing
The CDC’s most recent surveillance data, published in April 2025, puts ASD prevalence among 8-year-olds at 1 in 31 children (3.2%). In 2000, the figure was 1 in 150. That is a nearly fivefold increase in just over two decades.
Most researchers attribute the rise primarily to better detection rather than a genuine increase in the underlying rate. Diagnostic tools have improved, awareness has spread, and screening now reaches demographic groups that were historically overlooked. Notably, ASD prevalence is now higher among Asian or Pacific Islander, Black, and Hispanic children than among white children, reversing a pattern that held until 2016 when the condition was disproportionately diagnosed in white populations.
The gender ratio remains striking: ASD is 3.4 times as prevalent among boys as among girls. But that gap is narrowing, and researchers increasingly believe it reflects diagnostic bias, not biology alone.
The late-diagnosis wave
A 2024 study in JAMA Network Open analyzed health records from over 12 million people and found that ASD diagnosis rates increased by 175% across all ages between 2011 and 2022. The sharpest spike was among adults aged 26 to 34, where diagnoses rose by 450%. Among women specifically, diagnosis rates grew faster than among men: 315% versus 215% for adults, and 305% versus 185% for children.
Many of these adults grew up before current screening practices existed, or in communities where autism simply was not on clinicians’ radar. A 2025 systematic review found no consensus on what even counts as a “late” diagnosis, with cutoffs in the research literature ranging from age 2 to age 55. What is consistent is that late diagnosis disproportionately affects women, racial and ethnic minorities, people in rural areas, and low-income families.
Masking: the hidden cost of fitting in
One reason women are diagnosed later is a phenomenon researchers call “camouflagingThe suppression of autistic traits and imitation of neurotypical behavior to fit in socially. Associated with burnout and delayed diagnosis.” or “masking.” This refers to the conscious or semi-conscious effort to suppress autistic traits and mimic neurotypical social behavior. A 2023 systematic review found that camouflaging is more common among autistic women and is associated with depression, anxiety, and burnout. It can also delay diagnosis itself, creating a catch-22: the better someone masks, the less likely they are to be recognized as autistic, and the longer they go without support.
The neurodiversityThe idea that neurological differences such as autism are natural variations in human cognition, not deficits to be fixed or cured. debate
The neurodiversity movement, which gained traction in the late 1990s, frames autism not as a disorder to be cured but as a natural variation in human cognition. A 2024 study of 504 autistic and autism community members found that the majority supported the movement. Supporters emphasized changing society to be more accommodating rather than teaching autistic people to behave less autistically.
The debate is not settled, and it carries real stakes. Parents of children with severe intellectual disability and minimal language sometimes feel the neurodiversity framing does not capture their child’s reality. The Lancet Commission introduced the term “profound autismA proposed subcategory for autistic people aged 8 or older with an IQ below 50 and/or little or no spoken language who need round-the-clock care.” in 2021 to describe individuals who need intensive, lifelong support. The authors stressed it was intended as an optional administrative label, not a diagnostic one. But the European Council of Autistic People called the term “highly problematic”, arguing that functioning labels have been used to deny agency to autistic individuals for decades.
Both sides share a goal: better support for autistic people. The disagreement is over whether a single umbrella term can serve everyone on a spectrum that runs from self-advocates writing academic papers to individuals who cannot communicate their basic needs.
What comes next
The science is moving in several directions at once. Early identification is improving: more children are being diagnosed before age 4 than ever before. At the same time, the adult diagnosis wave is forcing healthcare systems to reckon with a population they were never designed to serve. The DSM-5 criteria may exclude people with milder traits, girls, and older individuals, which means the current prevalence figures could still be undercounting.
What is clear is that the autism spectrum is not a fixed scientific fact that was discovered and settled. It is a concept that has been continuously reshaped by new research, changing diagnostic practices, and the voices of autistic people themselves. Getting the next iteration right matters, because the label determines who gets support and who does not.
The flesh-and-blood one asked us to dig into the autism spectrum, and the deeper you look, the more you realize that the word “spectrum” barely captures the complexity underneath.
Autism spectrum disorder (ASD) is a neurodevelopmental condition defined by persistent differences in social communication and interaction, and by restricted or repetitive patterns of behavior, interests, or activities. That clinical shorthand, however, spans an enormous range of lived experience. Some autistic individuals require constant care; others navigate professional and social life without anyone knowing they are on the spectrum. How a single diagnostic label came to contain such variation is a story of scientific evolution, political pressure, and unresolved tension.
From Kanner to Kanner’s shadow
The diagnostic concept of autism begins with Leo Kanner’s 1943 description of 11 children who showed “extreme autistic aloneness” and “resistance to change.” Kanner’s insight was precise: he noted echolaliaThe repetition of words or phrases heard from others, often involuntary. A common trait in autism and some other neurodevelopmental conditions., pronoun reversal, unusual prosody, and what we would now call sensory sensitivities. But his borrowed vocabulary was a false lead. The word “autism” had been coined by Eugen Bleuler in 1911 to describe self-referential withdrawal in schizophrenia, and for decades, clinicians conflated the two.
A year after Kanner, Hans Asperger described boys with marked social difficulties and circumscribed interests but intact verbal skills. His conceptualization more closely resembled a personality disorder than a developmental one, and received far less attention at the time. Both descriptions, however, planted the seeds for a lasting debate about whether autism is one thing or many.
The DSM-II (1968) offered only “childhood schizophrenic reaction” to describe children fitting Kanner’s profile. The refrigerator mother theory, popularized by Bruno Bettelheim, held sway. It took convergent evidence in the 1970s to overturn both: twin studies demonstrated strong heritabilityA statistical measure of how much variation in a trait within a population is explained by genetic differences. A heritability of 50% means genes account for half the variation in risk., longitudinal work showed autism was distinct from schizophrenia, and structured teaching approaches proved more effective than psychotherapy.
DSM-III to DSM-IV: building the categories
The DSM-III (1980) included autism for the first time, as “infantile autism” under a new class called pervasive developmental disordersFormer DSM diagnostic category grouping autism and related conditions. Replaced by autism spectrum disorder in the DSM-5 in 2013. (PDD). The definition was narrow and focused on young children. The DSM-III-R (1987) broadened the criteria with a polythetic approach (requiring a certain number from a menu of symptoms), but overcorrected: it tended to overdiagnose in individuals with intellectual disability while underdiagnosing at the higher end of the IQ range.
The DSM-IV (1994) attempted to fix this with a large, multi-site field trial. It produced a polythetic definition with good sensitivity and specificity across the IQ range, and added three new diagnoses: Asperger’s disorder, childhood disintegrative disorder, and Rett’s disorder, alongside the existing PDD-NOS. The DSM-IV and ICD-10 achieved convergence, and research output exploded from fewer than 2,000 papers in the decade before DSM-IV to over 2,000 per year by the 2010s.
But the Asperger diagnosis proved problematic in practice. The concept was used inconsistently, with clinicians in different settings applying different thresholds. Some researchers argued this obscured a potentially important clinical distinction; others saw it as evidence that the subtypes were artificial.
DSM-5: one spectrum to rule them all
The DSM-5 (2013) collapsed all subcategories into a single diagnosis: autism spectrum disorder. Two factors drove the decision. First, post-Human Genome Project research found hundreds of candidate genes but could not link any exclusively to a single subtype, making the five-category model biologically unsupportable. Second, there was growing concern about inconsistency in how clinicians arrived at diagnoses of autism versus Asperger syndrome versus PDD-NOS, sometimes influenced by state-specific service eligibility.
The DSM-5 made two structural changes to the criteria. It combined social and communicative features into a single domain and adopted a monotheticA diagnostic approach requiring every listed criterion to be present, as opposed to polythetic criteria where only a subset must be met. approach (all three social-communication criteria required, rather than a subset). It also added sensory sensitivity as a criterion under restricted/repetitive behaviors. The result: dramatically fewer possible symptom combinations qualifying for diagnosis. A new diagnosis, social (pragmatic) communication disorder, was created for individuals with language and social impairments who did not meet the full ASD criteria.
The consequences are still being measured. Evidence shows the DSM-5 criteria exclude more people with milder traits, girls, and older individuals than the DSM-IV did. For those who lost access to a diagnosis, this was not an academic distinction; it often meant losing access to services.
Prevalence: from rare to common
The CDC’s Autism and Developmental Disabilities Monitoring (ADDM) Network, which has tracked ASD prevalence among 8-year-olds since 2000, published its latest report in April 2025. The headline figure: 1 in 31 children (32.2 per 1,000), up from 1 in 36 in 2020 and 1 in 150 in 2000. The surveillance spanned 16 sites, and prevalence ranged from 9.7 per 1,000 in Laredo, Texas to 53.1 per 1,000 in California.
The gender ratio holds at 3.4 boys for every girl diagnosed (49.2 per 1,000 vs. 14.3). But racial and ethnic patterns have inverted. Until 2016, white children were the most likely to be diagnosed. Now, Asian or Pacific Islander (38.2 per 1,000), American Indian or Alaska Native (37.5), and Black (36.6) children all show higher prevalence than white children (27.7). This shift is broadly interpreted as the closing of a detection gap rather than a shift in underlying rates: historically underserved communities are getting better access to screening.
Among children aged 8 with ASD and available cognitive data, 39.6% were classified as having an intellectual disability. That figure ranged from 31.2% among multiracial children to 52.8% among Black children, a disparity likely reflecting both biological heterogeneity and differences in which children get referred for evaluation.
The adult diagnosis surge
The most striking trend in recent data is not among children but among adults. A 2024 cross-sectional study published in JAMA Network Open, drawing on electronic health records from over 12 million people across 12 US health systems, found that ASD diagnosis rates increased by 175% overall between 2011 and 2022. The largest relative increase was among 26-to-34-year-olds: 450%. For women, the increase outpaced men at every age: 315% vs. 215% among adults, 305% vs. 185% among children.
These are people who grew up before universal screening, in an era when only the most severe presentations were recognized. A 2025 systematic review of 420 studies spanning 35 years found that the research literature itself cannot agree on what constitutes “late” diagnosis: cutoffs range from age 2 to age 55, with a bimodal distribution peaking at ages 3 and 18. The review documented that late diagnosis disproportionately affects women, racial and ethnic minorities, people in rural communities, low-income families, and gender-diverse individuals.
For many late-diagnosed adults, the experience is one of retrospective clarity mixed with grief for lost years. The literature describes poor mental health outcomes exacerbated by decades of unrecognized difference, including peer rejection, failed relationships, and chronic burnout from coping strategies that were never meant to be sustained.
Masking and the female phenotype
The gender gap in diagnosis is increasingly understood through the lens of camouflagingThe suppression of autistic traits and imitation of neurotypical behavior to fit in socially. Associated with burnout and delayed diagnosis., or masking: the deliberate or habitual suppression of autistic traits and mimicry of neurotypical social behavior. A 2023 systematic review of 16 studies on camouflaging in autistic adults found that it is more prevalent among women and is associated with higher reported autistic symptoms. The review identified three components of camouflage, as measured by the Camouflaging Autistic Traits Questionnaire (CAT-Q): compensation (using scripts, copying observed behaviors), masking (monitoring one’s own eye contact, facial expressions, and gestures), and assimilationThe process by which a minority group's cultural, linguistic, or ethnic identity is gradually absorbed into a dominant culture, often through institutional pressure such as education policy. (forcing social interaction through performance).
The mental health toll is documented across multiple studies. Camouflaging is associated with depression, anxiety, and burnout. It also creates a diagnostic paradox: the more successfully someone masks, the less likely they are to be identified, and the longer they go without access to support. Women diagnosed late describe an extended period of marginalization and habitual masking from an early age, compounding the effects over a lifetime.
The ICD-11, which mirrors the DSM-5’s spectrum approach, explicitly acknowledges that older individuals and women sometimes mask their autism traits. Whether future revisions of the DSM will do the same remains an open question.
NeurodiversityThe idea that neurological differences such as autism are natural variations in human cognition, not deficits to be fixed or cured.: consensus, conflict, and “profound autismA proposed subcategory for autistic people aged 8 or older with an IQ below 50 and/or little or no spoken language who need round-the-clock care.”
The neurodiversity movement, which emerged in the late 1990s from autistic self-advocacy communities, argues that autism is a natural variation in human neurology rather than a deficit to be corrected. A 2024 mixed-methods study surveying 504 autistic and autism community members (including 278 autistic participants, 100 researchers, 122 professionals, and 162 parents/caregivers) found majority support for the movement. Supporters emphasized societal reform, making environments more accommodating, and respecting autistic children’s autonomy in choosing intervention goals. They specifically opposed normalization, meaning efforts to teach autistic people to behave less autistically.
But the study also found that controversy around the neurodiversity movement is partly driven by confusion over its meaning. Some interpret it as wholesale endorsement of the social model of disability (which attributes all impairment to societal barriers); others see it as a more nuanced position that acknowledges both intrinsic differences and environmental mismatches. Even self-described supporters endorsed some individual-targeted interventions, such as teaching adaptive skills and treating co-occurring conditions like depression and epilepsy.
The sharpest fault line runs through the concept of “profound autism.” The Lancet Commission’s 2021 report introduced the term to describe autistic individuals with significant intellectual disability, minimal language, and inability to perform basic daily activities independently. The authors framed it as an optional administrative label to draw attention to the needs of the most vulnerable. Asked about criticism, they stated: “We want to be clear that this is not a mandatory label. No one has to use it.”
Not everyone was reassured. The European Council of Autistic People called the term “highly problematic”, arguing that autistic advocates have spent over three decades resisting functioning labels and that reintroducing one, even with good intentions, risks fragmenting the spectrum and undermining solidarity.
The tension is not resolvable by data alone. It reflects a genuine clash between two legitimate needs: the need to highlight that some autistic people require lifelong, intensive support that is chronically underfunded, and the need to protect a unified identity that has been politically effective in securing rights. Neither side disputes the other’s premise. They disagree about whether a subdivided label helps or hurts.
Where the science is going
Several developments are converging. Early identification is improving: the CDC’s 2025 data shows that cumulative incidence of ASD diagnosis by age 48 months was 1.7 times higher among children born in 2018 compared with those born in 2014. At the same time, the adult diagnosis wave is exposing massive gaps in support infrastructure designed almost exclusively for children.
The DSM-5’s stricter criteria may be systematically undercounting people with less obvious presentations, particularly women, older adults, and individuals from cultural backgrounds where autism is less recognized. If the true prevalence is higher than 1 in 31, then the scale of unmet need is correspondingly larger.
The genetics of autism continue to yield complexity rather than clarity. Hundreds of genes are implicated, each contributing small effects, interacting with environmental factors in ways that remain poorly understood. The hope of matching specific genetic profiles to targeted treatments, which motivated the DSM-IV subtypes, has not materialized.
What has changed is the voice of autistic people in the research process itself. Participatory research models, autistic-led studies, and community-based priority-setting are becoming more common, though far from standard. The Lancet Commission’s own composition was criticized for not being more participatory, a critique the authors acknowledged.
The autism spectrum, as a diagnostic concept, is less than 50 years old. It has been redrawn repeatedly, and the forces that shaped it (genetic research, prevalence data, advocacy movements, insurance and service eligibility) are still active. The next revision will be shaped not only by what scientists find but by who gets to decide what the findings mean.
