The autism constellation and neurodiversity: Long-term and adult outcomes in autism spectrum disorder

Autism spectrum disorder (ASD)¹ is a neurodevelopmental condition characterized by difficulties in social communication and the presence of repetitive behaviors, unusually strong and sometimes unusual interests, and sensory hyper and hypo-sensitivities. Epidemiological estimates² suggest a global autism prevalence of 1/100 with precise estimates ranging from 1.09/10,000 to 436.0/10,000. The wide variation in estimates reflects community-specific differences in access to services, awareness, and sociodemographic factors; for example, one recent survey of a racially and ethnically diverse caregiver cohort (n = 744) reported that Latinx and multiracial caregivers were significantly more likely than African-American/Black, White, and Asian parents to report that their autistic child had support needs that were unmet^3,4. Some 33% of autistic individuals have co-occurring intellectual disability²; thus, approximately two-thirds of autistic individuals have cognitive abilities that are broadly commensurate with chronological age.

Adult outcomes in autism.

The transition into adulthood for autistic individuals is a difficult one, even in comparison to outcomes for other developmental disabilities^5-8. The educational system provides many forms of support; after exiting school at age 18 or 21 years, autistic adults struggle significantly with underemployment or unemployment⁹, non-independent living arrangements¹⁰, unmet service needs¹⁰, psychiatric comorbidities¹¹, and a lack of meaningful social relationships and loneliness⁵. One study reported that more than 50% of autistic adolescents were not engaged in either employment or educational activities during the two year period following high school graduation⁵. Some American parents have described this difficult experience as “falling off the cliff”¹², reflecting the sudden suspension of services upon which they previously relied.

Most autism research focuses on childhood and adolescence; only limited research has quantified the experiences and outcomes of autistic adults. Reviewed below, this research includes studies that code and quantify broad outcomes, that typically measure multiple outcome variables, such as occupation, friendships, independent living, and physical health, and activities, and collapse these into a global outcome score. Other research examines trajectories of change, typically by focusing on a small number of focal outcome variables.

Broad outcomes in cohorts of adults.

Studies have used a global outcomes rating, including employment, living situation, and friendships developed by Howlin¹³, and modified by Gillespie-Lynch ¹⁴. Results of these studies suggest that most autistic adults were “very dependent”; few lived alone, had close friends or permanent employment; stereotyped behaviors and interests persisted into adulthood. More than half of individuals with ASD who had left high school in the past 2 years reported no participation in employment or education⁶. Social isolation was described as a particularly prevalent and painful concern, with one-half to two-thirds of adults with autism reporting no real friendships ^13,15,16. In a 2008 study, few individuals (4%) are described as having a “very good” outcome; some 17% had good outcomes, and the remainder had fair to very poor outcomes¹⁰. Social isolation, stress, depression, and anxiety were frequent concerns even with high IQ; see¹⁷ for a review. There is evidence that these difficulties tend to increase with age in ASD ^18,19. Some negative outcomes are mitigated by higher family income, better adaptive skills in activities of daily living, fewer behavioral issues and co-occurring psychiatric conditions, milder autism symptoms, and lack of intellectual disability ^{5,6,9,11,20,21}. Even so, adult outcomes of individuals diagnosed in the 1970s and 1980s (forming the bulk of the scant evidence base) are often poor²². While evidence is too limited to draw firm conclusions, research hints at improvement in global outcomes over time, as shown in Table 1. For example, the most recent reports suggest positive outcomes in approximately half of participants²³.

Table 1.

Longitudinal and adult outcomes as a function of publication year

Reference	Sample	Outcomes	Predictors of better outcomes
Global outcomes
Howlin, Goode, Hutton & Rutter (2004). J Child Psychol Psychiatry, 45, 212-29.	68 autistic adults with IQ > 50, from age 7 to 29	Outcomes rated as: 12% = Very Good, 10% = Good, 19% = Fair, 46% = Poor, 12% = Very Poor	Not reported
Billstedt, Gillberg & Gillberg (2005). J Autism Dev Disorders, 35(3), 351-360.	120 autistic children followed to ages 17-40	Outcome rated as: 78% = Poor	Higher child IQ, phrase speech at age 6 years
Billstedt, Gillberg & Gillberg (2007). J Child Psychol Psychiatry, 48(11), 1102-1110.	105 autistic adults with rigorous childhood diagnosis	Social impairments most consistent over time	Speech < age 5, higher IQ, male gender, no medical conditions, and no seizures in early childhood, predicted social, communication, and self-direction skills
Eaves & Ho (2008). J Autism Dev Disorders, 38 (4), 739-747.	48 autistic adults, age 24	Outcomes rated as: 50% = Good to Fair, 46% = Poor. Co-morbid conditions, obesity and medication use were common.	IQ and CARS score at age 11
Shattuck, Narendorf, Cooper, Sterzing, Wagner & Taylor (2012). Pediatrics, 129(6), 1042-1049.	680 autistic youth ages 13-16	35% had attended college and 55% had paid employment; however, more than 50% had no employment or education in the two years after high school graduation	Higher income and higher functional ability were associated with postsecondary employment and education
Mhatre, Bapat & Udani (2016). J Autism Dev Disorders, 46(3), 760-772.	80 autistic children over 10 years	80% = fluent verbal skills; 43% = spoken abilities with significant challenges; 25% = meaningful friendships; 46% = age-appropriate adaptive abilities	Lower symptomatology, parent participation, and higher maternal education
Longitudinal outcomes
Piven, Harper, Palmer & Arndt (1996). J Amer Acad Child Adol Psychiatry, 35(4), 523-529.	38 autistic adolescents and adults with age-appropriate IQ, from age 5 to ages 13-28	Improvements in social, communicative, and repetitive behaviors; 13% of these individuals no longer met criteria for ASD, though had significant impairment. Health, psychiatric and behavioral problems were common	IQ of 70 or higher
Turner, Stone, Pozdol & Coonrod (2006). Autism, 10(3), 243-265.	26 autistic children from age 2 to age 9	12% lost the diagnosis, 88% remained autistic; 67% had significant cognitive improvements (average increase: 23 IQ points)	Age of diagnosis, cognitive and language scores at 2 years, and hours of speech-language therapy at ages 2-3
Fountain, Winter & Bearman (2012). Pediatrics, 129(5), e1112-1120.	6975 autistic children, from age 2 to 14	10% of the sample displayed a very steep trajectory, with significant delays at Time 1 but age-appropriate levels of functioning at Time 2	Parental education, ethnicity/race, and lesser severity of ASD symptoms at diagnosis, predicted faster improvements
Fein et al. (2013). J Child Psychol Psychiatry, 54(2), 195-205.	Compared 34 LAD, 44 autistic individuals with rigorously evaluated ASD histories, and 34 non-autistic individuals	The LAD group had no symptoms of ASD and was in the non-autistic range for social and communication skills. LAD / non-autistic groups did not differ in social communication, language, or face recognition	Milder early social symptoms and earlier engagement in intervention predicted LAD outcome
Anderson, Liang & Lord (2014). J Child Psychol Psychiatry,, 55(5), 485-494.	85 autistic children from age two to 19	Of those with age-appropriate cognitive abilities, 9% no longer met criteria for ASD. 62% had verbal IQ (VIQ) scores under 70, and 38% had VIQ scores of 71+	Early behavioral intervention, and a sharp reduction in repetitive and stereotyped behaviors between ages 2-3
Gillberg, Helles, Billstedt & Gillberg (2016). J Autism Dev Disorders 46(1), 74-82.	50 adult males with Asperger syndrome, IQ > 70, over 20 years	54% of full sample had co-occurring ADHD, depression or both. 22% had no ASD symptoms at follow-up; of these, 6% had no co-occurring psychiatric conditions	Absence of co-occurring psychiatric conditions associated with better outcomes
Mukaddes, Mutluer, Ayik & Umut (2017). Pediatrics International, 59, 416-421.	26 autistic children who lost the diagnosis followed for 2–8 years	92% lifetime, 81% current co-occurring psychiatric conditions (ADHD, specific phobia, obsessive–compulsive disorder)	Not reported
Torenvliet, Groenman, Radhoe, Agelink van Rentergem, Van der Putten & Geurts (2023). Psychiatry Res, 321, 115063	128 autistic with IQ > 70 and 112 non-autistic adults at 3 timepoints	No group differences in verbal memory, visual working memory, prospective memory, theory of mind, fluency, response speed, inhibition, planning, or switching	Not reported

Note. All ages are shown in years.

Longitudinal changes.

Research conducted in the early 2000’s examined developmental change for individuals with a childhood diagnosis of autism. Piven and colleagues described adult outcomes in n = 38 autistic individuals with high IQ, reporting that communication and social abilities were more likely to improve compared to repetitive and stereotyped behaviors²⁴. Psychiatric, behavioral, and health difficulties were common; however, 13% of these individuals no longer met criteria for ASD (though still had unspecified “significant impairments”). Research by Turner, Stone, and colleagues ^25,26 followed 48 children diagnosed at age two years to follow-up at ages four and nine years. They found that 38% no longer met criteria at age four, and that earlier age at diagnosis was a predictor of this outcome, suggesting earlier treatment is more effective. In a different sample of 25 children followed until age nine years, 88% retained the ASD diagnosis, and IQ scores were over 70 for 72% (and in the average range or higher for 56%); this was a significant improvement from age two years. An estimated 32% of participants had conversational language skills, 56% had multiple words but were not able to engage in conversational exchanges, and 12% had very limited productive speech; none of the participants who were verbal at age two years fell into the limited speech group at age nine. A study of 80 children followed over 10 years, conducted in India,²³ indicated that 80% had fluent verbal skills; 43% had spoken abilities with significant challenges; 25% had meaningful friendships; and 46% displayed age-appropriate adaptive abilities. Fountain and colleagues followed a very large cohort of children (n = 6975) from ages two to 14 years and identified six distinct statistical trajectories ²⁷. One group, comprising 10% of the sample, displayed a very steep trajectory, with significant delays at time one and age-appropriate levels of functioning by age 14 years. These individuals were more likely to have been white and non-Latinx, and to have parents with more education. Altogether, results of more recent studies suggest that most autistic children make meaningful gains in functioning. The variability in outcomes reflects the timing of treatment onset (with better outcomes when treatment is initiated at younger ages), as well as individual differences in communication skills, milder autism symptoms, higher cognitive abilities, stronger adaptive skills, and parental involvement in intervention²⁸.

One of the strongest predictors of better outcomes is engagement in early intervention (EI). Most such interventions are grounded on the principles of applied behavioral analysis (ABA); current approaches (e.g., Early Intensive Behavioral Intervention, EIBI²⁹, and the Early Start Denver Model (ESDM)³⁰ are designed to address developmentally appropriate and socially relevant treatment goals, and are oriented towards a child’s own interests and choices. These approaches aim to motivate a child to participate and engage with the interventionist via play, home routines, naturalistic teaching, and structured learning environments; most are highly flexible and responsive to a child’s current behavior and affect. Related Naturalistic Developmental Behavioral Interventions (NDBI)³¹ incorporate a mixture of both therapist-led and child-initiated approaches; these include programs such as Incidental Teaching³², Social Communication/Emotional Regulation/Transactional Supports (SCERTS)³³, Enhanced Milieu Teaching³⁴, Joint Attention, Symbolic Play, Engagement & Regulation (JASPER)³⁵ and Early Achievements³⁶; see Waddington, van der Meer and Sigafoos (2016) for more detailed description.³⁷ Children in both ESDM and EIBI programs make significant gains in general cognitive ability after one year, with average gains in Developmental Quotient of 7.8 points (ESDM) or 14.8 points (EIBI), with no significant difference between the two approaches²⁸. A systematic review of EIBI intervention studies compared to “care as usual” found significant gains in spoken language with a mean effect size of g = 0.26 and a confidence interval of 0.11 to 0.42 ³⁸; effects were largest when parents and clinicians collaborated in the intervention. While there more double-blind randomized treatment trials are needed, the evidence clearly demonstrates that gains are greater when individuals participate in EIBI.

The results of outcomes research describing individuals diagnosed in earlier decades may not generalize fully to individuals who are diagnosed now. We know that effective early detection tools have led to steady decreases in the current age at diagnosis. For example, the MCHAT-R ^41,42 is a 23-item parent report checklist that can be completed in 5–10 minutes, ideally during pediatric well-child visits. When paired with a follow-up phone call for cases that fail the written screening, the MCHAT has excellent reliability (α = .85), with a positive predictive value (sensitivity) of .80 and a negative predictive value (specificity) of .99. Effective early screening leads to earlier diagnosis and thus to earlier engagement in behavioral intervention. Indeed, epidemiological work suggests that, currently, significantly more autistic children are being identified early (defined as prior to age 4 years) than in the past, with significant changes observed during the period of 2002 to 2016 ⁴³. Children were four times more likely to be identified by age 48 months in 2016, compared to 2002, with the largest increases observed for cases without concurrent intellectual disability (ID). Of course, the median age of diagnosis varies widely from state to state in the U.S., from (e.g.) 36 months in California to 63 months in Minnesota⁴⁴, due to differential access to services ⁴⁵.

More recent outcomes research.

Given significant changes in the timing of diagnosis and the provision of early intervention and other services, updated information about long-term and adult outcomes in ASD is of high interest. A recent meta-analysis reported that 20% had ‘good’ outcomes, with typical or near-typical social lives and satisfactory school/work functioning, 31% had a fair outcome, and 48% had a poor outcome ⁴⁶. Outcomes were defined as follows: “very good” indicated a high level of independence; “good” indicated that some degree of support in daily living was required; “fair” indicated some degree of independence, where support and supervision was needed, but not placement outside the home; “poor” indicated that residential placement and a high level of support was needed; and “very poor” indicated a need for high-level institutional care. These findings represent an improvement compared to older studies. An extended longitudinal study followed autistic individuals from 2-13 years to ages 29-64 years⁴⁷. Measures of language and cognitive abilities suggested stability or some improvement in 75%, but ongoing difficulty in language for 23%. Regarding aging in autism, a recent study found no evidence for heightened risk of cognitive decline or dementia in autistic adults without ID⁴⁸. Several groups have been following individuals who were diagnosed with ASD prior to age 5 years, using gold-standard measures, into adulthood, with strongly positive outcomes. Anderson and colleagues ³⁹ saw 85 autistic children diagnosed with ASD at age two years when those children were age 19 years. Of these, 62% had verbal IQ (VIQ) scores under 70, and 38% had VIQ scores of 71 or higher, and 9% no longer met criteria for ASD. These latter individuals were more likely to have had early behavioral intervention, and to have shown a reduction in repetitive and stereotyped behaviors between two and three years.

Loss of the autism diagnosis (LAD).

Although autism is generally assumed to be lifelong, there is evidence that 3 - 25% of children who meet clear diagnostic criteria for ASD will enter the typical range of cognitive, adaptive and social skills by adolescence, displaying no symptoms of autism⁴⁹. Mukaddes and colleagues ⁴⁰ described a sample of 39 children who met clear diagnostic criteria at age two years, but who no longer met criteria, according to expert clinical judgement, at age six years. Gillberg et al. reported that individuals who moved off the spectrum were largely free of other psychiatric disorders¹¹ though there were increased rates of psychiatric comorbidity. One study has followed individuals who were rigorously diagnosed early in development, who had no detectable symptoms of ASD by the time of adolescence⁵⁰; a series of papers describing these individuals has examined multiple domains of functioning, and reported that individuals who had lost the ASD diagnosis (LAD) were generally indistinguishable from their non-autistic peers. Studied domains include social skills ⁵¹, personality and traits of the “Broader Autism Phenotype” ⁵², academic skills ⁵³, standardized assessments of language ⁵⁴, and experimental assessments of subtle pragmatic language skills ^55-57. Executive Global Executive Composite scores as reported by parents using the BRIEF ⁵⁸ were in the clinical range for significantly more autistic participants (50%) compared to LAD (4%) and non-autistic (0%) peers⁵⁹. On direct assessment using the D-KEFS⁶⁰, all executive functioning scores for LAD individuals were in the average range (though they had lower scores than non-autistic peers on measures of impulsivity, set-shifting, problem-solving, working memory, and planning). While present in childhood, restricted and repetitive behaviors in LAD individuals had resolved at the time of study⁶¹. Our recent work indicates that LAD individuals maintain their non-ASD status in adulthood⁶⁷, according to brief impressions during the first few minutes of an interaction, and according to gold-standard expert clinician diagnoses. LAD young adults display heterogenous language outcomes (see also⁶⁸); while most have structural (morphosyntactic) abilities that are in the average range or higher, they display significantly more language impairments than non-autistic peers ^69,70.

A study of psychiatric profiles indicated equivalent lifetime prevalence of psychiatric diagnoses in LAD and autistic groups ⁶²; while concurrent diagnoses persisted in the autism group, they waned in the LAD group, leaving the latter with elevated rates only of ADHD (inattentive or combined types) and specific phobias. Consistent with findings of meaningful syndromic overlap between ADHD and autism at both behavioral and genetic levels⁶³, there is evidence that autism can “evolve” into ADHD, with age-appropriate social and communication skills alongside impairments and delays in executive function and adaptive skills⁶⁴. Recent work by our group finds that, as young adults, LAD individuals have elevated rates of depression and anxiety relative to their neurotypical peers⁶⁵, suggesting a relatively greater vulnerability to other psychiatric conditions in those with a history of autism; ongoing work is examining the role of treatment, but the predictors of this vulnerability are unknown to date. A careful ascertainment of internalizing disorders is warranted, given their higher prevalence in autistic young adults; estimates suggest that 23-37% of autistic young adults experience depression⁶⁶.

Neural functioning in LAD.

Although behaviorally indistinguishable from typically developing peers, functional neuroimaging results suggest that “normalized” behavior in LAD individuals reflects the action of compensatory neural systems⁷¹. Results of a functional MRI study performed during a sentence comprehension task indicated similar activations in frontal and temporal regions (left middle frontal, left supramarginal, and right superior temporal gyri) and posterior cingulate in LAD and autistic participants, where both differed from neurotypical peers. In addition, the LAD group had heightened activation in left precentral/postcentral gyri, right precentral gyrus, left inferior parietal lobule, right supramarginal gyrus, left superior temporal/parahippocampal gyrus, left middle occipital gyrus, and cerebellum. These results suggested that in the context of highly typical language abilities, brain functioning differed.

Predictors of positive long-term outcomes.

Acquiring useful language by age five years is often listed as a predictor of positive outcomes in ASD; for example, in a study of 119 children, producing first words (other than “mama” or “dada”) by age 24 months was a particularly strong predictor of better outcomes early in life ⁷². Important predictors of clinically or functionally meaningful adolescent and adult outcomes, such as higher IQ scores and age-appropriate and unimpaired social and language abilities, include age-appropriate or higher cognitive abilities; relatively stronger verbal and motor imitation skills; better pretend play; milder earlier symptoms; and stronger motor skills; see Helt et al. for a review⁴⁹. Note that earlier age of diagnosis, earlier engagement in intervention, and quicker response to intervention, are each associated with more age-appropriate outcomes and greater reductions in deficits and impairments. The presence of seizures, intellectual disabilities, and genetic syndromes are predictors of greater developmental delays and more ASD symptomatology in the long term. Interestingly, gender is not an important predictor of long-term outcomes.

Neurodiversity perspective.

What does a change in diagnostic status mean for an individual’s functioning and well-being¹? Advocates of the neurodiversity perspective have argued that autistic individuals need support and acceptance, rather than treatment and change, and that autism should be treated as a continuum of abilities rather than a categorical condition (as it is under the medical model). This argument is compelling given the enormous heterogeneity in abilities and long-term outcomes that characterizes autism as a clinical condition. Social impairments can range from an apparent disinterest in others (including family members), to a desire to socialize but with poor social judgment. Language abilities can range from a complete lack of spoken language with very limited comprehension, to language that is fluent and structurally within the normal range, but with poor pragmatic abilities. Cognitive functioning can encompass severe disability to the superior range. Repetitive behaviors and stereotyped interests encompass both debilitating self-injurious behaviors and idiosyncratic but functional strong interests.

While the neurodiversity approach to autism as a continuum of abilities rather than a clinical diagnosis is appealing for these and other reasons, we argue that the categorical medical model presents important practical advantages ⁷³. Relevant constructs are the notions of impairment and distress, which are central to DSM definitions of disorders; thus, in a DSM diagnosis of an anxiety disorder but not in “everyday anxiety” individuals struggle to function and feel pervasive, disabling distress due to their symptoms. Following this logic, if LAD individuals display no symptoms, experience no functional impairments, and experience little or no distress, they no longer fall into the diagnostic category of autism. Autistic self-advocates and others have argued that calling the loss of an ASD diagnosis a “positive outcome” implies that meeting criteria for the diagnosis is ipso facto a negative outcome⁷⁴. We agree that society needs to accommodate both the strengths and weaknesses of autism and other neurodiverse conditions; the primary goal of diagnosis and intervention should not be the loss of the ASD diagnosis, but rather, maximizing an individual’s autonomy, relationships, and daily living skills (including the ability to participate in employment or activities outside the home), in a manner that is consistent with an individual’s preferences and interests ⁷⁵. We believe that the LAD outcome reflects one such positive outcome. We also note that services that support autistic individuals are sometimes helpful to all. For example, a qualitative interview study of autistic adults and their experiences in the healthcare system ⁷⁶ reported concerns about logistical barriers to accessing services (e.g., insurance, fees, transportation); concerns about the clinical environment (e.g., wait times, sensory qualities, and anxiety-provoking procedures); and concerns about health-care provider knowledge about autism, communication and rapport, individualized care, and approaching the relationship as a partner. Improving clinical care in these dimensions would better serve not just autistic individuals, but everyone.

Conclusions.

Because of the enormous heterogeneity in long-term and adult outcomes It would be clinically irresponsible to promise a particular prognosis to a child diagnosed with autism early in development. That said, the results reviewed here indicate that, because of success in earlier screening and diagnosis, and provision of early behavioral interventions, autistic children can currently anticipate more positive outcomes than their counterparts in past decades, including the possibility of losing the diagnosis altogether. Clinicians can encourage parents to engage children in treatment and intervention, with the realistic expectation that many children will achieve significant independence later in life; as our group’s work indicates, a significant minority will function at the level of their non-autistic peers in adulthood, and can realistically hope to engage in meaningful social relationships, daily activities, and to be healthy and happy. Individuals who receive earlier behavioral interventions are likely as adults to display age-appropriate skills, fewer social communication challenges, and fewer repetitive behaviors or interests that present difficulties to those individuals. One important caveat is the likelihood of anxiety and depression symptoms in young adulthood for autistic individuals. Clinicians can recommend with confidence that individuals and families consider early behavioral interventions as a critical component of developmentally appropriate care⁷⁷.

Key Points:

• Research on autistic adults suggests significant heterogeneity in outcomes. A significant proportion of individuals struggle with intellectual disability and limited communication skills.

• Of the 67% who have age-appropriate cognitive skills, around half are expected to attain a college education, and 25% are likely to hold a full-time job.

• Outcomes have been improving over time, in part because of earlier diagnosis and earlier intervention. Indeed, an estimated 10-20% are expected to lose all symptoms of autism by adolescence.

Synopsis.

Research on autistic adults suggests significant heterogeneity in outcomes. A significant proportion of individuals struggle with intellectual disability and limited communication skills. Of the 67% who have age-appropriate cognitive skills, around half are expected to attain a college education, and 25% are likely to hold a full-time job. Outcomes have been improving over time, in part because of earlier diagnosis and earlier intervention. Indeed, an estimated 10-20% are expected to lose all symptoms of autism by adolescence.

Clinical Care Points.

• Producing first words (other than “mama” or “dada”) by age 24 months is a strong predictor of better childhood outcomes in autism.

• The best predictors of adult outcomes are early diagnosis, early engagement in intervention, milder early symptoms, and quicker response to intervention. Other important predictors of positive adult outcomes are higher IQ scores, using verbal communication in childhood, motor skills, motor imitation skills, pretend play, milder earlier symptoms. The effect size of early intervention on later outcomes is of large magnitude, though more randomized controlled trials are needed.

• Parents and caregivers seek guidance on maximizing communication abilities, social relationships with peers, adaptive skills, managing challenging behaviors and, in the long term, a child’s prospects for employment, education, independence, and having a family. In addition, families prioritize access to early intervention and other critical supports; the financial means to cover these supports is a significant stressor for many families.

• Adult outcomes are improving over time, due to earlier diagnosis and engagement in intervention. Reflecting input from community stakeholders, clinical priorities of adults include improving access to informed and individualized health care, management of highly prevalent co-occurring psychiatric conditions including depression, anxiety, and ADHD, and promoting broader social acceptance of autistic behaviors.