The blogosphere has been abuzz with the news that a minority of correctly-diagnosed autistic preschoolers may grow up to no longer meet threshold for autism symptoms as adolescents or adults. In short: they started out obviously autistic as kids (though with milder "social reciprocity" symptoms than similar kids who kept their symptoms)--but no longer looked autistic as adults. The authors call this an "optimal outcome." (They don't call it "recovery," although an accompanying editorial in the same journal does).
But, was this really an "optimal outcome?" Let's unpack this study (link goes to PDF) and figure out what these people were actually like once they lost their autism symptoms. What exactly do the authors consider to be an "optimal outcome," and should we agree with them?
You'll probably notice that, even though participants were adolescents and adults with largely above-average IQ and good verbal ability, the researchers did not bother to ask for their input. Self-report measures could easily have been included alongside all the parent and clinician-based ones, and would have added a lot of information about subtler aspects of functioning that the study didn't address.
1) Let's get the obvious out of the way: the optimal outcome group no longer met criteria for an autism label, based on the DSM criteria and based on the "gold standard" diagnostic test, the Autism Diagnostic Observation Schedule (ADOS, Lord et al 2000). They went through countless measures to make sure both that the initial autism diagnosis was accurate and that the later non-diagnosis was, too. Arguably, for people who no longer meet threshold because they have learned to compensate, but have the same underlying difficulties, losing this label could be a mixed blessing at best, as Emily Willingham points out. The study did not investigate whether losing the autism label helped or hurt these people as they were simply trying to establish that they lost the label; this would be a great topic for a follow-up.
1.5) Optimal outcome participants could not be receiving special education services to address autism-specific deficits. I think the authors assumed this would imply that symptoms had disappeared enough that participants no longer needed these services. However, an adolescent or adult could not be getting services for other reasons--such as lack of money or lack of services in the area for these age groups.
2) "Optimal outcome" groups really did seem to learn a number of skills, performing at the level of typically developing people on socialization, communication, face recognition, and most language measures. These included: ADOS Socialization and Communication subtests; Vineland Adaptive Behavior Scale (VABS) Socialization, Communication, and Daily Living Skills subtests; Social Communication Questionnaire (completed by parents), Benton Facial Recognition Test, and Clinical Evaluation of Language Fundamentals (CELF) concepts/following directions, word definitions, formulating sentences, recalling sentences, and word classes subtests.
The ADOS is designed for diagnostic purposes rather than for in-depth evaluation of an individual's functioning, but is often used that way in autism studies. The VABS is a standard test of functioning in a variety of areas used in many studies with neurodiverse groups (autism, ADHD, Sensory Processing Disorder, etc.). The CELF is one of the standard language tests used, broken down into many specific areas of functioning. I'm less familiar with the SCQ and the Benton Facial Recognition Test, but I've seen them in a lot of studies and as far as I know, they're standard measures.
The "optimal outcome" group still had difficulties with one skill: 8.8% of the optimal outcome group still had trouble recognizing faces, scoring in the below-average range. Another exception: on the VABS test of daily living skills (e.g., self-care--grooming, feeding and cleaning up after oneeself, etc.), 15% of the optimal outcome group had low scores. However, this was not judged to be significant as 12% of the typically developing people did, too.
3) On a less formal note, to be included in the study, optimal outcome individuals had to have typically developing friends. Some might interpret this criterion as implying that only typically developing friends are worth having or that typically developing friends are better for autistic people than other autistic or atypical people. That may follow, but I don't think it's what the authors meant. I think they were trying to ensure that "optimum outcome" groups had the skills necessary to make and keep typically developing friends. Since we don't all have the good fortune to spend our days around people exactly like ourselves, being able to befriend typically developing people could potentially help autistic people--even if their best, most supportive friends are other autistics.
4) Participants had to perform well enough in school to be in regular education classrooms without one-on-one assistance. However, they could receive limited special education services or psychological support for things like attention or academic difficulties. In short, they could be performing fairly well, but still struggling to do so. In my mind, this is a good thing, but not necessarily what autistic people themselves aspire to or what parents want for their children. In other words, not an optimal outcome.
5) The researchers considered subthreshold social and communication disabilities, not in a real-life context, but by examining the ADOS results with a fine-toothed comb. Thus, we know very little about how the "optimal outcome" participants functioned socially in real life. The researchers looked at items concerning eye contact, gesture, and facial expressiveness. Over a fifth of the optimal outcome participants (7/34) had slight impairment in expressive nonverbal communication--either facial expressiveness, gesture, or eye contact. Unlike an additional five potential participants who were rejected from the "optimal outcome" group, these participants' mild difficulties were judged to be caused by "inhibition, anxiety, depression, inattention and impulsivity, embarrassment, or hostility" rather than autism. The authors do not elaborate on how they made this decision.
6) That 20% of the "optimal outcome" group appeared to exhibit "inhibition, anxiety, depression, inattention, impulsivity, embarrassment, or hostility" already indicates that even without autism, these were not happy, typically-developing children. The authors did not rule out or even measure the rates of other learning disabilities, ADHD, or emotional problems. Yet all of these lead to poorer outcomes in both kids and adults--lower education levels, more employment difficulties, more substance abuse, greater risk of suicide, less self-confidence and happiness.
The Autistic Self Advocacy Network (ASAN) argues that it's more optimal to be a happy autistic person with persistent social difficulties than a depressed, suicidal one who happens to have good social communication skills and typically developing friends. Arguably, there's some relationship between getting along with others and happiness, and the true optimal outcome would involve both happiness and skill development, but they have a point.
When I think of an optimal outcome, I don't imagine unhappy teenagers and adults with academic difficulties.
Optimalitiy isn't just about learning new skills or losing a diagnostic label, it's about quality of life.
Not all researchers make the same mistakes as Fein's team, by the way. Catherine Lord is also troubled by the emotional problems seen in 20% of the optimal outcome group. She points out that the social and communication skillls measured may not be the ones important for real-life social relationships, and that obviously autistic people can have happy, productive lives without losing either their symptoms or their diagnosis.
Sometimes you learn as much from the questions people don't ask and the measures they don't take as from what they do. It's disheartening to see what low aspirations some researchers have for autistic people.