W. David Lohr, MD
Assistant Professor of Pediatrics
Division of Child & Adolescent Psychiatry and Psychology
Co-Clinical Director University of Louisville Autism Center
University of Louisville School of Medicine

Do some outgrow autism?

An eye-catching headline recently appeared in The New York Times: “Some With Autism Diagnosis Can Overcome Symptoms, Study Finds.” The basis for the article was a recent study by Deborah Fein at the University of Connecticut that evaluated 34 people diagnosed with autism before the age of 5 years who no longer met the criteria. This group did not differ from matched typically developed controls on measures of socialization, communication, facial recognition abilities and most measures of language. The group with optimal outcome was also compared with a higher-functioning group of patients with autism so that some early clues for prognosis can be considered.

This is a crucial study to understand because it conflicts with the typical message we teach to our trainees and families that autism is a lifelong condition. Cures of autism have been reported historically, and the reports from Lovaas, in which 47% of his patients were cured by his intensive behavioral treatments, have been mentioned frequently. To sort out and measure these claims, we must fall back upon the strength of the evidence. In this case, this study is well designed, using gold standard instruments with steps in place to foil the expert looking for autism. In my opinion, the strength of a research finding can be judged by the questions it raises, and the study by Fein is very provocative in this regard. It is crucial to understand more about these outliers who fit the phenomenon of optimal outcome.

How was the original diagnosis of autism made? These were high-functioning children at the time of diagnosis, with milder social symptoms than those children who retained the diagnosis of high-functioning autism. Diagnostic schemes may be a factor. DSM-IV was the diagnostic criteria used at the time of the study, and questions have been raised about its reliability in the diagnosis of high-functioning autism. For example, a recent study (Lord and colleagues) showed that even experts using Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview - Revised (ADI-R) to obtain data differ in how they apply clinical diagnosis to a group of patients. In many ways, the study provides ongoing support for the move to a more specific diagnostic scheme proposed by DSM-5.

How often might children grow out of autism? The researchers said this outcome occurs in a minority of children diagnosed with autism and is a recruited sample. A previous report (Sutera and colleagues) indicates that 18% of children diagnosed at age 2 years and receiving intensive behavioral treatment lost the diagnosis by age 4 years. In a long-term study (Howlin and colleagues) of adult outcome in children with a older mean age of diagnosis, 12% showed few or no problems with autism-related behaviors.

Is the group with optimal outcome normal? Those children with optimal outcome had ADOS scores and measures of facial recognition significantly lower than those judged to have high-functioning autism. Yet, based on parent reports from the ADI-R, the children with optimal outcome still had similar impairments compared with those with high-functioning autism in communication and repetitive behavior. Also, at follow-up, there are clues these children are not “identical” to the typically developing control. The optimal outcome group still has markedly elevated social communication dysfunction with a mean score of 17.1 on the Social Communication Questionnaire-lifetime scale compared with 1.5 for the normal control. Also, language fundamentals evaluated by the Clinical Evaluation of Language Fundamentals (CELF-IV) show significant differences in scores of formulated sentences and core language composite. Also, what psychiatric comorbidity might we expect in this optimal group? The optimal outcome group may turn out to have higher prevalence of other psychiatric conditions such as attention-deficit/hyperactivity disoirder, anxiety, and depression. Some point out social impairment in autism exists on a continuum (Constantino and colleagues), so this group of patients seems to exist in the boundary region between affected and unaffected. Maybe these children will fit the newly proposed DSM-5 diagnosis of social communication disorder.

What will be their long-term outcome? It will be important to follow this cohort over time to determine levels of education obtained, occupational functioning, independence, parents’ wish to know if their child will get a degree, hold a job, and live on his/her own.

Does the group achieve optimal outcome by the effect of intensive treatments or do they represent from the beginning a milder, self-limited condition? We will want to know if the treatments received or additional factors make the outcome more likely. The authors suggest their outcome may be influenced by a higher socioeconomic group of highly motivated parents. Ongoing research will evaluate the effect that intense therapies provide, and only prospective studies will provide these answers.

Are there any biomarkers in the optimal outcome group we can identify to help ongoing prognosis of newly diagnosed patients with autism? Recent advances in understanding white matter fiber tract development, genetic risks, and familial risks should all be applied to this group.

The phenomenon of outgrowing autism is fascinating and worthy of close examination of the evidence to teach and guide us all.

For more information:

Constantino JN. Arch Gen Psychiatry. 2003;60:524-530.
Fein D. J Child Psychol Psychiatry. 2013;54:195-205.
Howlin P. J Child Psychol Psychiatry. 2004;45:212-229.
Lord C. Arch Gen Psychiatry. 2012;69:306-313.
Lovaas OI. J Consult Clin Psychol. 1987;55:3-9.
Sutera S. J Autism Dev Disord. 2007;37:98-107.