change the world badge

change the world badge


Child Psychology Blogs

Concerned About Unconventional Mental Health Interventions?

Concerned About Unconventional Mental Health Interventions?
Alternative Psychotherapies: Evaluating Unconventional Mental Health Treatments

Tuesday, October 13, 2015

Very Early Diagnosis of Autism Is Still Not an Option

It’s so sad to see the number of queries that come to this blog from terrified young mothers who have become addicted to Googling about autism and want to know whether the (usually) typical infant behaviors they see mean that their children will be autistic. Other than cursing these insufficiently-informative “red flag” sites, there’s not much I can do except to tell them that they will just have to wait to get a definitive answer, because their babies are too young for anyone to tell whether there is an autism diagnosis in their future.
I recently came across two articles in the very helpful Brown University  Child and Adolescent Behavior Letter (CABL) that I would like to summarize--  not because they tell how to diagnose autism early, but because they underline the problems of trying to do so.

The first of these articles, by Stephen Sheinkopf, is titled “Autism in infancy: Advances and implications for clinical practice” (CABL, March 2013, pp. 1.4-5). Sheinkopf notes that “the presentation of autism in infancy is subtle, with more obvious signs and symptoms emerging over the first two to three years of life… these early developmental differences are difficult to detect, making it a challenge to screen for the risk of autism in early infancy. The gradual emergence of autism symptoms during development also makes diagnosis challenging in later infancy and toddlerhood.” Even when symptoms like poor communication development become evident, it may be far from clear whether there are hearing difficulties, specific speech and language disorders, or even mental retardation at work. Sheinkopf goes on to say that there are “few reliable findings reported in infants younger than 12 months of age”, even when children studied are known to be at risk for autism because they have autistic siblings--  and as Sheinkopf point out, it may not even be possible to generalize from those at-risk children to the rest of the population.

Even when toddlers are diagnosed with autism, a longitudinal follow-up is needed to be sure that diagnosis was correct. Sheinkopf describes two composite cases (in other words, these are not specific real children, but a narrative based on several similar cases). One case involved a 30-month-old boy who had been born at less than 30 weeks’ gestation, had been in intensive care for some time, and had normal motor development, no cerebral palsy, and no hearing or vision impairment. He was quite delayed in language and had some cognitive delays as well. He also showed behaviors  that can be symptoms of autism—repetitive movements, little eye contact, poor communicative abilities, and he received a tentative diagnosis of autism. However, following early intervention services he improved in language skills, performed fewer repetitive movements, and became more social and playful with adults as well as becoming more interested in other children than he had been. Although he continued to have some delays, by the time he was 40 months old, the autism diagnosis was ruled out.

In Sheinkopf’s second case, a 20-month-old boy had one older brother with autism, another with language delays, and a member of the extended family for whom autism was a possibility. He did not speak at all and did not vocalize much. He could solve problems that required no verbal ability but did not respond to verbal instructions.  He was affectionate and playful with his parents and showed some interest in other children. He flapped his hands when excited, but not in ways different from typical children of his age [I hope readers will notice the fact that hand-flapping  is common in toddlers]. It was not clear at this point whether his diagnosis should be autism or language delay. Ten months later, after early intervention services were received, it was seen that although his language had improved it was repetitive and not necessarily directed to other people. His social pragmatic skills had worsened compared to where he was at 20 months. He had an increase in repetitive and stereotyped actions and had begun to examine objects intensely in a way that interfered with play or social activities. By his third birthday, the diagnosis of autism given at 30 months continued to seem appropriate.

Notice that in both these cases the children were seen to have factors that put them at risk for autism, but that even well past the first birthday it was not clear whether their abilities and behaviors were indicators of autism—it was simply seen that autism was one of the possibilities. The first child became less “autistic-like” over time, but the autism diagnosis was not ruled out until he was well over three years old. The second child improved in language, but showed increasing evidence of autism and was finally diagnosed at age 2 ½. The child diagnosed with autism became more different from typically-developing children with age--  the differences were not clear-cut earlier in his life.

In a second relevant piece in CABL, Alison Knopf (“Brief observation not adequate to detect autism risk in young children”, March 2015, pp. 3-4) described a study by Gabrielsen, Farley, Speer, et al., (in Pediatrics, 2025, Jan. 12). Knopf notes that “autism spectrum disorder doesn’t manifest itself suddenly; symptoms appear over time, with gaps between typical and atypical development widening as the child gets older. Children have both typical and atypical behaviors [a most important point—JM] but little is known about what the ratio of these behaviors is during the 10-minute timespan of an average primary care visit.” In the study by Gabrielsen et al., clinicians were asked to see whether they could identify autism in children 15 to 33 months old, by watching two 10-minute video clips of each child.  Children who were previously identified as autistic were missed in 39% of the cases, suggesting that longer observations and more information than most pediatricians get are needed for more accurate diagnosis.

In the Gabrielsen study, a very important point concerned the fact that both typically-developing and autistic children had some typical behaviors and some atypical behaviors. The autistic children showed typical behaviors 89% of the time and atypical behaviors only 11%, while the typically-developing children showed atypical behaviors only 2% of the time. All the children were prompted to respond to their names at least once, and 50% of the children in the typical group failed to respond once or more, while 59% of children in an additional language-delay group did not respond on one or more occasions. Of the children in the autism group, 58% responded  to their name at least once, while 80% of the typically-developing children responded at least once.

Making eye contact, responding to their name, hand-flapping, repetitive behaviors—all of these are found in both autistic and typically-developing toddlers. There is no single symptom or list of symptoms that is a certain indication of autism even in children this old, and much less can such a symptom be found for infants in the first year.

Although specialized early intervention services can be helpful for toddlers and preschoolers, the best thing parents of infants can do is to encourage good development by sensitive, responsive care, by plentiful and playful social interaction that follows the child’s lead, and by minimizing the screen use that so easily keeps parents from these first two development-supporting activities. We would all like to think that a magic device could let us detect autism early, and some other magic ritual could allow us to prevent atypical development, but so far--  it just isn’t so, and all parents can do is the best parenting they can.          


  1. What about this computer program that tracks eye movement and allowed to detect a very high correlation between a strongly marked preference for shapes rather than persons and autism?

    1. First of all, there are going to be a lot of variations in this association. Autistic children act "typical" a large proportion of the time, and typical children act "autistic" quite often. Second, before such a test could be useful, it would be necessary to have tested large numbers of babies at different ages and established norms for visual preferences, which are going to be different at different ages. Keep in mind, too, that children who are diagnosed as autistic at an early age are quite often found to be typically-developing in another year or two-- many false positives, but I confess I know nothing about false negatives here.