For Better or for Worse? Later Diagnoses of Autism Spectrum Disorder in Some Younger Siblings of Already Diagnosed Children

Later diagnoses of autism spectrum disorder (ASD) represent a theoretical challenge for researchers, including myself, who have conceived of ASD as a very early arising developmental disorder that affects learning and behavior from infancy onward.¹ Such diagnoses also are a concern because of disparities in access to early intervention, which we know are associated with ethnic minority status and less parental education.² If we believe that early intervention makes a difference, then finding children with ASD who cannot be identified until early school age is problematic. Three recent studies, before the present study by Ozonoff et al.,³ found relatively small, but not insignificant, numbers of children who were not diagnosed until after 6 years of age.^4–6 The present study by Ozonoff et al.³ uses the opportunity provided by a unique prospective dataset of younger siblings of children with autism (often referred to as the Baby Siblings Research Consortium) to describe 14 children who had not been identified as having ASD at 3 years of age, despite detailed early assessments by a skilled research team, out of 113 children who received ASD diagnoses in middle childhood.

Seven of these 14 children, for most intents and purposes, seemed to have met or almost met criteria for ASD in the 3-year-old assessment, but for various reasons, which probably cannot be determined retrospectively, did not receive diagnoses of ASD. Like the 99 children who received 3-year-old ASD diagnoses (and unlike those who did not receive ASD diagnoses), these 7 children were rated as having more autistic symptoms by their parents than they were by clinicians using the Autism Diagnostic Observation Schedule (ADOS), with an equivalent degree of discrepancy as those who received diagnoses. In middle childhood, the same 7 children continued to show mild evidence of clinician-observed and parent-reported symptoms, which at that point resulted in new diagnoses of ASD. Thus, they were children whose scores at 3 and 7 years of age fell in the mildest ranges of possible autism on the ADOS and often in similar ranges on other parent-report measures. Because many other children without ASD diagnoses in the study fell in similar ranges on the ADOS at 3 years, this pattern was not particularly distinctive. It is not clear whether false positives on the ADOS continued to be as common in the typical group at 7 years.

For the most part, as a group, these 7 children had high average to above average verbal and performance IQs and adaptive scores in the average range. We do not know their ethnic status or about their social classes or parental education. Careful analyses by the authors suggest that these children were not being diagnosed with other disorders (eg, no diagnostic overshadowing) and they did not have different patterns of autism symptoms than the 99 children who received stable ASD diagnoses at 3 years and later. However, it is relatively easy to think of scenarios in which, if a child is very mildly symptomatic, especially if the child has an older sibling who is more impaired and parents who are already fairly knowledgeable but are not seeking an ASD diagnosis, clinicians might not make an ASD diagnosis at 3 years of age, especially in a child with strong cognitive abilities who seems to be developing well in everyday adaptive functioning, with no perceived impairment or need for specialized help. By middle childhood, parents’ perspectives might have changed and, with greater absolute discrepancies from typical development, a diagnosis might seem more important in terms of services. Other research has shown that parents’ expectations for a diagnosis and factors such as birth order and age differences between siblings affect various aspects of diagnoses.^7,8 This is one reason why we seek more objective measures such as biomarkers. In contrast, caregivers often know the needs of their children and the opportunities available better than professionals, so the fact that clinicians are taking parents’ perspectives into account would not necessarily be bad.

The other 7 children are more of a mystery. The authors made even more extensive attempts to figure out what might have been missed at 3 years, to no avail. Based on the researchers’ data for this group of children, except for parent-stated concerns that were generally not reflected in higher scores on standardized parent-report measures, these children did not show many signs of autism until they were older. In middle childhood, as a group, they looked very similar in terms of scores within milder ranges on the standardized autism scales to the late-diagnosed children described earlier. However, for these children, scores on autism measures in middle childhood were distinctly different than their earlier scores.

Four of these 7 children were girls; like the other 7, they mostly had high average IQs. What is particularly interesting is that it does not seem like either group with late diagnoses was particularly affected on codes from the ADOS that we associate with more complex social interactions, such as empathy or reciprocal conversations, as opposed to facial expressions and eye contact, which are symptoms that are more consistent across age groups.³ Because of the small sample, it is possible that the moderate effect size was not significant because of lack of power. If this difference was clearer, then we could hypothesize that there are some children, noted as at risk by their parents in preschool and biologically at risk because of family history, who move into autism as expectations increase for more complex social interaction. This is possible, but not supported here.

Overall, this study is a testimony to the ability of small, carefully collected datasets to raise and even answer questions that larger, less systematic data cannot address. One question that is left unanswered is whether the lack of an ASD diagnosis during preschool years made a difference for these children. Did they benefit from higher expectations and a lack of stigma or were they disadvantaged by not receiving special services? It could be that given their parents’ familiarity with ASD and approaches to it, the children were not slowed down by a lack of diagnosis. We await follow-up of later profiles in future articles about this unique, prospective dataset.

The 14 children are a diverse group; they met criteria for autism and were called not autistic for diverse reasons. This finding fits with many other kinds of evidence that autism is not 1 concept and that it does not have 1 pathway in early childhood or later on.⁹

For clinicians, we need to be conscious that children, at least younger siblings of children with ASD, with early mild difficulties on standard measures, when accompanied by parental concerns, should be monitored carefully. Their parents’ concerns should be addressed, with or without a diagnosis. In addition, an awareness that later diagnoses and later-appearing symptoms can emerge reminds us there is much we do not know. A very early history of typical development should not negate observation of clear symptoms of ASD later on, particularly in girls and in children with average or above intelligence.