The American Academy of Pediatrics1
and the American Psychological Association2
have recommended an approach to the identification of ASD that involves step-wise, and at times, recursive surveillance. Starting at pediatric (e.g.; “well baby”) appointments, the approach calls for formal screening if behaviors of concern are noted during surveillance (see Patel’s review of screening recommendations in this issue). If screening, including any caregiver concern, indicates cause for attention, this is to be followed by formal diagnostic assessment. In the years following the publication of these guidelines, the identification of ASD appears to have improved somewhat.3
However, many children continue to be first identified by their educational programs4,5
, and a significant minority of children with ASD are likely to be undiagnosed.4, 6
A contributing factor to the problem of under-identification likely results from the variability of symptom expression in ASD. As we have learned, the presentation of ASD can range from a child who is nonverbal and unlikely to make social initiations, to a child who is verbally fluent, but overly reliant on previously learned scripts of speech and social behavior (Ghaziuddin provides a detailed review of the clinical features of ASD in this issue). Because of this variability in symptom type and severity, diagnostic decision-making is a complex process; no singular algorithm can be applied to the diagnosis of ASD. In a clinical sense this has meant that no one behavior, such as responding to name or joint attention, excludes a diagnosis of ASD.7
For example, though responding to his or her name and responding to joint attention are important characteristics of ASD in toddlers, most children with ASD can carry out both of these actions by older preschool.8
Even as a toddler, a very intelligent child who understands his name and follows a point may still merit a diagnosis of ASD because he or she does not seek to share enjoyment with others, smile back to people except during intense physical activity, show any interest in his or her siblings or same-age peers and use language to answer questions or make socially-directed comments.
Diagnosis in ASD can be difficult because behaviors seen in a child are often dependent on a number of non-autism-specific factors, including cognitive functioning and age.9, 10, 11
The diagnosis of ASD is further complicated because of the interactions that occur between development and ASD symptoms. At certain ages, a number of characteristics, especially when defined by informants, that are common to ASD are not actually specific to the diagnosis and may occur in other disorders. In one study, as reported by parents and caregivers, stereotyped language was no more prevalent in children with ASD than in typically-developing children and children with other non-spectrum diagnoses who were under age 4 and in children who did not have complex language.12
These findings indicate that the types of behaviors to which a clinician must attend to for diagnosis are very much dependent on developmental factors such as age and language level, as well as the source of information (i.e., caregiver report through interview, questionnaire or clinician observation).
To address these challenges, the National Research Council Committee on Educational Interventions for Children with Autism advised that each child suspected of having ASD have an evaluation that incorporates the following standards: the assessment of multiple areas of functioning including adaptive skills, an appreciation that variability in performance and ability is common in autism, and the use of a developmental perspective when assessing behavior and synthesizing results.13
This is typically beyond the scope of usual pediatric practice and so, most cases will depend on appropriate referral. A wealth of recent studies, particularly those dedicated to the careful phenotyping of children who span the full spectrum of autism, have not only yielded data in support of these practices, they have also informed clinical practice with new methods and tools to implement these recommendations. These findings, and the practical implications of the guidelines, are discussed below.