To reduce the average age of autism diagnosis,screen preschoolers in primary care

In 2006, the American Academy of Pediatrics (AAP) first recommended routine autism spectrum disorder (ASD) screening of toddlers at the 18- and 24-month well-child visit. This recommendation was based on research showing that early identification and treatment of ASD is associated with earlier access to interventional treatment, which subsequently leads to improved language and cognitive outcomes. In addition, early identification of ASD can lead to decreased family stress and increased understanding of a child’s strengths and challenges (Hyman et al., 2020).

The AAP re-affirmed its ASD screening recommendation in an updated clinical report this year (Hyman et al., 2020). In the 14 years since the initial AAP screening recommendation, ASD prevalence has increased dramatically (Maenner, 2020), and screening for ASD in primary care has become commonplace. Some estimates suggest that that over 50% of US toddlers are screened for ASD in primary care (Arunyanart et al., 2012), although significant local variability exists (Carbone et al., 2020; Wallis et al., 2020). Increased screening has likely contributed to a decrease in the age of autism diagnosis in the United States: recent data from the Early Autism and Developmental Disabilities Monitoring Network suggest that the percentage of children with ASD diagnosis by the age of 4 years has changed from 58% in 2014 to 71% in 2018 (Shaw et al., 2020). The median age of ASD diagnosis among 8-year-old children in the United States was 4.25 (Maenner, 2020), which suggests that at least half of US children are being diagnosed with ASD prior to school entry.

However, a median age of diagnosis of just over 4 years, while encouraging, is hardly a cause for celebration when ASD could be detected much earlier for most children. The fact remains that despite widespread autism screening, the majority of children are not being identified with ASD until well past the toddler ages, and this age is even higher for children from low-income and or racial/ethnic minority backgrounds (Durkin et al., 2017; Maenner, 2020). Studies have demonstrated a variety of reasons for diagnostic delays such as (1) not all children attend primary care as toddlers, (2) not all of those who attend are screened for ASD, and (3) not all those who are screened are referred for ASD evaluation (Carbone et al., 2020; Guthrie et al., 2019; Wallis et al., 2020). Furthermore, emerging evidence suggests that ASD screening tools have limited sensitivity—meaning that some children who screen negative on toddler screens go on to receive an ASD diagnosis (Guthrie et al., 2019). Predictive values of ASD screening questionnaires are low (Carbone et al., 2020; Guthrie et al., 2019; Robins et al., 2014), particularly for cases near the cutoff score—which may contribute to the fact that children with positive screening test results often are not referred (Carbone et al., 2020; Monteiro et al., 2019; Sheldrick & Garfinkel, 2017; Wallis et al., 2020). As a result, screening of toddlers is only a partial solution to delays and disparities in age of ASD diagnosis. Though improving the test characteristics of screening tools for toddlers, as well as improving referral processes, is part of the solution, it nonetheless seems likely that a meaningful portion of children with ASD will not be identified via screening in the toddler years.

As experts in primary care–based screening for autism, we are often asked by our primary care colleagues which tools we recommend for ASD risk identification among children who are “too old” for infant and toddler-age autism screeners such as the Modified Checklist for Autism in Toddlers-Revised with Follow-up (MCHAT-R/F) or the Communication and Symbolic Behavior Scales Developmental Profile Infant Toddler Checklist. Unfortunately, we have little to recommend. There are no tools for preschool-aged children that are designed for a busy primary care setting, which requires rapid assessment, simple scoring algorithms, and provider decision support for further evaluation and treatment. Though some evaluation tools do exist for children aged above 3 years (e.g. Childhood Autism Rating Scale, Gilliam Autism Rating Scale-3, Social Communication Questionnaire, and Social Responsiveness Scale-2), these all consist of at least 35 items and many require clinical judgment to score: administration time is on average 15–20 minutes, the length of an entire primary care visit. Other tools (e.g. Childhood Autism Screening Test and Social Communication Questionnaire) are somewhat briefer but have a minimum age that does not span the preschool years. In fact, there are no tools in public domain that span the time between toddler screening and school age (30–60 months), and there are almost no tools of any kind for ages—30–36 months. Thus, if a primary care provider wishes to screen for autism in children who are “too old for the MCHAT-R/F,” using a validated tool, he or she would have to invest in multiple, rather lengthy proprietary tools. For this reason, standardized autism screening in children over 30 months is almost unheard of in primary care.

The situation is much rosier for other developmental conditions in primary care. For instance, for attention deficit hyperactivity disorder (ADHD), primary care providers have use of the Vanderbilt ADHD Assessment Scales, which span a large age range, are in the public domain and are simple to score. If a parent presents to a primary care provider with an ADHD concern, providers can administer and score the Vanderbilt and start the ADHD diagnostic and treatment process. Similarly, the Screen for Child Anxiety Related Disorders (SCARED) is a brief, publicly available, and easily usable screening tool for children who present with anxiety symptoms. There is no similar tool for children who present with autism symptoms in the preschool years. As a result, many primary care providers are faced with the uncomfortable responsibility of determining a preschooler’s ASD risk with very little support. This situation leads primary care providers to make autism referral decisions based on their clinical impressions alone, which is problematic since autism knowledge is known to be low in pediatric primary care (McCormack et al., 2019). In addition, when providers lack evidence-based assessment tools, biases based on patient race/ethnicity or social background may be more likely to occur, which can perpetuate autism disparities.

If a population goal is to decrease the age of autism diagnosis, we need tools that can not only detect autism symptoms among the very youngest but also evaluate these symptoms in older children. Doing so could shift the curve of median ASD diagnosis toward a younger age (Figure 1) (Rose, 2001). As the figure indicates, if screening can facilitate earlier evaluations for children who would otherwise have been diagnosed well after the median age, then it can help to improve the age of diagnosis at a population level. Furthermore, the signs of ASD are often less ambiguous in older children, so it is possible that ASD screening tools might be more accurate (i.e. higher sensitivity/specificity) among older children. In particular, by preschool age, more complex peer interactions can be observed, and information is frequently available from preschool teachers or other caretakers, which might aid assessment of social communication and play skills. Finally, having another “check-in” point for ASD in primary care would give another formal opportunity for providers and parents to reflect specifically on a child’s ASD risk and make appropriate referrals, which itself might improve early access to diagnostic and treatment services.

OPEN IN VIEWER

Figure 1.

Lowering age of ASD diagnosis by shifting the curve child age (years).

But just as early screening represents only part of the solution, so does later screening—that is, other tools and strategies will also be needed to meaningfully improve ASD diagnosis at a population level. In their report, Improving Diagnosis in Healthcare, the National Academy of Medicine (National Academies of Sciences, Engineering, and Medicine, 2015) emphasizes team-based approaches to achieve diagnoses that are not only technically accurate but also effectively communicated to the patient. These recommendations are highly relevant to ASD, in which so many diagnoses begin with a primary care provider and culminate with a specialist while involving others, such as nurses, early intervention/early childhood special education providers, and child care providers along the way. In addition, the responsibility of seeking care after a positive screening test typically rests with a parent, thus underscoring the critical importance of convincingly communicating a the test result after an ASD screening. In addition to screening questionnaires, tools are therefore needed to facilitate communication and shared decision-making with families as well as among providers.

What would a process for screening preschoolers in primary care ideally look like? Screening of preschoolers for ASD could take several forms. One option would be universal screening of low-risk preschoolers, for instance, at the 4-year-old well-child visit, which is gaining momentum as a “school readiness checkup.” Unlike in the toddler years, there are few other standardized screenings recommended at these ages, so the addition of ASD screening to these visits might not be burdensome for primary care physicians and families. However, routine preschool screening might be a big step for pediatric practices, especially considering the current non-support of ASD toddler screening by the US Preventive Services Task Force (Siu et al., 2016). A first step, which might have broad acceptability in primary care, could be a formal recommendation for an additional screening visit at the 3- and/or 4-year-old well-child visit for children who are identified at higher likelihood of ASD due to risk factors (e.g. prematurity, affected family members), borderline toddler screening results, and/or parent/clinician concerns. Such a recommendation would need to be accompanied by new screening tools targeted at preschoolers in primary care. Such tools need to cover the entire preschool age range (30–60 months), be quick to use (e.g. less than 10 min), written at a low reading level, easily score-able by non-clinicians, available in multiple languages, and provide clear guidance regarding what to do with a positive screening test result and how to talk with parents about next steps. It is also critical that such tools are validated in diverse pediatric populations among children with and without comorbid developmental conditions. Having evidence-based tools for ASD identification might help pediatric primary care providers offer a more informed answer when a parent comes to them asking, “I’m worried my preschooler has autism. What do you think?”