Abstract
Timely identification of autism in children offers crucial opportunities for the caring environment to be adjusted to meet the child’s developmental needs. In many jurisdictions, a medically confirmed autism diagnosis further “opens the door” for the family and the autistic children to receive tailored intervention and support. However, the demand for childhood autism diagnostic assessment continues to outpace the supply of subspecialist assessors and teams across the globe, leading to long wait times. For instance, the current wait time in British Columbia, Canada, is over 80 weeks (BC Provincial Health Services Authority, 2023), and less than 9% of children in England receive a diagnostic appointment within 13 weeks (NHS England, 2023). Assessments for autism are exceedingly difficult to access in low- and middle-income countries (Pillay et al., 2021) and are further hindered by the lack of linguistically and culturally appropriate instruments (Tekola et al., 2016). Irrespective of location, wait times impede access to early therapeutic supports, prolong a stressful time for families (Bejarano-Martín et al., 2020), and for older individuals, delay understanding of their autistic identity (Corden et al., 2021).
Given the demand for assessments, the role of general pediatricians in autism diagnostic practices is undergoing increasing study. Researchers at Vanderbilt University Medical Center, USA, conducted foundational work in this area, developing and evaluating a training program for general pediatricians to make autism diagnoses (Swanson et al., 2014; Warren et al., 2009). Members of this group developed the TELE-ASD-PEDS, an easy-to-administer tool to support community-based diagnosis (Corona et al., 2023). Furthermore, Extension for Community Healthcare Outcomes (ECHO) Autism Screening Test for Autism in Toddlers (STAT) has incorporated training on the STAT to support pediatricians in diagnosing autism (Mazurek et al., 2019). A recent study, wherein a small sample of 20 children assessed by general pediatricians through this program also underwent in-person multi-disciplinary assessment with standardized tools, showed 95% diagnostic agreement between the two assessment models (Sohl et al., 2023). Penner et al.’s (2023) recent study on the concordance of autism diagnosis between general pediatricians and a “gold standard” multi-disciplinary team also showed a very high positive predictive value (0.89) for predicting an autism diagnosis in young children (1.5–5.5 years of age). A recent publication from McNally Keehn et al. (2023) reported on diagnostic accuracy of primary care providers in Indiana’s Early Autism Evaluation Hub and found very similar results (i.e., a positive predictive value of 0.93). In interpreting these results, it is important to note that there are no available data on agreement between expert teams to use as a comparator, as well as no commonly agreed-upon standards for diagnostic accuracy. These emerging findings in North America suggest that community-based assessors, such as general pediatricians, present a feasible and effective option to expand the assessment capacity to catch up with the needs for autism diagnostic evaluations in children.
Critically, we need to identify which children are best suited for an autism diagnostic assessment by general pediatricians. First, the children in the aforementioned studies have been in the toddler to early school year age range (i.e., oldest children 66 months for Penner et al. and 48 months maximum for McNally Keehn et al.). A consistent finding from the studies by Penner et al. and McNally Keehn et al. was the comparatively lower negative predictive values (0.6 and 0.62, respectively, compared to positive predictive values ~0.9), indicating that ruling out autism may require more expertise. Both studies suggested that co-occurring developmental delays are associated with higher diagnostic agreement between general pediatricians and expert teams. In the former study, children who were non-speaking and had a lower Visual Reception score on the Mullen Scales of Early Learning (MSEL) were more likely to be correctly identified as autistic. The latter study similarly found significant differences in the MSEL and the Vineland Adaptive Behavior Scales Third Edition Adaptive Behavior Composite between “true positive” versus “false negative” groups. Finally, both studies found that clinician diagnostic certainty ratings were significant predictors of accuracy. A potentially relevant correlate is “frank” presentations of autism, as introduced by de Marchena and Miller (2017), who surveyed clinicians and described phenotypic features of “clear-cut” presentations including motor mannerisms, atypical gait/posture, repetitive language, minimal social overtures, decreased eye contact, and unusual quality of speech. Features contributing to diagnostic certainty or “frank” presentations should not necessarily be conflated with support needs; indeed, many of these features may be visible signs of autism not requiring therapeutic intervention. Taken together, young children with accompanying developmental delays who engender a high degree of clinician certainty seem to be the best candidates for general pediatrician assessment, although further delineation of specific phenotypic features is needed.
More broadly, we should also understand which general pediatrician assessors are best for this model. Existing studies have relied on pediatricians who were either willing to undergo training (McNally Keehn et al., 2023; Sohl et al., 2023) or who were interested in participating in a study on diagnostic accuracy (Penner et al., 2023). As such, these results cannot be generalized to all pediatricians or primary care physicians; instead, this group could be considered a “coalition of the willing” who might also have a higher baseline knowledge of autism. Penner’s group has done companion qualitative work with both urban and rural pediatricians in Canada to identify factors influencing whether they diagnose autism in their practices (Das et al., 2022; Penner et al., 2017). Important features here included obtaining additional training for autism diagnosis in the community context, having time for consultations (versus only primary care), practicing in a group with other pediatricians, and having access to experts to discuss clinical scenarios. Indeed, both the ECHO Autism STAT and the Indiana model are communities of practice that provide participants with opportunities to discuss clinical scenarios from their practices and learn from each other as well as experts. How general pediatrician assessment applies to health systems outside of North America remains unknown—especially in systems that do not have a clear general versus specialty delineation in the area of children’s healthcare or in regions with different socio-cultural compositions and needs. Foundational work in India has used community health visitors to conduct screening and intervention; such models could potentially be extended to support autism diagnosis (Naithani et al., 2022).
There are additional aspects to consider for autism diagnosis for children beyond accuracy. A family-centered process is essential. It is important for diagnosticians to ensure families are equipped to navigate the complexities of gaining access to funding and services once a child is clinically diagnosed with autism, and general pediatricians also need to be supported in helping families after diagnosis. We should not assume, however, that a specialist assessment with standardized tools to integrate caregiver interviews and assessments with the child is in itself family-centered. A recent scoping review showed that parental satisfaction is tied to the manner of the diagnostician, information provided through the process, the diagnostic report, and provision of post-diagnostic supports (Makino et al., 2021). In the ECHO Autism STAT evaluation, families were in favor of receiving a diagnostic assessment in their community, as opposed to traveling to a tertiary center (Sohl et al., 2023). Importantly, family-centered care also needs to be culturally sensitive and safe. There are long-standing disparities in timeliness of diagnosis among Black and Latinx families in the United States (Mandell et al., 2009); the models that expand diagnostic capacity must ensure that access and quality of care are optimized particularly for equity deserving groups. Not only do cultural norms need to be considered for the purpose of diagnostic accuracy, but families also need to trust their assessor, which can be challenging if they have encountered discriminatory or abusive care in the past. One example of an excellent resource is the South Asian Autism Awareness Centre’s “Bridge the Gap” toolkit for diagnosticians working with South Asian and Black families (Shanmugathasan & Sivapalan, 2022).
Making an autism diagnosis is only one part of understanding the child’s strengths and needs. Traditionally, multi-disciplinary input was required as part of the diagnostic assessment to provide a complete profile of the child’s strengths and support needs, helping to identify relevant interventions as well as characteristics that can be celebrated and cultivated. While this represents an ideal scenario (and is sometimes necessary to determine the autism diagnosis), this model can create bottlenecks for diagnostic access. Instead, we suggest taking a lifespan approach to assessment for all autistic children and youth. In such an approach, we recognize the ongoing need to obtain and synthesize knowledge about the child’s strengths and needs, allowing additional assessments to occur when their results will directly inform therapeutic decisions and care recommendations. Such an approach requires two key elements: (1) timely access to psychoeducational and other health and developmental assessments, including through the education system and (2) integrated care approaches, including a single point of medical care co-ordination for autistic children (Carbone et al., 2010) and development of technology to support informational continuity (Green et al., 2022). The development of such an approach needs to be tailored to the local socio-cultural contexts and health systems structure, and thus, the engagement of community members, including families and caregivers, autistic individuals, and service providers, in this process is integral to ensure feasibility and successful implementation.
As autism researchers and clinicians, we are excited by the many opportunities for discovery as we seek to understand how diagnostic pathways can be improved. Hybrid effectiveness-implementation trials (Curran et al., 2012) of new diagnostic models can accelerate systems-level knowledge for further policy development on expanding autism diagnostic capacity. Co-design approaches can help to ensure that the resulting service models balance the needs of children, families, clinicians, and the broader system. Importantly, autistic perspectives on diagnostic pathways have been understudied to date. Finally, the diagnostic systems will need to adapt to advances in understanding of neurodevelopmental differences and precision health approaches. Integrating these approaches while considering the unique constraints and opportunities in different local contexts is challenging but essential to mobilize systems-level quality improvement initiatives. Given the many challenges and developments facing autism diagnosis, researchers and clinicians alike must be ready to adapt, evaluate new ideas, and keep the needs of children and families at the forefront.