Trained Community-Based PCPs Can Make Accurate Autism Dx

Community-based primary care providers (PCPs) who receive specialty training can make accurate autism spectrum disorder (ASD) diagnoses in most cases, researchers found.

Among 126 kids ages 14 to 48 months, evaluation of the diagnostic accuracy of the Early Autism Evaluation (EAE) Hub system — a statewide network in Indiana providing specialized training to community PCPs — showed that ASD diagnosis was consistent between the EAE Hub clinicians and ASD experts in 82% of cases, reported Rebecca McNally Keehn, PhD, of Indiana University School of Medicine in Indianapolis, and colleagues.

Sensitivity, or correct classification of ASD diagnosis, was 81.5% (95% CI 72.4-88.1), while specificity, or correct classification of non-ASD diagnosis, was 82.4% (95% CI 66.5-91.7), they noted in Pediatrics.

The positive predictive value was 92.6% (95% CI 84.8-96.6), and the negative predictive value was 62.2% (95% CI 47.6-74.9). Overall, 60% of cases were true positive, 5% were false positive, 22% were true negative, and 14% were false negative.

“We have a national public health problem where the number of young children needing autism evaluation exceeds the capacity of specialists trained to provide this service,” McNally Keehn told MedPage Today in an email. “There is a pressing need for innovative solutions because delayed diagnosis leads to missed opportunities for interventions at the optimal time of impact.”

These findings are “highly promising, especially as we advocate for acceptance of tiered primary care models of diagnosis with payers and other stakeholders who are concerned about the implications of inaccurate diagnosis,” she noted.

“Our study provides evidence that a one-size-fits-all model of autism evaluation is not needed to achieve high rates of diagnostic accuracy,” she added. “Further, if we were to scale our model to serve all young children who need autism evaluations within their local primary care setting, burden on specialty care would be reduced, allowing for improved access for children who require a higher level of diagnostic expertise.”

There was no difference between EAE Hub sites in overall accuracy (P=0.89) or proportion of false-negative cases (P=0.67). Similarly, there was no difference in overall accuracy (P=0.24) or proportion of false-negative cases (P=0.09) by EAE Hub clinician for those submitting data for five or more children.

In an accompanying commentary, Susan Hyman, MD, and Abigail Kroening, MD, both of Golisano Children’s Hospital at the University of Rochester in New York, wrote that the researchers “demonstrated that with appropriate training and tools and the collaborative efforts of PCPs and specialists within a structured and committed system, that PCPs can accurately diagnose autism in most preschool children, thus expanding workforce capacity and decreasing wait time for evaluation.”

“Increasing access to diagnosis increases demands on the medical system for etiologic, medical, and behavioral care and the educational system for appropriate and equitable intervention,” they continued. “To meet the goal of earlier entry into services, improvements in access to diagnosis must be accompanied by increased access to effective interventions and family support.”

Of 182 referred children, 131 were enrolled, and 126 kids with index and reference evaluations from six EAE Hubs were included in the final analysis. The mean age of children was 2.6 years, 77% were boys, 66% were white, and 14% were Hispanic/Latinx.

Children’s scores on measures of developmental and adaptive skills “fell well below the average range and ASD symptom severity was in the moderate range,” McNally Keehn and team noted. Of the included children, 75% had a reference diagnosis of ASD, while 10% had global developmental delay, 10% had language delay, and 5% had another emotional, behavioral, or medical concern.

An exploratory analysis of differences between true-positive and false-negative cases showed a significant difference in dichotomized (“highly” or “completely certain” versus all other ratings) EAE clinician diagnostic-certainty ratings between the two groups (P=0.002), with a higher proportion of “highly-completely certain” ratings for the true-positive group (95%) compared with the false-negative group (65%).

Similarly, EAE Hub clinicians flagged a significantly higher proportion of false-negative cases (69% of false negatives, and 17% of true positives) for specialty follow-up evaluation (P=0.001).

Across measures of developmental and adaptive skills, there were significant differences between true-positive and false-negative cases (P<0.001), with the true-positive group showing significantly greater impairment as compared with the false-negative group.

McNally Keehn and colleagues noted that a primary limitation of their study was the high proportion (75%) of reference ASD diagnoses in the sample.

Furthermore, the inclusion of only children with English-speaking caregivers affected the generalizability of the study’s findings, they said. They also asked clinicians to flag kids who required specialty follow-up evaluation, but since the study was designed to force clinicians to make a binary choice about ASD outcome, that may have resulted in “artificially deflating accuracy indices because of caution against overdiagnosis.”