A recent JAMA Pediatrics article concludes that the Autism Diagnostic Observation Schedule (ADOS) is “generally not required” for diagnosing autism in young children. The authors report 90 percent agreement between a clinician’s initial diagnosis and the diagnosis made after reviewing ADOS results. They argue that experienced developmental-behavioral pediatricians (DBPs) can diagnose autism accurately without structured tools.

That conclusion may hold inside the eight academic centers included in the study. It does not hold in the real world where most American children receive care.

I spent forty years diagnosing autism in settings far removed from the curated environments of DBPNet. I worked in Peoria. I worked at Madigan Army Medical Center. I worked in corporate pediatrics, where time is rationed, staffing is inconsistent, and diagnostic rigor varies widely. In those settings, clinical judgment alone did not produce reliable diagnoses. Not even close.

The study assumes a level of expertise that barely exists anymore. Only about 350 to 400 board-certified DBPs remain in the United States with 20-plus years of clinical experience like the clinicians in this article. Most practice in academic centers. Most children with suspected autism never see one. Their diagnoses come from general pediatricians, family physicians, APRNs, and PNPs. The study never acknowledges this. It describes a diagnostic ecosystem that applies to a tiny fraction of American families.

The authors also fail to specify how clinicians quantified DSM-5 criteria. They do not describe how many A-criteria or B-criteria behaviors were documented. They do not explain how those behaviors were operationalized. They do not state how clinicians distinguished between “present,” “absent,” or “insufficient information.” They simply state that clinicians used a DSM-5 checklist.

That is not how accurate diagnosis works.

For decades, I used a structured DSM-5 table that broke each criterion into its component behaviors. I documented specific examples from history and direct observation. I listed behaviors that supported the diagnosis and behaviors that argued against it. I quantified severity. I integrated developmental, language, adaptive, and social data. When I worked with diagnostic teams, we reviewed every criterion together. When I worked alone, I followed the same process. That level of specificity is not optional. It is the foundation of diagnostic accuracy.

The study never explains how its clinicians reached their decisions. It assumes that “clinical judgment” is a uniform, reliable construct. It is not.

In Peoria and at Madigan, I saw the opposite of what this study implies. I saw DBPs who were excellent and DBPs who were not. I saw children diagnosed with autism because of language delay alone. I saw children dismissed as “behavioral” when they met full DSM-5 criteria. I saw toddlers labeled with autism based on a 10-minute observation in a chaotic clinic. I saw school-aged children denied services because a clinician “didn’t see it today.” I saw both false positives and false negatives. I corrected those errors with structured tools, including the ADOS.

The study’s children had a mean age of 39.9 months. Many could have been assessed with the Screening Tool for Autism in Toddlers-Primary Care Providers (STAT-MD), a 15-minute structured tool designed for toddlers. Both tools are exceptionally good at catching true positives. In head-to-head testing, if a toddler meets the autism threshold on the ADOS-2, there is an approximate 91 percent mathematical probability that they will also fail the STAT-MD. Instead, the study assumes that clinicians have 60 to 90 minutes to observe a child directly. That is a luxury of academic centers, not community practice. In corporate pediatrics, 30 minutes of observation is rare. The study treats time as if it is infinitely available. It is not.

The Madigan model

The study also ignores operational models that already solved the problem it claims to address. At Madigan Army Medical Center, we built a diagnostic system that was fast, accurate, and reproducible. A PhD-level pediatric nurse practitioner performed the initial evaluation using a structured developmental history, targeted behavioral probes, and the STAT-MD (a 15-minute, developmentally tuned observational tool for toddlers). I then confirmed the diagnosis using a DSM-5 table that broke each criterion into its component behaviors, with specific examples supporting or opposing the diagnosis. Every decision was anchored in observable data, not impressionistic judgment.

That model produced consistent, high-quality diagnoses in under 30 minutes. It did not require 60 to 90 minutes of observation. It did not require a full ADOS battery for every child. It did not require a DBP with 20 or 30 years of experience. It required a disciplined workflow, a structured tool, and a clinician who understood how to integrate history, observation, and DSM-5 criteria with precision.

The JAMA Pediatrics study never mentions models like this. It never acknowledges that structured, interdisciplinary systems can outperform solo clinical judgment. It never acknowledges that most DBPs do not have the experience level the study assumes. And it never acknowledges that the majority of autism diagnoses in the United States are made by general pediatricians, APRNs, and PNPs, not by the 350 to 400 DBPs who remain in practice.

The Madigan model demonstrates that accuracy does not come from time or pedigree. It comes from structure, training, and accountability. Those elements are missing from the diagnostic environments where most American children are evaluated.

The authors report that clinician diagnostic certainty was the strongest predictor of accuracy. That finding is not surprising. Experienced clinicians who have the time, training, and institutional support to perform comprehensive evaluations tend to be accurate. But that is not the system most families encounter. In many clinics, diagnostic certainty is a euphemism for diagnostic confidence, and confidence is not the same as competence.

The study’s sample, forty clinicians from DBPNet sites, does not represent the median DBP in America. It represents the best-resourced, most academically aligned subset of the field. DBPNet is not a cross-section of real-world practice. It is a self-selected network of clinicians who already believe in their model. Their outcomes reflect their environment, not the national landscape.

The danger of this study is not its data. The danger is its generalization. If policymakers, insurers, or administrators interpret this as evidence that structured assessments are unnecessary, the children most at risk for misdiagnosis will be the ones who suffer. The study’s conclusion may be valid for the eight centers that participated. It is not valid for the broader system in which most children live.

The ADOS is not perfect. It was never meant to be a stand-alone diagnostic test. But in the hands of clinicians working under real-world constraints, it remains an essential part of a reliable diagnostic process. It provides structure when clinical observation is rushed. It provides objectivity when clinician bias creeps in. It provides a common language when teams disagree. And it provides a corrective lens when clinical judgment is flawed.

Academic centers may not need the ADOS. The rest of the country still does.

If the authors had titled their paper “ADOS Not Required in Elite Academic Centers With Ample Time and Highly Trained Clinicians,” I would have no objection. But that is not the title. And that is not how the study will be interpreted.

Autism diagnosis is too important to be reduced to a binary debate about tools. The real question is not whether the ADOS is required. The real question is whether the system in which most children are evaluated is capable of producing accurate diagnoses without it.

Ronald L. Lindsay is a retired developmental-behavioral pediatrician whose career spanned military medicine, academic leadership, and national advocacy for dignity-centered neurodevelopmental care. His NIH-funded work with the RUPP Autism Network helped define evidence-based approaches to autism and related developmental disorders.

He directed the LEND Program at The Ohio State University and founded JBLM CARES, a $10 million autism resource center for military families. His writing spans clinical scholarship and long-form fiction. He is the author of The Mercy Directive and the six-novel Cassandra series, a completed political and medical fiction saga tracing the rise of the Cassandra system from its origins to its national and international legacy. His forthcoming memoir, The Quiet Architect, examines how conscience and structure collide in modern medicine.

He shares updates on LinkedIn.