In this post, I discuss the study Mazurek et al. (2017). A Prospective Study of the Concordance of DSM-IV and DSM-5 Diagnostic Criteria for Autism Spectrum Disorder. I then chat with three of our clinical and research leaders at the Seattle Children’s Autism Center – Dr. Soo Kim, Dr. Raphael Bernier, and Dr. Gary Stobbe – about how the new criteria for Autism Spectrum Disorder in the DSM-5 have impacted clinical and research practices.

Background References: The Diagnostic and Statistical Manual of Mental Disorders

The Diagnostic and Statistical Manual of Mental Disorders (DSM) is published by the American Psychiatric Association. For a brief history of the DSM: See the American Psychiatry Association Website (DSM History) or Shorter (2015). For a brief history of the DSM related to autism diagnosis, see King et al. (2014). The current DSM-5 criteria for Autism Spectrum Disorders can be found on the CDC website. Generally speaking, the DSM is meant to give clinicians and researchers concrete descriptions of behaviors and difficulties (which are called “diagnostic criteria”) for various diagnoses, so that they can apply those diagnoses in a more consistent and reliable way. Every so often, those diagnostic criteria are reviewed and updated by experts in the field using the knowledge that’s been gathered in recent years. There were four important changes that came about with the DSM-5 during the most recent update: First, the DSM-IV, under the heading of Pervasive Developmental Disorders, included the sub-categorization of Autistic Disorder (AD), Asperger’s Disorder (ASP), and Pervasive Disorder-Not Otherwise Specified (PDD-NOS). Thus, children with autism features were grouped into one of three different sub-categories based on symptom presentation and their development early in life (e.g., whether they’d had delays in language development). In the DSM-5, a single diagnosis of Autism Spectrum Disorder (ASD) is used.

Second, the criteria within the autism diagnosis changed from three domains ([1] social interaction, [2] communication & [3] restricted repetitive and stereotyped behaviors) to two domains ([1] social communication and social interaction; [2] restricted, repetitive behaviors). This change reflects an understanding that difficulties or unique use of language may be a key part of social communication but also can be part of the restricted or repetitive profile of children with autism.

Third, sensory difficulties were included in the restricted, repetitive behaviors domain. As reviewed in Robertson & Baron-Cohen (2017), sensory difficulties have been a significant issue for individuals with autism for a long time, but in the past were considered secondary to the social differences and weren’t included in the diagnostic criteria. As part of the DSM-5, sensory traits are considered a core or primary issue in autism.

Lastly, severity of symptoms is now specified in the DSM-5. Why? From a research perspective, findings from a decade of research had suggested significant behavioral overlap in the DSM-IV subcategories, poor predictive relation between early diagnostic category and later ability, and there was little relation to treatment approach or genetic etiology.

A Prospective Study of the Concordance of DSM-IV and DSM-5 Diagnostic Criteria for Autism Spectrum Disorder

In the Mazurek et al. study, the authors investigate how the criteria from the DSM-5 (published in 2013) versus the DSM-IV (published in 1994) impacts eventual diagnosis. To do this, across multiple clinical locations, the authors enrolled 439 children (aged 2 to 18 years) who were receiving a (first time) autism diagnostic evaluation. The clinicians used a standard protocol for diagnosis across sites, which included the ADOS (Autism Diagnostic Observation Schedule, a standardized clinician-child interaction) and cognitive and adaptive skills evaluations. After all clinical information was collected, the clinician evaluated the child’s behaviors using the DSM-IV and DSM-5 diagnostic criteria. For each patient, some clinicians filled out the DSM-5 first, and others filled out the DSM-IV first.

In 93% of the cases, there was agreement between both versions of the DSM about whether the individual did or did not have a clinical diagnosis of autism. More specifically, for 248 patients, there was an agreement in both systems that the individual had autism; and for 160 patients, there was an agreement that the individual did not have autism. For children who received a diagnosis of Autistic Disorder (via the DSM-IV), there was a strong likelihood of being given a diagnosis of ASD (via the DSM-5).

Table 1. Concordance / Discordance for the DSM-IV and DSM-5 checklists

Total Patients



YES- Diagnosis of AD, ASP or PDD-NOS


NO- Diagnosis of AD, ASP or PDD-NOS


YES- Diagnosis of ASD




NO- Diagnosis of ASD



However, there were 31 patients where the DSM-IV and the DSM-5 did not agree, most were individuals who were diagnosed with the DSM-IV but not the DSM-5 (n=30). Children who were discordant for diagnosis were more likely to be older, score higher on IQ tests, or be female. Of the 30 children who met criteria on the DSM-IV but not DSM-5, many received other diagnoses. Alternative diagnoses included ADHD, Anxiety Disorder, Social Communication Disorder, and Global Developmental Delay or Intellectual Disability. In summary, agreement was very high overall, but individuals who might have been diagnosed with ASP or PDD-NOS under the DSM-IV, were less likely to receive an ASD diagnosis with the DSM-5.

One benefit of the DSM-5 was consistency across clinicians and sites. Both the Mazurek et al. study and another multisite study (Lord et al., 2012) suggest that it was difficult for clinicians to reliably decide between DSM-IV subcategories. That is, even highly trained clinicians, using the same evaluation measures, did not provide the same DSM-IV sub-diagnosis. One interpretation is that the DSM-5 criteria and lack of sub-categories is more reliable across clinical communities.

Does the DSM-5 change the criteria for the diagnosis autism? The simple answer is yes. But each update of the DSM has changed the diagnostic criteria for autism, and there has been substantial change from the first published case studies in the 40s to our current criteria. Ideally, each set of changes improves how the DSM works for clinicians, researchers, individuals with ASD and families so that it captures ASD more and more accurately.

Implementing the DSM-5 at Seattle Children’s Autism Center.

To find out more about the impact of the DSM-5, I’ve asked 3 of our clinical/research leaders to weigh in on how this has impacted their clinical practice and research programs.

(1) In moving from DSM-4 criteria to DSM-5 criteria, what has changed in your clinical assessments?

“I would say one change when I do clinical evaluations is the consideration of sensory differences in category of repetitive behaviors/restricted interests. Most impactful, I think, is the better characterization of individuals – getting rid of Asperger’s/PDD-NOS and adding the inclusion of with or without co-occurring intellectual disability and language impairment has been useful in communicating to the family and to other providers.” – Dr. Gary Stobbe

 “I agree with Dr. Stobbe about DSM-5 vs. DSM-IV. While individuals with PDD-NOS were grandfathered into DSM-5 ASD diagnosis, PDD NOS was applied very differently based on the clinician and the clinic.” – Dr. Soo Kim

(2) What have been the biggest challenges for families or advocates related to the diagnostic criteria change? 

“I think, on the negative side, has been in accessing services through the Developmental Disabilities Administration (DDA). With DSM-5, the DDA is now also requiring IQ criteria to access DDA services, which can be challenging. Although in some ways, DSM-5 it is an improvement, as the DDA did not accept the DSM-IV subcategories of PDD-NOS or Asperger’s.” – Dr. Gary Stobbe

 “I’d agree that the biggest issue is that state servicing guidelines currently don’t directly follow along with the DSM-5 and at times that can result in additional steps that have to be taken to access services.” – Dr. Raphael Bernier

 (3) Has the DSM-5 changed the way you do research?

“I’m not sure it has changed my research at all. Most research continues to use the ADOS (a clinician – patient interaction) and ADI (a clinician – parent interview) to confirm diagnosis, so there has been consistency in the “gold standard” measurement of autism symptoms. But the change has set the stage to think about subgrouping based on biology instead of further searching for behaviorally defined subgroups.” –Dr. Raphael Bernier

“In regards to research, depending on the type of study, changes in diagnosis criteria definitely impact on inclusion/exclusion. Who gets into research will change if you require participants to have a clinical diagnosis before enrollment. — Dr. Soo Kim