Experimental Test Could Allow At-Home Autism Assessment

A novel computerized adaptive test (CAT) quickly identified children and adolescents with suspected autism with comparable accuracy to conventional assessments, a new study showed.Using data from more than 10,000 children, researchers developed and calibrated a test that distinguishes children and teens with autism spectrum disorder (ASD) from neurotypical youth with high accuracy.The CAT-Autism test relies on parents and caregivers’ responses to an average of 13 questions — far shorter than standardized ASD assessments that can contain hundreds of questions. While more research is needed, investigators said the ultimate goal would be for parents and caregivers to complete the assessment at home and work with clinicians to understand the results and review next steps.The tool may be useful for earlier detection and intervention, said Suma Jacob, MD, PhD, professor, Della Martin Chair and director of Child and Adolescent Psychiatry at the University of California, Los Angeles (UCLA), lead investigator and co-developer of the test.“Our data shows that the CAT-Autism offers a rapid, accurate, and scalable approach to identifying autism spectrum symptoms in children and adolescents. We’re excited that this is our first step in seeing how useful it is for families and for clinicians who may need this in their practices,” Jacob told Medscape Medical News.The study was published online on July 22nd in JAMA Network Open.A Tool for Earlier Diagnosis?In the US, about 1 in 31 children are diagnosed with ASD. The rate of diagnoses has increased over the past decade, which research suggests is largely due to broader diagnostic criteria.Even though accurate ASD diagnosis is possible in children as young as 18 months, many children aren’t diagnosed until they reach school age. And existing gold-standard diagnostics processes can be lengthy, costly, and require trained clinicians, researchers noted.“Families are waiting to get an appointment when their kids could already be benefitting from some services if ASD is identified,” Jacob said. “The standard diagnostic assessments are still important to evaluate and plan but we aimed to bridge a gap in the pipeline for quick, accurate, and accessible identification that will help refer/guide families sooner.”CAT screening models work by adapting in real time to respondents’ questions to follow up with the next most informative question. Previous studies have shown that CATs are reliable assessments for psychiatric disorders. Among the investigators’ primary goals in this new study was to explore the testing format’s potential utility for identifying ASD quickly and accurately.Using data from the National Institute of Mental Health Data Archive, the investigators included 10,309 children and adolescents (mean age, 9.3 years; 74.9% men) in their analysis. Of these, 19.9% were neurotypical and 80.1% had ASD.To develop the test, the investigators included 490 sample questions that assessed symptoms across 10 domains: medical and psychiatric history, cognitive ability, executive function, language, motor skills, stereotyped behavior, emotional traits, recognition of norms, interpersonal interactions, and adaptive skills.The test questions were based on parent or caregiver response forms on a number of standardized tools, including the Child and Adolescent Symptom Inventory, Repetitive Behavior Scale-Revised, and Social Responsiveness Scale.Of these, 424 questions were identified as informative for their bifactor multidimensional item response theory model.To validate CAT-Autism, the researchers used CAT estimates from participants that answered 200 or more questions and compared them to scores from the full bank of questions.They also compared results from their screen to a confirmed ASD diagnosis by a clinician.High Accuracy With CAT-AutismThe test yielded high diagnostic predictive accuracy for distinguishing children aged 1-5 years with ASD from those who were neurotypical, with an area under the curve (AUC) of 0.95 (95% CI, 0.92-0.98). Similar accuracy was observed among 6-18 year-olds, with an AUC of 0.94 (95% CI, 0.92-0.97).In comparison, the longer full item assessment with 200 plus questions had an AUC of 0.93 for both age groups, suggesting that the CAT test was superior at identifying ASD, researchers said.On average, 13 questions were required for the test to be accurate (range, 6-45 questions).The investigators also assessed the utility of the tool in identifying the varying ASD severity. In the most severe cases, CAT-Autism distinguished children with ASD from neurotypical children with 100% classification accuracy in both age groups. The test also showed high accuracy for detecting moderate cases but was less specific for distinguishing between children with mild ASD and those who were neurotypical.Based on prior research, CAT-Autism may only take up to 3 minutes to complete and predict ASD with high accuracy, the investigators noted.Clinical Interpretation KeyThe tool could be a potential game changer for ASD diagnosis, said Audrey Brumback, MD, PhD, an assistant professor of neurology and pediatrics at the Dell Medical School at The University of Texas at Austin who was not involved in the study.“To be able to essentially democratize autism diagnosis so that it can be done by people who aren’t subspecialists and so that it can be something that insurance providers will hopefully recognize as a valid diagnostic tool…it could be transformational,” Brumback told Medscape Medical News.However, she cautioned that just because a child may score a negative result on this test doesn’t mean they do not have autism.“The value of this is that it’ll be able to take cases that are very clear and provide answers to the family quickly and then cases that are not so clear can have further evaluation through specialists,” she said.Limitations of the study were that the test relies on parent and caregiver reports that can be subject to bias and that further testing is needed to validate efficacy in clinical practice.While parents or caregivers can fill out the test on their own, Brumback and Jacob both advised that results should be discussed in person with a clinician.“I think it’s always preferable for a clinician to be involved since they can ask questions about the context and help the family interpret findings,” Jacob said. When a concern arises, families need to know more than whether the result is valid — they need to know what steps to take next, she noted.While the initial results are promising, further research is needed to collect more data in various settings and communities to ensure applicability and accessibility, develop the test for a broader age range (eg, adults), and integrate the CAT-Autism into established clinical protocols, Jacob said.They plan to pilot the test in-clinic and online for research purposes as well as develop a rollout plan in the coming year.“Parents can work with clinicians, researchers, and educational institutions to complete this online at home or at their convenience, before reviewing next steps with professionals,” Jacob said. “We’d like its use to be connected with teams that know options for where families can go next to meet their child’s needs.”Jacob and Brumback reported having no relevant disclosures. The study was funded by the Semel Institute for Neuroscience and Human Behavior at UCLA and the Center for Health Statistics at the University of Chicago, Chicago.