You are currently viewing Reviewing the ADOS for the Diagnosis of Autism

Reviewing the ADOS for the Diagnosis of Autism

By Dr Michelle Garnett and Professor Tony Attwood

Within their clinical practice, allied health individuals and teams may be asked by medical practitioners to provide an ADOS assessment to determine if the person is autistic. Schools may suggest to parents that they have their child assessed using the ADOS. What is an ADOS and why is it being recommended? 


Evaluation of the ADOS  

The ADOS refers to the Autism Diagnostic Observational Schedule, Second Edition and is a diagnostic instrument  for use in the diagnostic assessment of autism. Approximately two decades ago, the ‘gold standard’ in the diagnosis of autism was suggested to be the combination of two instruments, the Autism Diagnostic Interview – Revised (ADI-R), and the ADOS. The term ‘gold standard’  stuck. However, many health practitioners and members of the community are not aware that the two tests were created to measure autism as defined in previous editions of the diagnostic manuals, i.e. prior to the Diagnostic and Statistical Manual for Mental Illness (DSM-IV-TR, APA, 2000). Neither the ADI-R not the ADOS have been adapted for our evolving  understanding about what autism is, and how to recognise the expressions and adaptations, such as camouflaging,   that have emerged in the intervening 20 or so years, and thus these instruments have been found to miss more subtle presentations of autism,  for example those autistic individuals with  an additional personality disorders or  high intellectual ability (Langmann et al, 2017).   

Additionally, the suggested ‘gold standard’ was meant to apply if the person administering the ADOS was trained to research standard, where many clinicians are not trained to this standard, but instead to clinical standard. Taylor and colleagues (2017) examined inter-rater reliability for 27 Australian health professionals trained to clinical standard who watched nine videotaped ADOS assessments. Only 24% of the clinicians achieved ‘good’ or ‘excellent’ levels of agreement on the diagnosis. Inter-rater agreement for diagnostic status was low at 33%. 

Diagnostic assessment tools are evaluated using a standardisation sample, which is a group of people chosen because they represent the people that the tool will be used to assess. The group needs to be representative, and large enough to give statistical weight to the results. In the development of Module 4, the part of the ADOS that would be used to assess a speaking adult of at least average intelligence, the sample was only thirty people, and of this sample only five were female. The algorithms for determining if the person is autistic are based on previous definitions of autism and the predictive validity of the ADOS was only approximately 50%.  

A recent study (Barberesi, Cacia & Friedman, 2022) examined whether the ADOS was needed as part of the diagnostic assessment for 349 children aged 18 months to 5 years, 11 months. The results of the ADOS were compared to the results of experienced developmental paediatricians who did not use the ADOS. Interestingly, nearly 80% of the sample were male. The clinician’s diagnosis with and without the ADOS was consistent for 90% of cases. Researchers concluded that the ADOS is generally not required in the diagnostic process by experienced developmental paediatricians. 


What in the current ‘gold standard’ for diagnosing autism? 

The current ‘gold standard’ for the diagnosis of autism is a clinician assigned diagnosis, as provided by experienced clinicians. . An experienced clinician keeps abreast of the current diagnostic and other tools available, and chooses the best tools, based on the referral information, the age, the differential diagnoses, and the gender of the person to be assessed. They need to know what data to collect and the various ways that autism can present. When assessing for autism it is also important to assess for co-occurring conditions, for e.g. anxiety, depression and ADHD, as well as differential diagnoses, for e.g. trauma and attachment disorders, as well as personality and learning disorders. Assessment for these conditions requires a more in-depth assessment than the ADI-R or ADOS can provide. 

Where to from here? 

We strongly encourage clinicians who wish to increase their knowledge, expertise, and skill in diagnostic assessment methods for autism seek expert training in what data to collect, what to look for, and the best methods to collect the data. We recommend that, in addition to expert training, the clinician see many clients who are, and who are not, autistic. Once trained in the diagnostic assessment for autism, we highly recommend receiving clinical supervision for the first few years of practice. 

We have created an online Masterclass for health professionals wishing to add to their discipline-specific knowledge to increase their skill in diagnostic assessment and provision of support and therapy for autistic children and teenagers. The Masterclass focuses entirely on diagnostic assessment, including assessment for co-occurring conditions and differential diagnosis, and provides training on strategies and ideas for support and therapy.



American Psychiatric Association (2000). Diagnostic and statistical manual of mental disorders (4th ed., Text Revision). Washington, DC: Author. 

Barbaresi W, Cacia J, Friedman S, et al. Clinician Diagnostic Certainty and the Role of the Autism Diagnostic Observation Schedule in Autism Spectrum Disorder Diagnosis in Young Children. JAMA Pediatr. 2022;176(12):1233–1241. doi:10.1001/jamapediatrics.2022.3605 

Langmann et al (2017). Research in Autism Spectrum Disorders 34 pp 34-43 

Taylor LJ, Eapen V, Maybery M, Midford S, Paynter J, Quarmby L, Smith T, Williams K, Whitehouse AJ. Brief Report: An Exploratory Study of the Diagnostic Reliability for Autism Spectrum Disorder. J Autism Dev Disord. 2017 May;47(5):1551-1558. doi: 10.1007/s10803-017-3054-z. PMID: 28233080.