"specificity for diagnosis was relatively low": the psychometric properties of autism diagnostic measures

The quote accompanying this fairly brief post - "specificity for diagnosis was relatively low" - comes from the findings reported by Sarah Wigham and colleagues [1] who undertook a systematic review of various "structured questionnaires and diagnostic measures" used in the assessment of autism in adults.

Their conclusions, based on some 20 studies identified in the current peer-reviewed literature, suggest that 'could do better' is a phrase best suited to various measures currently used to identify adults with autism, particularly in the context of an often complicated clinical picture (see here).

Similar things have already been discussed on this blog (see here for one example). In particular, how individual self-report 'are you autistic?' screening instruments whilst making good 'pop psychology' (see here) are absolutely no match for a thorough professional clinical assessment, save other important diagnoses/conditions being overlooked and going unmanaged (see here and see here). I know this puts the concept of 'self-diagnosis' as a result of the use of such instruments in some hot water, but as in many other branches of medicine and psychiatry, professionals and the assessments they conduct are there for a very good reason. Whether you can access such assessments in a timely fashion is an entirely different issue...

When I first tweeted about this paper being published, I emphasised one author on the Wigham paper in particular: Dr Tom Berney. The reasoning behind this was because of his involvement/link to research that has looked at how we identify adults with autism here in Blighty on the back of some headlines a few years back on estimating how many adults have autism here (see here). He, alongside some other notable authors who highlighted that '1% of adults with autism' figure, also talked about how some of the screening/assessment instruments used in that study weren't really cutting the epidemiological mustard [2]. It appears they might have been right.

So what lessons can be learned from this recent review? Well first, that whilst autism-related behavioural dimensions are vitally important to a diagnosis of autism, they are not universally specific to a diagnosis of autism, is important. Second is the need to perhaps move away from often very brief autism screening instruments that seem to provide a 'quick snapshot' to something rather more far-reaching and comprehensive. I know we all want a 'quick answer' that uses as few finite resources as possible, but sometimes, to get something right, you need to spend time and resources looking at it carefully. And diagnosing professionals also need to be mindful of notions of 'frank autism' too (see here). Finally, I'd like to re-emphasise that autism plus [3] does seem to be more typical these days, over autism appearing in some sort of diagnostic vacuum. As Wigham et al opine: "Robust autism spectrum disorder assessment tools specifically for use in adult diagnostic health services in the presence of co-occurring mental health and neurodevelopmental disorders are a research priority." Indeed they are.

----------

[1] Wigham S. et al. Psychometric properties of questionnaires and diagnostic measures for autism spectrum disorders in adults: A systematic review. Autism. 2018 Feb 1:1362361317748245.

[2] Brugha TS. et al. Validating two survey methods for identifying cases of autism spectrum disorder among adults in the community. Psychol Med. 2012 Mar;42(3):647-56.

----------