Trick Question: "Is the ADOS a Good Test?"

[Author's Note: This was originally a PED-NPSY post in response to a very long, evolving thread between many participants about whether or not the ADOS is, or should be, the "gold standard" for diagnosing autism spectrum disorders. I have no idea if the post will make sense without that context. However, I'm posting it here in case anyone who is not on the listserv wants to read it. It's reproduced here almost as exactly as written, but with a few more paragraph breaks, and a few specific references to previous posts in the thread removed for clarity. Also, when I say "ADOS", I am of course referring to the Autism Diagnostic Observation Schedule, Second Edition.]


I wonder if a big part of the problem with this whole conversation about "Is the ADOS is a good tool?" is that what we're really trying to answer the question of "What is autism?" without plainly asking this question. I've been thinking that our personal, implicit answers to the question "What is autism" so color our feelings about the ADOS that it's impossible to have any kind of dispassionate conversation about the ADOS unless we look at this underlying question. So, I guess, here are some thoughts about that underlying question and how it relates to the utility of the ADOS. 

While lots of kids with autism have a broad array of difficulties across multiple domains, as it is currently defined, autism is really just two clusters of symptoms:

  1. Social communication difficulties that are pervasive across all contexts, and

  2. Restricted interests and repetitive behaviors that are pervasive enough that they limit functioning across contexts. 

The ADOS is actually a pretty nice way of quantitatively observing if individuals show these symptoms in a given context, with the underlying assumption being that since autism is a pervasive developmental disorder, the symptoms should show up pervasively (e.g., in the context of testing, too). We have few other tests in our arsenal that help us quantify our clinical observations in a way that actually contributes to diagnostic formulation and treatment planning. Yes, the correlations between ADOS rater, and between ADOS raters and other raters of behavior such as parents and teachers, are not ideal. But, this is true of all behavior observations (indeed, the typical strength of the relationship between behavior observations and others' ratings is, if I may summarize my dissertation in two words, "super bad").


In general, I'm probably going to trust an evaluation from a clinician that I don't know if the clinician at least tried to use a reasonable tool to quantify their behavioral observations, rather than just going on their gut or experience (or lack thereof), especially knowing what we know about clinical judgment. And I think giving a lot of ADOSes (or is the plural ADOSi?) would give you great experience with those primary symptoms of autism, and how to at least try to assess them in a standardized manner. 

Of course, you don't need the ADOS to see those two clusters of symptoms, especially if you have tons of experience in working with individuals with autism - there are other ways to get that info. And of course, if you're looking for other symptoms besides social communication difficulties or RRBs, then of course the ADOS won't help you. 

Wait - why would you be looking for other symptoms

Well. We all know there are lots of kids who do not really have the primary features of autism (or only have very subthreshold symptoms), yet are experiencing clear social and adaptive difficulties. While these kids don't have the primary symptoms of autism, they do have at least some of symptoms that are sometimes but by no means universally seen in children with autism. I'm going to call these the "related symptoms" of autism. 

Here are some of the "related symptoms" of autism I can think of off the top of my head (I'm sure you can think of many others). Again, I'm talking about symptoms you sometimes-but-definitely-not-always see in kids with autism: 

  • Social discomfort, social anxiety, or some sense that others know the "rules" to the social world that they themselves don't know

  • Generalized anxiety or mood difficulties (maybe like Markham's "Intense World" theory of autism)

  • Introversion, reduced need for social interaction, or more extreme social aloofness

  • Unusual use of social imitation - e.g., trouble with imitation, or conversely, over-imitating peers to "blend in" at the expense of remaining authentic 

  • Strong emotional reaction to change and/or to something unexpected/"unfair"

  • Alexithymia and/or difficulty talking about emotions

  • Atypical cognitive profiles (e.g., extremely high IQ; extremely low IQ; extremely discrepant like a 40 point split between verbal and nonverbal IQ; psychomotor slowing)

  • Abstract thinking problems; black-and-white thinking; focusing on the details with difficulties seeing the 'big picture' (like Firth & Happe's weak central coherence theory of autism)

  • Broader language difficulties

  • Attention problems; EF difficulties; hyperactivity; difficulties with allocation of social attention

  • Social quirkiness that reduces the pool of potential like-minded peers 

  • Negative symptoms like avolition, amotivation, emotional flatness (or even positive symptoms like paranoia or neologisms)

  • Prioritizing own ideas and interests over others'; solipsism; difficulties going along with others' agendas or the group agenda

  • Some sort of sensorimotor involvement

  • A genetic or medical condition with known social or behavioral correlates


These kids clearly struggle due to their combination of symptoms, and clearly need intervention and support. But do they have autism? As far as I can tell, clinicians seem to fall into one of two camps: 

1. YES, AUTISM: That is, diagnosing these children with some/many of these "related symptoms", but without the primary symptoms, with autism, with related concerns like: 

  • That "subtler" presentations of autism are being missed

  • That our current diagnostic criteria focus on the wrong symptoms, or only a subgroup of important symptoms, when looking to explain social problems

  • That a missed diagnosis does a disservice to these children as it deprives them of the explanatory power of a diagnosis, the impact of parent/teacher/self-knowledge about the source of the problem, and access to needed services

  • That the criteria of autism may be biased in some way (e.g., against girls, or against those with high IQ)

  • That children who received treatment for the primary symptoms, which then improved, risk having needed services taken away from them when they are still struggling with many of these related symptoms

OR 2. NO, NOT AUTISM: That is, not diagnosing these children with "related symptoms" but without the primary symptoms with autism, with related concerns such as: 

  • That these "softer" diagnoses dilute the very meaning of the diagnosis of autism 

  • That in turn, this dilution affects our ability to do research on the causes, correlates, and treatment of autism, since any effects will wash out if your research includes lots of kids without actual autism, and that this dilution confuses the public about "what is autism"

  • That the actual problem (e.g., anxiety, introversion, ADHD, alexithymia, language disorder, psychosis, subcortical involvement, giftedness, OCD, schizoid disorder, depression with interpersonal sensitivity, over-pathologizing normal variability, etc, etc) is being missed and going untreated in these children

  • That the treatments for autism are not truly appropriate and/or fully necessary for these kids, and risk shunting precious resources away from the kids who do need them

This divide seems so personal and political - and so unstated! - that honestly, I stay out of most conversations my colleagues engage in about autism, for fear of alienating someone in the other camp. But it's hard to have a conversation about "Is the ADOS a good tool" (or any conversation about autism, really, including case consultations!) without explicitly discussing this divide, and the differences in values that drive this divide.

The ADOS is a fine tool - better than a lot of other alternatives really - for looking at the primary symptoms of autism. But for these kids with just the "related symptoms", well, there's the rub. Because, If you think these children should be diagnosed with autism, the ADOS is a horrible tool, because you're going to get way too many false negatives. If you think these children should not be diagnosed with autism, then the ADOS is generally unnecessary at best, and prone to too many false positives at worst. Either way, you're going to worry that giving the ADOS is trying to simplify a complex problem, and that the time spent giving it is taking away evaluation time that could be better spent assessing for these related symptoms (and that in the future, all we'll be given by insurance companies is authorization to do just the ADOS).

But these complaints aren't problems with the ADOS or its utility in identifying the primary symptoms of autism. The ADOS is a reasonably good tool for that job. These are problems with trying to use a hammer to make an incision.  It seems like we would all be better served by just agreeing "hey you know what the ADOS is a pretty OK hammer" and turning our attention to these kids with the "related symptoms" who need something other than a hammer.


For example, I'd much prefer to see more and better research on causes, correlates, subgroups, treatment response, and outcomes for kids with clear social and adaptive difficulties who DON'T score above the cut-off on the ADOS. What happens to these children long-term, and are the outcomes better, worse, or the same as for children with the primary symptoms of autism? Does it differ by subgroup, or by cluster of "related symptoms"? Does diagnostic accuracy and matching diagnosis to treatment really matter as much as I [as someone who does diagnostic evaluations for a living] think and hope it does, or is that an unnecessary concern? 

I'd also love to see better advocacy as a field for access to services for these kids with the "related symptoms." What if we put out a position paper saying, for example, that ABA is a great, theoretically-driven, research-tested intervention for LOTS of things that cause adaptive functioning problems, and the diagnosis of autism should not be used as a gatekeeper for services?


Or, for those who are connected with the school system, what about something looking at which students actually benefit from the services provided to students under the IEP category of ASD, and based on that, could that category be expanded and/or refined to ensure that it identifies all kids who would actually benefit from those services (and maybe even renamed while we're at it)?


Or for those in the pure research world, what if we got more involved in looking at kids using the RDoC Social Processes domain (or other criteria) to identify cross-disorder, underlying domains of social impairment that can be identified, evaluated, and treated regardless of diagnosis?


Or if you work in test construction, what about better ways to evaluate these "related symptoms", since some of them are pretty elusive to tease out with our current tools? E.g., now that I've got a reasonably functional hammer, can you give me some scalpels please?  To me, these seem potentially more useful than yet another 30+ post thread about whether or not the ADOS is the "gold standard." (By the way, I have no problem with calling the ADOS the "gold standard" if it's taken to mean the standard against which future tests should be measured. If you come up with a new test for autism (and I hope you do! Please do!), I think it ought to work better than the ADOS before I'll start using it.)


Just my two cents for the one person who might still be following this thread.



©2018 by Stephanie Nelson, Ph.D.