Podcast: Play in new window | Download
Subscribe: RSS
The question of whether or not autism is a difference or a true dysfunction in brain development has been debated for years. A new study from Canada demonstrates that within an autism diagnosis, there is less of a difference in symptoms in the last 5 years than there was 30 years ago. This raises a lot of questions about what autism researchers have been studying lately. It also reinvigorates the discussion of whether the heterogeneity issue in autism has become such an issue that those with a true dysfunction are not getting the services they need, because of an interpretation that autism is just a difference. The podcast includes thoughts of the community voiced on social media as well as opinions of the ASF scientific advisory board.
Article is open access here: https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2747847
The thing to remember, in my mind, is that with autism, uneven profiles of ability are the rule rather than the exception. So of course when you take an aggregate of a heterogeneous bunch of people, who have uneven profiles across a fairly broad range of possible deficits, the individual differences in each category are going to look small.
If four people each have a severity score of 4 in one domain of four different symptom domains, but a score of 0 in the other three domains, the average severity score for the entire sample in each domain is 1.
Yet, each of these people had a score of 4 in at least one domain! The average would be exactly the same as if they’d all scored 1 in every domain. In both cases the same number of “points” are being allocated but the interpretation should be quite different. So why assume it’s the latter case, and not the former?
What the study suggests to me is that many years ago, perhaps there were more people with deficits spanning the whole range of possible deficits, and now there are some people who have deficits in some areas but not in others. Our definition of autism allows for this; you don’t need to check every single box. But years ago it seems that, for whatever reason (I can conjure a few), more people were checking every box (like Rain Man). Thus, there is not enough evidence for me to accept the premise that this study actually reflects a greater number of people who have no real deficits receiving a diagnosis.
I think the lack of services is a real issue that strikes across the board. Individuals with a few highly impairing deficits are routinely overlooked because they lack other deficits which service gatekeepers believe (usually wrongly, and for no particular reason other than personal bias) to be essential while individuals with many deficits across the board remain underserved as whole. People are far too assumptive. They assume that having or not having one particular deficit is a good predictor of what other deficits that person may or may not have; generally, research has shown this to be untrue, yet it persists.
I’m a fan of the more RDoC-inspired solutions, which I see as the least assumptive: see the symptoms first, and forget using categorical claims (of a continuously distributed construct!) to define individuals. This is also a good approach for researchers, who are typically interested in a particular symptom domain over others. Diagnostically, I think what we need is not per se subtypes, and it’s definitely not hierarchical levels, as some seem to suggest. To me, that would be going backwards; if the issue being captured here is indeed that there is a higher level of heterogeneity across many potential domains of dysfunction (rather than simply less dysfunction across domains), how exactly do you plan on splitting people into such levels? Is having a score of 1 in each of four domains more or less severe overall than having a score of 4 in only one domain? Or is it the same thing? I actually don’t think any of those conclusions would be valid ones.
What I think would be useful is specifiers. Many more specifiers: “with _______”. You name it, there could be a specifier for it, but unless you named “intellectual impairment”, “language impairment”, or “catatonia” (which are all very good things to communicate about), there isn’t. Why can’t we broaden our lexicon of specifiers, to more easily demarcate exactly what the key deficits are for each individual?
Also… let’s stop using the AQ to screen participants. It’s terrible at doing its thing. I’m just saying. My apologies to Simon Baron-Cohen.
Thank you for those insights and for listening to the podcast.
It’s almost like there should be classifications, such as, I don’t know, “Aspergers” or “PPD-NOS” to be able to better research these things…