I think there's also another critical issue that plagues every single observational study - the direction of causality. Our behaviors can and do physically change every single aspect of our body, including our brains. For instance chess masters not only activate different parts of their brain when playing chess than amateurs, but also have observable morphometric differences in their brains. [1]
If the direction of causality there was not overtly obvious (as nobody starts out anything even close to a master), one could easily assume those differences caused one to be a master, rather than vice versa. It's the exact same issue here, except the direction of causality is unclear. You're seeing a shared morphometric difference among people of some trait (ADHD) and then just assuming that that difference causes ADHD, rather than that ADHD causes the difference!
To even assume this is the case one would need to perform MRIs on random very young individuals prior to ADHD diagnoses, and then make a prediction, based solely on that MRI, on which ones would be diagnosed with ADHD with no input or information given to the participants before a follow study in a decade or so. That study would then help prove the direction of causality that's seemingly just simply being assumed in this one.
In a world where we had precision tools that could be used to deliberately alter brain morphology, the direction of causality would be highly relevant for trying to treat ADHD -- it would tell us whether there is any point in trying to use those tools. But for diagnosis only, I don't think the direction is actually relevant. That is, if we see a strong correlation between ADHD behaviour (as measured by, say, a standardised written test) and certain brain morphologies, then we're justified in concluding that ADHD has some objective physical manifestation -- that is, that the written tests aren't just measuring arbitrary collections of symptoms, they're estimating "something real". There might still be many different underlying causes.
Would this be useful? I think so: It would then be possible to assess the accuracy of a written test.
I think the risk of conflation with stimulant use that another poster mentioned is very real, though, and that problem is a close cousin of the "direction of causality" problem.
This is true and quite an interesting and nuanced point. But the connotation of this discussion, and something I expect applies to the overwhelming majority of people, is an association of physical manifestation (brain morphometrics or whatever) causes issue, rather than issue causes physical manifestation.
In particular this whole thread of discussion started back with somebody expressing a bit of skepticism about the definitive nature of ADHD (and other psychiatric disorders) and somebody responded with the brain morphometrics as proof of such, yet in reality that's mostly still just begging the question.
If the direction of causality there was not overtly obvious (as nobody starts out anything even close to a master), one could easily assume those differences caused one to be a master, rather than vice versa. It's the exact same issue here, except the direction of causality is unclear. You're seeing a shared morphometric difference among people of some trait (ADHD) and then just assuming that that difference causes ADHD, rather than that ADHD causes the difference!
To even assume this is the case one would need to perform MRIs on random very young individuals prior to ADHD diagnoses, and then make a prediction, based solely on that MRI, on which ones would be diagnosed with ADHD with no input or information given to the participants before a follow study in a decade or so. That study would then help prove the direction of causality that's seemingly just simply being assumed in this one.
[1] - https://www.frontiersin.org/journals/neuroscience/articles/1...