The quote heading this post - "Our findings beg the question, what is going on with these children who no longer have an ASD [autism spectrum disorder] diagnosis?" - comes from some media coverage of the findings reported by Lisa Shulman and colleagues . Shulman et al (bravely) set about examining an important phenomenon in autism research and practice circles: those who were previously diagnosed as being autistic / having autism but at a later date 'no longer met the diagnostic criteria for autism'.
I've talked about these so-called 'optimal outcomers' quite a bit on this blog (see here and see here and see here for examples). I know such discussions aren't everyone's cup of tea, particularly those who see autism as so much more than a diagnostic label, perhaps akin to an identity. The fact of the matter is however that there is what I would call 'substantial evidence' in the peer-reviewed science domain and beyond that the idea that 'autism is a lifelong condition/disorder' does not necessarily cover the huge heterogeneity encompassed under the label autism. Some people, for whatever reasons, do not reach critical diagnostic cut-off points for autism on a lifelong basis.
So, what did Shulman and colleagues do and find? They reviewed the clinical records of over 500 children who were diagnosed with autism or autism spectrum disorder (ASD) at a specific clinic. Most were aged around 3 years old when first diagnosed and were followed up about 3-4 years later. Importantly most of the children participated in one or more intervention programs aimed at improving skills and the like and (hopefully) quality of life. Again, although not everyone's cup of tea, the words 'applied behavioural analysis' (ABA) are also mentioned as an intervention; something that has been discussed in the context of optimal outcome before (see here).
Shulman et al noted that 38 children, equating to around 7% of their group (38/569), "subsequently experienced resolution of ASD symptomatology and no longer met diagnostic criteria for ASD at follow-up." This figure (7%) is not a million miles away from other figures noted in other independent studies (see here and see here).
Further examination of records however revealed that not meeting diagnostic cut-off points for autism did not necessarily mean 'symptom-free' as various other symptoms/conditions were noted in about two-thirds of their 'optimal outcomers'. This included language disorders, attention-deficit hyperactivity disorder (ADHD) and even the signs and symptoms of psychosis in a few. Three of the 38 optimal outcome children were noted to be completely symptom-free (described as 'recovered from autism' with no other issues); something that has again been noted in other studies too (see here).
Then back to that quote titling this post: what is going on with these children who no longer have an ASD diagnosis? I'm sure some people will put it wholly down to initial misdiagnosis. Y'know, something along the lines of 'they weren't autistic in the first place' despite the fact that they previously met clinical cut-off points for a diagnosis. Minus sweeping generalisations, misdiagnosing autism is not something that can be completely taken off the table as per other examples in the peer-reviewed literature and beyond (see here and see here). Indeed, if one ventures down the pathway of misdiagnosis as accounting for results such as those by Shulman and colleagues, one must logically then assume that such misdiagnosis is pretty widespread (at least in 7-12% of cases of autism). Such a situation also plays into other ideas too; particularly how self-diagnosis of autism is even more dangerous than has been hitherto suggested (see here and see here) with regards to the risk of misdiagnosis.
Other people might talk about things like 'masking' as accounting for such optimal outcome, where symptoms are merely being consciously hidden by those with autism (see here). It's an important area of study by all means but seriously ask yourself the question: how likely is it that a 6 or 7-year old child would be able to mask some fundamental signs and symptoms of autism so as to mislead a professional clinician that they didn't have autism having previously met cut-off points? Adults, yes perhaps some (see here). But young children? Be honest now...
Personally, I'm inclined to believe that at least some of those optimal outcome cases are genuine. That is, children (and adults) did meet the diagnostic criteria and clinical cut-off points for autism (including the criteria about symptoms significantly affecting day-to-day life) and then for whatever reason(s) symptoms abated. Intervention certainly could have played a role, but I'm also inclined to believe that behavioural intervention in particular, does not have the power to render someone who was autistic to be not-autistic. I know some big claims have been made about certain interventions down the years, but I've seen little [longitudinal] convincing evidence in the peer-reviewed literature yet.
There must be other factors at work. There must, for example, be a biological element to this. And as one example, just head back to all those discussions about certain types of infection potentially *leading* to the presentation of autism or autistic traits (see here and see here) as a possible template, and the outcomes mentioned for some. One possibility at least.
Much like discussions on another sometimes contentious topic - regression and autism (see here and see here) - there's enough peer-reviewed science literature to suggest that optimal outcome (or however you want to describe such 'growing out of' issues) is a very real scenario for some. Not all, but for some. And so once again the call goes out to start studying the genetics and biology of these so-called optimal outcomers, and then ascertaining whether any findings might have some important implications more widely for the [plural] label of autism...
 Shulman L. et al. When an Early Diagnosis of Autism Spectrum Disorder Resolves, What Remains? J Child Neurol. 2019 Mar 12:883073819834428.