When I started in Lambeth, I inherited the follow-up of a young man called Jamal. I was in a hurry the day I met him, and just read the last year or so of notes. I reviewed his ADHD control, and got to know him and his mum.
Jamal was charming, bright and funny. I mean properly, intentionally funny, in a wry, self-aware way that led me to jokingly suggest by the end of the visit that he consider a career in stand-up comedy (we don’t have enough Somali stand-ups from Stockwell).
So I was rather blown away, when leafing through his old notes, to find that he had been diagnosed with autism aged 4. We’ve always had a very good set-up for autism/Asd diagnosis, and this assessment seemed pretty robust.
Since Jamal wasn’t presenting as remotely autistic now, this raised a few questions for me. Was the diagnosis wrong? Had he been cured somehow? And what did this mean for our confident pronouncements at the point of diagnosis?
The accepted wisdom is that autism/Asd is, in the words of the National Autistic Society, a ‘lifelong neurodevelopmental disability’. What I want to argue is that what we know about the condition, about how children develop, and about the nature of diagnosis, do not support this unequivocal statement.
First, two disclaimers. I’m not talking about an autism cure- I’m not sure such a thing exists as most people understand it, as I hope will become obvious. Secondly, I’m not a researcher, but I hope researchers might read this and tell me what I’ve got wrong.
There are three points that to me lead to the conclusion that an autism diagnosis is not always for life.
1) Autistic behaviours and traits are continuously distributed in the population. This if a point that parents grasp instinctively- in pretty much every family with an Asd child’s there’s an obsessive uncle, or an anxious gran who wouldn’t fit with autism criteria, but have some difficulties. It’s an idea that had good evidence (see this paper for a good open-access example), and has spawned the concept of the.broader autism phenotype. One implication of this is that when we diagnose autism, what we are saying is that the traits and behaviours are so exaggerated that they cause significant distress and difficulty, not that the child is a member of a distinct sub-set of humanity. Another is that there is genuine grey area between ‘neurotypical’ and autistic brains. It amazes me that this idea is not universally accepted among professionals, but only last week Cathy Lord from the DSM task force refuted it (unconvincingly, in my view) at a conference in London. In the next section I need to explore why the idea of a discontinuity between autistic and neurotypical people persists.
2) We can only measure ‘autisticness’ indirectly
A huge but worthwhile amount of intellectual energy had gone into finding the common cognitive feature that underlies autistic behaviours. This effort may be, as Lynn Waterhouse argues, unsuccessful, but I think there has to be an essence of ‘autisticness’ in order for the diagnosis to hang together at all. There may be an occasion to deconstruct autism altogether, but this isn’t it.
The problem is that we can’t directly measure this essence, and instead rely on its behavioural manifestations to diagnose. We measure and observe difficulties in social communication and interaction, record rigid and repetitive behaviours, and, crucially, establish that these problems are disabling for the child.
Of course, this means that we are sometimes wrong, even with the very best assessments. After all, there is no scan we can do to confirm out clinical impression, and no one feature is unique to autism, so making the diagnose feels like doing a puzzle of an elephant: each piece is frustratingly grey until you finally get the big picture.
But the more important point is deeper. For a given degree of autisticness, the degree of observable behaviour and impairment vary, according to the supportiveness of the environment, and the child’s own state of mind. Combine this with the continuous distribution and grey area arguments and you start to see some fuzziness at the edges of autism.
So why the discontinuity idea? It seems to me from experience that as one progresses up a gradient of autisticness, there is a point where the levels of maladaptive behaviours sharply increase, as well as impairment. As the autisticness continues to increase the levels of behaviour continue to increase, but not as fast as at this crucial point which is, of course, our grey area for diagnosis. So if you take two populations, one diagnosed and one not, it can look very much like they are distinct, not continuous in their profiles, as long as you ignore the grey areas. I’ve tried to illustrate this here: Over-simplified attempt to capture the relationship between autisticness.
So I maintain that autisticness is a dimensional, not categorical, characteristic of developing brains. The next question is whether it is stable over time.
3) Children develop!
This seems like a candidate for most obvious statement ever made. The thing is that the accepted wisdom in paediatrics is that, because autism is a genetic disorder (albeit multi-gene), while the child will develop his capacities, the autism will remain unfixed and unmovable. This is based on the belief that autism is overwhelmingly (at least 90%) heritable, and that therefore an individual’s genes set their degree of autisticness in stone.
Unfortunately, recent data has challenged this assumption. If, as the recent JAMA paper suggests, the heritability is closer to 40%, the environmental influences become important, and some (but not all) may exert an ongoing and modifiable effect on autistic behaviours, impairment, and possibly even the elusive autisticness itself. The problem is we haven’t a clue what these environmental factors might be, how many are ante-natal, still less which might be modifiable. But there seems to be evidence that these traits are not entirely stable in the general population, and studies of diagnostic stability in autism have widely varied in their findings, being lower in younger children (who have more developing to do). While this doesn’t show definitively that autisticness can change over time, I think it’s hard to argue that it is fixed in all children.
So autisticness is a dimensional variable of human development, which we can measure only indirectly, and which leads to a variable degree of impairment. . The distinction between autistic and neurotypical is a nuanced one, with no objectively defined ‘cut-off’. Diagnosis is based on observation of behaviour and detection of impairment, and these vary over time, as does diagnostic stability and, potentially, autisticness itself.
What does this mean for Jamal? I suspect some of his autistic behaviours were more due to his inattention and language delay, but equally, it seems quite possible that he did have genuine autistic-like difficulties, that the diagnosis was both appropriate and helpful, and that his autistic difficulties have faded with time.
I’ve met a few other boys like Jamal, diagnosed with autism in early childhood and subsequently growing into a neurodevelopmental profile which doesn’t really fit with the diagnosis, although in all cases that I’ve seen, there are other issues.
Does this mean it’s a curable disorder? Well, no. A cure is an agent that will act on an identified pathology to eliminate it, and we are nowhere near that in any sense. What this model suggests is that there may be some movement along the ‘spectrum’, that a child may enter or leave the ‘grey area’ between diagnosis and normality, and that potentially we may get a handle on how to influence this movement. Potentially.
More importantly, what does this mean for what we should tell parents at the point of diagnosis? A diagnosis remains helpful in guiding therapy and behavioural intervention in the here and now, but I do think we need to be cautious about long-term prognostication, and avoid the word ‘lifelong’. As woolfenden et al point out, one needs to be particularly careful when diagnosing young children with associated cognitive impairment. I would add to this from my own experience that one also needs to be cautious when diagnosing very hyperactive young children, and able young people suffering social difficulties where an asperger’s profile is suspected.
So this is not a call for revolution, but a corrective to what I think of as a bit of medical hubris, the idea that we can discern from our clinical skills the deep structure of someone’s thoughts, and predict that this condition we detect will be with them to the grave.
As for Jamal, I asked him next time we met what he thought of this autism diagnosis. He look at his feet, and shrugged. I said that I didn’t think it applied to him anymore, and he grunted in a very un-Jamal way. But when he was leaving, he fixed me with his most dazzling smile:
‘so I ain’t autistic no more?’
‘no, I don’t think you are’
And off he went.