“I feel more and more under pressure to really think about this problem,” says Dame Uta Frith, referring to the exponential rise in the number of young people being diagnosed with autism.

According to the Institute for Fiscal Studies, the number of children with EHCPs for autism spectrum disorder almost tripled between 2015 and 2025, accounting for 40 per cent of the total increase in EHCPs.

It’s a rise that has placed unprecedented strain on schools - and led the government to include autism as a focus of an independent review into rising demand for services for mental health, ADHD and autism, which is set to report later this year.

That review is welcomed by Frith, who is emeritus professor in cognitive development at the Institute of Cognitive Neuroscience at University College London (UCL), and the person who pioneered much of the research that underpins our current understanding of autism.

She is, she says, “very happy to see that some action is being taken to understand why there has been such a dramatic increase in ASD diagnosis and why we have such long waiting lists. The current situation is dire, and something must be done.”

However, the issue of rising diagnoses is, Frith adds, something she has so far been “hesitant” to comment on, due to the sensitivity of debates around overdiagnosis and uncertainty about how an already stretched SEND system should best meet growing need.

In the wake of the government proposing changes to EHCPs and new “tiers” of support for pupils with SEND, we sat down with Frith to discuss how our understanding of autism has changed during the course of her decades-long career, why she no longer believes that autism is a “spectrum”, and what this all means for how teachers work with autistic children.

What is autism? How has it traditionally been defined?

Autism is a neurodevelopmental disorder, which is to say that there is some pathology in the brain existing from birth. Some people object to the word “disorder”, but that’s what I would call it.

It’s lifelong, and the main features have to do with distinct problems in social communication and interaction. There is also an additional problem with what’s called repetitive, restrictive behaviours. This can present as narrow interests or sensory issues.

Neither of these characteristics has been researched sufficiently. My lifetime’s work has been to try to explain how you could have these particular difficulties.

Twenty years ago, behaviour geneticists came to the conclusion that the symptoms to do with repetitive behaviour have a different genetic origin from the ones relating to social communication difficulties.

So, that is interesting, because I wonder how much the whole concept of autism will continue to hang together if we do find a different genetic basis for these two halves. But this is all still in flux.

Over the years, the basic definition of autism that I’ve given you - that it’s lifelong and neurodevelopmental, and that there are communication difficulties and restricted behaviour - has remained the same. It is generally accepted. But the interpretation of that definition is a different matter, because we have made it more inclusive.

Do you mean that the definition has come to include more people?

Yes. Nothing is a neat category, and we wanted to include the not-so-typical cases. And so we widened the criteria, and the idea came about that autism isn’t just a single category; it is a spectrum.

But that’s very difficult, because what’s notable about being part of a huge spectrum that we all belong to? We’re all neurodiverse; we can accept this because all our brains are different. But it makes a medical diagnosis completely meaningless.

With the purely categorical approach, you had a very rare number of cases that were quite precisely defined, and then lots of people who just missed that. It was felt this wasn’t quite right, which is how we came to a spectrum-based approach.

But that still wasn’t enough. Because of various cultural factors, the spectrum has gone on being more and more accommodating. And I think now it has come to its collapse.

This is something that I don’t think has been quite recognised, because people still hang on to the idea that there is something that unites all the people who are diagnosed as autistic. I don’t believe that any more.

What would be a more accurate way to think about autism, if not as a spectrum?

I think at least we have two big subgroups: the people who are diagnosed in early childhood - usually before age three or age five, depending on things like their intellectual abilities and language - and another group, diagnosed much later.

This population is different. It is made up of a lot of adolescents, and among them, a lot of young women. These are people without intellectual impairment, who are perfectly able to communicate verbally and non-verbally, but who might feel highly anxious in social situations. They are perhaps characterised mainly by a sort of hypersensitivity.

That group is increasing at a rate that is just frightening, while the rate of the first group is only moderately increasing. And in autistic children with intellectual disability, there has not been any real increase; that group seems quite stable.

Now, I think the people in the second group really do have problems. I would definitely not say they are “making it up”. But I would say that these are problems that can perhaps be treated much better than under the label of “autism”. I would fight for that label to be limited to the first group.

Could an explanation for later diagnosis be that some people, particularly girls, are able to successfully ‘mask’ their symptoms?

The masking idea has no scientific basis, yet everybody, including the researchers and the clinicians, has been enamoured with this idea.

It’s understandable, because they listen to the lived experiences of people who say they’ve been masking, that they spend their time imitating what neurotypical people do, and they’re exhausted every day because of this. So, the harm is not the masking, but the exhaustion afterwards. I can’t quite understand that, because exhaustion could arise from lots of other causes.

I expect we could say we are all masking, all the time, trying to adapt to our society’s norms. So, from this point of view, I’m very critical of this idea.

There have always been girls with autism. The ratio was assumed to be four to one, and more recently three to one. In children diagnosed before age 10, this ratio has remained the same over decades.

Is there a cultural bias against identifying girls and women as autistic? Have they been unjustly overlooked? I don’t think so. We know, for example, that psychopathy is overwhelmingly male. Do we ask whether psychopathic girls have been overlooked? I haven’t seen anything like that.

There are just diseases that are more common in males and diseases that are more common in females.

Another way that the category of autism has changed is that Asperger syndrome is now no longer used as a diagnostic term. What effect has that had on how we understand autism?

Asperger syndrome was introduced as an official diagnostic term in the 90s, as a way of including people who would not fit the original autism description. These were people who had fluent language, sometimes a high IQ, and yet still had very typical symptoms of autism.

This was a small group, but once the label was introduced, parents who wanted their children diagnosed were almost all pushing for a diagnosis of Asperger syndrome. The label had acquired great cachet, perhaps because it was associated with certain figures from history and some famous scientists. Perhaps the label held hope that even if a child needed significant support, they might have hidden superpowers that could enable them to do well in certain niche occupations.

So this category was becoming desirable for the wrong reasons, and it was eventually abandoned in the latest Diagnostic Manual. I think the consensus was: “We have the spectrum, so why do we want anything else?”

That was an interesting decision, because it assumed a person could meet the diagnostic criteria for autism, whether or not they had language difficulties or intellectual difficulties. However, it was assumed that one could specify the level of support needs.

But what actually happened is that, sadly, the spectrum was stretched over and over into including people who would never have met the criteria of Asperger syndrome.

The spectrum has collapsed, but we might still find meaningful subgroups, each with its own label. That’s my view. But I know that other people would be totally against this.

How exactly is autism diagnosed?

These days, many people are self-diagnosing before they are assessed. I have heard from clinicians that they feel themselves to be under severe pressure to give the diagnosis, if the person has waited months or even years and feels absolutely sure that they’re autistic.

I fear that the way assessments are made now may not be quite right. I get the sense that people are relying too much on detailed interviews that provide a lot of subjective experience, rather than on objective clinical observation. For example, observing the flow of a conversation to spot indicators of poor reciprocal interaction, say, in eye contact or body language. I would guess these are things that are difficult to mask.

I fear there also isn’t enough attention being paid to what I would call the contraindicators.

To me, for example, it’s a contraindicator for the diagnosis of autism if the person can interact smoothly with you in conversation. If a person is autistic, the conversation will often feel stilted or abrupt.

Another contraindicator, in my view, is being able to read between the lines in a conversation, and to get irony and humour. There could be a whole list of these contraindicators, if one paid attention to them - and if people accepted them.

How well do we understand the causes of autism?

Our understanding is still very primitive; we are so ignorant about how the brain works when interacting with other people. It’s such a complex field, due to the brain being as complex as it is.

There are massive reasons to believe that autism is a genetically caused disorder. However, it’s not one gene or two genes; it’s hundreds. That’s the difficulty. And all those genes work together in the context of your environment.

Many environmental causes have been studied and ruled out, such as [pain-relief medication] Tylenol, vaccines, and what you might call “toxic parenting”. There are some non-genetic causes that do lead to an increase in the likelihood of autism, such as Valproate [a medication used to treat epilepsy] taken early in pregnancy, but the effects are very small. In other words, we are led towards mainly genetic causes that affect the development of the brain, possibly with different effects at different stages of development.

There have been lots of brain-imaging studies, but it’s not like a traumatic brain injury, which can be seen very clearly. What is going on in the autistic brain is often written about as “under-connectivity” or “over-connectivity” between the millions of nerve cells, but it’s all completely vague and at a level that we can’t quite capture. Our tools are not sensitive enough, and we don’t have any very good hypotheses. So there is a lot of work to be done.

We do not have a biomarker, and I suspect that when we get a biomarker, we need more than one. There will be different kinds of autism - very many different kinds.

What does all of this mean for how teachers support pupils with autism in the classroom?

I think it’s useful to make clear distinctions between the different types of children who have been diagnosed as autistic.

Those who are intellectually impaired and are diagnosed early cannot very easily be accommodated in mainstream. They tend to have very challenging behaviour.

Then, there are children who have been diagnosed as autistic but who have good language, and might once have fitted into the Asperger syndrome category. They can often really benefit from mainstream education, but you have to be concrete in telling them what you want them to do, and you can’t assume they understand norms of behaviour, as other children do.

Finally, we have the extended spectrum. This will mainly be older children or children in later primary. We don’t currently have a separate label, but it might make sense to treat them as “hypersensitive”.

I’m not sure whether this group is easier to teach or not. If they are incredibly anxious, they might be the ones who won’t come to school any more. That’s a big problem. Schools will need to find ways to build trust with these pupils. Work on building resilience might help them - something that is less likely to work with the neurodevelopmental group.

One thing I would say is that teachers often intuitively know what a child needs, even without a diagnosis. Together with parents, they can work out what the needs are and act on them straight away, rather than waiting for an assessment. That’s what I would rather see than a long wait for a diagnosis.

Is there anything that mainstream teachers should avoid when working with autistic children?

If we think about the reasonable adjustments that schools are expected to make, some of the most common requests are around sensory sensitivities - for example, letting a child wear ear defenders or turning down the lights in the room. These adjustments are easy to make, but they have very little scientific grounding.

We don’t really know what causes the sensory issues, and I think people love the idea of being able to do something about it. Giving a child ear defenders feels intuitive; it feels plausible. But I’m sceptical.

As far as I know, there hasn’t been any research into finding out whether these sensory adjustments actually improve lives. I’ve no idea whether it’s good for the child or not. It may feel good to them, but not actually be good.

One additional thing I want to say is this: our cultural understanding of the concept of autism, which has changed so much, is not just affecting ordinary people who might be interested in who they are; it equally affects teachers and other professionals, including me.

I’ve been quite swept up by the autism spectrum idea, and it’s only in the past 10 years or so that I have felt things have gone too far, and very slowly I have come to say, “No, this is not right”.

You can now get the UK’s most-trusted source of education news in a mobile app. Get Tes magazine on iOS and on Android