Autism Diagnosis Risks “Losing Its Meaning," Says Veteran Researcher. Others Say Her Arguments Just Don’t Stack Up

“Ever since I started doing research in autism – which was about 60 years ago, would you believe? – we even then thought autism was a heterogeneous thing, that every individual was somewhat different,” Frith tells me. “But there came a point, I have to say, in my own life, and after decades [that] this heterogeneity, which we always paid lip service to, […] became just too much, at least for me.”

As she asserts in a recent editorial, Frith now believes too many people are being diagnosed with autism when some may not truly fit the diagnosis, at least as she sees it. 

But let’s back up for a moment: what actually are the diagnostic criteria for autism?

“When you’ve met one autistic person, you’ve met one autistic person” 

One tool used to diagnose autism, as well as other neurodevelopmental conditions and psychiatric disorders, is the Diagnostic and Statistical Manual of Mental Disorders (DSM), published by the American Psychiatric Association. It’s currently on its fifth edition, referred to as DSM-5. 

The DSM-5 includes criteria for diagnosing “autism spectrum disorder” or ASD. It’s worth noting here that while this terminology is in widespread use among scientists and clinicians, research has indicated that many autistic people and their families don't consider autism to be a disorder and prefer identity-first language: that is, someone is an “autistic person,” not a “person with autism." 

The DSM criteria include the following signs of autism:

• Difficulties in social communication and social interaction
• Restricted, repetitive behaviors or interests
• Signs of the condition being present from early childhood
• Significant difficulties socializing, at work/school, or in other important areas of life
• No other intellectual disability that could better explain the difficulties

There’s also another tool called the International Classification of Diseases 11th Revision (ICD-11), which is broadly similar.

Some, but not all, autistic people also have an intellectual disability. Some of these individuals may have degrees of language impairment, ranging all the way up to some people who are non-speaking. 

Although it’s not mentioned specifically above, many autistic people also report sensory processing difficulties, whether that be struggling with too much sensory input (such as bright lights and loud noises) or seeking sensory input (such as banging or tapping objects in a quiet environment).

There’s an expression I’ve heard some autistic people use: “When you’ve met one autistic person, you’ve met one autistic person.” 

This is what Frith means when she describes it as heterogeneous – there are many different ways that autism can manifest, and the level of support that different autistic individuals need can be massively different.

Asperger syndrome: a cautionary tale

One attempt to conceptualize some of this heterogeneity is the autism "spectrum." 

Individuals are not “more or less autistic” depending on where on the spectrum they are placed; being autistic or not is a binary category. But the spectrum helps frame the different strengths and challenges each person might have and what kind of tailored support they may need.

Another “category” that once existed to help differentiate between autistic individuals was Asperger syndrome. Frith says she was initially in favor of Asperger’s being listed as a separate diagnosis. These individuals, she says, were “verbally fluent people, quite different from the original group that we studied.”

Asperger syndrome was only included in one edition of the DSM, the fourth. By the time DSM-5 came out, it had been removed. The story now serves as a microcosm of the way definitions of autism have shifted over the years.

I was perfectly happy then, but now I'm coming to think, was it really right? I mean, are [these different presentations of autism] really the same thing?

Dame Uta Frith

As one 2016 paper on the topic outlines, many people who met the criteria for a potential Asperger's diagnosis also fit the picture of autism, and it was more difficult to separate the two in clinical practice than had been assumed.

Add to that the controversy around the legacy of Hans Asperger and his debated links with the Nazi regime, and the term has fallen out of widespread use. There are still those who were diagnosed with Asperger syndrome who choose to continue to use it to refer to their own condition, but it’s not used as a diagnosis today.

This is just one of many meanders on the journey to the DSM-5 criteria we have now, which illustrates the difficulty in trying to separate out different “types” of autism.

But for Frith, who has watched all this unfold from the perspective of someone right in the midst of this research, it’s now time for another rethink, perhaps heading back towards defining separate "subtypes" of autism.

“I was perfectly happy then, but now I'm coming to think, was it really right? I mean, are they really the same thing?” she muses. “You can see some commonalities, but at the moment with the latest, sort of, wave of this huge increase in cases, I think […] that we can't accommodate that anymore under the same spectrum.”

“There's still work to do to ensure autism is recognized across its full spectrum”

The statistics are clear: more people are now receiving autism diagnoses than at any time since the condition was named.

A 2021 analysis of 7 million UK children, for example, found that 1.76 percent were on the autism spectrum. Another study from the same year concluded that the UK saw "a 787 percent, exponential increase in recorded incidence of autism diagnoses" in the period between 1998 and 2018.

The British Medical Association now estimates that around 700,000 people in the UK have an autism diagnosis. 

Across the pond in the US, rates are thought to be higher, with an estimated 3.2 percent of children aged 8 years having been identified as autistic according to the Centers for Disease Control and Prevention (CDC).

Globally, according to a 2022 systematic review, approximately 1 in 100 children have an autism diagnosis. 

Many scientists believe the major reason for this is not an actual real-terms increase in how many autistic people there are. Rather, it is greater awareness and recognition of the condition leading to more diagnoses.

For example, there has always been a large disparity between diagnosis rates in men and boys and those in women and girls. The CDC states that autism is over three times more common among boys than girls. 

“Historically, a lot of what we knew about autism came from studies of boys, so our understanding was incomplete,” said Dr Emily Casanova, co-author of a recent study that looked at autism incidence between 2016 and 2024 in a large US health system covering Illinois, Missouri, Oklahoma, and Wisconsin.

“We're now recognizing that autistic girls and women may present differently on average. On the other hand, autism is highly diverse, and there's no single ‘female presentation,’ so we also need to be cautious about making broad generalizations.”

Casanova and her co-authors found that within their dataset, female individuals received their autism diagnoses on average 3.4 years later than male individuals, being diagnosed at an average age of 15.7 years. This, for Casanova, was “one of the most striking findings."

“Without an explanation for their experiences, many people blame themselves for difficulties they don't understand. Delayed diagnosis can also mean delayed access to support, accommodations, and appropriate care,” Casanova tells me. 

“Despite growing awareness, there's still work to do to ensure autism is recognized across its full spectrum.”

I’m actually very sad and very angry sometimes [at] the way that the discussion has been usurped by the people who talk about autism [as if] it was a sort of lifestyle choice.

Dame Uta Frith

But when I raise the question of the gender disparity with Frith, she disagrees: “[I believe] that we never had a bias against girls being diagnosed. They were always there.”

She cites the example of Dr Temple Grandin, the American academic and animal behavior expert who also became a prominent author and speaker on her life as an autistic woman – one of the first autistic people to openly document their experiences in this way.

Grandin is, of course, just one person. In addition to the evidence from studies like Casanova’s, a quick search of Reddit or other online platforms will provide you with countless accounts from women who say their autism was not recognized when they were young, and that they missed out on crucial support because of this.

However, Frith suggests that many of the women and girls now being diagnosed have what she calls a “relatively mild form of autism," unlike Grandin. 

“When you think of the girls – the ‘hidden girls’ as they are sometimes called – they have behaviour that doesn’t actually remind even me of anything autistic. And all I’m told is, they’re masking.”

Masking refers to neurodivergent people consciously or subconsciously suppressing behaviors or characteristics in order to conform to patterns of behavior and interaction they observe in neurotypical people around them. 

That could look like choosing to make eye contact with other people, even if this is something you find very challenging, or not performing repetitive behaviors such as hand-flapping, even if you find this comforting. 

Masking is thought to contribute to delayed diagnosis in some autistic people because it can cause traits to go unrecognized. It can also be very difficult to maintain masking in the long term, which may lead to stress and anxiety. 

“There is growing clinical awareness of 'masking,' and I believe this is helping to identify more people, but there's still room for improvement,” Casanova told me.

While masking is often associated particularly strongly with autistic women and girls, other researchers have echoed Casanova in cautioning against using it to denote a “female type” of autism, instead emphasizing that stereotypes around autism are unhelpful. 

Frith, however, rejects the notion of autistic masking altogether. 

“To me it's a totally unscientific concept. Where are the data? [Is it] entirely self-selected people who say that this is the case? That's not science that you can't rely on, but that's what it comes down to.”

From the background research I've done to write this report, this appears to be a minority view. 

Charities, advocacy organizations, and health services all discuss masking as part of the lived experience of some autistic people, though it has been pointed out that it’s not exclusive to autistic people. We all modify our behavior, language, and mannerisms to fit different social situations, even unconsciously. 

As I’m listening to Frith’s perspective though, it strikes me that there’s a difference between a neurotypical person temporarily changing their behavior to suit one time-limited social situation versus an autistic person feeling like they must constantly perform an uncomfortable action like eye contact, without a break or relief, in all the social interactions they face.

A “broad umbrella” 

Frith tells me that she views the current DSM-5 criteria as a "simplification." In future editions, she feels it’s time to try splitting autism up again, as was attempted with Asperger syndrome.

One reason is that she feels this “broad umbrella” could actually be holding back scientific research.

“We just know too little about the brain and the mind,” she tells me. “I would foresee in the future ways of getting towards some biomarkers for particular phenotypes.”

“They could be genetically defined, or they could be what I would call cognitive phenotypes – because behavioural phenotypes are more tricky because behaviour is determined by many, many different things and it's all a continuum.”

“But when you look at the cognitive level […] you may be able to go to mechanisms,” she adds. 

The lack of a diagnostic biomarker – a biological “thing” that doctors can test for to make a diagnosis – is absolutely not something that is unique to autism and doesn’t preclude diagnosis of many other conditions. 

What Frith is suggesting is that should biomarkers be identified in the future, it may be possible to group autistic people together based on these underlying genetic or neurobiological features. Scientists are continually exploring these questions, but there are still a lot of unknowns. 

As to why she feels this shift in thinking around autism is important, she is emphatic that her goal is not to remove support from people who may not fit the new criteria, but instead to ensure that they get “more precise help."

“We had great efforts in destigmatisation, and I was one of the people who was very much working towards that in emphasising [strengths] as well as difficulties. But now, it’s almost as if you could only talk about strengths and not about difficulties.”

“It seems the kind of groups that I used to know [are] almost overshadowed,” she tells me, referring to autistic children with intellectual disabilities. 

“And I do hear from the parents of these children that they feel completely unrepresented by the charities and organisations. And I’m actually very sad and very angry sometimes [at] the way that the discussion has been usurped by the people who talk about autism [as if] it was a sort of lifestyle choice or something like that.”

“Substantial concerns”

Despite Frith’s insistence to me that she isn’t seeking to withdraw from people the support that an autism diagnosis can unlock, the publication of her recent editorial and other comments she has made previously and in the wake of it have drawn strong criticism. 

“Research about subtyping autism is growing, however these ideas remain entirely theoretical. They currently have no clinical value, diagnostic relevance or practical application,” said Mel Merritt, Head of Policy and Influencing at the UK’s National Autistic Society, in a statement. 

“There are substantial concerns about further categorising of autism and the inaccuracy of subtypes in the past meant they ceased be used. We fear that it’s unhelpful and may lead to more stigma and discrimination, labels being misinterpreted and jeopardising access to support.”

The goal [is] to help people understand themselves earlier and ensure they receive the supports they need to thrive.

Dr Emily Casanova

Educational psychologist Dr Sue Franklin also responded to Frith’s suggestions in an article for Tes Magazine, which had previously interviewed Frith in March.

“Frith’s challenge to our collective thinking about the spectrum, and her views on describing subgroups to better understand the varied presentations in children and adults, makes sense to me,” wrote Franklin.

“However, Frith’s perspectives, based in a medical model, appear to seek a return to a narrow and pathologising diagnosis of autism, and with this, a concerning approach to the female presentation of autism.”

On the masking question specifically, Franklin echoes my own misgivings: “Frith seems to underestimate the very real impact of children imitating, all the time, what neurotypicals do.”

“I believe Frith’s view that masking is not harmful is reductionist.”

Franklin concludes that while it may well be time for a rethink of autism diagnosis, this should be done with awareness of the condition in women and girls at the forefront, so as not to threaten such progress as has been made to address the disparity that Casanova told me about.

Writing for Psychology Today, Dr Ludmila Praslova at Vanguard University of Southern California spoke about the epistemic injustice that autistic people face and how Frith’s comments can be read as feeding into that.

“Epistemic injustice sounds abstract, but for many in the autistic community, this is the story of being told, over and over, ‘We know you better than you know yourself.’”

“Autistic people pay for epistemic injustice with their health, their careers, and far too often, their lives.”

Another dissenting voice was that of Professor Sue Fletcher-Watson, a developmental psychologist who describes Frith as “an influential figure” in her own career – Frith was even one of the examiners of Fletcher-Watson’s PhD thesis. 

In a blog post, Fletcher-Watson pushed back on Frith's assertion that data on things like masking may be somehow less valid because it's largely based on self-reports: “Without autistic people there is no autism, so saying that their experience is somehow beside the point, or even actively incorrect, is bizarre to me.”

Given the current lack of biomarkers, as Frith herself lamented in her editorial and to me, Fletcher-Watson contends that “the big breakthroughs for the autism field since my undergraduate days have nearly all been generated through autistic scholarship, developed using coproduction projects, and/or drawn from self-report measures."

She believes that far from being, as Frith called it, "unscientific," autistic people’s descriptions of their lived experience of things like masking are crucial parts of the evidence base that scientists and clinicians rely on. 

A later-life diagnosis provided answers to questions that had haunted me for years

Dr Steven Shorrock 

In Frith's March interview with Tes Magazine, she laid out some of the same views expressed in her later editorial – including the idea that the autism spectrum has now become "too broad."

In response, Dr Patrick Dwyer – an autistic researcher at La Trobe University in Melbourne, Australia – wrote a piece on his blog, Autistic Scholar.

"I do genuinely believe there are common themes and experiences that run through the broad autistic population, so we would need to consider the cost of some populations potentially losing access to concepts and ideas with broad explanatory value," Dwyer wrote.

"Moreover, surely there are common advocacy goals we ought to be able to agree and work collectively on, like promoting autism acceptance and improving inadequate disability benefits."

Dr Steven Shorrock also responded in a detailed post on his blog Humanistic Systems. Shorrock is a psychologist who wrote that he has "lived and family experience of autism, ADHD, and related conditions," and he writes that getting an autism diagnosis in later life "provided answers to questions that had haunted me for years."

Shorrock also echoed calls for the perspectives of autistic individuals to be included in any discussions around the future of diagnosis, not sidelined: "Advancing the field calls for blending scientific rigour with autistic-led insights, allowing for refinements such as evidence-based subgroups without unjustified gatekeeping."

“The goal isn't simply to diagnose more people”

Lots of commentators have spoken about the increase in autism diagnosis through a lens of concern. 

Frith’s thesis appears to be that the broadness of the diagnostic criteria is leading to a larger number of people than before falling under the umbrella of autism and thus “qualifying” for a diagnosis, and that this is a problem.

But after weeks of reading varying perspectives on this complex topic, I’m forced to ask myself: is it, actually?

“The goal isn't simply to diagnose more people – it's to help people understand themselves earlier and ensure they receive the supports they need to thrive. It's ultimately about quality of life,” Casanova told me.

I found my own thoughts resonating with the conclusion reached last year by BBC journalist Michael Blastland at the end of a five-part podcast series called The Autism Curve. In the series, Blastland sought a range of different perspectives on the question of why autism diagnoses have increased so much.

“We feel it often, the desire for answers, clarity – but maybe we can’t have that with autistic experience, any more than we can have a single vision of human experience. So, the surprising conclusion of a series that set out to understand the numbers and their meaning is that maybe we should ease off,” Blastland said in episode five.

“Ease off the definitional heartache. Definitions matter – they’re just not going to settle things the way we’d like.”

I can understand why people like Frith may want to develop a set of criteria for autism that is exhaustive and clear. But I can also hear and understand the views of those who think this simply isn't possible in the way Frith suggests, and that the collateral damage may be too great.

Frith’s editorial is published in Psychological Medicine.