Hidden in Plain Sight: The Gender Gap in Autism Screening and Diagnosis

Written By Emily Winter

What is Autism?

Autism Spectrum Disorder (ASD) is said to have been first recognised by the medical field in the 1940s. Leo Kanner, the famous Child Psychologist, was the first to use the term ‘autism’ in his paper titled “Autistic Disturbances of Affective Control”. Kanner described a collection of characteristics including “extreme loneliness”, “preservation of sameness”, “excellent rote memory”, “repetitive sounds”, “hypersensitivity to stimuli”, “limited spontaneity”, “good cognition” and “intelligent families” which together he defined as autism¹. The collection of characteristics was further developed in the 1970s by Wing and Gould with the introduction of the triad of impairments: difficulties in social interaction and communication, as well as impairment of imagination (foundational to the theory of mind account seen later)². It was at this point that the ‘autism spectrum’, an idea that remains pervasive to this day, was first proposed, suggesting that individuals simply had more or less intense forms of these characteristics². Today, we recognise that this neurodevelopmental condition is a lot more complex than its 20th-century definition. Although communication difficulties, sensory processing differences and stereotyped behaviours and interests remain core to the diagnosis as defined by the ICD-11³ and DSM-5⁴, there is a greater understanding that all autistic individuals are completely different from one another, each with a unique combination of characteristics. 

Despite this improved understanding of the condition, statistics illustrate that ASD remains largely a male diagnosis. According to Brickhill et al. (2023), male autistics are three times more likely to receive an autism diagnosis than females⁵. Autistic males are also more likely to be diagnosed in childhood, with women making up a greater proportion of late-diagnosed autistic individuals⁵. This has called into question whether diagnostic criteria and screening tools for ASD are biased towards a ‘male phenotype’, resulting in underdiagnosis of autistic females rather than it truly being more prevalent in males. This article identifies where this systemic gender bias has come from, the consequences of it and how we may be able to improve the current underrecognition of autism in women.

Gender Stereotypes and Autism

Gender stereotypes are embedded within society. Humans have a natural tendency to categorise themselves, placing each other into boxes to make sense of our social environments⁶. Although contemporary legislation, such as the 2010 Equality Act⁷, aims to prevent discrimination based on nine protected characteristics, including sex, Okin argues we live in a society with “false gender neutrality”⁸, where laws and policies create a false sense of equality whilst deeper structural inequalities remain. Notably, these inequalities are just as prevalent in medicine as they are in broader society. For example, women are still largely underrepresented in clinical trials, and female-specific disease and female presentations of diseases are understudied. In the case of ASD, this is no different, with early screening and diagnostic tools based almost unanimously on male presentations. The current ‘gold standard'⁹ tools for autism diagnosis are the Autism Diagnostic Observation Schedule-2¹⁰ and the Autism Diagnostic Interview-Revised¹¹. While these are psychometrically strong - tested by scientifically valid psychological methods - they were not created with female autistic traits in mind and are derived from studies using male autistic samples. As a result of this, clinical expectations of autism have been largely determined by male presentations. Clinicians expect and have a greater understanding of male-coded autistic traits. The continued use of these diagnostic tools leads to confirmation bias and encourages clinicians to maintain their androcentric view of autism. 

This male view of autism is further reinforced by social expectations of men and women. Traditionally, men have been stereotyped as apathetic, emotionally detached and assertive, whilst women are viewed as more emotional, submissive and socially driven¹². These gender stereotypes, whilst more implicit than they were a century ago, enforce the idea that women and girls need to be sociable and outgoing, and this includes women with autism. In contrast to male autistics, for whom social isolation is less scrutinised, autistic women are forced to mask and imitate their peers to conform to social expectations. Ultimately, greater camouflaging (a term which encompasses the act of using compensatory strategies to overcome difficulties in social situations and assimilation strategies to fit in with a neurotypical group) and masking of autistic traits (the ability to hide autistic characteristics or appear as non-autistic) makes it even harder for autistic females to be spotted, especially through an androcentric diagnostic lens.

One glaring example of male-centred autism is the widely taught concept of “Theory of Mind”¹³. Theory of Mind is defined as the ability to recognise and understand that others may have thoughts, beliefs, and feelings different from one’s own and is frequently presented as a universal deficit present in autistic individuals. Medical school teaching typically suggests that autistic children struggle to interpret and predict others’ behaviour, which leads to impaired social functioning. However, while this impairment in mentalisation is often observed in autistic boys, research shows that autistic girls may present differently¹⁴. 

Social expectations, as we have seen above, encourage girls to be communicative, empathetic, and socially engaged and, therefore, autistic girls may not display an overt impairment in Theory of Mind during childhood. Instead, many autistic girls develop compensatory strategies, including social masking and imitation, as well as intense interests in individuals rather than objects, which helps them to learn about those around them, mitigating their social difficulties¹⁴. Significantly, “Theory of Mind” impairment was popularised by Simon Baron-Cohen, the same researcher who defined autism as an extreme version of expected male characteristics, the so-called “Extreme Male Brain Theory”¹⁵. This theory oversimplifies differences between genders, utilising gender stereotypes to define autism as an extreme adherence to the stereotyped male behaviour of hyper-systemising and hypo-empathising. His Theory of Mind, while formed through empirical evidence, was likely biased by his prior conceptions of the extreme male brain and thus has come under significant criticism, not just for ignoring differences in female and male autism, but also for not accounting for learned behaviours that come with age and the ability to 'hack' Theory of Mind¹ through compensatory behaviours and masking. 

Furthermore, diagnostic criteria tends to assume autistic individuals have a low intrinsic motivation for socialisation³, a characteristic that once again aligns more closely with male social behaviours. In contrast, autistic women often report a strong desire for friendship and connection. Society views social isolation in women less favourably than in men, resulting in a greater inclination for autistic women to camouflage their neurodiversity through masking and imitating neurotypical social behaviour⁵. Although difficulties with understanding emotions and maintaining eye contact were reported as equal between the sexes, autistic females are typically able to overcome this by studying their peers and emulating them.¹⁶. With this in mind, the societal pressures and expectations of the female sex alter the phenotype of autistic women, preventing them from being recognised as autistic as their behaviours do not match with the expected “poor social skills” quoted in diagnostic criterion. 

So why does this matter?

The lack of female autism awareness in clinical settings leads to delayed diagnosis in women, greater likelihood for misdiagnosis, poorer mental health outcomes and autistic burnout. 

The greater the extent to which autistic females camouflage their autistic traits, the greater psychological distress they experience. Hence, camouflaging behaviour directly relates to poorer mental health in autistic women¹⁷. Additionally, masking and the lack of awareness of female autistic traits amongst clinicians can result in misdiagnosis.^18, Therefore, autistic women are less likely to have the correct support or accommodations in place when compared to autistic men, further increasing anxiety and the autistic overwhelm experienced by these individuals. Dr Pang of the National Autistic Society states that autistic women are more likely to be diagnosed with anxiety and/or depression¹⁹. This is both due to autistic burnout resulting in misdiagnosis, but also because these mental health conditions are more common in women than men⁵. Ultimately this diagnostic overshadowing  prevents autistic women from obtaining the right support and causing worsening mental distress. This pattern is also seen in the autistic community, with a large pool of research suggesting that mental health outcomes for autistic women are poorer than those of autistic men, with a greater prevalence of anxiety, stress, and exhaustion  in autistic women due to higher levels of masking²⁰.

Camouflaging and masking in social situations are one significant cause for poorer mental health amongst autistic women. When interviewing autistic women, the National Autistic Society determined that women feel more pressure to act in a socially acceptable way in comparison to autistic men, suggesting that gender-related social stigmatisation as well as the difficulties with socialising faced by autistic individuals contribute to increased stress for autistic females and thus explain why autistic women experience worse mental health outcomes^{19, 20}. Therefore, a clinical understanding of the specific characteristics and social expectations of autistic females is crucial to ensure the medical field can provide the right accomodations and support for autistic women in order to improve mental health outcomes.

Psychometric Tools for ASD

Despite the differences between phenotypically male and female autism, screening tools often attempt to be universal across both sexes and cultures. This technique has been criticised for not recognising phenotypically female autistic traits and thus leading to a greater proportion of undiagnosed or misdiagnosed females compared to males. This lack of female trait recognition likely stems from the use of predominantly male samples when forming psychometric scales for autism²¹. Due to this, many of these scales can appear scientifically valid while failing to work effectively for the entire population for which they are intended. Currently, NICE encourages the use of Autism Spectrum Quotient 10 (AQ10)²². 

Considering the differences in presentation between the sexes, which are now recognised by the ICD-11, it is obvious that the AQ10 targets, deliberately or not, phenotypically male autistic traits. As we have seen, autistic females are often better at camouflaging their autistic traits and social difficulties, which would make questions 5, 6, 9 and 10²² more difficult to answer. The AQ10 fails to investigate any masking or imitation behaviour, which, although adopted more often by female autistics, is also seen in male autistic individuals. Furthermore, even when autistic females exhibit intense or ‘restricted’ interests, these are often related to people or animals rather than objects¹⁶, with only object-related interests being identified in diagnostic criterion³ (see question 8 of the AQ10)²². Moreover, autistic women may not experience the stereotypical socialisation difficulties as quoted in the diagnostic criterion, with their difficulties focused on friendship maintenance rather than initiation, which is not explored by the AQ10 at all^16,22. 

Not only do the flaws with this tool identify reasons behind underdiagnosis of female autism, they highlight a broader issue: while medicine - with its human-made guidelines, screening and diagnostic tools - is supposed to be bias-free, it is often marred by the same biases that exist in broader society.

Due to the growing awareness of differences in autistic traits between the sexes, and the underdiagnosis of autistic women, screening tools have been developed that specifically target more female-typical autistic traits, such as camouflaging. The Camouflaging Autistic Traits Questionnaire (CAT-Q)²³ and the Autistic Women's Experiences Questionnaire²⁴ were created with female autism specifically in mind, and have thus been evaluated to have greater recognition of female-typical autistic traits or female autistic life experiences. The CAT-Q is particularly important in our discussion of camouflaging and emulating peers as a method for overcoming social impairments. Hull et al. have demonstrated that females have differing camouflaging behaviours when compared to males. Using the CAT-Q, they identified that female autistics scored significantly higher than both neurotypical individuals and their male autistic counterparts on two out of three of the CAT-Q subscales²³. This further supports the critique of the AQ10 in that it does not adequately explore the camouflaging or masking techniques learned by autistic women especially when considering that autistic women tend to score higher on the CAT-Q in comparison to men and lower on the AQ10. It is due to this that autistic women often evade diagnosis much longer or require further questioannires and scales in order for their traits to be recognised which delays diagnosis, access to adequate support and prolongs distress. However, due to being newer, these more comprehensive screening tools require further psychometric evaluations and clinical validation before they can be readily used in research and medicine.

Looking to the future…

So, how can we improve autism screening and diagnostic tools to better fit both male and female autistic individuals? Once again, there is difficulty here, and by defining a ‘male-phenotype autism’ and a ‘female-phenotype autism’ we are falling at the same hurdle of putting people into boxes. In light of this, I will propose how we could move forward.

In the ICD-11, the diagnostic criteria for personality disorders was dramatically changed from the ICD-10. It moved from a categorical system which, similarly to the phenotype model of autism, placed individuals in categories based on whether set criteria was present or absent, to a dimensional model that recognises multiple domains. This overhaul was performed to better reflect clinical reality in which people very rarely fit into neat boxes. The dimensional model recognises that diagnoses are most beneficial to individuals when they encapsulate nuance and are flexible, enabling treatment and support to be more patient-centred²⁵. 

Based on the issues we have identified with the current criterion for autism with regard to female autistics, a dimensional model would be well-suited. As we have seen, the current method of screening and diagnosis is based on a heavily gender-biased categorical system in which individuals either meet a set of pre-specified criteria or don’t. A dimensional model better handles heterogeneity in presentations which ensures individuality within autism is accounted for. This type of model would be able to assess a larger number of traits, including, but not limited to, social motivations, camouflaging and masking, sensory sensitivities and cognition. Adopting a dimensional model not only reduces gender bias but could also improve clinical outcomes, ensuring that less stereotypically autistic individuals, in particular autistic girls, go undiagnosed whilst also preventing misdiagnosis and ensuring there is appropriate support put in place for these individuals. Medicine prides itself on being evidence-based, so it's time the evidence includes everyone. 

Read the rest of the edition here