An initial sample of 795 papers was found by the database searches and were screened. A total of 13 quantitative, six qualitative, and one mixed-methods papers were included in this review (see Fig. 1, which summarises the numbers and reasons for exclusion at each stage). A total of 23,760 participants were included in this review; this included a total of 3197 females with a diagnosis of ASD, 92 parents/family members of girls/young women diagnosed with ASD, zero teachers, and three professional providers.
For quantitative papers, two out of 15 papers were excluded due to poor quality. Of those included, five papers were scored ‘high’, six papers ‘moderate’, and two papers ‘low’ for quality. For the qualitative papers, three were scored ‘high’, two scored ‘moderate’, and one scored ‘low’ for quality. The one mixed-method paper included was scored ‘moderate’ (Table 1).
Gendered Symptoms as a Barrier to Diagnosis
In relation to ASD behaviours constituting barriers to ASD diagnosis in girls and young women, six themes were identified from the literature, i.e. behavioural problems, social and communication abilities, language, relationships, additional diagnoses/difficulties and restricted and repetitive behaviours and interests (RRBIs) (Fig. 2).
Four papers described comparisons between genders and presented conflicting findings (Duvekot et al. 2017; Dworzynski et al. 2012; Giarelli et al. 2010; Petrou et al. 2018). When comparing females and males with high ASD traits, it was found that males were more likely to have aggression and hyperactivity and/or attention deficits, and females were more likely to exhibit staring and seizure-like activities (Giarelli et al. 2010). Within genders, it was found that ASD girls were more likely to have behavioural difficulties (e.g. hyperactivity), compared to high-trait girls not meeting diagnostic criteria, while boys showed no such difference (Dworzynski et al. 2012). This was also true for low IQ, with girls meeting diagnostic criteria having lower intellectual abilities than those who did not meet the criteria.
Within ASD samples, no gender differences in age of diagnosis were found if toileting or temper problems were present (Petrou et al. 2018), i.e. both males and females being diagnosed earlier if these problems were apparent versus absent. However, boys, but not girls, who had both ASD and eating problems, were diagnosed earlier than those without eating problems. However, Duvekot et al. (2017) found that in a clinic-based sample, having additional behavioural problems increased the likelihood of obtaining an ASD diagnosis in females, whereas this effect was not present for males; this reflects the findings from a high-trait sample (Dworzynski et al. 2012). Arguably, this could be framed as females diagnosed with ASD needing additional behavioural problems to improve their chances of receiving a diagnosis.
Social Communication Abilities
Three studies described social communication abilities as a possible influencing factor of diagnosis of females with ASD (Mandy et al. 2018; Rynkiewicz et al. 2016; Tillmann et al. 2018).
Mandy et al. (2018) examined the developmental trajectories of autistic social traits (AST) for males and females included in the population-based Avon Longitudinal Study of Parents and Children (ALSPAC), assessed with the parent-report Social Communication Disorders Checklist. Significant differences emerged for the developmental course of AST between genders in this general population sample; at earlier ages (7 years), males showed higher levels of AST. Yet, by early to mid-adolescence (13, 16 years), the gender gap narrowed, irrespective of IQ. It would seem that social difficulties became more pronounced in girls over time, or that compensatory mechanisms such as camouflaging are less successful, against the higher social demands of teenagers.
Rynkiewicz et al.’s (2016) study included the use of the Faces Test to assess emotion recognition in a clinical sample of 5- to 10-year olds and found that girls with ASD made significantly more mistakes than boys, indicating poorer emotion recognition skills. It should be noted that the sample size for this study was small (n = 16 females), and all participants were of average or above average IQ and verbally fluent—factors that may limit the generalisability of this finding. The study also included the parent-completed Social Communication Questionnaire (SCQ), on which current communication skills were reported to be significantly better for ASD boys than girls. This again indicates greater problems for diagnosed girls and the possibility of girls needing more ASD traits than boys to gain an ASD diagnosis.
Tillmann et al. (2018) pooled datasets from across European clinical and research sites (n = 18 sites across nine countries), thereby providing one of the largest sample sizes to date, to assess the impact of sex and age in terms of scores on the Autism Diagnostic Observation Schedule (ADOS-G, Lord et al. 1989) and Autism Diagnostic Interview (ADI-R Rutter et al. 2003). This study found no significant sex differences relating to current social communication symptoms for either the ADOS or the ADI-R, indicating a similar phenotype in terms of social communication deficits for males and females meeting ASD diagnostic criteria, irrespective of age, in a clinical sample.
Three studies addressed the issue of language and the role that this might play in terms of diagnosis for males and females (Dworzynski et al. 2012; Petrou et al. 2018; Salomone et al. 2016). Dworzynski et al. (2012) found that in a population-based sample, girls, but not boys, meeting diagnostic criteria for ASD showed significantly lower levels of verbal cognitive ability than peers with similarly high CAST (Childhood Autism Spectrum Test) scores who did not meet criteria for ASD. Salomone et al.’s (2016) finding reflects that of Dworzynski et al.’s (2012) in a lower age group (< 7 years compared with 10–12 years old). Salomone et al. (2016) found that for verbally able individuals (i.e. those with phrase speech) in a population-based sample, girls had a significantly higher age of diagnosis than boys. This gender difference was not seen for non-verbal or minimally verbal children. These results indicate that verbally able girls may be waiting longer for a diagnosis, thereby supporting Dworzynski et al.’s (2012) suggestion that additional language (or other) difficulties are often necessary for girls to receive an ASD diagnosis in contrast to boys. However, in the final paper to examine language skills in relation to diagnosis, Petrou et al. (2018) found that in a clinic-based sample, boys, but not girls, with lower language levels were diagnosed earlier.
Only one study specifically addressed relationships in females compared to males with ASD under 21 years of age (Head et al. 2014). The Australian study with 25 male and 25 female adolescents with high-functioning ASD demonstrated that females with ASD showed similar social skills on a (self-report/parent-report) questionnaire of friendship and social function as typically developing males. The authors go on to identify that if diagnostic criteria in the social communication domain considered these striking gender differences, a larger number of females may go on to be diagnosed with ASD. This is an important finding, however, study limitations should not be ignored; the friendship questionnaire (FQ, Baron-Cohen and Wheelwright 2003) was modified from its original form for adults, and this had not been validated and had relatively low reliability. However, the study also reported that adolescents with ASD scored lower on the FQ compared to neurotypical adolescents, which is strongly supported by the literature, and study findings were consistent when using the FQ reported by adolescents compared to parent-report, providing further validation of this study’s findings.
RRBIs and Sensory Sensitivities
Three papers reported gender differences in RRBIs relevant to diagnostic barriers (Duvekot et al. 2017; Rynkiewicz et al. 2016; Tillmann et al. 2018). Generally, studies indicated that males show more RRBIs than females (Duvekot et al. 2017; Tillmann et al. 2018), and that RRBIs are more predictive of an ASD diagnosis in males compared to females.
Interestingly, Rynkiewicz et al.’s (2016) cross-sectional study found that the number of stereotyped behaviours in boys significantly decreased between five and 10 years of age, whereas it remained at a consistent level across these ages in girls with ASD. Tillmann et al. (2018) found that girls with an ASD diagnosis exhibited fewer RRBIs than boys, however non-verbal intellectual functioning accounted for and attenuated these differences.
Only one paper reported on sensory sensitivities between genders in relation to diagnosis (Duvekot et al. 2017). This study found that parent-reported sensory symptoms significantly predicted an ASD diagnosis irrespective of gender.
Four papers discussed the impact of additional (co-occurring) diagnoses on ASD diagnosis (Cridland et al. 2014; Giarelli et al. 2010; Petrou et al. 2018; Watson 2014). Watson (2014) found that all participants (n = 13 females with a clinical ASD diagnosis) reported having a co-occurring condition (e.g. ADHD), with 10 out of the 13 participants receiving their co-occurring diagnosis prior to ASD. Additionally, Cridland et al. (2014) showed parents reported misdiagnosis as a factor leading to delayed diagnosis in females. Petrou et al. (2018) found that boys who had an additional diagnosis were diagnosed significantly later than boys who did not. Although the model was not significant for girls (girls represented 18% of the sample), a trend was observed, whereby girls with an additional diagnosis were diagnosed later than girls who did not have an additional diagnosis.
It has also been suggested that cognitive impairment increases the likelihood of having a documented ASD diagnosis for boys, but not for girls (Giarelli et al. 2010). Girls with an IQ of 70 or less were less likely than boys with an IQ of 70 or less to have a documented ASD diagnosis. This may suggest that once a cognitive impairment had been identified in a female, it is less likely that an ASD assessment will take place.
Perceived Barriers to Diagnosis
Five themes were identified from the literature, i.e. compensatory behaviours, parental concerns, others’ perceptions, lack of information/resources and clinician bias (see Fig. 2).
Six papers discussed compensatory behaviours as barriers to diagnosis for females (Beteta 2008; Cook et al. 2018; Cridland et al. 2014; Dean et al. 2017; Rynkiewicz et al. 2016; Watson 2014). These papers discussed how females with ASD might go unnoticed due to their behaviours appearing similar to their neurotypical peers.
Dean et al. (2017) observed and compared the playground activity of male and female neurotypical and ASD children (n = 73; mean age, 7 years). They highlighted that when observing social interactions from a distance, girls with ASD behaved like neurotypical girls, i.e. spending a significant amount of time talking and weaving in and out of groups; yet, it was only upon closer inspection of the quality of interaction with peers that social challenges were perceived. By contrast, it was easier to identify social challenges in boys with ASD at a distance. The authors argued that using camouflaging techniques to mask social difficulties makes girls with ASD more vulnerable and less likely than boys to be identified within a school setting.
This is further supported by Rynkiewicz et al. (2016), in which gestures during ADOS-2 assessments were analysed in 10 girls and 16 boys with high-functioning ASD; girls with ASD tended to use gestures more ‘vividly’ (i.e. with more energy) than boys with ASD (a non-significant trend). The authors hypothesised that girls with ASD are effective at camouflaging such diagnostic features, and this may thereby increase the risk of under-diagnosis in autistic girls. While qualitative differences in gesture use by children with ASD are thought-provoking, they must be considered with caution due to small numbers and lack of a significant group effect.
A paper employing in-depth interviews with females with an Asperger’s diagnosis reported ‘social echolalia’, the act of mimicking socially skilled peers, as a factor that might contribute to the mis- and missed diagnosis of ASD females (Beteta 2008). Another paper using in-depth interviews, this time with parents, adolescent autistic daughters, and siblings, suggested that ASD females are less likely to be identified until the social demands they experience exceed their compensatory strategies (Watson 2014).
During interviews, parents also described their daughters masking their autistic behaviours, adjusting their behaviours to fit in with their peers (Cook et al. 2018). Parents suggested ASD girls were less likely to be identified as autistic by their teachers, due to masking of differences. One parent stated ‘At that stage she was masking and covering up quite well, and although we were aware that there was something not quite right, every time I raised with her teachers at the primary school, … they just said ‘she’s fine’ you know, ‘you’re expecting too much, she’s fine’ (p. 310). Mothers interviewed in one further paper, by Cridland et al. (2014), reported negative consequences from their daughters imitating social behaviours during assessment, and clinicians therefore being unable to identify their autistic behaviours.
Five papers mentioned parental concerns as barriers to diagnosis for females (Duvekot et al. 2017; Little et al. 2017; Navot et al. 2017; Ramsey et al. 2018; Watson 2014). Concerns related to five main themes, i.e. RRBIs, emotional/behavioural problems, social interaction, labelling ASD (i.e. parents considering the possibility of ASD as explaining their daughter’s symptoms), and additional diagnoses.
Ramsey et al. (2018) examined the issues flagged by parents of toddlers at higher likelihood of ASD, who were screened for ASD, and they reported two main effects for gender, in terms of RRBIs and labelling ASD. For RRBIs, parents of boys were found to express concern 1.74 times more than parents of girls, regardless of subsequent diagnosis of the child. The finding for labelling ASD did not reach significance, but this may have been due to small numbers, as it was expressed by the parents of four girls and 22 boys, highlighting that parents of boys were 2.43 times more likely than parents of girls to label ASD itself as a concern for their toddlers who screened positive on ASD questionnaires (again regardless of subsequent diagnosis). When looking at all concerns as a whole, the study found that overall parents expressed one or more concerns about ASD 1.46 times more often for boys than for girls.
Duvekot et al. (2017) also found evidence of differences in terms of parental concerns regarding RRBI type symptoms. Parental reports of RRBI symptoms were significantly less predictive of an ASD diagnosis for females than for males, indicating that even with high scores in this domain, there was less of a link to diagnosis for females. The study also found that one area of parental concern that seemed important for females was emotional and behavioural problems, with females being more likely to receive a diagnosis if levels of these two problem areas were high. Little et al.’s (2017) results indicated that parents of boys with ASD were more likely to express concerns about social interaction prior to diagnosis than parents of girls with ASD.
Two studies dealt exclusively with the aspect of labelling ASD itself for females and recognising symptoms (Navot et al. 2017; Watson 2014). Using semi-structured interviews, Watson (2014) found that parents perceived ASD to be a ‘boy’s disorder’. This perception led to delays in picking up on ASD symptoms. Similarly, Navot et al. (2017) in their study of 11 mother-daughter dyads found that parents expressed concern that they failed to recognise symptoms of ASD in their daughters, for example; ‘I didn’t listen to her. She would just say how she hated school and how she hated visiting my parents and how loud everything is .... I forced her to do all those things. I arranged play dates for her and forced her to go and of course it just made things worse. It was a total failure. And then I got so angry. I was actually furious with her for years .... Although things are very different today, I can’t just push the delete button and pretend that it didn’t happen. We both know I didn’t listen’ (p. 16).
Three qualitative studies identified the perceptions of others to be a barrier to female ASD diagnosis, based on interviews conducted with adolescent girls with ASD and their mothers (Cook et al. 2018; Cridland et al. 2014; Navot et al. 2017). The difficulty of having a daughter with a disorder primarily associated with boys was described, for example ‘Most people, once I said, she’s got Asperger’s, would look at me like they didn’t believe me’; ‘We live in a small town. Most people don’t know anyone that has autism, and if they do it’s probably going to be a boy, because autism is sort of a boy kind of issue. And my daughter was so sweet, and people would just say, ‘What’s your problem? Why are you labelling her?’ (Navot et al. 2017, pg. 14). Parents found their concerns were often met with scepticism, however, some adolescent girls with ASD also highlighted positive aspects of ASD being considered a male disorder, for example ‘being surrounded by boys’, because adolescent boys were easier to get along with than adolescent girls (Cridland et al. 2014).
Lack of Information/Resources
One study indicated that a lack of information was a barrier for the diagnosis of females (Navot et al. 2017). In a qualitative study examining 11 mother-daughter dyads, Navot et al. (2017) found that parents expressed concerns over the lack of information that existed in terms of female ASD, for example ‘I had a hard time finding information that would help me because she was a girl. Everything I read was so much about boys. It was so frustrating and irrelevant. There was just nothing there that could help me figure her out’ (p. 14). Lack of information was seen as detrimental to the diagnostic process, as without relevant and tailored information, parents may not be so inclined to seek diagnosis for their daughters.
Finally, one paper described financial resources as a barrier to ASD diagnosis in females; in focus groups, parents expressed concerns about being able to cover services for their adolescent daughters with ASD (Mademtzi et al. 2018).
Clinician bias was identified by six studies as a barrier to females obtaining an ASD diagnosis (Beteta 2008; Cridland et al. 2014; Giarelli et al. 2010; Mademtzi et al. 2018; Navot et al. 2017; Watson 2014). A large, population-based study, which used case definition to examine the prevalence of ASD (Giarelli et al. 2010), found that boys were more likely than girls to have a diagnosis of ASD even when both sexes had documented ASD symptoms in educational and clinical records. They concluded that this may result from an ‘interpreting bias’, where the observed experiences differ from the expected behaviours dependent on sex bias. They highlighted that clinicians evaluating girls, compared to boys, with a complex developmental profile may be more likely to exclude a classification of ASD if other conditions are present, due to this bias.
Four further papers, using interviews with adolescent girls with ASD and their parents, explored this same theme (Beteta 2008; Cridland et al. 2014; Navot et al. 2017; Watson 2014). Watson (2014), for example, found that parents perceived medical professionals to be hesitant in giving girls a full ASD diagnosis, and instead, they seemed to opt for other diagnoses. Parents also expressed frustration, knowing their daughter was developing atypically, yet not receiving validation from medical professionals about their concerns: ‘We started bringing concerns with our paediatrician really early. A lot of times they were like, oh, she is a late bloomer, but I still felt like there was more going on’; ‘I kept asking to have her evaluated, but with her being a girl, it was even less likely that the paediatrician would refer us. I remember her saying that this is usually a boys’ thing and she is only a little different’ (Navot et al. 2017, pg. 14). Furthermore, findings from Cridland et al. (2014) included reports of healthcare professionals being reluctant to diagnose a female as autistic and a lack of awareness of ASD in females due to a perceived higher incidence of ASD in males.
An additional qualitative paper, this time using focus groups with families, explored the impact of clinician bias. Mademtzi et al. (2018) reported a parent feeling the need to exaggerate their daughter’s impairments to gain a diagnosis; ‘I felt that I needed to make my daughter look more impaired than she actually was, in order to get diagnosis and needed services’. Strict diagnostic criteria also led to delayed diagnosis for females, with one parent saying their daughter was declined diagnosis because ‘she’s two points above the cut-off score’.