Assistant Professor Clare Harrop My name is  Dr Clare Harrop, and I’m an Assistant Professor   at the University of North Carolina at Chapel  Hill. My programme of research for many years   has focused on autistic females, from early in  development, through to adulthood. And that’s   what I’m going to talk to you about today,  autism in females, what we know, and where   do we go? This talk will cover sex differences  in the diagnosis of autism in females, factors   that influence delayed or misdiagnosis, and  phenotypic differences between males and females. A brief note on terminology. I typically  try and use identity-first language. That is   autistic males, autistic females, based on recent  commentaries in the field. However, I recognise   that there is diversity in terminology preference.  I also use World Health definition – World Health   Organization definitions of sex and gender.  That is assigned sex at birth and gender. Most   of what I will talk about today will refer to  assigned sex at birth. That is the sex you are   given when you are born. But I will, also, touch  on gender, which is really important in autism. So, the central problem of my work, and  many others, is that autistic females are   underdiagnosed and underserved. So, first,  there are sex difference in the prevalence   and diagnosis of autism. Autism has historically  been a male dominant diagnosis. This is arguably   one of the most replicated findings in  autism research. The first descriptions   of autism described it as “a variant of  male intelligence.” Original case reports   of autism were all from males, and as such,  females were rarely diagnosed with autism. For many years, the sex ratio in autism sat around  four-to-one, however, recent estimates have slowly   begun to reduce. The recent Center for Disease  Control estimate in the US currently sits at 3.6   males to one female receiving a diagnosis. This  has gone down steadily, from 4.5 males in 2012,   four males in 2016, and now, more recently, 3.6. Depending on the methods used however,   the ratio can range from three-to-one to even  one-to-one. There was a recent meta-analysis   that found that when aggregated together the sex  ratio was closer to three males to one female.   And in a study of infants at higher likelihood  for an autism diagnosis, by virtue of having an   older sibling already with a diagnosis, the ratio  for children in the severe group was one-to-one. There is also variation depending on other  factors. We know that individuals with delays   in language, or co-occurring intellectual  disabilities, are more likely to be diagnosed   earlier with autism, and the sex ratio in  this group is closer to two-to-one. However,   when we think about individuals who  have language in a normal range,   we see a sex ratio of anywhere between  eight males to 16 males to one female. We also know that females are more likely to  be diagnosed later than males. This is another   really consistent finding, and studies from the  US and Europe report females being diagnosed 18   to 24 months later. This is even later for  females when their language falls in what we   think of as the normative range. Interestingly,  these delays in diagnosis persist, despite very   few different – differences reported in early  milestones between males and females. That is,   parents, when asked to report, reflect  back on their child’s development,   there is very few differences, despite  this difference in the age of diagnosis. Females are alm – also more likely to  be classed as “late diagnosed,” this is,   receiving their diagnosis in adolescence  or adulthood. We also know that females are   more likely to be misdiagnosed. This male  view of autism has certainly continued to   cloud Teacher and Clinician judgement.  Females are more likely to experience   diagnostic overshadowing than males. This is when  a previous diagnosis, or a co-occurring diagnosis,   overshadows a Clinician’s judgement, and they may  be less open to considering an autism diagnosis,   explaining challenges and differences through  those existing or co-occurring diagnoses. Another interesting fact is that females are  more likely to drop their prior diagnosis   than males. That is, say if they come in with  an ADHD diagnosis, or an anxiety diagnosis,   but go on to receive an autism diagnosis, they  will shed their prior diagnosis more likely   than males. Males are more likely to retain  all of their diagnosis and have co-occurring   diagnosis. Males are up to ten times more  likely to be referred for an autism diagnosis,   and Clinicians report feeling less confident  diagnosing autism in females. Parents have   often discussed that their children need  to look “more impaired,” and there is a   general sense of females needing to tick  more boxes to receive an autism diagnosis. It’s important to note that with delayed or  misdiagnosis comes later support for females,   or supports that may not be appropriate.  We currently know very, very little about   autis – about intervention and  supports for autistic females,   but this is an area of great importance. A number  of factors may contribute to delayed, missed,   or misdiagnosis in females. A recent  colleague – by Estrin and colleagues,   identified factors that are potential barriers  to diagnosis in females. These included the   behaviours that females may show, parental  concerns, and the perception of the others. We know that Clinician and Teacher  knowledge, or confidence surrounding   identifying autism in females is lower  than for males. A study by Whitlock and   colleagues reported that Teachers were less  likely to assign an autism label to vignettes   describing males versus females, despite  the same behaviours described in these. The historic view of autism may continue to cloud  both Clinician and parent judgement. There are   generally just fewer resources available about  autism in females, and these do not come from   the extensive years of peer reviewed journals that  we’ve had for males. So, we are beginning to build   our literature base, but, also, there is a lot  of misinformation out there regarding in autism   in females. And the differences that we see  in the phenotypic presentation of autism,   between males and females, may contribute  to this missed, delayed or misdiagnosis. That leads me to section two is there is a  female autism phenotype? It’s really important   to acknowledge that as research has started  to include more females, key differences have   emerged between males and females. For many,  many years females were excluded from research,   just one of the early eye tracking studies  of autism have just males in them. And,   also, we, for many years, have nu – very small  numbers of autistic females in our samples. So,   this meant that women or girls were grouped with  males, and we did not study sex differences. But there is increasing evidence that autism  may present differently in females, and, also,   at different stages across the lifespan.  It is possible that often subtle and   nuanced differences contribute to delayed or  misdiagnosis. It is also important to consider   these differences across the lifespan, because  women are more likely to be diagnosed later,   particularly in adolescence and adulthood.  And there are key differences in normative   development between males and females, for  example, women go through different pubertal   stages, they go through different stages  of life, such as pregnancy and menopause,   that have not been studied in autism, due to  the male view that we have previously adopted. So, starting in infancy, this is prior to when  children receive a diagnosis, so thinking from   birth through to around two years, many  children are not diagnosed ‘til they’re   four or above. But there is some evidence that  girls that go on to receive an autism diagnosis   may meet milestones earlier. Now, again, most of  this is based on retrospective parent reports.   This isn’t – a lot of these studies are not  prospective, following children up over time. The parents of females later diagnosed  with autism report that their fe – their   daughters have earlier first words and  phrases. There is potential differences   in the type of initial first concern  expressed by parents. This is more   likely to be language delay for males,  or more likely motor delay for females,   which is not a core autism diagnostic  feature, though does co-occur at a high rate. In sibling cohorts, where individuals are  followed up prospectively over time because   they have an older sibling with autism, there  are some differences in male and female infants,   though, they really vary. Some  studies show on differences,   others have noted more joint attention,  attention to faces, in females. When we’re shifting now to differences  in toddlerhood, the data is very,   very mixed. It is also important to note that  children studied during the toddlerhood and   preschool period received their diagnosis  early, between the ages of two to five. So,   the girls that we’re studying at this age  range may be more aligned with more what we   consider the autism phenotype in males, as  they are getting diagnosed very young. So,   again, parent report suggests that girls  may neach – reach language milestones   earlier. This varies though. It does not  differ on clinical measures of language,   and does seem to be based on parent report. This  is girls saying phrases earlier and words earlier. Girls may have more complex play behaviour.  They may play at a more advanced level,   and with more gender typical toys, such as dolls  or toy – or tea sets. They have more gestures,   and they use these more vividly. This  is things such as pointing, shrugging,   giving. The strongest differences emerging  at this age range seem to be for restrictive   and repetitive behaviours. This is restricted  patterns and behaviours that we see in autism,   such as interest, motor mannerisms, language,  repetitive object use, they are a core diagnostic   feature of autism, and young females may show  fewer of these, or different types of behaviours. It’s again important to note that the detection  of these differences does not come out from global   clinical report measures. It appears to hinge on  more nuanced and fine grained analysis. My work,   for example, has coded videotapes, looking at  minute behaviours as they occur, rather than a   rating scale. Again, everything highlighted here  in bold is a core diagnostic feature of autism.   These are behaviours that we are looking for  with DSM-5 and ICD-10 diagnosis of autism. So,   if girls are showing subtle differences, this may  contribute to their delayed or missed diagnosis. When we shift to middle childhood, this  is what I think about for primary school   or elementary school, girls and boys going  into formal schooling. And this is often the   first stage when girls are diagnosed, if they are  diagnosed on average 18 to 24 months later, they   are often getting diagnosed as they are entering  formal schooling. What we see here is that girls   seem to have greater social motivation, across  different modalities. That is, we may see it in   how they play with their peers on the playground,  how their parents report, they’re more likely   to be reported as a friend, they also attend  to faces more in studies using eye tracking. They are reported to have more typical friendships  and friendship behaviours. They weave in and out   of playground situations more. They are  more likely to be named as a best friend,   and are more likely to be able to list  friends and what friendship entails. Again,   we continue to see these differences in  restrictive and repetitive behaviours,   that girls may show fewer or they are different. As we shift to adolescence and adulthood,  females face unique challenges, and, again,   I mentioned that we may not understand the  complexities of these, as we have typically   just studied males previously. So some of  these may be unique to females going through   these developmental stages. We know that there  is an increased likelihood of late diagnosis,   getting diagnosed in adulthood or adolescence.  Autistic characteristics, in general, increase   a lot across the lifespan for females. This  has been found in large longitudinal studies   of individuals in the general population,  so we see these traits increase over time. Change in nature of friendships and  social participation may lead to more   social exclusion and isolation for autistic  females. As friendships change, in females   they change more rapidly. They are more linked  to trust and sharing, whereas adolescent males   may continue to play sports on the playground, or  play videogames. They’re more the smaller groups,   and they’re more personal for females, and  this may be a challenge for autistic females. In autism overall in adolescence and adulthood,  we see higher rates of depression, anxiety,   suicidal ideation, and eating disorders, but we  see these at a higher rate for autistic females,   mirroring what we see in the general population.  We also see poorer outcomes in adulthood. A really   interesting study found that women more  – were more likely to receive job offers,   but they were less likely to retain their  jobs. So, they were having trouble within   the workplace that we were not seeing  to the same extent in autistic males. Shifting gear slightly to talk about  camouflaging. Some of the things that   I have talked about seem to almost be  a protective fect – effect of females   using behaviours in a way to mitigate some of  their challenges. I really like this image,   as it conveys really nicely what autistic  camouflage is. So, “Don’t forget your mask   and the other mask. Hello, nice to see you,  I love to make small talk and eye contact.” So, we know that autistic females are more  likely to employ compensatory behaviours,   such as eye contact, small talk, gestures, to  mitigate some of their social challenges. It’s   really important to note that camouflaging is  not unique for autism, and it’s not unique to   autistic females, but we see it at a higher rate  in this group. And these behaviours have been   found to include gestures, linguistic markers,  such as “um,” and “ah,” pragmatic language,   and friendship behaviours. And, again, these  are all core diagnostic markers of autism. I’m going to touch slightly on beh – on studies  that have looked beyond behaviour. Some of my work   has done this, and there’s other people in the  field that have used really great neuroscience   methods to look at differences between males and  females beyond just behavioural assessment. So,   differences between males and females appear  beyond what we see on the surface. In my own work,   and those of others, we find that autistic  females attend more to social stimuli when   using eye tracking. This is where we follow  individuals’ eyes when they’re looking at   images. This has been found for pictures of  faces and, also, videos that are more complex. We find that autistic females attend less to  images that have been reported previously to   be very attention grabbing for males. These  are images such as computers, trains, Lego,   things that we associate with the male phenotype,  and particularly the male autism phenotype,   these do not attract the attention of  females to the same degree. Interestingly,   studies using EEG have found different results.  So while my work, and that of others, has found   that females attend more to faces, work of others  have found that females show attenuated responses   to faces. This has been indexed by the N170,  which is a well validated face ERP component. So, while individuals may look at faces,  their processing may be different. So,   autistic females may look at faces, but they  are not responding in the brain in the same   way. Other studies have found that the brain  at rest, that is resting state EEG activity,   varies by sex, specifically the association  between resting state activity and behaviour. So,   power across various frequency correlated  with social skills, non-verbal IQ,   and repetitive behaviours for autistic males, but  these associations were not found for females. And a recent systemic review of over 50  imaging studies reported potential sex   by age effect. What they found in this review,  and what we have found in eye tracking studies,   is that autistic females seem to  fall somewhere between autistic   males and non-autistic males. They did  not differ from these neurotypical males. And this is really important,  because we need to make sure that   we’re including both autistic males and au –  non-autistic males and females in our study,   so that we could combine these studies  to look at the importance of both sex and   diagnosis. This systemic review also found  that the large age ranges in our study may   mask some of the differences that we’re  finding. So, really bringing home the   importance of having large samples that  have a concentrated developmental period. So, I’m going to touch very briefly on gender  in autism. This study here is from the NIH,   and it shows the differences between  sex and gender. Sex is considered a   biological construct. It is what we think of  as what we are assigned at birth. Whereas,   gender is only relevant for humans,  not for other animals, and it’s broadly   defined as “a multidimensional construct that  encompasses gender identity and expression,   as well as social and cultural expectations  about status, characteristics and behaviour.”   These are often highly associated with  specific biological sex, so male or female. In autism, and in general research, sex and gender  are used interchangeably. I am guilty of this,   you can look at some of my own research,  where I talk about gender differences,   rather than sex differences, but sex and gender  are really important in autism, and both should   be considered. Up to 15% of autistic individuals  do not affiliate with their assigned sex at birth,   and this number may be higher for autistic  females. We know that in general development,   and in autism, gender incongruence is associated  with poorer mental health and suicidality. And in autism, gender is important for a  number of reasons. Our diagnostic tools   are often play-based, so are they missing girls  based on the assumptions of how we’re building   our tools? We have different parental,  Clinician and Teacher expectations for   males and females. We have a number of play-based  interventions in early development that have been   largely developed on male samples. And it’s  really important that in early development,   and in middle childhood, a lot of the  differences that we have seen between   males and females fall along traditional gender  lines, so we may also have to consider gender. But where do we go from here? I’ve given you  a whistlestop tour of the state of the field,   and there are still – there is still so much  to do with regard to autistic females. So,   research into autistic females have  exploded in the past ten years,   but there is still so much to learn. Research  must consider both sex and gender. This does   not relate just to the participants,  but, also, the design of our studies,   the toys we’re picking, the images we’re  picking, these have all been shown to be   influenced by sex and gender, and we should  consider them in our design of our studies. We really need robust, longitudinal studies  that study children over time and through   key developmental periods, which we  have not done with females, to date.   Researchers should endeavour to recruit more  females, whether this is a three-to-one ratio,   or a one-to-one ratio, if you are specifically  studying things related to sex and/or gender. And we need to learn about the  support autistic females want or   need across the lifespan. These supports may vary,   based on their unique experience, such  as puberty, motherhood, menopause,   which we have not studied in significant detail  due to the male bias in prior research. So,   thank you for your time today, and I hope  you’ve enjoyed learning about autistic females.