ArticlePDF Available

Elevated rates of autism, other neurodevelopmental and psychiatric diagnoses, and autistic traits in transgender and gender-diverse individuals

Authors:

Abstract and Figures

It is unclear whether transgender and gender-diverse individuals have elevated rates of autism diagnosis or traits related to autism compared to cisgender individuals in large non-clinic-based cohorts. To investigate this, we use five independently recruited cross-sectional datasets consisting of 641,860 individuals who completed information on gender, neurodevelopmental and psychiatric diagnoses including autism, and measures of traits related to autism (self-report measures of autistic traits, empathy, systemizing, and sensory sensitivity). Compared to cisgender individuals, transgender and gender-diverse individuals have, on average, higher rates of autism, other neurodevelopmental and psychiatric diagnoses. For both autistic and non-autistic individuals, transgender and gender-diverse individuals score, on average, higher on self-report measures of autistic traits, systemizing, and sensory sensitivity, and, on average, lower on self-report measures of empathy. The results may have clinical implications for improving access to mental health care and tailoring adequate support for transgender and gender-diverse individuals. It is unclear if rates of autism and other neurodevelopmental and psychiatric diagnoses are elevated in transgender and gender-diverse individuals compared to cisgender individuals. Here, the authors use data from five different large-scale datasets to identify elevated rates of autism diagnoses, diagnoses of other neurodevelopmental and psychiatric conditions, and elevated traits related to autism in transgender and gender-diverse individuals, compared to cisgender individuals.
ORs and 95% CIs for autism in transgender and gender-diverse individuals compared to cisgender males, cisgender females, and cisgender individuals altogether a This figure provides the unadjusted Odds Ratios (ORs, point) and 95% CIs for autism in transgender and gender-diverse individuals compared to either cisgender males, cisgender females, or cisgender (cisgender males and cisgender females) individuals in five datasets (C4: N = 514,100; MU: N = 85,670; APHS: N = 2312; IMAGE: N = 1803; and LifeLines: N = 37,975). b This figure provides adjusted ORs (point) and 95% CIs for autism in transgender and gender-diverse individuals compared to cisgender males, cisgender females, or all cisgender individuals in five datasets (C4: N = 514,100; MU: N = 85,670; APHS: N = 2312; IMAGE: N = 1803; and LifeLines: N = 37,975). ORs have been adjusted for age, educational attainment, and in the case of IMAGE dataset, an additional dummy variable for study (see “Supplementary Methods”). The y-axis is on the same scale for both the panels. The differences in ORs for the IMAGE dataset between Models 1 and 2 is primarily due to the inclusion of “study” group as a covariate. Specifically, the IMAGE dataset consists of individuals recruited into a study of mathematics and autism (“Methods”). Whilst the mathematics group is predominantly male and have higher educational attainment (all have at least an undergraduate degree), the case–control group had a more balanced ratio and a wider range of educational attainment. Covarying for the study the participants have been recruited into (mathematics or autism case–control) changes the ORs.
… 
Content may be subject to copyright.
ARTICLE
Elevated rates of autism, other neurodevelopmental
and psychiatric diagnoses, and autistic traits in
transgender and gender-diverse individuals
Varun Warrier 1, David M. Greenberg1,2, Elizabeth Weir 1, Clara Buckingham1, Paula Smith1,
Meng-Chuan Lai 1,3,4, Carrie Allison1& Simon Baron-Cohen1
It is unclear whether transgender and gender-diverse individuals have elevated rates of
autism diagnosis or traits related to autism compared to cisgender individuals in large non-
clinic-based cohorts. To investigate this, we use ve independently recruited cross-sectional
datasets consisting of 641,860 individuals who completed information on gender, neurode-
velopmental and psychiatric diagnoses including autism, and measures of traits related to
autism (self-report measures of autistic traits, empathy, systemizing, and sensory sensitivity).
Compared to cisgender individuals, transgender and gender-diverse individuals have, on
average, higher rates of autism, other neurodevelopmental and psychiatric diagnoses. For
both autistic and non-autistic individuals, transgender and gender-diverse individuals score,
on average, higher on self-report measures of autistic traits, systemizing, and sensory sen-
sitivity, and, on average, lower on self-report measures of empathy. The results may have
clinical implications for improving access to mental health care and tailoring adequate sup-
port for transgender and gender-diverse individuals.
https://doi.org/10.1038/s41467-020-17794-1 OPEN
1Autism Research Centre, Department of Psychiatry, University of Cambridge, Douglas House, 18B Trumpington Road, Cambridge CB2 8AH, UK.
2Interdisciplinary Department of Social Sciences and Department of Music, Bar-Ilan University, Ramat Gan 5290002, Israel. 3Child and Youth Mental Health
Collaborative, Centre for Addiction and Mental Health and The Hospital for Sick Children, Department of Psychiatry, University of Toronto, 80 Workman
Way, Toronto, ON M6J 1H4, Canada. 4Department of Psychiatry, National Taiwan University Hospital and College of Medicine, No. 7, Zhongshan South Rd.,
Taipei 10002, Taiwan. email: vw260@medschl.cam.ac.uk;sb205@cam.ac.uk
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 1
1234567890():,;
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Autism is a group of neurodevelopmental conditions
characterized by early-emerging difculties in social-
communication, unusually repetitive behavior and narrow
interests, and atypical sensory sensitivity1. Approximately 12%
of the general population is estimated to be autistic based on
large-scale prevalence and surveillance studies, although these
numbers vary between countries, age at the time of assessment
and other criteria28. Whilst several studies have investigated
rates of autism in individuals who are birth-assigned as males and
females, there still is limited information on rates of autism in
transgender and gender-diverse individuals in the general popu-
lation. Gender identity is a different construct from sex assigned
at birth, which is typically classied as male or female primarily
based on external genitalia. Some individuals are born with
chromosomal, genital, or hormonal sex-characteristics which vary
from the malefemale binary (intersex individuals) and who may
be assigned as or raised as males or females. Gender identity is a
persons sense of their own gender, which may or may not
coincide with sex assigned at birth. Following current recom-
mended practice, we use the term cisgenderto refer to indivi-
duals whose gender corresponds to their sex assigned at birth.
However, there is a diversity of gender identities including
transgender, non-binary, genderuid, agender, genderqueer, two-
spirit, bigender or others. Again, based on current recommended
practice, we collectively refer to these and other diverse gender
identities as transgender and gender-diverse(i.e., individuals
whose gender does not always correspond to the sex they were
assigned at birth). Currently, 0.41.3%911 of the general popu-
lation is estimated to be transgender and gender-diverse, although
the numbers vary considerably based on how the terms are
dened11.
A few studies, mostly clinic-based, typically with small sample
sizes, and in individuals with gender dysphoria (GD, dened as
persistent distress arising from a mismatch between sex assigned
at birth and gender identity), have investigated the link between
autism/traits related to autism and gender diversity12,13. These
studies have identied increased rates of gender diversity in
autistic children and adolescents1418, and adults19,20, compared
to the general population. Most of these studies in children and
adolescents have used a single item on the Child Behavior
Checklist (CBCL), a caregiver-report measure for behavioral
problems, to quantify gender variance, and these have identied
that between 4% and 5.4% of autistic children may potentially be
transgender or gender-diverse, compared to 0.7% of non-autistic
children1416. The largest of these, conducted in nearly 300,000
children, identied a fourfold likelihood of GD clinical diagnoses
in autistic compared to non-autistic children (i.e., 0.07% of
autistic children and 0.01% of non-autistic children)17. Despite
the differences in percentages of transgender and gender-diverse
identities in the studies using CBCL and clinical GD information,
the relative rates are largely similar (between 5.7 and 7.7). A
second set of studies has investigated rates of autism in both
children and adolescents2123 and adults24,25 with GD. These
studies have identied that between 4.8% and 26% of individuals
who present at GD clinics have an autism diagnosis based on
several different criteria. The largest of these studies (N=53224,
and N=54025) identied that 6.0% and 4.8% respectively of these
individuals are autistic, based on review of clinical and medical
records. Although none of these studies have used a matched
control sample to investigate the relative rates of autism diag-
noses, using a baseline population estimate of 12%28suggests
that autism diagnoses are signicantly elevated in individuals
presenting at GD clinics. A third group of studies have identied
elevated traits related to autism in individuals with gender
diversity24,2634 compared to cisgender individuals. These studies
have not investigated whether atypical sensory sensitivity (now
dened as a core feature of autism1) is elevated in transgender
and gender-diverse individuals.
The existing literature is heterogeneous, conducted using dif-
ferent methods, across age ranges and nationalities. These studies
demonstrate an increased occurrence of autism in gender-diverse
individuals or individuals from GD clinics. However, almost all
studies were conducted using modest sample sizes (a typical
sample size is in a few hundreds). Whilst these have the advan-
tage of carefully characterizing gender identity, they may not
correctly estimate the effect sizes as the Odds Ratios (ORs) may
be biased away from zero35,36. Larger samples would minimize
the bias, but a bias will likely exist in most samples. Additionally,
most studies have focused on individuals from GD clinics.
However, not all transgender and gender-diverse individuals have
GD, and the rates of autism in GD individuals may be different
from rates of autism in transgender and gender-diverse indivi-
duals. It is also likely that young people attending GD clinics
represent young people with the most intense gender dysphoria,
such that it warrants a referral for clinical care, and/or those
young people who can access this care (e.g., with parents who are
more tolerant of difference, or who have greater resources, etc.).
Therefore, it is important to understand what the odds are of
being diagnosed as autistic in transgender and gender-diverse
individuals at large, not solely in those recruited through GD
clinics.
In parallel, studies have also investigated the rates of mental
health conditions and mental distress in transgender and gender-
diverse individuals, including individuals with GD (e.g.,
references3744). The literature is heterogeneous with varying
research methodologies and sample sizes45. Two recent reviews
identify higher rates of mental health conditions and mental
distress (notably depression, anxiety, and substance use disorders)
in transgender and gender-diverse individuals compared to cis-
gender individuals40,45. Most of this research has focused on
depression, substance misuse, and anxiety, with limited research
on neurodevelopmental and other psychiatric conditions. It is
unclear how the elevated rates of autism diagnosis in transgender
and gender-diverse individuals compare to other neurodevelop-
mental and psychiatric conditions. To our knowledge, barring
one study16, none of the existing studies of autism and gender
identity have compared the rates of other related neurodevelop-
mental and psychiatric conditions in transgender and gender-
diverse individuals versus cisgender individuals, making it dif-
cult to estimate if the observed effects are specic to autism.
The availability of large datasets to investigate the link between
autism and gender identity is currently limited to internet-based
surveys. As far as we are aware, there is no large-scale national or
regional registry with information available on both gender
identity40 (not limited to individuals with gender dysphoria) and
autism diagnosis. We address these issues using four large-scale
cross-sectional, internet-based datasets, and one longitudinal
dataset, all sampled using a convenience framework. Using these
ve datasets, we investigate if transgender and gender-diverse
individuals, compared to cisgender individuals, have: (1) elevated
rates of autism diagnosis; (2) elevated autistic traits, systemizing
traits, sensory hypersensitivity traits, and reduced empathy traits,
all related to autism; and (3) elevated rates of any of six neuro-
developmental and psychiatric conditions that commonly co-
occur with autism (attention-decit/hyperactivity disorder
(ADHD), major depressive disorder (depression), bipolar dis-
order, obsessive-compulsive disorder (OCD), learning disorder
(also known as specic learning disorder), and schizophrenia)46,47
(Fig. 1). Finally, whilst the previous literature has provided com-
pelling evidence that autism is under-diagnosed (or mis-diagnosed
as other conditions) in cisgender females, it is unclear if this is true
of transgender and gender-diverse individuals4850. So, as an
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
2NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
exploratory analysis, we investigate whether transgender and
gender-diverse individuals are more likely to suspect that they
have undiagnosed autism compared to cisgender individuals.
Results
Rates of autism diagnosis.Werst investigated whether rates of
autism diagnosis differed by gender in the C4 dataset. A χ2test
identied a signicant difference in autism diagnosis based on
gender (χ2(2) =3316, φ=0.08, pvalue < 2 × 1016). Transgen-
der and gender-diverse individuals had higher rates of autism
diagnosis compared to cisgender males (OR =4.21, 95%
CI =3.854.60, pvalue < 2 × 1016), cisgender females
(OR =6.80, 95%CI =6.227.42, pvalue < 2 × 1016), and cis-
gender individuals altogether (i.e., cisgender males and cisgender
females combined) (OR =5.53, 95%CI =5.066.04, pvalue < 2 ×
1016) (Fig. 2). After accounting for age and educational attain-
ment, transgender and gender-diverse individuals had higher
rates of autism diagnosis compared to cisgender males
(OR =3.88, 95%CI =3.544.25, pvalue < 2 × 1016), cisgender
females (OR =5.31, 95%CI =4.855.82, pvalue < 2 × 1016), and
cisgender individuals altogether (OR =4.59, 95%CI =4.205.03,
pvalue < 2 × 1016) (Fig. 2).
Given the limitations of the C4 dataset, we investigated this
hypothesis in four other independently recruited datasets: MU,
IMAGE, APHS, and LifeLines (Methods). χ2tests identied
signicant gender-based differences in autism diagnosis rates
(pvalue < 1 × 105in all datasets). Transgender and gender-diverse
individuals had higher rates of autism diagnosis compared to
cisgender males (MU: OR =5.5, 95%CI =4.107.28, pvalue < 2 ×
1016;IMAGE:OR=6.36, 95%CI =3.7510.93, pvalue =6.32 ×
1014;APHS:OR=4.46, 95%CI =2.956.96, pvalue =3.6 × 1013;
LifeLines: OR =3.63, 95%CI =1.1211.73, pvalue =0.02), cisgender
females (MU: OR =9.92, 95%CI =7.3213.20, pvalue < 2 × 1016;
IMAGE: OR =5.35, 95%CI =3.149.24, pvalue =5.23 × 1011;
APHS: OR =6.66, 95%CI =4.4510.29, pvalue < 2 × 1016;Life-
Lines: OR =6.88, 95%CI =2.2720.85, pvalue =1×10
4),
and cisgender individuals altogether (MU: OR =7.08,
95%CI =5.289.30, pvalue < 2 × 1016;IMAGE:OR=5.90,
95%CI =3.5210.02, pvalue =1.80 × 1013;APHS:OR=5.77,
95%CI =3.888.86, pvalue < 2 × 1016; LifeLines: OR =5.50, 95%
CI =1.6016.60, pvalue =0.002). These results were statistically
signicant after accounting for age and educational attainment in
three of the four cohorts (transgender and gender-diverse vs.
cisgender: MU: OR =6.07, 95%CI =4.568.08, pvalue < 2 × 1016;
IMAGE: OR =6.36, 95% CI =3.3412.13, pvalue =1.08 × 109;
APHS: OR =6.28, 95%CI =4.139.53, pvalue < 2 × 1016). In
addition, we identied concordant effect direction in the LifeLines
cohort (LifeLines: OR =3.03, 95% CI =0.7212.76, pvalue =0.13),
though this was not statistically signicant due to the low statistical
power (Supplementary Note). Supplementary Table S3 provides the
results for all three genders.
Additional sensitivity analysis in the MU dataset conducted by
separating the cisgender group into cisgender males and
cisgender females and the transgender and gender-diverse group
into transgenderand otherindicated that both the non-
cisgender groups had higher rates of autism diagnosis compared
to both cisgender males and cisgender females (Supplementary
Table S4).
Given that we did not collect information on sex and gender
separately in the MU and the C4 datasets, we further investigated
if the adjusted ORs (transgender and gender-diverse vs.
cisgender) were signicantly different for the APHS, IMAGE,
and LifeLines datasets when compared to the MU and the C4
datasets. We used a subsampling bootstrap approach (10,000 sub-
samples) to test this and calculated empirical p-values (Meth-
ods). Empirical pvalues suggested that the ORs for the APHS
(pvalue =0.078), IMAGE (pvalue =0.11), and LifeLines (pvalue
=0.84) datasets were not statistically different from the ORs
observed in the 10,000 samples generated from the C4 dataset.
Similarly, empirical pvalues for the APHS (pvalue =0.56),
IMAGE (pvalue =0.44), and LifeLines (pvalue =0.85) datasets
suggested that the ORs were not statistically different from that
observed in the 10,000 permuted samples generated from the MU
dataset.
We also investigated if rates of transgender and gender diversity
are higher in individuals diagnosed with autism using a logistic
regression framework after accounting for age and educational
attainment. We identied signicant associations in four of the ve
dataset (C4: OR =4.66, 95%CI =4.265.10, pvalue < 2 × 1016;
MU: OR =6.05, 95%CI =4.558.05, pvalue < 2 × 1016;IMAGE:
OR =6.35, 95%CI =3.3212.11, pvalue =2.1 × 108;APHS:
OR =6.31, 95%CI =4.149.62, pvalue < 2 × 1016)anda
Channel 4
N = 514,100
Musical Universe
N = 85,670
IMAGE
N = 1803
APHS
N = 2312
Gender identity and autism diagnosis
Gender identity and autistic traits
Channel 4: AQ-10, EQ-10, SQ-10, SPQ-10
IMAGE: AQ-50; LifeLines: AQ-10
Gender identity and other neurodevelopmental
and psychiatric conditions
ADHD, bipolar disorder, depression, OCD,
schizophrenia, and learning disorder
LifeLines
N = 37,975
Fig. 1 Schematic diagram of the study. This gure provides a schematic overview of the study. In this study we investigated three questions, presented in
the red boxes. For each question, the primary dataset was the Channel 4 dataset (pink box). We used four validation datasets to validate the results
Musical Universe (cyan box), LifeLines (orange box), IMAGE (yellow box), and APHS (purple box). Colored arrows from the dataset boxes to the questions
indicate which questions were investigated in which datasets. AQ-10 (Autism Spectrum Quotient-10), SQ-10 (Systemizing Quotient-10), EQ-10 (Empathy
Quotient-10), SPQ-10 (Sensory Perception Quotient-10), AQ-50 (Autism Spectrum Quotient-50), ADHD (Attention-Decit/Hyperactivity Disorder), OCD
(Obsessive-Compulsive Disorder).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1 ARTICLE
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 3
Content courtesy of Springer Nature, terms of use apply. Rights reserved
concordant effect direction in the LifeLines dataset (OR =2.91, 95%
CI =0.6912.20, pvalue =0.14).
Traits related to autism. As seen in cisgender individuals51,
autistic transgender and gender-diverse individuals scored higher
on the AQ-10, SQ-10, and SPQ-10, and lower on the EQ-10
compared to non-autistic transgender and gender-diverse indi-
viduals (Cohens D: 0.540.72, p-value < 2 × 1016, Supple-
mentary Tables S5 and S6).
We next investigated gender differences in scores on the AQ-
10, SQ-10, EQ-10, and SPQ-10 in autistic and non-autistic
individuals separately in the C4 dataset. In both autistic and non-
autistic individuals separately, ANOVA identied signicant
differences based on gender on all four measures (pvalue < 2 ×
1016 in all comparisons). Post-hoc t-tests indicated signicant
differences between groups across all measures: transgender and
gender-diverse individuals scored higher on the AQ-10, SQ-10,
and SPQ-10, and lower on the EQ-10 compared to both cisgender
males and cisgender females. The effect sizes for differences in
scores were larger for the cisgender male vs. transgender and
gender-diverse as well as cisgender female vs. transgender and
gender-diverse tests compared to the cisgender male vs. cisgender
female tests across all four measures in both non-autistic and
autistic individuals (Supplementary Tables S5 and S6).
For both cisgender male vs. transgender and gender-diverse as
well as cisgender female vs. transgender and gender-diverse
comparisons, effect sizes were larger in autistic individuals
(Cohens D: 0.551.05) compared to the same analyses in non-
autistic individuals (Cohens D: 0.320.96). This contrasts with
cisgender male vs. cisgender female gender differences for these
measures, which are attenuated in autistic individuals compared
to non-autistic individuals (Supplementary Tables S5 and S6 and
Fig. 3).
We repeated the analyses after accounting for autism diagnosis,
age, and educational attainment. Transgender and gender-diverse
individuals scored higher (pvalue < 2 × 1016 for all) than both
cisgender males and cisgender females on the AQ-10 (cisgender
males: Beta =0.89 ± 0.02, cisgender females: Beta =1.05 ± 0.02),
the SQ-10 (cisgender males: Beta =0.66 ± 0.02, cisgender females:
Beta =0.99 ± 0.02), and the SPQ-10 (cisgender males: Beta =
0.66 ± 0.02, cisgender females: Beta =0.55 ± 0.02), and lower on
the EQ-10 (cisgender males: Beta =0.33 ± 0.02, cisgender
females: Beta =0.70 ± 0.02) (Fig. 3and Supplementary Fig. S1).
We replicated this in two datasets: the IMAGE dataset using the
AQ-50 and the LifeLines dataset using the AQ-10. In the IMAGE
dataset, transgender and gender-diverse individuals scored
higher than both cisgender males (Beta =0.45 ± 0.11, pvalue =
3.09 × 105) and cisgender females (0.52 ± 0.11, pvalue < 1.80 ×
106). In the LifeLines dataset, transgender and gender-diverse
individuals scored higher than cisgender females (Beta =1.23 ±
0.25, pvalue =1.4 × 106) and nominally higher than cisgender
males (Beta =0.51 ± 0.25, pvalue =0.045).
The previous analyses investigated the association between
gender identity and traits related to autism individually. We next
investigated if there are differences in the standardized dis-
crepancy between the EQ-10 and the SQ-10 in the three gender
categories using Brain Types. Compared to both non-
autistic cisgender males and non-autistic cisgender females,
non-autistic transgender and gender-diverse individuals were
signicantly more likely to be classied as Type S (cisgender
males 40.23%, cisgender females 25.58%, transgender and gender-
diverse 53%) or Extreme Type S (cisgender males 4.14%,
cisgender females 1.69%, transgender and gender-diverse
13.15%) (pvalue < 2 × 1016). This was more pronounced in
autistic transgender and gender-diverse individuals compared to
autistic cisgender individuals (Extreme Type S: cisgender males
11.42%, cisgender females 7.55%, and transgender and gender-
5
10
15
20
C4
MU
APHS
IMAGE
Lifelines
OR
0
5
10
15
20
25
C4
MU
APHS
IMAGE
Lifelines
OR
Reference category Cisgender male Cisgender Cisgender female
Fig. 2 ORs and 95% CIs for autism in transgender and gender-diverse individuals compared to cisgender males, cisgender females, and cisgender
individuals altogether. a This gure provides the unadjusted Odds Ratios (ORs, point) and 95% CIs for autism in transgender and gender-diverse
individuals compared to either cisgender males, cisgender females, or cisgender (cisgender males and cisgender females) individuals in ve datasets (C4:
N=514,100; MU: N=85,670; APHS: N=2312; IMAGE: N=1803; and LifeLines: N=37,975). bThis gure provides adjusted ORs (point) and 95% CIs for
autism in transgender and gender-diverse individuals compared to cisgender males, cisgender females, or all cisgender individuals in ve datasets (C4:
N=514,100; MU: N=85,670; APHS: N=2312; IMAGE: N=1803; and LifeLines: N=37,975). ORs have been adjusted for age, educational attainment,
and in the case of IMAGE dataset, an additional dummy variable for study (see Supplementary Methods). The y-axis is on the same scale for both the
panels. The differences in ORs for the IMAGE dataset between Models 1 and 2 is primarily due to the inclusion of studygroup as a covariate. Specically,
the IMAGE dataset consists of individuals recruited into a study of mathematics and autism (Methods). Whilst the mathematics group is predominantly
male and have higher educational attainment (all have at least an undergraduate degree), the casecontrol group had a more balanced ratio and a wider
range of educational attainment. Covarying for the study the participants have been recruited into (mathematics or autism casecontrol) changes the ORs.
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
4NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
diverse 34.73%; Type S: cisgender males 50.97%, cisgender
females 42.29%, transgender and gender-diverse 51.79%) (p
value < 2 × 1016). (Supplementary Table S7 and Supplementary
Fig. S2). Cumulatively, in autistic individuals, 86.52% of
transgender and gender-diverse individuals were classied as
Type S or Extreme Types S compared to 62.39% of cisgender
males. In both autistic and non-autistic transgender and gender-
diverse individuals, observed values were signicantly shifted
towards Type S and Extreme Type S compared to what is
expected (pvalue < 2 × 1016).
Rates of other neurodevelopmental and psychiatric conditions.
We next investigated if rates of six other neurodevelopmental and
psychiatric conditions (ADHD, bipolar disorder, depression,
learning disorder, OCD, and schizophrenia) differed by gender in
the C4 dataset. Compared to cisgender individuals, transgender and
gender-diverse individuals had elevated rates of all these conditions,
with the highest effect size for schizophrenia (OR =28.52, 95%
CI =24.1733.66, pvalue < 2 × 1016) and the lowest for learning
disorders (OR =3.48, 95%CI =3.093.91, pvalue < 2 × 1016)
(Supplementary Table S8). Including age and educational attain-
ment as covariates (Model 2) attenuated the ORs only modestly
(ORs: 3.08 (learning disorders) to 19.73 (schizophrenia)). However,
the ORs were substantially attenuated when autistic individuals
were excluded, i.e., Model 3 (1.92 (learning disorders) to 6.39
(schizophrenia)) (Supplementary Table S8). Notably, there was a
considerable attenuation in the OR for schizophrenia. The ORs for
autism, ADHD, bipolar disorder and depression were similar to
each other. In comparison, the ORs for OCD and LD were about
half that for autism. Supplementary Table S9 provides results of the
analyses repeated for the three genders (cisgender male, cisgender
female, and transgender and gender-diverse).
We repeated the analyses for ve of the six conditions tested
above in the MU dataset. Compared to cisgender individuals,
transgender and gender-diverse individuals reported higher rates
of all ve conditions (Model 1; OR: 2.15 (schizophrenia) to 3.83
(depression)), with the results for schizophrenia not being
statistically signicant, possibly due to small sample size (Fig. 4).
These results were similar after accounting for educational
attainment and age (Model 2; OR: 1.81 (schizophrenia) to 3.89
(depression)), and additionally, after excluding autistic indivi-
duals (Model 3 OR: 1.11 (schizophrenia) to 3.91 (depression))
(Supplementary Table S8). In contrast to the C4 dataset, in the
MU dataset, the ORs for autism was the largest, followed by the
two mood disorders (depression and bipolar disorder). Notably,
the OR for depression was similar in both the C4 and the MU
datasets. Supplementary Table S9 provides results of the analyses
repeated for three genders (cisgender male, cisgender female, and
transgender and gender-diverse).
0.0
0.1
0.2
0.3
0.4
0.5
0246810
AQ-10
Density
0.00
0.05
0.10
0.15
0 5 10 15 20
SQ-10
Density
0.00
0.02
0.04
0.06
0.08
0 5 10 15 20
EQ-10
Density
0.00
0.02
0.04
0.06
0102030
SPQ-10
Density
Non-autistic cisgender females Non-autistic cisgender males Non-autistic transgender and gender-diverse individuals
Fig. 3 Kernel density plot of scores on the four self-report measures in the C4 Dataset for non-autistic individuals only. This gure provides kernel
density plots for scores on the four self-report measures (AQ-10, EQ-10, SQ-10, and SPQ-10) for non-autistic participants from the C4 dataset
(N=514,100) based on their gender (cisgender males, cisgender females, transgender and gender-diverse individuals). Scales on the axes are different
between the panels. See Supplementary Fig. S1 which provides kernel density plots for all four measures for both autistic and non-autistic individuals. The
non-autistic transgender and gender-diverse kernel density plots appear smoother due to the relatively low number of participants included, hence
providing less resolution in the kernel density estimates when compared to the non-autistic cisgender males and non-autistic kernel density plots.
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1 ARTICLE
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 5
Content courtesy of Springer Nature, terms of use apply. Rights reserved
To further clarify the role of autism compared to other
neurodevelopmental and psychiatric conditions, we conducted
multiple regressions to investigate the relative effects of associa-
tion of autism on transgender and gender-diverse identities
compared to other neurodevelopmental and psychiatric condi-
tions. In the C4 dataset, depression had the highest OR
(OR =3.55, 95%CI =3.843.29, pvalue < 2 × 1016) followed
by autism (OR =3.43, 95%CI =3.793.11, pvalue < 2 × 1016).
In the MU dataset, we obtained very similar ORs. Autism had the
highest OR (OR =3.94, 95%CI =5.612.77, pvalue < 2 × 1016)
followed by depression (OR =3.50, 95%CI =4.252.89, pvalue <
2×10
16). ORs for other conditions are provided in the
Supplementary Table S10.
Exploratory analysis: rates of suspected autism. In the IMAGE
dataset, we also investigated if transgender and gender-diverse
individuals were more likely to suspect they had undiagnosed
autism compared to cisgender individuals. A χ2test identied a
signicant difference between genders (χ2(2) =42.087, φ=0.15,
pvalue =7.52 × 1010). Transgender and gender-diverse indivi-
duals were more likely to suspect they had undiagnosed autism
compared to cisgender males (OR =4.32, 95%CI =1.9410.10,
pvalue =2.51 × 104), cisgender females (OR =7.99, 95%
CI =3.5418.92, pvalue =3.13 × 108), and cisgender male and
female individuals altogether (OR =5.47, 95%CI =2.4712.72,
pvalue =9.01 × 106).
Discussion
In this study, we investigated three primary questions, and an
additional exploratory question using ve different, large-scale
datasets. First, across all ve datasets, transgender and gender-
diverse individuals were 3.03 to 6.36 times as likely to be autistic
than were cisgender individuals, after controlling for age and
educational attainment. Second, transgender and gender-diverse
individuals scored signicantly higher on self-report measures of
autistic traits, systemizing and sensory sensitivity and scored
signicantly lower on empathy traits compared to cisgender
individuals. Third, in two datasets with available data, transgen-
der and gender-diverse individuals had elevated rates of multiple
other neurodevelopmental and psychiatric conditions. Finally,
exploratory analysis identied that transgender and gender-
diverse individuals were more likely to report that they suspected
they had undiagnosed autism.
These associations between gender identity and autism diag-
noses are unlikely to be false positives for multiple reasons. First,
we observe consistent effect directions across multiple datasets
with very different recruitment strategies, ascertainment biases,
cultural backgrounds, and age ranges. The effects after accounting
for age and educational attainment were statistically signicant
for four of the ve datasets, and in the same direction for the fth
(i.e., LifeLines cohort). The lack of statistical signicance is due to
the low statistical power of the LifeLines dataset, because parti-
cipants were older and healthier as individuals with severe mental
health conditions were excluded at the time of recruitment, and
individuals with higher genetic likelihood for mental health
conditions are likely to drop out from longitudinal studies52,53.
Second, comparing the ORs of the three smaller samples
(IMAGE, APHS, and LifeLines) to bootstrapped ORs from
10,000 subsamples in the two largest samples (C4 and MU) did
not identify statistically signicant differences in ORs. This
indicates that the ORs are similar regardless of different recruit-
ment strategies and different methods used to ascertain gender
and autism. Third, sensitivity analysis in the MU dataset did not
identify differences in the rates of autism diagnosis between
participants who indicated Othervs. Transgender. Fourth, the
ORs observed in this study are similar to those observed in
participants from GD clinics17, suggesting that ORs observed
using an internet-based convenience sampling framework is
similar to ORs observed in GD clinic-based samples.
Supporting the association between gender identity and autism
diagnoses, transgender and gender-diverse individuals also had
higher scores on self-report measures of autistic traits, sensory
sensitivity, and systemizing, and lower scores on a self-report
measure of empathy traits, compared to cisgender individuals.
The transgender and gender-diverse vs. cisgender effect sizes are
equivalent to or larger than the autism vs. non-autism effect sizes
and the cisgender male vs. cisgender female effect sizes in non-
autistic individuals. Importantly, these effects were also observed
when investigating the discrepancy of scores on the EQ-10 and
SQ-10 using the Brain Typesanalyses. In addition, in a rela-
tively smaller sample (IMAGE), transgender and gender-diverse
individuals were more likely to suspect they had undiagnosed
autism. Taken together, our analyses indicate that transgender
and gender-diverse individuals are more likely to be autistic
compared to cisgender individuals, and further that undiagnosed
autism may also be higher in transgender and gender-diverse
individuals.
0
10
20
30
Autism
ADHD
Bipolar
Depression
LD
OCD
OR
0.0
2.5
5.0
7.5
Autism
ADHD
Bipolar
Depression
OCD
Schizophrenia
OR
Model Model 1 Model 2 Model 3
a
b
Schizophrenia
Fig. 4 ORs and 95% CIs for other neurodevelopmental and psychiatric
conditions in transgender and gender-diverse individuals compared to
cisgender individuals. a This gure provides the Odds Ratios (ORs, point)
and 95% CIs for diagnosis of autism and six other neurodevelopmental and
psychiatric conditions in transgender and gender-diverse individuals
compared to cisgender individuals in the C4 dataset (N=514,100). We did
not employ Model 3 for autism as it was conducted after excluding autistic
individuals in the dataset. ORs have been calculated using three models
(see Methods). ADHD =Attention-Decit/Hyperactivity Disorder; OCD =
Obsessive-Compulsive Disorder; LD =Learning Disorder. bThis gure
provides the same, but for the MU dataset (N=85,670). Information on
LD was not available in the MU dataset. The y-axis is on a different scale
from the panel above.
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
6NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
However, this association with gender identity is not specicto
autism. In two datasets, transgender and gender-diverse indivi-
duals also had elevated rates of ADHD, bipolar disorder,
depression, OCD, learning disorders, and schizophrenia com-
pared to cisgender individuals. In one of the two datasets, we
tested and conrmed that transgender and gender-diverse indi-
viduals had higher rates of learning disorders compared to cis-
gender individuals. In the C4 dataset, we identied elevated rates
of schizophrenia in transgender and gender-diverse individuals
compared to cisgender individuals but were unable to replicate
this in the MU dataset.
Our multiple regression analyses helped clarify the relative
association strengths of these conditions with transgender and
gender-diverse individuals. In both the MU and the C4 datasets,
autism and depression had the highest effect sizes. Notably, in the
MU dataset, none of the other conditions were signicantly ele-
vated in transgender and gender-diverse individuals after con-
trolling for autism and depression, which is discordant with the
results identied in the C4 datasets. This discrepancy in the
results may be due to differences in sample sizes, ascertainment,
or other cohort characteristics. For instance, the C4 study directly
recruited participants to an autism study. This may oversample
individuals with other co-occurring mental health conditions. In
contrast, the MU dataset is a convenience sample collected over
many months. There is some evidence to suggest that individuals
with elevated genetic liability for schizophrenia, ADHD, and
depression may be less likely to participate in studies52,53, and, as
a result, they may be underrepresented in the MU dataset. In
addition, most of the participants in the C4 are from the UK,
whilst most of the MU participants are from the US. Differences
in diagnostic practices may also contribute to sampling differ-
ences. A more comprehensive investigation of the relative rates of
neurodevelopmental and psychiatric conditions in transgender
and gender-diverse individuals compared to cisgender individuals
is warranted.
The elevated rates of autism and other conditions must be
considered against other hypotheses that may explain the
observed results due to the non-probabilistic nature of the sam-
ple. Specically, for autism, one alternative hypothesis is that
transgender and gender-diverse individuals may be more likely to
report higher rates of autistic traits due to long-standing experi-
ences and feelings of not tting in socially, with true levels of
autistic traits being comparable between cisgender and trans-
gender and gender-diverse individuals. Although this is possible,
other studies have reported elevated autistic traits measured using
parent- or teacher-report instruments in individuals with
GD31,33. Importantly, in our study, we note that the shift in scores
in transgender and gender-diverse individuals is observed across
both social (EQ-10) and non-social (SPQ-10 and SQ-10) mea-
sures of traits related to autism, which themselves are only partly
correlated51,54,55. Notably, transgender and gender-diverse indi-
viduals also score higher on the SPQ-10, a measure of sensory
sensitivity, and response to items on this measure are unlikely to
be inuenced by social gender norms.
Another alternative hypothesis is that autistic transgender and
gender-diverse individuals may be more likely to participate in
these studies compared to autistic cisgender individuals (i.e.,
selection bias). However, this is unlikely: the datasets were not
collected to specically investigate the links between gender and
rates of autism diagnosis. Whilst autistic individuals may be more
likely to participate in the autism-related studies (C4, APHS, and
IMAGE), it is unlikely that this will be biased towards autistic
transgender and gender-diverse compared to autistic cisgender
individuals. In addition, two of the datasets (MU and LifeLines)
were not collected specically for an autism-based study. Further,
the LifeLines also has a healthy volunteer bias, which is likely to
attenuate ORs. In other words, a strength of this study is that
none of the datasets were collected to specically test the asso-
ciation between autism and gender identity. Furthermore, similar
ORs have been observed in a large-scale study of autism in par-
ticipants of GD clinics which are unlikely to be affected by this
specic type of selection bias17, providing further corroboration
to our ndings.
Whilst our study does not test causality, a few hypotheses may
explain the over-representation of autism and other neurodeve-
lopmental and psychiatric conditions in transgender and gender-
diverse individuals. First, autistic individuals may conform less to
societal norms compared to non-autistic individuals, which may
partly explain why a greater number of autistic individuals
identify outside the stereotypical gender binary. Second, prenatal
mechanisms (e.g., sex steroid hormones) shaping brain develop-
ment have been shown to contribute to both autism (and asso-
ciated neurodevelopmental conditions) and gender role
behavior5660. It is unclear if prenatal sex steroid hormones also
contribute to gender identity and this should be investigated in
future studies. Neurodevelopmental conditions such as ADHD
and learning disorders frequently co-occur with autism47, and
genetic evidence suggests a shared underlying liability for many of
the co-occurring neurodevelopmental and psychiatric condi-
tions61,62. Finally, an alternative but not mutually exclusive
explanation is that transgender and gender-diverse individuals
have elevated vulnerabilities for multiple psychiatric challenges
related to stressful life experiences in the contexts of unfriendly
environments, discrimination, abuse and victimization, explain-
ing the elevated rates of mental health diagnoses63,64.
These ndings must be interpreted in light of the lived
experiences, rights, and clinical and daily life needs of transgender
and gender-diverse individuals. Both autistic individuals and
transgender and gender-diverse individuals are marginalized
groups where the currently available support and understanding
is inadequate65. Both groups are also more likely than others to
engage in self-harm, suicidal ideation and suicidal behaviors, and
to have other vulnerabilities63,6668. This intersection of autism
and gender diversity can be doubly distressing if adequate safe-
guarding and support are not provided. A recent study demon-
strated that a third of autistic individuals had their gender
identity questioned because they were autistic65. There is a need
to ensure that autistic transgender and gender-diverse individuals
have the right to express their gender, live with dignity, and
receive social and legal recognition of their gender69 (also see:
https://autisticadvocacy.org/wp-content/uploads/2016/06/joint_
statement_trans_autistic_GNC_people.pdf). Additionally, recent
studies demonstrate that autistic characteristics partly differ
between cisgender males and cisgender females50,70,71. However,
it is still unclear if autistic characteristics differ in transgender and
gender-diverse individuals compared to cisgender individuals.
This co-occurrence requires gender-informed and neurodiversity-
informed clinical care for autistic transgender and gender-diverse
individuals.
There are caveats to this study. First, in two of the datasets we
excluded intersex individuals, but this was not an option in other
datasets (C4, LifeLines and MU). Second, there is a possibility
that some nonbinary, gender-neutral, or other gender-diverse
individuals may not identify with the transgenderterm in the
C4 dataset as we did not concurrently provide the transgender
and otheroptions. Third, some gender-aware individuals may
respond by providing their sex rather than their gender. It is
difcult to disentangle this. However, the magnitude of the
sample size suggests that the effects of such misclassication will
have a minimal effect on the analyzes and ndings. Supporting
this, subsampling bootstrap analyses indicate that the ORs are
similar across the different datasets. Additionally, the ORs are
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1 ARTICLE
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 7
Content courtesy of Springer Nature, terms of use apply. Rights reserved
similar between the ve internet-based datasets in this study and
a study based on GD-clinic based samples17. This similarity
suggests that regardless of recruitment (internet-based vs. clinic-
based) or ascertainment criteria (self-report gender identity vs.
clinically ascertained gender dysphoria) or age (adults vs. chil-
dren), the results converge on similar ORs. Fourth, individuals
with severe mental health conditions and intellectual disability are
less likely to participate. Finally, these datasets are not statistically
well-powered to investigate rates of autism diagnosis in trans-
gender and gender-diverse individuals after stratifying by sex
assigned at birth; thus, we have not investigated this.
In conclusion, our study demonstrates that transgender and
gender-diverse individuals have elevated rates of autism diag-
nosis, related neurodevelopmental and psychiatric conditions,
and autistic traits compared to cisgender individuals. This study
has clinical implications by highlighting that we need to improve
access to care and tailored support for this under-served
population.
Methods
Overview of the datasets. We used ve datasets for this study. The largest of
these (Channel 4 dataset, C4) consists of N=514,100 individuals who completed
online questionnaires as a part of a UK Channel 4 television program about autism.
These participants self-reported their autism diagnosis, and indicated their gender
based on three options Male,Female, and Transgender. To address autism-
related self-selection bias in this dataset, we used a second dataset (Musical Uni-
verse, MU, N=85,670) recruited through a website for research about musical
behavior, personality and cognition. Participants completed information about
their autism diagnosis and selected their gender from four options: Male,
Female,Transgenderand Other. However, neither of these two datasets have
separately recorded information on sex at birth and gender, and in both datasets,
participants were asked to choose their Sex, although we acknowledge that the
information collected is primarily of gender. To address this, we used two addi-
tional datasets where information was collected separately for sex at birth and
gender. In the third dataset (APHS, N=2312), participants were recruited for an
internet-based physical health survey. Participants completed information on their
autism diagnosis including when they were diagnosed and who diagnosed them,
their sex at birth, and their current gender identity. The fourth dataset (IMAGE,
N=1803) consists of participants who were recruited for a genetic study of autism
and mathematical ability. Participants completed information on their autism
diagnosis, their sex at birth, and their gender. In addition, all autistic participants
provided a copy of their diagnostic report to verify their diagnosis. The fth and
nal dataset consists of a subset of participants from the LifeLines Cohort and
Biobank72 (N=37,975) who provided information on sex assigned at birth and
gender, autism diagnosis, and completed a measure of autistic traits. This dataset
consists of individuals who are considerably older than those in the other four
datasets, and who were recruited primarily through GP clinics. None of the ve
datasets were recruited specically to investigate the association between gender
diversity and autism, which limits gender-based self-selection bias.
Channel 4 dataset: overview. The Channel 4 dataset (C4 dataset) comprises
participants who completed self-report measures as a part of the Channel 4 doc-
umentary titled Are you autistic?, in Spring 201751. A mobile-friendly website
was developed and advertised on the Channel 4 TV website (https://www.channel4.
com/). Participants indicated if their results could be used for research purposes. A
total of 758,916 entries were recorded. Participants provided information on
demographics (gender (see below for details), age, educational attainment, geo-
graphical region, handedness, occupation, autism and other neurodevelopmental
or psychiatric diagnosis) and completed four self-report measures. Participants
who consented to share their data for research were asked: Have you taken this
survey before? To make sure our data are as accurate and as useful as possible
please tell us if youve taken this survey before.If participants indicated that they
had taken the survey before, they were marked as duplicates. After removing
duplicates, we were left with a total of 695,166 participants. We were unable to use
IP addresses to identify duplicates due to ethical constraints. We included parti-
cipants aged 15 to 90 years, in line with previous research51. Participants were
asked to indicate their Sexusing one of four options: Male,Female,
Transgenderand Prefer not to say. Whilst Sexwas asked in the survey, we
recognize that the information provided here is of sex or gender, or both and we
refer to this as gender throughout the manuscript. Whilst designing the survey we
did not make a distinction between gender and sex as these terms are often used
interchangeably in the general population. We further removed individuals who
did not provide information on gender (Prefer not to say), resulting in
N=675,360 individuals.
Channel 4: ascertaining gender identity. During data collection, information on
gender was initially collected using four options listed above. However, towards the
end of the data collection phase, the Transgenderoption was modied to Other
to make it more inclusive. For this study, we restricted our analysis to only those
participants from the rst phase of data collection who could choose from Male,
Female,Transgenderand Prefer not to say, as this makes it clearer for
interpreting the data. This resulted in 514,100 individuals whose gender was either
Male(N=193,398), Female(N=317,891), or Transgender(N=2811 or
0.55%).
Channel 4: ascertaining diagnosis of autism and other conditions. 27,919
participants (5.4%) indicated they had an autism diagnosis (cisgender
males =13,317; cisgender females =13,934, transgender and gender-diverse =
668). Diagnoses of autism and other psychiatric conditions were asked using the
question: Have you been formally diagnosed with any of the following (please
click all that apply?). For other psychiatric conditions, participants could choose
from ADHD, bipolar disorder, depression, learning disorder, schizophrenia, and
OCD. The wording of the question should typically preclude (though not com-
pletely eliminate) self-diagnosed individuals. Participants indicated they had the
following diagnoses: ADHD (N=19,300), bipolar disorder (N=9025), depression
(N=122,829), learning disorder (N=18,559), OCD (N=13,115), and schizo-
phrenia (N=1321). These were not mutually exclusive, as individuals could
endorse several diagnoses. In addition, participants provided information on their
educational attainment and age (Supplementary Tables S1 and S2).
Channel 4: measures of traits related to autism. All participants completed four
short, self-report psychological trait measures: the Autism Spectrum Quotient-10
(AQ-10)73, a widely-used measure of autistic traits; the Empathy Quotient-10 (EQ-
10)51, a measure of empathy traits; the Systemizing Quotient-10 (SQ-10)51 (10
items from the Systemizing QuotientRevised74, but referred to here as System-
izing Quotient-10), a measure of systemizing traits (the drive to analyze or build a
system75); and the Sensory Perception Quotient-10 (SPQ-10)51, a measure of
sensory sensitivity. Using the SQ-10 and the EQ-10 data, we calculated Brain
Types51, which refer to an individuals cognitive prole based on the discrepancy
of their scores on empathy and systemizing traits. Individuals may be classied into
one of ve different Brain Typesbased on the standardized discrepancy between
their systemizing and empathy scores51,76.
Musical Universe dataset: overview of dataset. The Musical Universe (MU)
dataset consists of a total of 89,218 individuals who completed measures on
musical behavior, personality, and cognition, in exchange for feedback about their
scores at www.musicaluniverse.org. We identied duplicates rst using IP
addresses, and then, among individuals with identical IP addresses, using demo-
graphic variablesgender (see below for further information about this), age,
educational attainment, occupation, and diagnosis. A total of 85,670 unique
records were identied. Participants ranged in age from 18 to 88 years old (Sup-
plementary Table S1).
Musical Universe: ascertaining gender identity. Similar to C4, the MU data
collection did not make a clear distinction between gender and sex. Participants
were asked for their Sexwhere they could choose one of four options: Male
(42,291 non-autistic and 666 autistic), Female(41,659 non-autistic and 365
autistic), Transgender(361), and Other(328) (Supplementary Table S1).
However, we recognize that participants have actually provided information on
their gender and we refer to this as gender throughout the manuscript. In the
primary analyses, we combined participants who chose the Transgenderand
Otheroption into the transgender and gender-diverse group (634 non-autistic
and 55 autistic individuals) and conducted further sensitivity analyses using only
individuals who chose the Transgenderoption. We decided to combine the two
groups as some individuals who are transgender and gender-diverse in the broad
sense (i.e., their gender is different from their sex assigned at birth) may not
identify as transgender and may interpret the term transgender more narrowly (i.e.,
their binary gender identity is opposite to the binary sex assigned at birth).
Musical Universe: ascertaining diagnosis of autism and other conditions.
Participants were asked if they had a formal diagnosis of autism from a profes-
sional. This should typically preclude (though not completely eliminate) self-
diagnosed autistic individuals from participating. A total of 1,086 participants
indicated that they had an autism diagnosis (Supplementary Table S1). In addition,
they were asked if they had a formal diagnosis of additional mental health con-
ditions. A subset of participants (N=54,127) indicated if they had a formal
diagnosis of: 1. ADHD (N=3189, 5.89%); 2. Bipolar disorder (N=1532, 2.83%); 3.
Depression (N=11,919, 22.02%); 4. OCD (N=1419, 2.62%); and 5. Schizophrenia
(N=202, 0.37%).
Autism Physical Health Survey: overview of dataset. The Autism Physical
Health Survey (APHS) dataset consists of 2312 individuals aged 1690 years who
were recruited via the Cambridge Autism Research Database (CARD), autism
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
8NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
charities and support groups, and social media as a part of a study investigating the
association between autism and physical health conditions. The study employed an
anonymous, online self-report survey via Qualtrics. Participants were asked
questions regarding their demographics, lifestyle factors (including diet, exercise,
sleep, and sexual/social history), personal medical history, and family medical
history for all rst-degree, biological relatives. As the study was anonymous (and
we did not collect IP addresses), we excluded records that we determined were
likely to be duplicates. We excluded all records that matched a previous record
across 11 categories: whether or not they had an autism diagnosis, specic autism
diagnosis, type of practitioner who diagnosed them, year of diagnosis, syndromic
autism (if applicable), country of residence, sex assigned at birth, current gender
identity, age, maternal age at birth, paternal age at birth, and educational
attainment.
Autism Physical Health Survey: ascertaining gender identity. Participants were
asked for their sex assigned at birth (Male,Female,Other) and for their
current gender identity (Female(N=1383), Male(N=766), Non-binary
(N=109), and Other(N=20)). We removed participants who indicated Other
for their sex assigned at birth (N=1), and who did not complete information on
gender identity (N=3). Additionally, 33 individuals had discordant sex and gender
information (7 individuals of male sex but female gender, and 26 individuals of
female sex and male gender). As we did not provide a transgender option in the
gender identity column, we classied these individuals as transgender. Thus, in
total there were 162 individuals who were included in the transgender and gender-
diverse group (Supplementary Table S1).
Autism Physical Health Survey: ascertaining autism diagnosis. Participants
were asked to indicate if they had an autism diagnosis. Whilst we did not require
participants to upload a copy of their diagnostic report, they had to provide further
information about which type of clinician diagnosed them as autistic (general
practitioner, neurologist, pediatrician, psychiatrist, psychologist or other (free text
box)), what their specic diagnosis was, and when they were diagnosed. A total of
1082 individuals indicated that they had an autism diagnosis (Supplementary
Table S1).
The IMAGE study: overview of dataset. The Investigating Mathematics and
Autism using Genetics and Epigenetics (IMAGE) dataset consists of individuals
recruited into a genetic study of autism and mathematical ability. This was done
using two different research designs. The rst targeted autistic and non-autistic
individuals as a part of a casecontrol design (N
nal
=292) by advertising in
research databases, autism-related magazines, and on social media. The second
targeted individuals who studied or were studying mathematics or a related degree
(N
nal
=1803) by advertising in universities, mathematics societies, in mathematics
specic or alumni magazines, or on social media. Participants registered at a
bespoke website and provided contact details, demographics, and completed var-
ious questionnaires. As participants provided both their names and their contact
details, we used this information to remove duplicate records.
The IMAGE study: ascertaining gender identity. Participants were asked for
their sex at birth (Male,Femaleor Intersex) and their gender (Man
(N=994), Woman(N=747), Transgender Man(N=7), Transgender
Woman(N=3), Nonbinary(N=35), Gender Neutral(N=10), Other
(N=7), and Prefer not to say(N=15)). We excluded individuals who chose
Intersex(N=2) for their sex, and Prefer not to say(N=15) for their gender.
Of the remaining, we combined individuals who chose Manand Womanas the
cisgender group (N=1741), and the remaining into the transgender and gender-
diverse group (N=62). Further details are provided in Supplementary Table S1.
The IMAGE study: ascertaining autism diagnosis. Participants were asked if
they had a diagnosis of autism on the autism spectrum (e.g., autism, Asperger
Syndrome). As a part of this, we indicated that diagnosis must have been made by a
qualied professional (e.g., clinical psychologist or psychiatrist). Participants were
also asked when they received an autism diagnosis and who diagnosed them. In
addition, autistic individuals in this study were asked to provide a copy of their
diagnostic report that we used to conrm their autism diagnosis. A total of 1082
individuals indicated that they had an autism diagnosis (Supplementary Table S1).
A subset of participants (N=1787) provided information on educational attain-
ment. 1417 participants indicated if they suspected they had undiagnosed autism
(Yesor No). This was used to investigate if transgender and gender-diverse
non-autistic individuals were more likely to suspect they had undiagnosed autism
compared to non-autistic cisgender individuals.
The IMAGE study: measures of traits related to autism. All participants
completed the AQ-5077.
LifeLines: overview of dataset. The LifeLines Cohort is a Netherlands-based
population cohort study, recruited between 2006 and 201372. Participants were
invited through their general practitioners in three northern provinces in the
Netherlands (Freisland, Groningen, and Drenthe). Notably, participants were not
invited if they had a severe mental health condition, which suggests that this
dataset will be biased towards healthy participants. A total of 167,729 participants
aged between 6 months and 93 years completed the baseline survey. The LifeLines
dataset used in this study consists of 37,975 individuals from the cohort, who
responded to an online questionnaire on autistic traits in summer 2019. All par-
ticipants were at least 18 years of age. The participants in the LifeLines cohort were,
on average, about twice as old as the participants in the C4 and the MU cohorts,
and this may in part explain the relatively low number of transgender and gender-
diverse individuals in this dataset. In addition, 37,574 participants provided
information on their highest level of educational attainment (Supplementary
Table S2).
LifeLines: ascertaining gender identity. Information on gender was collected
using one question: Please choose which description ts you best. This was
followed by ve options: At birth I was registered as female and I am female,At
birth I was registered as male and I am male,At birth I was registered as female,
but I am male,At birth I was registered as male, but I am female, and Different
from the options above, namely…”. Participants who chose the nal option were
required to ll in a short box describing their gender identity. In total, there were
15,527 cisgender males, 22,375 cisgender females, 18 transwomen, 17 transmen and
18 individuals who chose the other option and identied with other gender
identities (e.g., genderuid). Thus, in total, there were 53 transgender and gender-
diverse individuals (Supplementary Table S1).
LifeLines: ascertaining autism diagnosis. Autism diagnosis was ascertained using
the question: Do you have an autism diagnosis?followed by In what year was
this diagnosed. 439 individuals indicated that they had an autism diagnosis (252
cisgender males, 184 cisgender females, and 3 transgender and gender-diverse
individuals) (Supplementary Table S1).
LifeLines: measures of traits related to autism. All participants also completed
the AQ-1073, provided the age when they completed the AQ-10.
Ethics. The Human Biology Research Ethics Committee, University of Cambridge,
provided ethical approval for the collection and use of data for both the APHS and
the IMAGE cohorts. They also provided ethical approval to access de-identied
data from the LifeLines cohort. The Psychology Research Ethics Committee of the
University of Cambridge conrmed that formal ethical review was not needed for
use of the C4 dataset since it was secondary use of deidentied and anonymized
data. The same was conrmed for the MU dataset by the Ethical & Independent
Review Services. Informed consent was obtained for all participants included in
this study.
Statistical analyses: rates of autism diagnosis. In all ve datasets, we investi-
gated if rates of autism diagnosis signicantly differed by gender by rst conducting
χ2tests (Model 1, unadjusted), and then by conducting logistic regressions adjusted
for age and educational attainment as covariates (Model 2, adjusted). Both age and
educational attainment were associated with autism diagnosis, with younger
individuals more likely to receive an autism diagnosis78,79, and educational
attainment typically negatively correlated with autism51. Further, these two vari-
ables were measured across all ve datasets. In addition, for the IMAGE dataset, we
included a dummy variable for the two studies participants were drawn from
(mathematical ability and casecontrol) to account for potential confounding
effects of recruitment.
Each model was conducted rst by using three gender categories (transgender
and gender-diverse, male, and female), and then by using two gender categories
(transgender and gender-diverse and cisgender). Regression betas were converted
to ORs. As an additional sensitivity analysis, only in the MU dataset, we repeated
the analyses after dividing the cohort into four groups (Male,Female,
Transgender, and Other), to investigate if these results differed by gender
identity.
Additionally, we also investigated if rates of transgender and gender-diverse
individuals vary by autism diagnosis. This was done by using a logistic regression
comparing transgender and gender-diverse individuals to cisgender individuals
(dependent variable). Autism diagnosis was the independent variable, and
educational attainment and age were included as covariates.
Whilst information for this study from all ve datasets were collected using
internet-based surveys, there are differences between them. Of importance is that
sex, gender, and autism diagnosis information were all collected differently in the
ve datasets. In the C4 and MU datasets, gender information was collected using a
single question whereas in the IMAGE and APHS datasets, gender information was
collected using two questionsone for sex assigned at birth and another for gender
identied with. In the LifeLines dataset, gender information was collected using a
single question, but this included options about sex assigned at birth alongside
gender. Further, information on autism diagnosis was also collected differently
with deeper information provided by participants in the IMAGE, LifeLines, and
APHS datasets. There are other cohort-based differences as well. For example, the
MU dataset was aggregated over a long period of time and primarily collected from
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1 ARTICLE
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 9
Content courtesy of Springer Nature, terms of use apply. Rights reserved
the US, whilst three datasets (C4, APHS, and IMAGE) were collected over a shorter
period of time and primarily from the UK. The LifeLines dataset used here was a
subset of a cohort study, where participants were invited through general
practitioner clinics rather than via the internet. This was collected in the
Netherlands and consists of older participants.
Given the heterogeneity in these datasets, we wanted to investigate if the ORs
obtained across the ve datasets are comparable. Two factors affect ORs: winners
curse which inate ORs in smaller cohorts35,36, and lower precision, i.e., higher
standard errors of ORs in smaller cohorts80. Thus, ORs are not directly comparable
between the datasets. In order to make the ORs comparable, we generated sub-
datasets of equivalent sample sizes to the three smaller datasets (IMAGE, APHS, and
LifeLines) in the two larger datasets (C4 and MU). We used a subsampling bootstrap
approach to compare ORs in the two larger datasets with ORs in the smaller datasets.
We generated six sets of 10,000 random subsamples each from the C4 and the MU
datasets. Each of the 10,000 subsamples was matched to the numbers of cisgender
males, cisgender females and transgender and gender-diverse individuals in the
IMAGE, APHS, and LifeLines datasets. Thus, we sampled 10,000 times from the C4
and MU datasets with each sample consisting of 766 cisgender males, 1383 cisgender
females, and 162 transgender and gender-diverse individuals to match the APHS
dataset. In addition, we also sampled 10,000 times from the C4 and MU datasets with
each sample consisting of 994 cisgender males, 747 cisgender females, and 62
transgender and gender-diverse individuals to match the IMAGE dataset. Finally, we
sampled 10,000 times from the C4 and MU datasets with each sample consisting of
15,527 cisgender males, 22,375 cisgender females, and 52 transgender and gender-
diverse individuals to match the LifeLines dataset. In each sample, we calculated
adjusted ORs using logistic regression. We then calculated the empirical pvalues for
the adjusted ORs for the IMAGE, APHS, and LifeLines samples from the distribution
of ORs generated in the 10,000 samples from MU and C4. We corrected for the six
tests using Bonferroni correction (empirical pvalue alpha =0.008).
Statistical analyses: rates of other neurodevelopmental and psychiatric con-
ditions. In the C4 and MU datasets we investigated if diagnosis of six neurode-
velopmental and psychiatric conditions differed by gender using χ2tests (Model 1)
and logistic regression accounting for educational attainment and age (Model 2).
Additionally, we repeated Model 2 after excluding autistic individuals (Model 3), as
there may be an autism-based ascertainment bias in these cohorts. Each model was
conducted rst by using three gender categories (transgender and gender-diverse,
cisgender male, and cisgender female), and then two categories (transgender and
gender-diverse and cisgender).
We also investigated the relative association between each neurodevelopmental
and psychiatric conditions to gender identity. Gender identity (transgender and
gender-diverse versus cisgender) was the dependent variable. The independent
variables were diagnosis of ADHD, autism, bipolar disorder, depression, learning
disorder (only in C4 dataset), OCD, and schizophrenia. Age and educational
attainment were included as covariates.
Statistical analyses: traits related to autism. In the C4 dataset, we investigated
differences in scores by gender (cisgender males, cisgender females, and trans-
gender and gender-diverse) on the four measures using ANOVA and then con-
ducted post-hoc T-tests. We repeated the analyses using linear regression
accounting for age and educational attainment. Distributions in Brain Types
between the three genders were investigated using χ2tests. Validation using the
AQ-5077 was conducted in the IMAGE dataset, and using the AQ-10 was con-
ducted in the LifeLines dataset.
Statistical analyses: calculation of Brain Types. Calculation of Brain Types
was only done in the C4 dataset. We rst calculated the standardized scores of the
SQ-10 and the EQ-10. This was done by subtracting the mean of the SQ-10 and the
EQ-10 (means were calculated using only non-autistic individuals from the C4
dataset) from each individuals score and then dividing by the maximum possible
score (20 for both the SQ-10 and the EQ-10). We next calculated a D-scoreby
subtracting the standardized EQ-10 score from the SQ-10 score. We then divided
individuals into ve Brain Types based on D-score percentiles. The lowest 2.5th
percentile was Extreme Type E and the highest 2.5th percentile was Extreme Type
S. Those scoring between the 35th and 65th percentiles were classied as Type B.
Participants who scored between the 2.5th and 35th percentiles were Type E, and
Type S was dened by scoring between the 65th and 97.5th percentile.
Statistical analyses: multiple testing correction. Across all the datasets and the
three aims and the exploratory aim, we conducted at least 182 different analyses.
Given the size of the datasets used, the standard errors are low. We thus dene a
study-wide p-value of 0.0002 to correct for all the tests. Details of the tests con-
ducted are provided in Supplementary Table S11.
Statistical analyses: power calculations in the LifeLines dataset. Given the
relatively low number of transgender and gender-diverse individuals, we conducted
power calculations to investigate if the LifeLines cohort had sufcient statistical
power to identify effects. We used effect sizes obtained from the results of the C4
dataset as this was the largest dataset, and hence, likely to have effects that are least
affected by winners curse (Supplementary Methods). Power calculations sug-
gested that we were underpowered to detect effects at an alpha of 0.05 for calcu-
lating ORs using logistic regression, with power achieved between 0.62 (reference
group: cisgender males)0.69 (reference group: cisgender females). However, we
proceeded with the analyses to identify if the effects observed were in the same
direction as those observed in other datasets.
Reporting summary. Further information on research design is available in the Nature
Research Reporting Summary linked to this article.
Data availability
As participants did not consent for their data to be publicly shared, even anonymized,
data will be made available to only potential collaborators with ethical approval after they
submit a research proposal to the Autism Research Centre, University of Cambridge, UK
for four of the datasets (C4, MU, IMAGE, and APHS). Data for LifeLines can be obtained
by making an application to the LifeLines Biobank (https://www.lifelines.nl/researcher).
A reporting summary for this Article is available as a Supplementary information le.
Code availability
Scripts are provided at: https://github.com/autism-research-centre/Atypical-gender-and-
autism. All analyses were conducted using R version 3.4.4 (2018-03-15).
Received: 27 August 2019; Accepted: 15 July 2020;
References
1. American Psychiatric Association. The Diagnostic and Statistical Manual 5th
edn. (2013).
2. Elsabbagh, M. et al. Global prevalence of autism and other pervasive
developmental disorders. Autism Res. 5, 160179 (2012).
3. Baio, J. et al. Prevalence of autism spectrum disorder among children aged 8
yearsautism and developmental disabilities monitoring network, 11 Sites,
United States, 2014. Mmwr. Surveill. Summ. 67,123 (2018).
4. Xu, G., Strathearn, L., Liu, B. & Bao, W. Prevalence of autism spectrum
disorder among US children and adolescents, 20142016. JAMA 319,8182
(2018).
5. Baird, G. et al. Prevalence of disorders of the autism spectrum in a population
cohort of children in South Thames: the Special Needs and Autism Project
(SNAP). Lancet 368, 210215 (2006).
6. Brugha, T. S. et al. Epidemiology of autism spectrum disorders in adults in the
community in England. Arch. Gen. Psychiatry 68, 459 (2011).
7. Brugha, T. S. et al. Epidemiology of autism in adults across age groups and
ability levels. Br. J. Psychiatry 209, 498503 (2016).
8. Baxter, A. J. et al. The epidemiology and global burden of autism spectrum
disorders. Psychol. Med. 45, 601613 (2015).
9. Meerwijk, E. L. & Sevelius, J. M. Transgender population size in the united
states: a meta-regression of population-based probability samples. Am. J.
Public Health 107,e1e8 (2017).
10. Zucker, K. J. Epidemiology of gender dysphoria and transgender identity. Sex.
Health 14, 404 (2017).
11. Collin, L., Reisner, S. L., Tangpricha, V. & Goodman, M. Prevalence of
transgender depends on the casedenition: a systematic review. J. Sex. Med.
13, 613626 (2016).
12. Van Der Miesen, A. I. R., Hurley, H. & De Vries, A. L. C. Gender dysphoria
and autism spectrum disorder: a narrative review. Int. Rev. Psychiatry 28,
7080 (2016).
13. Øien, R. A., Cicchetti, D. V. & Nordahl-Hansen, A. Gender dysphoria,
sexuality and autism spectrum disorders: a systematic map review. J. Autism
Dev. Disord. 48, 40284037 (2018).
14. Janssen, A., Huang, H. & Duncan, C. Gender variance among youth with
autism spectrum disorders: a retrospective chart review. Transgender Heal. 1,
6368 (2016).
15. May, T., Pang, K. & Williams, K. J. Gender variance in children and
adolescents with autism spectrum disorder from the National Database for
Autism. Res. Int. J. Transgenderism 18,715 (2017).
16. Strang, J. F. et al. Increased gender variance in autism spectrum disorders and
attention decit hyperactivity disorder. Arch. Sex. Behav. 43, 15251533
(2014).
17. Hisle-Gorman, E. et al. Gender dysphoria in children with autism spectrum
disorder. LGBT Heal. 6,95100 (2019).
18. Nabbijohn, A. N. et al. Gender variance and the autism spectrum: an
examination of children ages 612 years. J. Autism Dev. Disord. 49, 15701585
(2019).
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
10 NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
19. Bejerot, S. & Eriksson, J. M. Sexuality and gender role in autism spectrum
disorder: a case control study. PLoS One 9, e87961 (2014).
20. George, R. & Stokes, M. A. Gender identity and sexual orientation in autism
spectrum disorder. Autism 22, 970982 (2018).
21. de Vries, A. L. C., Noens, I. L. J., Cohen-Kettenis, P. T., van Berckelaer-Onnes,
I. A. & Doreleijers, T. A. Autism spectrum disorders in gender dysphoric
children and adolescents. J. Autism Dev. Disord. 40, 930936 (2010).
22. Kaltiala-Heino, R., Sumia, M., Työläjärvi, M. & Lindberg, N. Two years of
gender identity service for minors: overrepresentation of natal girls with severe
problems in adolescent development. Child Adolesc. Psychiatry Ment. Health
9, 9 (2015).
23. Shumer, D. E., Reisner, S. L., Edwards-Leeper, L. & Tishelman, A. Evaluation
of asperger syndrome in youth presenting to a gender dysphoria clinic. LGBT
Heal. 3, 387390 (2016).
24. Heylens, G. et al. The co-occurrence of gender dysphoria and autism spectrum
disorder in adults: an analysis of cross-sectional and clinical chart data. J.
Autism Dev. Disord. 48, 22172223 (2018).
25. Cheung, A. S. et al. Sociodemographic and clinical characteristics of
transgender adults in Australia. Transgender Heal 3, 229238 (2018).
26. Akgül, G. Y., Ayaz, A. B., Yildirim, B. & Fis, N. P. Autistic traits and executive
functions in children and adolescents with gender dysphoria. J. Sex Marital
Ther.18, https://doi.org/10.1080/0092623X.2018.1437489 (2018).
27. Jones, R. M. et al. Brief report: female-to-male transsexual people and autistic
traits. J. Autism Dev. Disord. 42, 301306 (2012).
28. Kristensen, Z. E. & Broome, M. R. Autistic traits in an internet sample of
gender variant UK adults. Int. J. Transgenderism 16, 234245 (2015).
29. Pasterski, V., Gilligan, L. & Curtis, R. Traits of autism spectrum disorders in
adults with gender dysphoria. Arch. Sex. Behav. 43, 387393 (2014).
30. Shumer, D. E., Roberts, A. L., Reisner, S. L., Lyall, K. & Austin, S. B. Brief
report: autistic traits in mothers and children associated with childs gender
nonconformity. J. Autism Dev. Disord. 45, 14891494 (2015).
31. Skagerberg, E., Di Ceglie, D. & Carmichael, P. Brief report: autistic features in
children and adolescents with gender dysphoria. J. Autism Dev. Disord. 45,
26282632 (2015).
32. van der Miesen, A. I. R., de Vries, A. L. C., Steensma, T. D. & Hartman, C. A.
Autistic symptoms in children and adolescents with gender dysphoria. J.
Autism Dev. Disord. 48, 15371548 (2018).
33. Zucker, K. J. et al. Intense/obsessional interests in children with gender
dysphoria: a cross-validation study using the Teachers Report Form. Child
Adolesc. Psychiatry Ment. Health 11, 51 (2017).
34. Nobili, A. et al. Autistic traits in treatment-seeking transgender adults. J.
Autism Dev. Disord. 48, 39843994 (2018).
35. Zhong, H. & Prentice, R. L. Bias-reduced estimators and condence intervals
for odds ratios in genome-wide association studies. Biostatistics 9, 621634
(2008).
36. Zhong, H. & Prentice, R. L. Correcting winners cursein odds ratios from
genomewide association ndings for major complex human diseases. Genet.
Epidemiol. 34, 78 (2010).
37. Jones, B. A., Pierre Bouman, W., Haycraft, E. & Arcelus, J. Mental health and
quality of life in non-binary transgender adults: a case control study. Int. J.
Transgenderism 20, 251262 (2019).
38. Davey, A., Bouman, W. P., Arcelus, J. & Meyer, C. Social support and
psychological well-being in gender dysphoria: a comparison of patients with
matched controls. J. Sex. Med. 11, 29762985 (2014).
39. Arcelus, J., Claes, L., Witcomb, G. L., Marshall, E. & Bouman, W. P. Risk
factors for non-suicidal self-injury among trans youth. J. Sex. Med. 13,
402412 (2016).
40. Reisner, S. L. et al. Global health burden and needs of transgender
populations: a review. Lancet 388, 412436 (2016).
41. Nuttbrock, L. et al. Gender abuse and major depression among transgender
women: a prospective study of vulnerability and resilience. Am. J. Public
Health 104, 21912198 (2014).
42. Thorne, N. et al. A comparison of mental health symptomatology and levels of
social support in young treatment seeking transgender individuals who
identify as binary and non-binary. International Journal of Transgenderism
https://doi.org/10.1080/15532739.2018.1452660 (2018).
43. Nuttbrock, L. et al. Gender abuse, depressive symptoms, and substance use
among transgender women: a 3-year prospective study. Am. J. Public Health
104, 21992206 (2014).
44. Clark, T. C. et al. The health and well-being of transgender high school
students: results from the New Zealand Adolescent Health Survey (Youth12).
J. Adolesc. Heal. 55,9399 (2014).
45. Dhejne, C., Van Vlerken, R., Heylens, G. & Arcelus, J. Mental health and
gender dysphoria: a review of the literature. Int. Rev. Psychiatry 28,4457
(2016).
46. Rosen, T. E., Mazefsky, C. A., Vasa, R. A. & Lerner, M. D. Co-occurring
psychiatric conditions in autism spectrum disorder. Int. Rev. Psychiatry 30,
4061 (2018).
47. Lai, M.-C. et al. Prevalence of co-occurring mental health diagnoses in the
autism population: a systematic review and meta-analysis. Lancet Psychiatry 6,
819829 (2019).
48. Lai, M.-C., Lombardo, M. V., Auyeung, B., Chakrabarti, B. & Baron-Cohen, S.
Sex/gender differences and autism: setting the scene for future research. J. Am.
Acad. Child Adolesc. Psychiatry 54,1124 (2015).
49. Loomes, R., Hull, L. & Mandy, W. P. L. What is the male-to-female ratio in
autism spectrum disorder? A systematic review and meta-analysis. J. Am.
Acad. Child Adolesc. Psychiatry 56, 466474 (2017).
50. Lai, M.-C. & Szatmari, P. Sex and gender impacts on the behavioural
presentation and recognition of autism. Curr. Opin. Psychol. 33, 117123
(2020).
51. Greenberg, D. M., Warrier, V., Allison, C. & Baron-Cohen, S. Testing the
empathizing-systemizing theory of sex differences and the Extreme Male
Brain theory of autism in half a million people. Proc. Natl Acad. Sci. U.S.A.
201811032, https://doi.org/10.1073/pnas.1811032115 (2018).
52. Taylor, A. E. et al. Exploring the association of genetic factors with
participation in the Avon Longitudinal Study of Parents and Children. Int. J.
Epidemiol. 47, 12071216 (2018).
53. Adams, M. J. et al. Factors associated with sharing e-mail information and
mental health survey participation in large population cohorts. Int. J.
Epidemiol. https://doi.org/10.1093/ije/dyz134 (2019).
54. Happé, F., Ronald, A. & Plomin, R. Time to give up on a single explanation for
autism. Nat. Neurosci. 9, 12181220 (2006).
55. Warrier, V. et al. Systemizing is genetically correlated with autism and is
genetically distinct from social autistic traits. bioRxiv 228254, https://doi.org/
10.1101/228254 (2017).
56. Hines, M. Gender development and the human brain. Annu. Rev. Neurosci.
34,6988 (2011).
57. Baron-Cohen, S. et al. Elevated fetal steroidogenic activity in autism. Mol.
Psychiatry 20, 369376 (2015).
58. Auyeung, B. et al. Fetal testosterone and autistic traits. Br. J. Psychol. 100,122
(2009).
59. Baron-Cohen, S. et al. Why are autism spectrum conditions more prevalent in
males? PLoS Biol. 9, e1001081 (2011).
60. Baron-Cohen, S. et al. Foetal oestrogens and autism. Mol. Psychiatry 19,
https://doi.org/10.1038/s41380-019-0454-9 (2019).
61. Cross-Disorder Group of the Psychiatric Genomics Consortium. Genomic
relationships, novel loci, and pleiotropic mechanisms across eight psychiatric
disorders. Cell 179, 14691482.e11 (2019).
62. Niemi, M. E. K. et al. Common genetic variants contribute to risk of rare
severe neurodevelopmental disorders. Nature 562, 268271 (2018).
63. Winter, S. et al. Transgender people: health at the margins of society. Lancet
388, 390400 (2016).
64. Chew, D. et al. Youths with a non-binary gender identity: a review of their
sociodemographic and clinical prole. Lancet Child Adolesc. Heal. https://doi.
org/10.1016/S2352-4642(19)30403-1 (2020).
65. Strang, J. F. et al. They thought it was an obsession: trajectories and
perspectives of autistic transgender and gender-diverse adolescents. J. Autism
Dev. Disord.48, 40394055.
66. Grifths, S. et al. The Vulnerability Experiences Quotient (VEQ): a study of
vulnerability, mental health and life satisfaction in autistic adults. Autism Res.
aur.2162, https://doi.org/10.1002/aur.2162 (2019).
67. Cassidy, S., Bradley, L., Shaw, R. & Baron-Cohen, S. Risk markers for
suicidality in autistic adults. Mol. Autism 9, 42 (2018).
68. Cassidy, S. et al. Suicidal ideation and suicide plans or attempts in adults with
Aspergers syndrome attending a specialist diagnostic clinic: a clinical cohort
study. Lancet Psychiatry 1, 142147 (2014).
69. Strang, J. F. et al. Initial clinical guidelines for co-occurring autism spectrum
disorder and gender dysphoria or incongruence in adolescents. J. Clin. Child
Adolesc. Psychol. 47, 105115 (2018).
70. Head, A. M., McGillivray, J. A. & Stokes, M. A. Gender differences in
emotionality and sociability in children with autism spectrum disorders. Mol.
Autism 5, 19 (2014).
71. Frazier, T. W., Georgiades, S., Bishop, S. L. & Hardan, A. Y. Behavioral and
cognitive characteristics of females and males with autism in the Simons
simplex collection. J. Am. Acad. Child Adolesc. Psychiatry 53, 329340.e3
(2014).
72. Scholtens, S. et al. Cohort prole: LifeLines, a three-generation cohort study
and biobank. Int. J. Epidemiol. 44, 11721180 (2015).
73. Allison, C., Auyeung, B., Baron-Cohen, S., Bolton, P. F. & Brayne, C. Toward
brief Red Flagsfor autism screening: the Short Autism Spectrum Quotient
and the Short Quantitative Checklist for Autism in toddlers in 1000 cases and
3000 controls [corrected]. J. Am. Acad. Child Adolesc. Psychiatry 51, 202212.
e7 (2012).
74. Wheelwright, S. J. et al. Predicting Autism Spectrum Quotient (AQ) from the
Systemizing Quotient-Revised (SQ-R) and Empathy Quotient (EQ). Brain Res.
1079,4756 (2006).
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1 ARTICLE
NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications 11
Content courtesy of Springer Nature, terms of use apply. Rights reserved
75. Baron-Cohen, S., Richler, J., Bisarya, D., Gurunathan, N. & Wheelwright, S. J.
The systemizing quotient: an investigation of adults with Asperger syndrome
or high-functioning autism, and normal sex differences. Philos. Trans. R. Soc.
Lond. B. Biol. Sci. 358, 361374 (2003).
76. Baron-Cohen, S. Autism: the Empathizing-Systemizing (E-S) Theory. Ann. N.
Y. Acad. Sci. 1156,6880 (2009).
77. Baron-Cohen, S., Wheelwright, S. J., Skinner, R., Martin, J. & Clubley, E. The
autism-spectrum quotient (AQ): evidence from Asperger syndrome/high-
functioning autism, males and females, scientists and mathematicians. J.
Autism Dev. Disord. 31,517 (2001).
78. Schendel, D. E. & Thorsteinsson, E. Cumulative incidence of autism into
adulthood for birth cohorts in Denmark, 19802012. JAMA 320, 1811
(2018).
79. Lai, M.-C. & Baron-Cohen, S. Identifying the lost generation of adults
with autism spectrum conditions. Lancet Psychiatry 2, 10131027
(2015).
80. Altman, D. G. & Bland, J. M. Statistics notes: standard deviations and standard
errors. Br. Med. J. 331, 903 (2005).
Acknowledgements
This study was supported by the Medical Research Council (MRC), the Wellcome
Trust (214322/Z/18/Z), the Templeton World Charity Foundatio n, the Autism
Research Trust, and the National Institute of Health Research (NIHR) Collaboration
for Leadership in Applied Health Research and Care-East of England (CLAHRC-EoE).
The views expressed are those of the authors and not necessarily those of the NHS, the
NIHR or the Department of Health. The authors also received funding from the
Innovative Medicines Initiative 2 Joint Undertaking (JU) under grant agreement No
777394. The JU receives support from the European Unions Horizon 2020 research
and innovatio n program and EFPIA and Autism Speaks, Autistica, SFARI. Funding for
the Autism and Physical Health Survey was provided by the Autism Research Trust, the
Rosetrees Trust, the Cambridgeshire and Peterborough NHS Foundation Trust, and the
Corbin Charitable Trust. Thanks also to the Cambridge Autism Research Database,
Autisticas Discover Network, and various autism sup port groups and charitie s for
assisting our recruitment for the APHS. Varun Warrier is supported by the Bowring
Research Fellowship at St. Catharines College, Cambridge. D.M.G. was supported in
part by the Zuckerman STEM Leadership Program. M.-C.L. is supported by the
Academic Scholars Award from the Department of Psychiatry, University of Toronto,
the Ontario Brain Institute via the Provin ce of Ontario Neurodevelopmental Disorders
(POND) Network (IDS-I l-02), the Canadian Institutes of Health Research (CIHR)
(PJT 159578 and a CIHR Sex and Gender Science Chair, GSB-171373), and the Slaight
Family Child and Youth Mental Health Innovation Fund via the CAMH Foundation.
We are grateful to all the participants, and for Channel 4 for sharing the anonymized
data with us.
Author contributions
V.W. conducted the analyses. V.W. and S.B.-C. designed the study. V.W., D.M.G., E.W.,
C.B., P.L.S. collected the data. V.W., D.M.G., E.W., C.B., P.L.S., M.-C.-L., C.L.A., and
S.B.-C. interpreted the data, wrote, read, and edited the paper.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41467-
020-17794-1.
Correspondence and requests for materials should be addressed to V.W. or S.B.-C.
Peer review information Nature Communications thanks John Strang, Mark Stokes,
Traolach (Terry) Brugha and Lisa Wiggins for their contribution to the peer review of
this work. Peer reviewer reports are available.
Reprints and permission information is available at http://www.nature.com/reprints
Publishers note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the articles Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
articles Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
© The Author(s) 2020
ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-17794-1
12 NATURE COMMUNICATIONS | (2020) 11:3959 | https://doi.org/10.1038/s41467-020-17794-1 | www.nature.com/naturecommunications
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
... La condición transgénero forma parte de las diversidades sexuales y genéricas, a partir del concepto de la identidad de género (Basante Ballesteros & Ortiz Quevedo, 2021;Warrier, Greenberg, Weir, Buckingham, Smith, Lai, Allison & Baron-Cohen, 2020). La persona cuya identidad y expresión de género no coincide con la que es esperada según el sexo de hombre o mujer que le fue asignado al nacer, es una persona transgénero. ...
... De acuerdo a lo mencionado, debe entenderse que, a pesar de haber cambiado la categoría diagnóstica específica a la generalidad del TEA (Autismo Diario, 2019), todavía se entiende al síndrome de Asperger como un denominativo de tipo social e identitario. En ese sentido, como nos explica Warrier et al. (2020), el síndrome de Asperger es una condición de temprana aparición que, según la Confederación de Autismo de España (Autismo Diario, 2019), se define en dos áreas: 1) las barreras que existen en la comunicación social, en cuanto a los matices de la conversación y, 2) la dificultad para pensar y actuar con flexibilidad. ...
... Planteamos la nombrada intersección de temas -síndrome de Asperger y condición transgénero, desde un ángulo complejo: es necesario comprender que sobre ellos no existe bibliografía muy amplia y actualizada, habiéndose realizado un número reducido de estudios, la mayoría de tipo clínico y con poblaciones pequeñas (Warrier et al., 2020). No obstante, nuestro punto de partida es entender que, tanto las personas transgénero como aquellas que presentan el Síndrome de Asperger, perciben la realidad de manera distinta, tanto en relación a su propio cuerpo como a sí mismos (Attwood, 1997;Espinoza et al., 2019), en cuanto al sentimiento de extrañeza de 'no encajar' (Warrier et al., 2020) y al mundo que los rodea, acerca la presión social y los juicios negativos, que puede entenderse en el sentido de estigma social (Yee, 2021, Missé, 2018. ...
Article
Full-text available
Los objetivos del presente artículo están centrados en el estudio de caso de un adolescente transgénero y con Síndrome de Asperger, abordado desde el enfoque psicológico de la Terapia Alfa, desde sus respectivas etapas. La metodología de este texto es de tipo cualitativa, específicamente desde el enfoque interpretativo del interaccionismo simbólico para entender los conceptos articuladores del análisis: cuerpo, percepción y realidad. Entre las conclusiones tenemos que existe una interdependencia entre tales conceptos, a partir de la dualidad de identidades que presenta el sujeto, funcionando todo como un sistema sobre el que se sustenta el desarrollo del proceso terapéutico y el propio individuo.
... Particulièrement, et bien que cela puisse être discuté par certaines études (ex : [76]) 8 , les femmes autistes pourraient être moins susceptibles de présenter des comportements externalisants, tels que l'hyperactivité/impulsivité et les problèmes de comportement, et être, en revanche, plus enclines à présenter des troubles concomitants plus intériorisés, tels que l'anxiété, la dépression et les troubles de l'alimentation, avec bien préférentiellement l'anorexie bien que d'allure atypique [64,95], et ont davantage de symptômes sensoriels [78], comparativement aux hommes autistes. Par ailleurs, il existe des hypothèses suivant lesquelles les femmes TSA seraient plus susceptibles de développer des tendances dysphoriques liées au genre (personnes transgenres ou ayant une diversité de genre) ou des questionnements sur l'orientation sexuelle [55,76,[96][97][98]. ...
... Non-binary autistic young people were not represented in this study. Considering that more autistic people identify as non-binary than non-autistic people (Warrier et al., 2020) and report gender-diverse experiences (Kourti & MacLeod, 2019), future research should endeavour to include non-binary autistic young people. The majority of interviews were carried out remotely. ...
Article
Full-text available
Background: Many autistic young people use online devices for social connection and to share interests. However, there is limited research regarding autistic online safety behaviours. Compared with non-autistic children, parental surveys have indicated that autistic young people are less likely to block people and/or online sites. To date, no research has explored autistic young people’s perceptions of their online safety experiences. This qualitative research explored autistic young people’s experiences of communicating with others online, as well as their online safety experiences. Method: Semi-structured interviews were conducted with 14 autistic young people aged 11–17 years (M = 14.0, SD = 2.2), including 8 males (M = 13.9, SD = 2.1) and 6 females (M = 14.5, SD = 2.5). These were conducted face to face (n = 1), phone call (n = 2), or via Skype (n = 8) or live web chat (n = 3). Questions explored factors relating to autistic young people’s online safety experiences. Results: Interpretative Phenomenological Analysis was used to analyse the data. In line with previous studies, autistic young people reported being victims of cyberbullying. Young autistic females reported being subject to online sexual harassment. While participants’ online experi- ences varied, there were commonalities, including a desire for more support to block online comments and/or individuals. Conclusions: Our results support previous findings that autistic young people are subject to online harassment and are not confident blocking unwanted contact from others online. Future in- terventions will be more readily accepted and ecologically valid if they address the unique needs of autistic young people.
... We chose to use identity-first language for this paper. Gender is distinct from sex and has important variance, especially in the autistic community [86,87]. For language around sex and gender, we exclusively considered sex (meaning sex designated at birth). ...
Article
Full-text available
The complexity of autism’s phenotypic spectra is well-known, yet most genetic research uses case-control status as the target trait. It is undetermined if autistic symptom domain severity underlying this heterogeneity is heritable and pleiotropic with other psychiatric and behavior traits in the same manner as autism case-control status. In N = 6064 autistic children in the SPARK cohort, we investigated the common genetic properties of twelve subscales from three clinical autism instruments measuring autistic traits: the Social Communication Questionnaire (SCQ), the Repetitive Behavior Scale-Revised (RBS-R), and the Developmental Coordination Disorder Questionnaire (DCDQ). Educational attainment polygenic scores (PGS) were significantly negatively correlated with eleven subscales, while ADHD and major depression PGS were positively correlated with ten and eight of the autism subscales, respectively. Loneliness and neuroticism PGS were also positively correlated with many subscales. Significant PGS by sex interactions were found—surprisingly, the autism case-control PGS was negatively correlated in females and had no strong correlation in males. SNP-heritability of the DCDQ subscales ranged from 0.04 to 0.08, RBS-R subscales ranged from 0.09 to 0.24, and SCQ subscales ranged from 0 to 0.12. GWAS in SPARK followed by estimation of polygenic scores (PGS) in the typically-developing ABCD cohort (N = 5285), revealed significant associations of RBS-R subscale PGS with autism-related behavioral traits, with several subscale PGS more strongly correlated than the autism case-control PGS. Overall, our analyses suggest that the clinical autism subscale traits show variability in SNP-heritability, PGS associations, and significant PGS by sex interactions, underscoring the heterogeneity in autistic traits at a genetic level. Furthermore, of the three instruments investigated, the RBS-R shows the greatest evidence of genetic signal in both (1) autistic samples (greater heritability) and (2) general population samples (strongest PGS associations).
... However, this is not mirrored in other personality traits that characterise males more than females, such as aggression or sociopathy. It is also worth noting that autistic people also show higher rates of gender identity incongruence, as well as atypical sexual orientation, compared to their peers (Bejerot and Eriksson 2014;Warrier et al. 2020). In humans, the development of gender identity and sexual orientation depend on many different factors, including social learning, which may be altered in autistic people. ...
Thesis
Autism is a neurodevelopmental condition that is more frequently diagnosed in males than females. To explain this, in 2014, the prenatal sex steroid theory was proposed. This extended the fetal testosterone theory, published in 2004. The prenatal sex steroid theory proposes that exposure to higher levels of prenatal sex steroids (e.g., prenatal androgens and estrogens) that are on average higher in male fetuses are associated with higher likelihood for autism and elevated autistic traits. This background literature is reported in Chapter 1. In this thesis, eight novel studies are reported that test and extend the prenatal sex steroid theory by investigating perinatal factors related to sex differences in physiology. Study 1 (described in Chapter 2) reports a case-control analysis of steroid levels in the amniotic fluid of males who were later diagnosed as autistic, linked with the Danish Biobank (n = 98 cases, n = 177 controls). This included univariate analyses of both prenatal androgens and estrogens, as well as the aromatisation ratio. All estrogens, but not testosterone, on average were elevated in autistic males. Study 2 (described in Chapter 3) reports a prospective cohort study (the Cambridge Ultrasound and Pregnancy [CUSP] study) of pregnant women and their infants in Cambridge (n=219), who were assessed for their autistic traits during pregnancy and late infancy. Steroid hormone levels were assessed in maternal serum. Estradiol levels correlated with both maternal autistic traits and the male infants’ autistic traits, but there was no correlation with female infants’ autistic traits. Study 3 (described in Chapter 4) reports a large prospective cohort study in Rotterdam (Generation-R) that studied the levels of placental function markers in maternal serum (n=3469), their sex differences in the general population, their association with both autistic traits in childhood (assessed using the Social Responsiveness Scale - SRS), and with likelihood for autism in males. Male-like patterns in placental angiogenic markers, high placental growth factor (PlGF) and low soluble fms-like tyrosine kinase-1 (sFlt-1) levels, respectively correlated with higher autistic traits in females and an autism diagnosis in males. Chapter 5 describes Studies 4, 5, and 6, all based on a longitudinal cohort, the Cambridge Human Infant Longitudinal Development [CHILD] Study. This included prenatal (n=41) and postnatal (n=27) brain MRI imaging and salivary testosterone measurements during mini-puberty. Study 4 found that both male and female infants experienced transient increases in testosterone postnatally (2 to 6 months), but this did not correlate to their autistic traits at 18 months. Study 5 focused on total brain volume and surface area in infancy, as well as rate of brain growth perinatally, all of which correlated negatively with the infant’s autistic traits. Study 6 found that this was driven by low volume in regions that show sex differences and are involved in face recognition. Chapter 6 describes two genetic studies, which found that autism-related genetic variance (rare and common variance respectively) overlaps with X-linked genes that show sex differences in the placenta (Study 7) and correlates with the genetics for early age of menarche (Study 8). Chapter 7 brings all of the findings from Studies 1 to 8 together to draw conclusions and consider limitations and future directions. Based on these analyses, I then propose a new theory on the role of the placenta in mediating sex differences in human perinatal development and autism.
... This includes differences in core diagnostic features (core features) 1,3,4 and associated features such as IQ, adaptive behavior and motor coordination, all of which have an impact on life outcomes 3,5,6 . Furthermore, sex and gender 7,8 and co-occurring ID and developmental, behavioral and medical conditions 9,10 alter the presentation and measurement of core autism features. While a few studies have attempted to investigate the genetic influences on this heterogeneity [11][12][13][14][15][16][17][18] , substantial gaps remain. ...
Article
Full-text available
The substantial phenotypic heterogeneity in autism limits our understanding of its genetic etiology. To address this gap, here we investigated genetic differences between autistic individuals ( n max = 12,893) based on core and associated features of autism, co-occurring developmental disabilities and sex. We conducted a comprehensive factor analysis of core autism features in autistic individuals and identified six factors. Common genetic variants were associated with the core factors, but de novo variants were not. We found that higher autism polygenic scores (PGS) were associated with lower likelihood of co-occurring developmental disabilities in autistic individuals. Furthermore, in autistic individuals without co-occurring intellectual disability (ID), autism PGS are overinherited by autistic females compared to males. Finally, we observed higher SNP heritability for autistic males and for autistic individuals without ID. Deeper phenotypic characterization will be critical in determining how the complex underlying genetics shape cognition, behavior and co-occurring conditions in autism.
... That said, emerging research suggests that TGD individuals may be more likely to be diagnosed on the autism spectrum (Cheung et al., 2018;Heylens et al., 2018;Murphy et al., 2020) or with attention deficit/hyperactivity disorder (Cheung et al., 2018;Dawson et al., 2017) and that TGD individuals with neurodevelopmental disorders experience more severe depression and anxiety (Murphy et al., 2020;Warrier et al., 2020). Previous studies have also found that TGD individuals are more likely to experience trauma, including traumatic events that are most likely to lead to PTSD, and that they have greater incidence of PTSD (Breslau, 2001;Shipherd et al., 2011). ...
Article
Individuals who are transgender and gender diverse (TGD) are more likely to suffer from and to seek mental health services for mood disorders. Some literature suggests that TGD individuals, because of pervasive and systemic minority stress, may have more complex clinical presentations (i.e., psychiatric conditions and severity of symptoms) and may benefit from empirically based treatments to a lesser degree than their cisgender peers. However, research has yet to examine individuals who are TGD receiving treatment in specialized, intensive mood disorder treatment despite the propensity for them to be diagnosed with and treated for mood disorders. Using a sample of 1,326 adult patients in intensive mood disorder treatment (3.8% TGD), the clinical presentation and treatment outcomes were compared between patients who are TGD and cisgender. Contrary to previous research, TGD patients were largely similar if not healthier than their cisgender counterparts, including similar depression, quality of life, emotion dysregulation, and behavioral activation, and less severe rumination at admission. Despite similar to better reported mental health symptoms, TGD patients were diagnosed with more psychiatric conditions overall, including greater prevalence of social anxiety and neurodevelopmental diagnoses. Those who are TGD did not experience attenuated treatment response as predicted. Findings suggest that patients in intensive mood disorder treatment who are TGD may be more resilient than previously assumed, or supports may have increased to buffer effects of stigma on mental health, and emphasize the need to exercise discretion and sensitivity in diagnostic practices to prevent over-diagnosis and pathologizing of TGD individuals.
... It may also be pertinent to study sex-specific differences as we found some nuanced differences. Emerging evidence suggests that some autistic individuals do not fall into the confines of binary gender identification and thus, such identities should be accommodated if data are available (Warrier et al., 2020). Replication studies in cohorts of comparable or larger sample sizes and with less social stratification are recommended, particularly those with larger numbers of autism cases, to further assess whether autism is associated with increased or decreased risk of MRBs. ...
Article
Full-text available
Background: Multiple risk behaviours (MRBs), typically beginning in adolescence, are associated with increased risk of adverse health and social outcomes. The association between autism and MRBs is little understood. Methods: Data were from the Avon Longitudinal Study of Parents and Children, an UK-based longitudinal, birth cohort study. Exposures were diagnosed autism and four autistic traits: social communication difficulties, pragmatic language, repetitive behaviours and reduced sociability. Outcomes were participation in up to 14 risk behaviours, including alcohol consumption, smoking, risky sexual behaviours and physical inactivity. Outcome data were collected at ages approximately 12, 14, 16 and 18. Results: Up to 4300 participants were included in latent basis growth curve analyses with adjustment for confounders. Social communication difficulties were associated with an above average level of MRBs engagement at ~12 years (mean difference β 0.26; 95% CI 0.13-0.40), and above average rate of engagement from ages ~12-18 (β 0.08; 95% CI 0.02-0.13). Repetitive behaviours were associated with above average levels of engagement in MRBs at ~12 years (β 0.24; 95% CI 0.09-0.38). Contrastingly, reduced sociability was associated with a reduced rate of engagement in MRBs from ages ~12-18 (β -0.06; 95% CI -0.11 to -0.02). In sex-specific analyses, persisting differences in MRB engagement patterns from ages ~12-18 were observed in males with social communication difficulties and females with reduced sociability temperament. Conclusions: Having elevated levels of some autistic traits appear to have differentiated effects on MRB engagement patterns. These findings could reflect difficulties fitting in and/or coping mechanisms relating to difficulties with fitting in.
Article
Few studies have examined self‐reported perceived stress in autistic adults. Existing studies have included relatively small, predominantly male samples and have not included older autistic adults. Using a large autistic sample (N = 713), enriched for individuals designated female at birth (59.3%), and spanning younger, middle, and older adulthood, we examined perceived stress and its associations with independence in activities of daily living and subjective quality of life (QoL). Perceived stress for autistic adults designated male or female at birth was compared to their same birth‐sex counterparts in a general population sample. In addition, within the autistic sample, effects of sex designated at birth, age, and their interaction were examined. Regression modeling examined associations between perceived stress and independence in activities of daily living and domains of subjective QoL in autistic adults, after controlling for age, sex designated at birth, and household income. Autistic adults reported significantly greater perceived stress than a general population comparison sample. Relative to autistic adults designated male at birth, those designated female at birth demonstrated significantly elevated perceived stress. Perceived stress contributed significantly to all regression models, with greater perceived stress associated with less independence in activities of daily living, and poorer subjective QoL across all domains—Physical, Psychological, Social, Environment, and Autism‐related QoL. Findings are contextualized within the literature documenting that autistic individuals experience elevated underemployment and unemployment, heightened rates of adverse life events, and increased exposure to minority stress. This study looked at self‐reported perceived stress in a large sample of autistic adults. Autistic adults reported more perceived stress than non‐autistic adults. Autistic individuals designated female at birth reported higher stress than autistic individuals designated male at birth. In autistic adults, greater perceived stress is related to less independence in activities of daily living and poorer subjective quality of life.
Article
Full-text available
The suggested overlap between autism spectrum disorder (ASD) and gender dysphoria/incongruence (GD/GI) has been much disputed. This review showed a relationship between ASD traits and GD feelings in the general population and a high prevalence of GD/GI in ASD. Our meta-analyses revealed that the pooled estimate of the prevalence of ASD diagnoses in GD/GI people was 11% (p < .001) and the overall effect size of the difference in ASD traits between GD/GI and control people was significant (g = 0.67, p < .001). Heterogeneity was high in both meta-analyses. We demonstrated that the chances that there is not a link between ASD and GD/GI are negligible, yet the size of it needs further investigation.
Preprint
Full-text available
The core diagnostic criteria for autism comprise two symptom domains – social and communication difficulties, and unusually repetitive and restricted behaviour, interests and activities. There is some evidence to suggest that these two domains are dissociable, yet, this hypothesis has not been tested using molecular genetics. We test this using a GWAS of a non-social autistic trait, systemizing (N = 51,564), defined as the drive to analyse and build systems. We demonstrate that systemizing is heritable and genetically correlated with autism. In contrast, we do not identify significant genetic correlations between social autistic traits and systemizing. Supporting this, polygenic scores for systemizing are significantly positively associated with restricted and repetitive behaviour but not with social difficulties in autistic individuals. These findings strongly suggest that the two core domains of autism are genetically dissociable, and point at how to fractionate the genetics of autism.
Article
Full-text available
This is a correction to: International Journal of Epidemiology, Volume 48, Issue 3, June 2019, Pages 678–679j, https://doi.org/10.1093/ije/dyz073
Article
Full-text available
Elevated latent prenatal steroidogenic activity has been found in the amniotic fluid of autistic boys, based on measuring prenatal androgens and other steroid hormones. To date, it is unclear if other prenatal steroids also contribute to autism likelihood. Prenatal oestrogens need to be investigated, as they play a key role in synaptogenesis and corticogenesis during prenatal development, in both males and females. Here we test whether levels of prenatal oestriol, oestradiol, oestrone and oestrone sulphate in amniotic fluid are associated with autism, in the same Danish Historic Birth Cohort, in which prenatal androgens were measured, using univariate logistic regression (n = 98 cases, n = 177 controls). We also make a like-to-like comparison between the prenatal oestrogens and androgens. Oestradiol, oestrone, oestriol and progesterone each related to autism in univariate analyses after correction with false discovery rate. A comparison of standardised odds ratios showed that oestradiol, oestrone and progesterone had the largest effects on autism likelihood. These results for the first time show that prenatal oestrogens contribute to autism likelihood, extending the finding of elevated prenatal steroidogenic activity in autism. This likely affects sexual differentiation, brain development and function.
Article
Full-text available
Co‐morbid mental health conditions such as anxiety and depression are extremely common in autistic adults. Vulnerability to negative life experiences such as victimisation and unemployment may be partially responsible for the development of these conditions. Here we measure the frequency of negative life experiences in autistic adults and explore how these are associated with current anxiety and depression symptoms and life satisfaction. We developed the Vulnerability Experiences Quotient (VEQ) through stakeholder consultation. The VEQ includes 60 items across 10 domains. Autistic adults with a clinical diagnosis and non‐autistic controls completed the VEQ, screening measures for anxiety and depression, and a life‐satisfaction scale in an online survey. Likelihood of experiencing each VEQ event was compared between groups, using binary logistic regression. Mediation analysis was used to test whether total VEQ score mediated the relationship between autism and (1) depression (2) anxiety and (3) life satisfaction. Autistic adults (N = 426) reported higher rates of the majority of events in the VEQ than non‐autistic adults (N = 268). They also reported more anxiety and depression symptoms and lower life satisfaction. Group differences in anxiety, depression and life satisfaction were partially mediated by VEQ total score. This study highlights several important understudied areas of vulnerability for autistic adults, including domestic abuse, contact with social services (as parents) and financial exploitation and hardship. Improved support, advice and advocacy services are needed to reduce the vulnerability of autistic adults to negative life experiences, which may in turn improve mental health and life satisfaction in this population. Autism Res2019. © 2019 The Authors. Autism Research published by International Society for Autism Research published by Wiley Periodicals, Inc. This study investigated whether autistic adults are more vulnerable to certain negative life experiences, and whether these experiences are related to anxiety, depression and life satisfaction. We found that autistic adults are more vulnerable to many different negative life events, including employment difficulties, financial hardship and domestic abuse. Negative life experiences partially explained the higher rates of anxiety and depression symptoms and lower life satisfaction in autistic adults compared to non‐autistic adults. Improved support services are required to reduce the vulnerability of autistic adults. Reducing vulnerability may improve mental health and increase life satisfaction in this population.
Article
Full-text available
Background: People who opt to participate in scientific studies tend to be healthier, wealthier and more educated than the broader population. Although selection bias does not always pose a problem for analysing the relationships between exposures and diseases or other outcomes, it can lead to biased effect size estimates. Biased estimates may weaken the utility of genetic findings because the goal is often to make inferences in a new sample (such as in polygenic risk score analysis). Methods: We used data from UK Biobank, Generation Scotland and Partners Biobank and conducted phenotypic and genome-wide association analyses on two phenotypes that reflected mental health data availability: (i) whether participants were contactable by e-mail for follow-up; and (ii) whether participants responded to follow-up surveys of mental health. Results: In UK Biobank, we identified nine genetic loci associated (P <5 × 10-8) with e-mail contact and 25 loci associated with mental health survey completion. Both phenotypes were positively genetically correlated with higher educational attainment and better health and negatively genetically correlated with psychological distress and schizophrenia. One single nucleotide polymorphism association replicated along with the overall direction of effect of all association results. Conclusions: Re-contact availability and follow-up participation can act as further genetic filters for data on mental health phenotypes.
Article
Many of the considerable number of young people who identify as transgender or gender diverse do not conform to traditional binary notions of gender (male vs female), and instead have a non-binary gender identity. This narrative Review summarises literature related to the sociodemographic and clinical profiles of young people with a non-binary gender identity. Young people identifying as non-binary form a substantial minority of the general population. They experience lower levels of support and are at increased risk of experiencing abuse and victimisation than young people who are cisgender. Furthermore, compared with young people who are transgender and binary, people who identify as non-binary experience less access to trans-specific health care. Young people identifying as non-binary have poor mental health outcomes, with high rates of depression, anxiety, and suicidal ideation that were found to be similar if not higher than in those who are transgender and binary. This Review highlights that young people who identify as non-binary are highly vulnerable and likely to have important health-care needs.
Article
Genetic influences on psychiatric disorders transcend diagnostic boundaries, suggesting substantial pleiotropy of contributing loci. However, the nature and mechanisms of these pleiotropic effects remain unclear. We performed analyses of 232,964 cases and 494,162 controls from genome-wide studies of anorexia nervosa, attention-deficit/hyperactivity disorder, autism spectrum disorder, bipolar disorder, major depression, obsessive-compulsive disorder, schizophrenia, and Tourette syndrome. Genetic correlation analyses revealed a meaningful structure within the eight disorders, identifying three groups of inter-related disorders. Meta-analysis across these eight disorders detected 109 loci associated with at least two psychiatric disorders, including 23 loci with pleiotropic effects on four or more disorders and 11 loci with antagonistic effects on multiple disorders. The pleiotropic loci are located within genes that show heightened expression in the brain throughout the lifespan, beginning prenatally in the second trimester, and play prominent roles in neurodevelopmental processes. These findings have important implications for psychiatric nosology, drug development, and risk prediction.
Article
Purpose of review: With increasing awareness of potential differences of autism presentation in nonmale versus male individuals, this review summarizes the rapidly evolving literature on sex and gender impacts on autism across nosology, behavioural presentation, developmental change and contextual recognition biases. Recent findings: Most studies have not differentiated sex versus gender impacts. Regarding behavioural presentation, measurement invariance across sex/gender was found in several standard measures. On this basis, diagnosed females overall showed lower restricted/repetitive behaviour/interests/activities (RRBI) than males, with small and variable effects depending on age, developmental level and kinds of RRBI. Differences insufficiently captured by standard measures may include autistic females displaying female-gender-typical narrow interests, higher social attention, linguistic abilities, motivation for friendship and more camouflaging than autistic males. Regarding developmental change, diagnosed young girls were more likely to have better cognitive development, less intense autistic symptoms and reduction of symptoms over time. Difficulties in adaptive functioning and social challenges, however, may emerge more for females in adolescence. Regarding diagnosis, general expectancy biases and gender-stereotypes may impede timely recognition of autism in females. Summary: Appreciating the multilevel sex and gender impacts on presentation, development, and diagnosis is key to sex-equitable and gender-equitable care for autistic individuals. A holistic approach to understanding the person in the contexts of sex and gender is essential for timely and accurate diagnosis and support.
Article
Background: Co-occurring mental health or psychiatric conditions are common in autism, impairing quality of life. Reported prevalences of co-occurring mental health or psychiatric conditions in people with autism range widely. Improved prevalence estimates and identification of moderators are needed to enhance recognition and care, and to guide future research. Methods: In this systematic review and meta-analysis, we searched MEDLINE, Embase, PsycINFO, Scopus, Web of Science, and grey literature for publications between Jan 1, 1993, and Feb 1, 2019, in English or French, that reported original research using an observational design on the prevalence of co-occurring mental health conditions in people with autism and reported confirmed clinical diagnoses of the co-occurring conditions and autism using DSM or ICD criteria. For co-occurring mental health conditions reported with at least 15 datapoints (studies), we assessed risk of bias and we determined pooled estimates of prevalence for different co-occurring conditions in autism using random-effects models, and descriptively compared these with prevalence estimates for the general population from the literature (post hoc). We investigated heterogeneity in prevalence estimates using random-effects meta-regression models. This systematic review is registered with PROSPERO, CRD42018103176. Findings: Of 9746 unique studies identified, 432 were selected for full-text review. 100 studies were eligible for inclusion in our qualitative synthesis, of which 96 were included in our meta-analyses. 11 categories of co-occurring conditions were investigated, of which eight conditions were included in the meta-analyses and three were descriptively synthesised (ie, trauma and stressor-related disorders, substance-related and addictive disorders, and gender dysphoria). From our meta-analyses, we found overall pooled prevalence estimates of 28% (95% CI 25-32) for attention-deficit hyperactivity disorder; 20% (17-23) for anxiety disorders; 13% (9-17) for sleep-wake disorders; 12% (10-15) for disruptive, impulse-control, and conduct disorders; 11% (9-13) for depressive disorders; 9% (7-10) for obsessive-compulsive disorder; 5% (3-6) for bipolar disorders; and 4% (3-5) for schizophrenia spectrum disorders. Estimates in clinical sample-based studies were higher than in population-based and registry-based studies, and these estimates were mostly higher than those in the general population (post hoc). Age, gender, intellectual functioning, and country of study were associated with heterogeneity in prevalence estimates, yet remaining heterogeneity not explained was still substantial (all I2 >95%). Interpretation: Co-occurring mental health conditions are more prevalent in the autism population than in the general population. Careful assessment of mental health is an essential component of care for all people on the autism spectrum and should be integrated into clinical practice. Funding: Academic Scholars Awards, Department of Psychiatry, University of Toronto; O'Brien Scholars Program, Slaight Family Child and Youth Mental Health Innovation Fund, and The Catherine and Maxwell Meighen Foundation via the Centre for Addiction and Mental Health Foundation.
Article
Background: The social challenges that non-binary people experience, due in part to social intolerance and the lack of validation of non-binary gender identities, may affect the mental health and quality of life of this population. However, studies that have distinguished between non-binary and binary transgender identities are lacking. Aim: To compare the mental health and quality of life of a community sample of non-binary transgender adults with controls (binary transgender people and cisgender people) matched on sex assigned at birth. Method: A total of 526 participants were included. Ninety-seven were classified as non-binary and were compared with two control groups: 91 people classified as binary and 338 cisgender people. Only transgender people not on gender affirming hormone treatment or who had not undergone gender affirming surgery were included. Participants were invited to complete an online survey that included mental health and quality of life measures. Results: Non-binary people reported significantly better mental health than binary transgender people, but worse than cisgender people. Overall, there were no significant differences in quality of life between non-binary and binary transgender participants assigned male at birth and transgender females, but non-binary assigned males at birth had better scores on the psychological and social domains of quality of life than transgender males. Quality of life was better across all domains in cisgender people than transgender groups. Conclusion: There is an inequality with regard to mental health and quality of life between non-binary (and binary) transgender people and the cisgender population that needs to be addressed. The better mental health scores in non-binary people may reflect lower levels of body dissatisfaction among the non-binary population. Mental health problems and poor quality of life are likely to have social causes and hence legislative measures and broader government-led inclusive directives should be put in place to recognize and to validate non-binary identifying people.