611381

article2016
AUT0010.1177/1362361315611381AutismWilson et al.

Original Article
Autism

Does sex influence the diagnostic 2016, Vol. 20(7) 808­–819

© The Author(s) 2016
Reprints and permissions:
evaluation of autism spectrum sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1362361315611381

disorder in adults? aut.sagepub.com

C Ellie Wilson1,2,3, Clodagh M Murphy1,2, Grainne McAlonan1,2,

Dene M Robertson1,2, Debbie Spain1,2, Hannah Hayward1,2,
Emma Woodhouse1,2, P Quinton Deeley1,2,4, Nicola Gillan2,5,6,
J Chris Ohlsen2, Janneke Zinkstok1,2, Vladimira Stoencheva1,2,
Jessica Faulkner1,2, Hatice Yildiran2, Vaughan Bell2,7,
Neil Hammond2, Michael C Craig1,2,4 and Declan GM Murphy1,2

Abstract
It is unknown whether sex influences the diagnostic evaluation of autism spectrum disorder, or whether male and female
adults within the spectrum have different symptom profiles. This study reports sex differences in clinical outcomes for
1244 adults (935 males and 309 females) referred for autism spectrum disorder assessment. Significantly, more males
(72%) than females (66%) were diagnosed with an autism spectrum disorder of any subtype (x2 = 4.09; p = 0.04). In high-
functioning autism spectrum disorder adults (IQ > 70; N = 827), there were no significant sex differences in severity of
socio-communicative domain symptoms. Males had significantly more repetitive behaviours/restricted interests than
females (p = 0.001, d = 0.3). A multivariate analysis of variance indicated a significant interaction between autism spectrum
disorder subtype (full-autism spectrum disorder/partial-autism spectrum disorder) and sex: in full-autism spectrum
disorder, males had more severe socio-communicative symptoms than females; for partial-autism spectrum disorder, the
reverse was true. There were no sex differences in prevalence of co-morbid psychopathologies. Sex influenced diagnostic
evaluation in a clinical sample of adults with suspected autism spectrum disorder. The sexes may present with different
manifestations of the autism spectrum disorder phenotype and differences vary by diagnostic subtype. Understanding
and awareness of adult female repetitive behaviours/restricted interests warrant attention and sex-specific diagnostic
assessment tools may need to be considered.

Keywords
autism spectrum disorder, diagnosis, females, males, sex differences

Introduction
1King’s College London, UK
Autism spectrum disorder (ASD) is a neurodevelopmental 2South London and Maudsley NHS Foundation Trust, UK
condition diagnosed when there is evidence from early 3University of Seville, Spain
4Bethlem Royal Hospital, UK
childhood of impairments in social functioning and com-
5Bristol Autism Spectrum Service, UK
munication co-occurring with repetitive behaviours and 6University of Bristol, UK
restricted interests (International Classification of Diseases– 7University College London, UK
10th Revision (ICD-10R); World Health Organization
(WHO, 1993); Diagnostic and Statistical Manual of Mental Corresponding author:
C Ellie Wilson, Individual Differences, Language and Cognition
Disorders, 5th ed. (DSM-5); American Psychiatric Laboratory, Department of Developmental and Educational Psychology,
Association (APA), 2015). ASD is a common condition University of Seville, Avda. Camilo José Cela s/n, 41018 Seville, Spain.
with recent epidemiological studies in the United Kingdom Email: [email protected]
Wilson et al. 809

estimating prevalence at between 1% (Brugha et al., 2012) Mandy et al., 2011; Park et al., 2012; Solomon et al., 2012;
and 1.7% (Russell et al., 2014). Males are diagnosed for a review, see Van Wijngaarden-Cremers et al., 2014)
approximately four times more often than females in child- although a recent study with boys and girls under the age of
hood (Fombonne, 2005, 2009) although this ratio varies 5 years reported no significant differences in number of
with IQ and is reportedly as low as 2:1 when ASD is co- symptoms in this domain (Harrop et al., 2015).
morbid with intellectual disability, and as high as 6–8:1 in Reports of differences in ASD presentation between
high-functioning populations (Fombonne, 2005). The rea- male and female adults are far more limited. Three studies
son for the gender discrepancy is unclear. It has been pro- have included participants ranging in age from childhood
posed that females require a greater assault at the genetic through to adulthood. Of these, two reported no significant
level in order to develop ASD (Jacquemont et al., 2014; sex differences in any domain: one included a sample with
Levy and Perry, 2011), and that ASD in females may be less intellectual disability aged 3–30 years (Pilowsky et al.,
frequently diagnosed because females tend to be better at 1998) and the other included a group without intellectual
compensating for their difficulties (Attwood, 2007; Lai disability aged 5–20 years (Holtmann et al., 2007). The
et al., 2011). Additionally, males and females may present third, including a sample of ASD individuals aged
with different ASD phenotypes (Howe et al., 2015; Mandy 6–36 years, found age-related differences: in early devel-
et al., 2012; Van Wijngaarden-Cremers et al., 2014); this opment, males had more severe social difficulties than
may affect diagnostic rates since the female profile is less females, but in adolescence and adulthood, females exhib-
well understood and hence less easily detected (Kirkovski ited more severe social and communication difficulties
et al., 2013). It is important to establish how sex influences than males (McLennan et al., 1993). These studies were
the presentation of ASD because this has implications for important first steps in investigation of sex differences
understanding the biology of ASD in both sexes, and has across the lifespan, but conclusions were inconsistent and
implications for service design and clinical care. Therefore, may have been limited by small numbers (N < 42/group)
we report, to the best of our knowledge, the first large-scale and wide age ranges within the samples.
comparison of symptom profiles in men and women who Only one other study has investigated gender and diag-
were referred for an assessment of ASD for the first time in nosis in adults: Lai et al. (2011) investigated 62 adults
adulthood. (aged 18–45 years) with previous diagnoses of high-func-
Sex differences in children with ASD have been tioning autism or Asperger syndrome. They reported that
reported in several previous studies. The largest study to ASD females had fewer repetitive and stereotyped behav-
date included 304 girls and 2114 boys aged 4–18 years iours than males both in childhood (reported retrospec-
with a diagnosis of ASD (Frazier et al., 2014). They tively) and currently. Females exhibited fewer
reported that girls showed more social and communication social-communication symptoms in adulthood although
symptoms on the Autism Diagnostic Observation no significant sex differences were detected during
Schedule–Generic (ADOS-G; Lord et al., 2000) and had childhood.
fewer repetitive behaviour symptoms on the Autism There are other factors, besides age and level of intelli-
Diagnostic Interview–Revised (ADI-R; Lord et al., 1994). gence, which are an integral part of diagnostic evaluation
Most participants in the sample had an IQ below 80, but but have been largely overlooked in previous studies. First,
the ASD girls had lower average IQ in both verbal and sex differences in symptom profile may vary by diagnostic
performance domains than the ASD boys; lower IQ scores subtype, yet the majority of studies only include individu-
in females was advanced as an explanation for greater als that meet ‘full-ASD’ criteria – that is, they have a diag-
social impairment but did not mediate fewer symptoms in nosis of Asperger syndrome, childhood autism or
repetitive interests/restricted behaviours. high-functioning autism. In the clinical setting, a signifi-
A number of earlier studies investigated a similar demo- cant number of people have a ‘partial-ASD’ diagnosis –
graphic of participants using smaller samples although pervasive developmental disorder–unspecified
results were not entirely consistent. In agreement with (PDD-unspecified, or ‘other PDD’) or atypical autism –
Frazier et al. (2014), some studies have reported that girls and these individuals are part of the autistic spectrum as
have more socio-communication symptoms and lower cog- currently defined, and eligible for services and support.
nitive and language ability (Carter et al., 2007; Lord et al., One study examined symptom presentation in a sample of
1982). Hartley and Sikora (2009) reported that in a group of high-functioning children referred for an ASD assessment
toddlers (N > 200), girls were more impaired on the com- (Mandy et al., 2011) and reported that relatively more
munication domain, but less impaired on the restricted males were diagnosed with a full-ASD subtype and rela-
interests/repetitive behaviours domain, than boys. This tively more females were diagnosed with a partial-ASD
reduced impairment in females compared to males on the subtype although the difference narrowly missed signifi-
restricted interests/repetitive behaviours domain has been cance. Beyond relative rates of diagnostic subtypes, a sex-
replicated in several previous studies with children and subtype interaction may affect manifestation of autistic
adolescents (Bölte et al., 2011; McLennan et al., 1993; traits (Lai et al., 2015). For instance, a recent large-scale
810 Autism 20(7)

study pooling four datasets, each including multiple clini- Method

cal sites, demonstrated that symptomatic differences
between boys and girls on the autistic spectrum vary by Participants
dataset (Howe et al., 2015). The authors suggested that the The initial sample included 1244 individuals aged 18–
results could have been affected by ascertainment strate- 75 years (inter-quartile range of 22–39 years); 935 males
gies, such that the clinical samples included participants of and 309 females. These adults were consecutively assessed
varying degrees of autistic symptoms. Thus, the validity of for ASD for the first time in a specialist national tertiary
extrapolating results from studies of ‘text-book’ cases of ASD clinic between April 2003 and April 2014
childhood autism and Asperger syndrome to the wider (Behavioural Genetics and Adult Autism Clinic, The
autistic spectrum is uncertain in children, and has yet to be Maudsley Hospital). People can be referred by their local
examined in adults. Furthermore, ASD subtype diagnoses family physician/general practitioner or consultant psy-
may provide a useful basis for developing individualized chiatrist for assessment of possible ASD in adulthood and
treatment plans, and clarification of potential differences is referrals are accepted from both the local community and
pertinent because upcoming modifications to the ICD across the United Kingdom.
diagnostic system are expected to follow the lead of the Ethical approval was granted by the National Research
DSM-5 by collapsing diagnostic subtypes into one ‘ASD’ Ethics Committee, London (12/LO/07990). In an addi-
diagnosis; therefore, this information may not be available tional 54 cases, diagnosis was inconclusive due to severe
in the future. psychotic or depressive symptoms, non-compliance or his-
Second, co-morbid psychiatric conditions are common tory of major head injury; these individuals were excluded
in ASD (Hofvander et al., 2009; Joshi et al., 2013; Russell from the study. There were no significant differences in
et al., 2005, 2015) and symptoms of ASD are often diffi- sex distribution between the excluded cases and the full
cult to disentangle from additional or alternative condi- sample.
tions (Mazzone et al., 2012; Rydén and Bejerot, 2008).
This may lead to inaccurate diagnoses (Attwood, 2007) or
misguided referrals to specialist clinics. A recent epidemi-
Clinical assessment
ological study has demonstrated that diagnostic rates of Assessment included a detailed neuropsychiatric assess-
certain common psychiatric conditions – in particular ment by a multidisciplinary clinical team with expertise in
mood and anxiety disorders – are higher in women than ASD: a consultant psychiatrist, +/− junior doctor and a
men in the general adult population (Kessler et al., 2012), research-reliable ADI-R/ADOS-G administrator (nurse,
but it is unclear whether sex differences translate to the psychologist or doctor).
autistic spectrum, or whether additional mental health con- Each patient’s history and clinical information was
ditions influence sex differences in manifestation of autis- reviewed on the day of their appointment and they com-
tic symptomology (Lai et al., 2015). It is also of clinical pleted a psychiatric clinical interview and ADI-R/ADOS-G
importance to establish what mental health conditions are assessment (lasting 1–4 h, with breaks as necessary). The
commonly diagnosed in patients with suspected ASD, but ADI-R, lasting 1.5–3 h, is a semi-structured parent/car-
who do not go on to receive a diagnosis of ASD. egiver interview designed to assess and quantify a devel-
To summarize, in this study, we examine whether sex opmental history of autism-specific behaviours (Lord
influenced the diagnostic evaluation of ASD in a sample of et al., 1994). The ADI-R was completed if the patient pro-
individuals who were referred to a national specialist clinic vided consent, and if a parent/early childhood caregiver
for an ASD assessment for the first time in adulthood. We was available. If it was not possible to complete an ADI-R,
addressed the following four specific aims. or additional information was required to determine diag-
nosis, an ADOS-G (module 4) was completed. The
1. To compare the rates of positive ASD diagnoses, ADOS-G is a standardized assessment conducted with the
and characteristics (age, intelligence, ASD subtype patient that lasts 40–60 min. It involves a semi-structured
and additional mental health diagnoses), of men interview interspersed with activities and tasks intended to
and women referred for an ASD assessment. elicit behaviours associated with ASD. In all, 630 individ-
2. To examine sex differences in type and severity of uals were assessed using the ADI-R, 408 were assessed
ASD core-symptoms across the autism spectrum. with the ADOS-G and 206 were assessed using both
3. To examine the moderating effects of diagnostic sub- ADI-R and ADOS-G.
type, the presence of additional psychiatric condi- The presence or absence of an ASD diagnosis was made
tions and IQ, on any sex/core-symptom interactions. in a diagnostic meeting attended by all members of the
4. To compare characteristics (age, alternative mental clinical team that conducted the assessment, who deter-
health diagnoses) of males and females with sus- mined by consensus whether each criterion on the ICD-
pected ASD, but who did not receive a diagnosis of 10R ASD algorithm was fulfilled or not. In line with
an ASD. ICD-10R guidelines, for a patient to meet full ICD-10R
Wilson et al. 811

criteria for autism, a total of at least six symptoms must be diagnoses and presence of additional mental health condi-
present – either currently or by history – with at least two tions (Table 1).
from the ‘social interaction’ domain and one from each of To address Aim 2, sex differences in domain scores of
the ‘communication’ and ‘restricted and repetitive inter- the ADI-R (social, communication, repetitive and restricted
ests’ domains, and symptoms noted before the age of interests and behaviours) and ADOS-G (social, communi-
3 years. They were diagnosed with childhood autism or cation, stereotyped behaviours and restricted interests)
high-functioning autism (if they exhibited a language were examined using t-tests. To Bonferroni-correct for
delay) or Asperger syndrome (if there was no evidence of multiple comparisons, we considered p values of less than
a language delay). If a patient differed from the ICD-10R 0.006 to be significant (Table 2).
autism criteria either in age of onset (i.e. later than 3 years To address Aim 3, multivariate analyses of variance
of age) or number of symptoms (e.g. a lack of sufficient (MANOVAs) were conducted with sex (male/female) and
demonstrable abnormalities in one or two of the three ASD diagnostic subtype (full-ASD diagnosis/partial-ASD diag-
domains, despite characteristic abnormalities in other nosis) as fixed factors. First, ADI-R domain scores were
area(s)), they were diagnosed with atypical autism. If a entered as dependent variables. Post hoc t-tests were per-
patient’s history and presentation was in keeping with an formed on significant interactions. The same analyses
ASD but there was a lack of adequate information, they were conducted with scores from the ADOS-G (Figure 1).
were diagnosed with PDD-unspecified. The MANOVAs were repeated including only the ‘full-
Of the 1244 referrals, 874 (70%) were diagnosed with ASD’ group, contrasting participants with childhood/high-
an ASD. Of these, 219 (25%) participants were subtyped functioning autism (i.e. those with a language delay;
as childhood autism or high-functioning autism, 429 N = 429) and Asperger syndrome (i.e. no language delay;
(49%) as Asperger syndrome, 154 (18%) as atypical autism N = 154).
and 72 (8%) as PDD-unspecified. Except when stated oth- The presence of co-morbid conditions (any additional
erwise, for this article, participants with childhood autism, mental health condition; ADHD; phobic disorder; obses-
high-functioning autism and Asperger syndrome were sub- sive compulsive disorder (OCD); any anxiety disorder;
sumed into a ‘full-ASD’ diagnostic subgroup, and those any depressive disorder) were entered as covariates in the
with atypical autism and PDD-unspecified and were sub- multivariate models reported above (fixed factors: sex,
sumed into a ‘partial-ASD’ diagnostic subgroup. diagnostic subtype; dependent variables: ADI-R scores,
Additional mental health conditions were diagnosed in ADOS-G scores).
accordance with the ICD-10R (with the exception of adult Finally, an exploratory analysis including all ASD par-
attention deficit hyperactivity disorder (ADHD)) which, in ticipants with verbal IQ (VIQ) and performance IQ (PIQ)
keeping with UK guidelines, was assessed using Diagnostic data available (N = 279; this included those with an intel-
and Statistical Manual of Mental Disorders (4th ed., text lectual impairment) was conducted using a MANOVA
rev.; DSM-IV-TR). with sex and subtype as fixed factors, and VIQ and PIQ as
Neuropsychological testing was completed in 319 par- dependent variables (full-scale IQ (FIQ) was not included
ticipants either for their clinical care if intellectual disabil- because for many ASD individuals FIQ was not computa-
ity or a significant lacuna in cognitive function was ble due to the discrepancy between VIQ and PIQ).
suspected (248 participants completed the Wechsler Adult To address Aim 4, t-tests were used to compare age,
Intelligence Scale-III (WAIS-III; Wechsler, 1997)) or as ADI-R and ADOS-G scores, and presence of alternative
part of associated research projects (71 participants com- mental health conditions between men and women who
pleted the Wechsler Abbreviated Scale of Intelligence were not diagnosed with an ASD (Table 1; Figure 1).
(WASI; Wechsler, 1999)).
Results
Data analyses The outcome of the ASD assessments are presented in
To address Aim 1, chi-square analyses were employed to Table 1 (age, intelligence, ASD subtype and additional
compare rates of ASD diagnosis. T-tests were used to com- mental health diagnoses in ASD participants; alternative
pare age and IQ (where available) of ASD participants. In mental health diagnoses in non-ASD participants).
order to focus on symptom profile in high-functioning
adults with ASD, participants with confirmed IQ < 70 in
Aim 1: characteristics of ASD participants
any domain (N = 29), and participants where an intellectual
disability was suspected but testing could not be com- Of the initial 1244 participants, 70% were diagnosed
pleted (N = 17) were excluded from the following analy- with an ASD (671 males and 203 females; a ratio of
ses, and the final high-functioning ASD sample size 3.3:1). The proportion of participants who received a
included 639 males and 188 females. T-tests were then positive ASD diagnosis was significantly higher in males
used to examine sex differences in rates of ASD subtype (72%) compared to females (66%; x2 = 4.09, p = 0.04,
812 Autism 20(7)

Table 1. Outcome of ASD assessment: age, intelligence, ASD subtype and additional mental health diagnoses in ASD participants;
alternative mental health diagnoses in non-ASD participants.

N Males Females Gender difference Effect size

(Cohen’s d)
Outcome of ASD % of ASD positive M: 935, F: 309 71.8% 65.7% x2 = 4.09, p = 0.04 0.12
assessment referrals
Mean age of ASD M: 671, F: 203 31.2 years 30.2 years t = 1.09, p = 0.28 0.09
diagnosis
Suspected or M: 671, F: 203 4.8% 7.4% x2 = 2.10, p = 0.15 0.05
confirmed IQ < 70
Average VIQ M: 226, F: 56 101.1 (SD: 17.2) 96.3 (SD: 19.3) t = 1.79, p = 0.08 0.26
Average PIQ M: 223, F: 56 95.2 (SD: 17.9) 92.0 (SD: 19.1) t = 1.17, p = 0.24 0.17
Average FIQ M: 163, F: 41 99.4 (SD: 17.6) 92.4 (SD: 20.2) t = 2.18, p = 0.03 0.37
ASD subtype Full-ASD diagnoses M: 639, F: 188 75.4% 70.2% x2 = 2.07, p = 0.15 0.05
(as % of all ASD
diagnoses)
Partial-ASD 24.6% 29.8%
diagnoses (as % of all
ASD diagnoses)
Additional mental % with any additional M: 639, F: 188 57.6% 61.2% t = −0.99, p = 0.33 0.08
health diagnoses in mental health
ASD participants diagnosis
% with ADHD 12.5% 13.3% t = −0.28, p = 0.78 0.02
% with social phobia 11.2% 14.3% t = −1.15, p = 0.25 0.10
% with OCD 17.8% 19.7% t = −0.47, p = 0.64 0.10
% with any anxiety 40.8% 46.3% t = −1.46, p = 0.15 0.12
disorder
% with any 21.9% 20.2% t = 0.43, p = 0.67 0.04
depressive disorder
Alternative mental % with any M: 264, F: 106 54.1% 62.3% t = −0.66, p = 0.51 0.08
health diagnoses alternative mental
in non-ASD health diagnosis
participants % with ADHD 11.4% 6.6% t = 1.61, p = 0.11 0.19
% with social phobia 5.7% 17.0% t = 3.50, p = 0.001 0.40
% with OCD 11.0% 12.3% t = −0.25, p = 0.81 0.03
% with any anxiety 29.2% 41.5% t = 2.30, p = 0.02 0.27
disorder
% with any 22.7% 23.6% t = −0.17, p = 0.86 0.02
depressive disorder

VIQ: verbal IQ; PIQ: performance IQ; FIQ: full-scale IQ; ADHD: attention deficit hyperactivity disorder; OCD: obsessive compulsive disorder.
‘Full-ASD diagnosis’ includes Asperger syndrome, childhood autism and high-functioning autism. ‘Partial-ASD diagnosis’ includes pervasive develop-
mental disorder–unspecified and atypical autism. ‘Any anxiety disorder’ includes phobic disorders, OCD, generalized anxiety disorder, mixed anxiety
and depression, social anxiety. ‘Depressive disorders’ include bipolar affective disorder; mild, moderate or severe depressive episode; mixed anxiety
and depression; recurrent depressive disorder; dysthymia.

d = 0.12). The mean age was 31.0 years (SD = 11.1) and more males were diagnosed with full-ASD when com-
there was no significant sex difference in the age of ASD pared to those with partial-ASD although the difference
diagnosis (N = 874; Table 1). was not significant, (p = 0.15, d = 0.05). There were no sex
There was no significant difference between the pro- differences in proportion of ASD participants who received
portion of males and females who had an IQ below 70 in any additional mental health diagnosis (Table 1; all
FIQ, PIQ or VIQ; however, males had significantly higher ps > 0.3).
FIQ than females (p = 0.03, d = 0.37) and marginally higher
VIQ (p = 0.08, d = 0.26). Aim 2: sex differences in core-symptom profiles
ASD subtype and additional mental health conditions
in high-functioning ASD (N = 827): the ratio of males to
in high-functioning ASD (N = 827)
females in the full-ASD subtype was 3.7:1, and in the par- After Bonferroni corrections, the only difference that
tial-ASD subtype the ratio was 2.8:1. Therefore, relatively reached significance was in the repetitive behaviours and
Wilson et al. 813

Table 2. Core domain scores for high-functioning ASD males and females.

Number of participants Test and domain Male Female Gender difference Effect size (Cohen’s d)
ADI-R: N = 320 males, ADI-R social 13.2 (6.1) 12.6 (6.3) t = 1.0, p = 0.3 0.10
90 females ADI-R communication 9.9 (4.7) 9.7 (4.3) t = 0.4, p = 0.7 0.07
ADI-R repetitive behaviours 3.6 (2.1) 2.9 (2.1) t = 3.4, p = 0.001 0.33
and restricted interests
ADI-R total 26.7 (11.0) 25.1 (10.8) t = 1.4, p = 0.1 0.15
ADOS-G: N = 203 ADOS-G social 7.9 (2.8) 7.4 (2.8) t = 1.7, p = 0.1 0.21
males, 63 females ADOS-G communication 3.5 (1.8) 3.0 (1.6) t = 2.3, p = 0.02 0.22
ADOS-G restricted interests 1.6 (1.5) 1.6 (1.4) t < 0.1, p = 0.9 0.06
and behaviours
ADOS-G 11.4 (4.1) 10.4 (3.8) t = 2.2, p = 0.03 0.25
social + communication

ADI-R: Autism Diagnostic Interview–Revised; ADOS-G: Autism Diagnostic Observation Schedule–Generic.

Mean and standard deviation (in brackets) of scores.

restricted interests domain of the ADI-R, with males scor- The effect of sex was only significant for the repetitive
ing higher than females, t(526) = 3.27, p = 0.001, d = 0.33. behaviours/restricted interests domain (male > female;
2
All other comparisons were non-significant (ps > 0.02). F(1) = 9.18, p = 0.003, η p = 0.02 ). There were no signifi-
cant interactions between subtype and sex. In the ADOS-G,
there was a significant effect of subtype only in the repeti-
Aim 3: interactions between sex, diagnostic tive behaviours/restricted interests domain (Asper-
subtype and core-symptoms ger > childhood/high-functioning autism; F(1) = 6.26,
As expected, the MANOVA confirmed that on average the p = 0.01, η2p = 0.02 ). Significant effects of sex were evident
full-ASD participants scored significantly higher than par- in the communication domain (male > female; F(1) = 4.14,
2
tial-ASD participants in all ADI-R domains (all ps < 0.001, p = 0.04, η p = 0.02 ), but again, there were no significant
all η2p > 0.05 ). The effect of sex was only significant for interactions between subtype and sex.
the repetitive behaviours and restricted interests domain
(male > female; F(1) = 7.62, p = 0.006, η2p = 0.01 ). There Additional mental health conditions. The significance of
was a significant interaction between sex and diagnostic results of the multivariate models was unchanged when
subtype in ADI-R communication domain (F(1) = 5.28, additional mental health conditions were added as
p = 0.02, η2p = 0.01 ), and a marginal interaction in the covariates.
2
ADI-R social domain (F(1) = 3.52, p = 0.06; η p = 0.01 ;
Figure 1). The interaction in the repetitive behaviours IQ. This analysis was conducted with all ASD participants
domain was non-significant (p = 0.9). Post hoc t-tests where VIQ and PIQ data were available (N = 279), includ-
(Figure 1) confirmed that in the full-ASD group the aver- ing those with an intellectual impairment who were
age male score was higher than the average female score excluded from previous analyses. There was no significant
on the social and communication domains. Conversely, in effect of sex (male IQ > female IQ; F(2) = 2.47, p = 0.09,
the partial-ASD group, the average female score was η2p = 0.02 ), no significant effect of subtype (p > 0.3), and
higher than the average male on the social and communi- no significant interaction between sex and subtype
cation domains. In the repetitive behaviours and restricted (p > 0.4).
interests domain, the average male score was significantly
higher than the average female score in all ASD subtypes.
Aim 4: non-ASD participants (N = 370)
In the ADOS-G, the MANOVA confirmed the expected
2
effect of diagnostic subtype (p < 0.001, η p = 0.1 ) but no The non-ASD participants were significantly older than
significant effect of sex (p = 0.5). The interaction between the ASD participants (t(1242) = 4.70, p < 0.001, d = 0.3)
sex and diagnostic subtype was not significant (p = 0.14). (mean age = 34.3 years, SD = 12.0 years).
There were no significant differences on any domains
Asperger syndrome versus childhood/high-functioning autism. In of the ADI-R or ADOS-G between males and females
the ADI-R, there were significant effects of subtype in the who were not diagnosed with ASD (N = 370, all ps > 0.3;
communication domain (childhood/high-functioning Figure 1).
2
autism > Asperger; F(1) = 23.2, p < 0.001, η p = 0.05 ) and In all, 62% of non-ASD women and 54% of non-ASD
in the social interaction domain (childhood/high-function- men were diagnosed with at least one alternative mental
ing autism > Asperger; F(1) = 17.5, p < 0.001, η2p = 0.04 ). health condition. Significantly, more females than males
 814

Figure 1. Mean scores on the core domains of the ADI-R, split by sex and diagnostic subtype. Significant interactions were found between sex and diagnostic subtype (full-ASD/
partial-ASD) in the communication domain (p = 0.02) and in total ADI-R score (p = 0.04), and a marginal interaction was found in the social domain (p = 0.06).
Autism 20(7)
Wilson et al. 815

were diagnosed with social phobia, t = 3.5, p = 0.001 all diagnostic subtypes, males scored significantly higher
(d = 0.4). Females were also diagnosed with anxiety disor- than females on the repetitive behaviours/restricted inter-
ders more frequently than males, t = 2.3, p = 0.02 (d = 0.3) ests domain of the ADI-R. This is widely consistent with
although correcting for multiple comparisons renders this previous research (Bolte et al., 2011; McLennan et al.,
result non-significant. 1993; Mandy et al., 2011; Park et al., 2012; Solomon et al.,
2012; Van Wijngaarden-Cremers et al., 2013) although
contrasts with recent evidence from young children (Howe
Discussion
et al., 2015), and suggests an alternative explanation for
In this study, we examined sex differences in a clinical the results from the socio-communicative domains:
sample of adults referred for an assessment of ASD to females frequently have prominent symptoms in the socio-
determine whether sex influenced diagnostic evaluation. communicative domains but reduced symptoms in the
Participants had been referred for an ASD assessment for repetitive behaviours/restricted interests domain. This
the first time in adulthood and the majority had no intel- places them into the ‘partial-ASD’ diagnostic category and
lectual disability. In answer to our first aim, more men means that males and females with the same diagnostic
(72%) than women (66%) received a positive ASD diag- label often have very different symptom profiles. Of
nosis of any subtype; this difference was significant course, ASD is a highly heterogeneous condition so varia-
although the effect size was small. If we accept that the bility within subtypes is to be expected; however, these
higher rate of ‘incorrect’ referrals in women exists, it could results contribute to emerging evidence for sex-specific
either be due to general health practitioners/psychiatrists manifestations of the autism phenotype. Specifically, ASD
being less clear about how ASD manifests in adult females, females without an intellectual disability typically exhibit
or it could be due to an under-diagnosis of women and thus fewer repetitive behaviours and restricted interests than
a need to adjust the diagnostic criteria. The sex ratio in the their male counterparts with comparable socio-communi-
high-functioning ASD group was 3.4 men to 1 woman. cative impairment.
There were no sex differences in age or presence of addi-
tional mental health conditions (58% of men, 61% of
women had at least one co-morbid diagnosis), but males Sex differences in core-symptom profiles:
had marginally higher IQ scores and there was a trend implications for efficacy of diagnostic tools
towards a higher proportion of males in the full-ASD Our approach cannot rule out the possibility that women
subtype. do not exhibit ‘fewer’, but that they exhibit ‘different’,
In response to our second aim, there were notable dif- repetitive behaviours or restricted interests. This is because
ferences in core-symptom profiles; overall, both sexes current assessment tools, such as the ADI-R and ADOS-G,
exhibited a similar degree of socio-communicative symp- have been designed to measure the symptoms that define
toms but men exhibited more restricted behaviours and ASD, therefore only serve to confirm or reject the presence
repetitive interests than women. However, in response to of what we describe as ‘ASD traits’. If females (or males)
our third aim, sex differences in core-symptomology var- actually manifest symptoms not currently included in the
ied by subtype. algorithm, no current assessment tool or diagnostic algo-
rithm will detect that. This problem is referred to as the
Sex differences in core-symptom profiles: ‘nosological (how autism is defined) and diagnostic (how
evidence for differing manifestations of the ASD autism is identified) challenge’ of ASD research (Lai et al.,
2015).
phenotype Use of qualitative methods to investigate sex-typical
The finding of no significant sex differences in socio-com- traits could contribute useful information to this debate.
municative symptoms in the ASD group as a whole is in However, to date, few studies have documented how
line with a review by Van Wijngaarden-Cremers et al. repetitive behaviours and restricted interests actually differ
(2013), but contrasts with studies reporting that females between males and females. One possibility is that girls
have more (Carter et al., 2007; Frazier et al., 2014; Hartley are more likely to have socially accepted special interests
and Sikora, 2009; Lord et al., 1982; McLennan et al., that may mask the atypical nature of the interest (Kopp and
1993) or less (Lai et al., 2011) symptoms than males. This Gillberg, 1992; Lai et al., 2015). For example, a parent
discrepancy is potentially reconcilable by the fact that the may report that their daughter liked playing with dolls, but
studies listed include different age groups and different when probed about how they ‘played’ it could become
criteria for inclusion since our results indicate that socio- apparent that every session involved brushing the hair
communicative symptoms are not constant across the again and again, with little flexibility or imagination.
spectrum: in the full-ASD group, adult males exhibited Moreover, we propose that circumscribed interests in
more socio-communicative symptoms than females, but in males could actually be over-identified due to preconcep-
the partial-ASD group the reverse was true. By contrast, in tions about common interests in ASD boys. For example, a
816 Autism 20(7)

parent may report their son was very keen on trains or and intelligence. Regarding IQ, our results were largely in
dinosaurs, this could be over-interpreted as a ‘special inter- line with previous research reporting that females with a
est’, but on further questioning it may emerge that in this diagnosis of ASD tend to have a lower IQ than males
particular individual the trains/dinosaurs interest was little (Fombonne, 2005), and non-significant interactions
more than an age-appropriate phase that did not interfere between IQ and ASD subtype provided no evidence for
with other interests. Thus, clinicians should be careful of variation across the spectrum. Regarding additional men-
stereotyping observed behaviours. Identifying common tal health diagnoses, our results indicated no sex differ-
examples of restricted interests and repetitive behaviours ences in prevalence of additional psychopathologies in the
in both sexes across the spectrum in both childhood and ASD group, or interactions with core-symptomology, at
adulthood may alleviate this problem. the time of assessment. Nevertheless, we note that it is still
Additional future investigations could focus on devel- possible that sex differences exist regarding historical
opmental differences between males and females on the diagnoses, rates of previous misdiagnosis and patterns of
spectrum across the lifespan (Lai et al., 2015). In our evolving diagnosis across the lifespan. All these have
report, more prominent sex differences were identified by potential implications for diagnostic practice.
the ADI-R (focusing mainly on childhood symptoms) than We also contrasted core-symptom presentation between
the ADOS-G (which focuses on current symptoms). While Asperger syndrome (full-ASD with no language delay)
we note that the ADOS-G is not consistently sensitive to and childhood autism/high-functioning autism (full-ASD
repetitive and restricted behaviours in male or female with a language delay). Group differences were evident:
adults (hence, scores in this domain are not required for Asperger syndrome participants exhibited significantly
diagnosis), the results suggest further investigation into more social interaction symptoms, but fewer communica-
change in symptom presentation over the lifespan warrants tion symptoms than their childhood autism/high-function-
research. Longitudinal methods would be ideal to elimi- ing autism counterparts. This warrants further investigation
nate the effects of parental bias, whereby parental toler- and has implications for collapsing the two diagnostic sub-
ance and recall of perceived difficulties in early childhood types, and the two domain categories, in the DSM-5
may vary across gender which would influence results of (Wilson et al., 2013) and forthcoming ICD-11. However,
the ADI-R but not the ADOS-G. However, behavioural no sex-subtype interactions emerged; thus, we found no
adaptations and learned skills that may contribute to lower evidence that a language delay differentially affects the
present-state ADOS-G scores in some adults with ASD development of core-symptoms in late-diagnosed males
should be considered. Ultimately, we should aim to and females.
improve guidelines for general healthcare professionals, In general, our results raise the issue of ‘spread’ of
parents and teachers, and introduce clear examples for symptoms versus ‘severity’ of symptoms when using diag-
both sexes into diagnostic algorithms. nostic algorithms. Currently, an individual with moderate
This issue is also relevant to the manifestation and symptoms spread across all domains will qualify for a
development of socio-communicative symptoms in males diagnosis (perhaps a typical male profile), but those with
and females. Our result of ‘no overall differences in num- severe symptoms focused in one domain may not (perhaps
ber of socio-communicative symptoms’ was likely mask- a typical female profile). Some of these people may qual-
ing differences in more fine-grained socio-communicative ify for an alternative, possibly more appropriate, diagnosis
symptoms – hence the evident contrasts between subtypes. but others may miss out on a diagnosis altogether and
A recent paper demonstrated that boys and girls on the hence not receive any services or support. In the DSM-5
spectrum that were matched for overall level of core- (2014), social and communication symptoms are collapsed
symptomology contrasted in terms of what factors were to a single domain, and an individual must fulfil three out
associated with play skills (Harrop et al., 2014). The of three criteria (and two out of four criteria in the repeti-
authors reported that in boys, the social-communication tive/restricted behaviour domain) to qualify for the ASD
skill of ‘initiating behavioural requests’ was associated diagnosis. Thus, there is a strict cut-off for minimum
with non-verbal IQ and language ability, but in girls, it was ‘spread’ of symptoms. The impact of the new system is yet
‘responding to behavioural requests’ that was associated to be established, but a study analysing clinic outcomes
with non-verbal IQ. The authors note that the contrasting retrospectively for 150 adults suggested that 44% of par-
correlations did not survive Bonferroni corrections, but ticipants that met criteria for any ASD using the ICD-10R,
nevertheless they promote investigations into detailed and 22% that met DSM-IV-TR criteria for Asperger syn-
components of core-symptoms with attention to potential drome/autistic disorder, would not qualify for a diagnosis
sex differences in how these may relate to other cognitive of ASD under the DSM-5 (Wilson et al., 2013). The same
functions. study reported no differences in rates of men and women
In this study, we investigated how autistic traits related who would qualify for diagnoses under current and new
to additional cognitive and behavioural symptoms in two systems although this warrants replication with prospec-
respects: presence of additional mental health conditions tive data. Regarding varying levels of symptom severity
Wilson et al. 817

within the diagnostic category of ASD, the DSM-5 has and detection of ASD symptomology by general health
introduced three ‘severity levels’ to be allocated on the professionals does seem to be improving.
basis of accompanying intellectual impairment, language
impairment or known medical/genetic/environmental fac-
Strengths, limitations and future research
tors. However, this does not deal with differences in spe-
cific symptom severity and core-symptom profiles, which Concerning limitations, this sample included only people
may be a factor for consideration in future diagnostic tools. not diagnosed during childhood and therefore the sample
In addition, we note that the DSM-5 (and likely the forth- may be skewed towards people whose childhood symp-
coming ICD-11) relies heavily on retrospective data when toms were subtle, overcome by compensatory factors or
examining adults. Following the earlier proposal that undetected for other reasons. The extent to which data
parental recall may differ for girls and boys, females may reported here can be generalized to the ‘early-diagnosed’
again be at a disadvantage in terms of fulfilling criterion ASD population remains unclear. In particular, whether sex
and being adequately diagnosed. differences differ by early versus late-diagnosed individu-
als remains unknown and warrants further investigation.
Formal IQ testing was not completed for the majority of
Implications for service design participants, but instead intelligence levels were assumed
This report has implications for ASD services that con- to be in the normal range (IQ > 70) unless the clinicians –
tinue to evolve in the wake of the Autism Act (Her who were highly experienced in working with adults with
Majesty’s Government, 2009) and National Institute for intellectual disabilities – had any reason to suspect other-
Clinical Excellence guidelines (NICE, 2012; Wilson et al., wise. The relationship between ASD symptom profile and
2014). ASD is currently the only mental health disorder IQ level could therefore not be investigated within this
with dedicated legislation in the United Kingdom, but the high-functioning sample. Analysis of non-core-symptoms
resulting increase in demand for ASD services coincides was beyond the scope of this report; however, previous
with a reduction in available resources in the healthcare research has indicated differences between males and
system. Since specialized assessments are time-consuming females in several specific domains including executive
and costly to both patients and service providers (Murphy functioning (e.g. Bölte et al., 2011; Lai et al., 2012; Lemon
et al., 2011), it is useful to know what could underlie et al., 2011), perceptual attention to detail and motor func-
‘errors’ – that is, referrals that do not result in an ASD tion (Lai et al., 2012), adaptive skills (Frazier et al., 2014),
diagnosis. In response to our fourth aim to compare char- autobiographical memory (Goddard et al., 2014) and sleep
acteristics of those patients who were not diagnosed with habits (Hartley et al., 2009). Such factors could interact
ASD, 54% of men and 62% of women were diagnosed with core-symptom presentation and may contribute to a
with an alternative condition; this discrepancy was driven definition of sex-specific manifestations of the ASD
by a significantly higher rate of social phobia in women. phenotypes.
There is clear overlap between symptoms of social anxiety Finally, we acknowledge that despite the expertise of
and ASD; for example, behaviours common to both the clinical team and the use of adequate instruments,
include social withdrawal and being quiet in social situa- diagnostic misses of ASD and/or a failure to detect certain
tions (Eriksson et al., 2013). However, important distinc- symptoms was possible. Not all alternative disorders could
tions can be made, for example, adults with social phobia be assessed since appointments were completed in a single
may be anxious that they are socially inept, but actually day. Attachment disorder/personality disorder, for exam-
these skills and knowledge are not lacking. By contrast, ple, requires further investigation and was likely to be an
adults with ASD may lack knowledge about how to act accurate diagnosis for some of the non-ASD participants.
appropriately in social situations, and they may or may not The possibility that differing rates of attachment disorder
have insight into this deficit (Bejerot et al., 2014). in men and women who are referred for an ASD assess-
Continued investigation into how symptoms of social pho- ment remains open.
bia and ASD differ, in particular in females, along with Concerning strengths, this study included a relatively
more sensitive and readily available screening tools, could large sample size in comparison to the existing literature,
help general practitioners and psychiatrists avoid errant and every participant underwent an assessment with spe-
referrals and appropriately identify diagnoses and access cialist clinicians using best available diagnostic tools. The
to management options. sample is from a national tertiary clinic and is likely to be
Nevertheless, we stress that in this sample around 70% representative of the adult population presenting with
of patient referrals were accurate – that is, assessment con- autism-like mental health problems in the United Kingdom.
firmed the suspected ASD diagnosis – this is a substantial The inclusion of adults with diagnoses across the autistic
improvement on the 50% accuracy rate (Murphy et al., spectrum provides confidence that results are applicable to
2011) and 56% accuracy rate (Russell et al., 2015) that real clinical settings of adult-diagnostic clinics – an advance
were reported from the same national clinic with data from on most previous studies that have included only ‘full-
4 years ago and 3 years ago, respectively. Thus, awareness ASD’ subtypes. Moreover, the inclusion of participants
818 Autism 20(7)

who were referred for an ASD assessment but were not on the-prevalence-of-autism-spectrum-conditions-in-adults-
the autistic spectrum allowed us to explore under what cir- extending-the-2007-adult-psychiatric-morbidity-survey/
cumstances ASD may be incorrectly identified in males Carter AS, Black DO, Tewani S, et al. (2007) Sex differences in
and females. toddlers with autism spectrum disorders. Journal of Autism
and Developmental Disorders 37(1): 86–97.
Eriksson JM, Anderson LM and Bejerot S (2013) RAADS-14
Conclusion Screen: validity of a screening tool for autism spectrum dis-
order in an adult psychiatric population. Molecular Autism
We report sex differences in symptom profile in late- 4: 49.
diagnosed individuals with ASD, and suggest that men Fombonne E (2005) Epidemiology of autistic disorder and other
and women may present with different manifestations of pervasive developmental disorders. Journal of Clinical
the ASD phenotype. Sex appears to influence the diag- Psychiatry 66: 3–8.
nostic evaluation of adults, and further research should Fombonne E (2009) Epidemiology of pervasive developmental
investigate how this impacts on clinical care, in particular disorders. Pediatric Research 65(6): 591–598.
whether males and females respond differently to Frazier TW, Georgiades S, Bishop SL, et al. (2014) Behavioral
treatment. and cognitive characteristics of females and males with
autism in the Simons Simplex Collection. Journal of the
American Academy of Child and Adolescent Psychiatry
Acknowledgements
53(3): 329.e1–3–340.e1–3.
The views expressed are those of the author(s) and not necessar- Goddard L, Dritschel B and Howlin P (2014) A preliminary
ily those of the NHS, the NIHR or the Department of Health. study of gender differences in autobiographical memory
D.G.M.M. and G.M. acknowledge support from the Sackler in children with an autism spectrum disorder. Journal of
Centre for Translational Neurodevelopment at King’s College Autism and Developmental Disorders 44(9): 2087–2095.
London. Harrop C, Shire S, Gulsrud A, et al. (2014) Does gender influence
core deficits in ASD? An investigation into social-commu-
Funding nication and play of girls and boys with ASD. Journal of
Autism and Developmental Disorders 45(3): 766–777.
This work was supported by the Medical Research Council
Hartley SL and Sikora DM (2009) Sex differences in autism
(MRC, UK), the EU Autism Imaging Study (AIMS) network
spectrum disorder: an examination of developmental func-
(grant no. 115300) and the National Institute for health Research
tioning, autistic symptoms, and coexisting behavior prob-
Biomedical Research Centre for Mental Health at King’s College
lems in toddlers. Journal of Autism and Developmental
London, Institute of Psychiatry and South London and Maudsley
Disorders 39(12): 1715–1722.
National Health Service Foundation Trust. This research received
Her Majesty’s Government (2009) Autism Act. London:
infrastructure support from the National Institute for Health
HMSO.
Research (NIHR) Mental Health Biomedical Research Centre at
Hofvander B, Delorme R, Chaste P, et al. (2009) Psychiatric and
South London and Maudsley NHS Foundation Trust and King’s
psychosocial problems in adults with normal-intelligence
College London.
autism spectrum disorders. BMC Psychiatry 9: 35.
C.E.W. is supported by the European Union’s Seventh
Holtmann M, Bölte S and Poustka F (2007) Autism spectrum
Framework Programme via the Marie Curie Action, ‘Co-funding
disorders: sex differences in autistic behaviour domains and
of Regional, National and International Programs’ to stimulate
coexisting psychopathology. Developmental Medicine and
research activities without mobility restrictions, co-financed by
Child Neurology 49(5): 361–366.
the Junta de Andalucía and the European Commission under
Howe YJ, O’Rourke JA, Yatchmink Y, et al. (2015) Female
Talentia Postdoc grant agreement no. 267226.
autism phenotypes investigated at different levels of lan-
guage and developmental abilities. Journal of Autism and
References Developmental Disorders. Epub ahead of print 23 June.
American Psychiatric Association (APA) (2015) DSM-5 DOI: 10.1007/s10803-015-2501-y.
Development (Autism Spectrum Disorder). Available at: Jacquemont S, Coe BP, Hersch M, et al. (2014) A higher muta-
http://www.dsm5.org tional burden in females supports a “female protective
Attwood T (2007) The Complete Guide to Asperger’s Syndrome. model” in neurodevelopmental disorders. American Journal
London: Jessica Kingsley Publishers. of Human Genetics 94(3): 415–425. DOI: 10.1016/j.
Bejerot S, Eriksson JM and Mörtberg E (2014) Social anxiety in ajhg.2014.02.001.
adult autism spectrum disorder. Psychiatry Research 220: Joshi G, Wozniak J, Petty C, et al. (2013) Psychiatric comor-
705–707. bidity and functioning in a clinically referred population of
Bölte S, Duketis E, Poustka F, et al. (2011) Sex differences in cog- adults with autism spectrum disorders: a comparative study.
nitive domains and their clinical correlates in higher-func- Journal of Autism and Developmental Disorders 43(6):
tioning autism spectrum disorders. Autism: The International 1314–1325.
Journal of Research and Practice 15(4): 497–511. Kessler RC, Petukhova M, Sampson NA, et al. (2012) Twelve-
Brugha T, Cooper S and McManus S (2012) Estimating the month and lifetime prevalence and lifetime morbid risk
prevalence of autism spectrum conditions in adults: extend- of anxiety and mood disorders in the United States.
ing the 2007 adult psychiatric morbidity survey. Available International Journal of Methods in Psychiatric Research
at: http://leftbrainrightbrain.co.uk/2012/02/01/estimating- 21(3): 169–184.
Wilson et al. 819

Kirkovski M, Enticott PG and Fitzgerald PB (2013) A review of the National Institute for Health and Clinical Excellence (NICE)
role of female gender in autism spectrum disorders. Journal (2012) Autism: Recognition, Referral, Diagnosis and
of Autism and Developmental Disorders 43: 2584–2603. Management of Adults on the Autism Spectrum (CG142).
Kopp S and Gillberg C (1992) Girls with social deficits and London: NICE.
learning problems: autism, atypical Asperger syndrome or Park S, Cho S-C, Cho IH, et al. (2012) Sex differences in children
a variant of these conditions. European Child & Adolescent with autism spectrum disorders compared with their unaf-
Psychiatry 1(2): 89–99. fected siblings and typically developing children. Research
Lai M-C, Lombardo MV, Auyeung B, et al. (2015) Sex/gender in Autism Spectrum Disorders 6(2): 861–870.
differences in autism: setting the scene for future research. Pilowsky T, Yirmiya N, Shulman C, et al. (1998) The Autism
Journal of American Academy of Child and Adolescent Diagnostic Interview–Revised and the Childhood Autism
Psychiatry 54(1): 11–24. Rating Scale: differences between diagnostic systems
Lai M-C, Lombardo MV, Pasco G, et al. (2011) A behavioral and comparison between genders. Journal of Autism and
comparison of male and female adults with high functioning Developmental Disorders 28(2): 143–151.
autism spectrum conditions. PLoS ONE 6(6): e20835. DOI: Russell AJ, Mataix-Cols D, Anson M, et al. (2005) Obsessions
10.1371/journal.pone.0020835. and compulsions in Asperger syndrome and high-function-
Lai M-C, Lombardo MV, Ruigrok ANV, et al. (2012) Cognition ing autism. The British Journal of Psychiatry: The Journal
in males and females with autism: similarities and differ- of Mental Science 186: 525–528.
ences. PLoS ONE 7(10): e47198. Russell AJ, Murphy CM, Wilson CE, et al. (2015) The men-
Lemon JM, Gargaro B, Enticott PG, et al. (2011) Executive func- tal health of individuals referred for assessment of autism
tioning in autism spectrum disorders: a gender comparison spectrum disorder in adulthood: A clinic report. Autism.
of response inhibition. Journal of Autism and Developmental Epub ahead of print 15 October 2015. DOI: 10.1177/
Disorders 41(3): 352–356. 1362361315604271.
Levy A and Perry A (2011) Outcomes in adolescents and adults with Russell G, Rodgers LR, Ukoumunne OC, et al. (2014) Prevalence
autism: a review of the literature. Research in Autism Spectrum of parent-reported ASD and ADHD in the UK: findings
Disorders 5(4): 1271–1282. DOI: 10.1016/j.rasd.2011.01.023. from the millennium cohort study. Journal of Autism and
Lord C, Risi S, Lambrecht L, et al. (2000) The Autism Diagnostic Developmental Disorders 44: 31–40.
Observation Schedule–Generic: a standard measure of Rydén E and Bejerot S (2008) Autism spectrum disorder in an
social and communication deficits associated with the adult psychiatric population. A naturalistic cross sectional
spectrum of autism. Journal of Autism and Developmental controlled study. Clinical Neuropsychiatry 5: 13–21.
Disorders 30(3): 205–223. Solomon M, Miller M, Taylor SL, et al. (2012) Autism symp-
Lord C, Rutter M and Couteur A (1994) Autism Diagnostic toms and internalizing psychopathology in girls and boys
Interview–Revised: a revised version of a diagnostic with autism spectrum disorders. Journal of Autism and
interview for caregivers of individuals with possible per- Developmental Disorders 42(1): 48–59.
vasive developmental disorders. Journal of Autism and Van Wijngaarden-Cremers PJM, Van Eeten E, Groen WB, et al.
Developmental Disorders 24(5): 659–685. (2014) Gender and age differences in the core triad of impair-
Lord C, Schopler E and Revicki D (1982) Sex differences in ments in autism spectrum disorders: a systematic review
autism. Journal of Autism and Developmental Disorders and meta-analysis. Journal of Autism and Developmental
12(4): 317–330. Disorders 44(3): 627–635.
McLennan JD, Lord C and Schopler E (1993) Sex differences in Wechsler D (1997) Wechsler Adult Intelligence Scale-III (WAIS-
higher functioning people with autism. Journal of Autism III). San Antonio, TX: Psychological Corporation.
and Developmental Disorders 23(2): 217–227. Wechsler D (1999) Wechsler Abbreviated Scale of Intelligence.
Mandy W, Charman T, Gilmour J, et al. (2011) Toward specify- Harcourt Assessments. San Antonio, TX: Psychological
ing pervasive developmental disorder – not otherwise speci- Corporation.
fied. Autism Research 4: 121–131. Wilson CE, Gillan N, Spain D, et al. (2013) Comparison of
Mandy W, Chilvers R, Chowdhury U, et al. (2012) Sex differ- ICD-10R, DSM-IV-TR and DSM-5 in an Adult Autism
ences in autism spectrum disorder: evidence from a large Spectrum Disorder Diagnostic Clinic. Journal of Autism
sample of children and adolescents. Journal of Autism and and Developmental Disorders 43(11): 2515–2525.
Developmental Disorders 42(7): 1304–1313. Wilson CE, Roberts G, Gillan N, et al. (2014) The NICE guide-
Mazzone L, Ruta L and Reale L (2012) Psychiatric comorbidities line on recognition, referral, diagnosis and management of
in Asperger syndrome and high functioning autism: diagnos- adults on the autism spectrum. Advances in Mental Health
tic challenges. Annals of General Psychiatry 11(16): 1–13. and Intellectual Disabilities 8(1): 3–14.
Murphy DGM, Beecham J, Craig M, et al. (2011) Autism in adults. World Health Organization (WHO) (1993) The ICD-10
New biological findings and their translational implications Classification of Mental and Behavioural Disorders:
to the cost of clinical services. Brain Research 1380: 22–33. Diagnostic Criteria for Research. Geneva: WHO.