Distinguishing Features of Autism in Boys With Fragile

Journal of Intellectual Disability Research
894
doi: 10.1111/j.1365-2788.2010.01315.x
volume 54 part 10 pp 894905 october 2010
Distinguishing features of autism in boys with fragile

X syndrome
jir_1315
894..905
M. Brock & D. Hatton

FPG Child Development Institute, University of North Carolina, Chapel Hill, NC, USA
Abstract
Background Males with fragile X syndrome and
autism (FXS/autism) represent a distinct subgroup
of males with FXS at risk for markedly poorer outcomes. Early identification and intervention can
improve outcomes for males with autism spectrum
disorder.
Method To advance the development of a specialised autism screening tool for young males with
FXS that could assist in early identification, backward regression was used to identify the combination of parent-report questionnaire items that best
predicted autism symptoms in a sample of 60 males
with FXS, ages 418 years old.
Results Both social and repetitive behaviours distinguished males with FXS/autism, with repetitive
behaviours playing a more prominent role than previously documented in the literature.
Conclusions Healthcare workers and early interventionists may be able to interview parents about a
few key behaviours to determine if young child with
FXS should be formally evaluated for autism.
Evidence-based practices identified for children
with autism spectrum disorder can be implemented
as early as possible.
Correspondence: Mr Matthew Brock, University of North Carolina, Campus Box 8040, Chapel Hill, NC 27599-8040, USA
(e-mail: matthew.brock@unc.edu).
Keywords autism, autism screening, early

identification, fragile X syndrome
Introduction
Males with fragile X syndrome (FXS) are at higher
risk for autism compared with typically developing
children (Rogers et al. 2001; Demark et al. 2003;
Hagerman 2006; Hatton et al. 2006; Lewis et al.
2006; Clifford et al. 2007) and children with other
common (>1/10 000) developmental disabilities
(Collacott et al. 1992; Kent et al. 1999; Abrahams &
Geschwind 2008). Males with comorbid FXS and
autism (FXS/autism) are likely to have poorer outcomes than males with only FXS (Bailey et al.
2000, 2001; Rogers et al. 2001; Hatton et al. 2002,
2006; Kau et al. 2004; Lewis et al. 2006) or only
autism (Bailey et al. 2000; Rogers et al. 2001).
Because research demonstrates that early intervention improves outcomes for young children with
autism (National Research Council 2001), it is
imperative that males with FXS/autism be identified
as early as possible.
The purpose of this study was to review research
literature to identify behaviours that appear to distinguish males with FXS/autism from those with
FXS only and to determine if these behaviours
actually identify a subgroup with FXS/autism in a
sample of 60 males with FXS. Based on a literature
2010 The Authors. Journal of Intellectual Disability Research 2010 Blackwell Publishing Ltd
volume 54 part 10 october 2010
895
M. Brock & D. Hatton Distinguishing features
review of specific behavioural features that appeared

to distinguish males with comorbid FXS/autism
from males with only FXS, specific items from
behavioural questionnaires were selected as potential predictors of autism, and backwards regression
analysis was used to identify a combination of items
that best predicted autistic status.
Fragile X syndrome, the most common inheritable
genetic cause of intellectual disability (ID), affects
approximately 1 in 2500 males (Crawford 2001).
FXS is caused by an abnormal nucleotide repetition
at Xq27.3 on the X chromosome that impairs the
production of fragile X mental retardation protein,
resulting in abnormalities in brain development and
function. Physical characteristics in males with FXS
may include unusually high foreheads, asymmetrical
faces, large jaws, long protruding ears and large testicles after puberty. Males, having only one X chromosome, are more frequently affected by FXS and
tend to have more severe impairments (Crawford).
Because of its known genetic underpinnings, FXS
can be detected through DNA tests using blood
samples, or prenatally through chorionic villus sampling or amniocentesis (Crawford).
Unlike FXS, a definitive genetic marker for
autism has not been identified; current research
suggests that autism most likely involves multiple
genes (Happ et al. 2006). Without a known aetiology, autism is defined behaviourally by impairments
in social interaction and communication, and by
restricted interests and/or repetitive behaviours
(APA 2000).
The reported prevalence of autism in males with
FXS ranges from 18% to 47%, depending on the
diagnostic tool and sample size (Rogers et al. 2001;
Demark et al. 2003; Hagerman et al. 2006; Hatton
et al. 2006; Lewis et al. 2006; Clifford et al. 2007).
When the category is broadened to include autism
spectrum disorders (ASDs), estimates are as high as
67% (Clifford et al.). Ninety percent of males with
FXS display at least one behaviour that is characteristic of autism (Hagerman 2002), suggesting that
most boys with FXS, even those who do not meet
criteria for autism, display some autistic behaviours
such as gaze aversion (Cohen et al. 1989; Hessl
et al. 2006).
Males who meet diagnostic criteria for autism
represent a distinct subgroup at risk for markedly
poorer outcomes (Bailey et al. 2000, 2001; Rogers
et al. 2001; Hatton et al. 2002, 2006; Kau et al.

2004; Lewis et al. 2006). Males with FXS/autism
had poorer social and communication skills and
greater cognitive impairment (Bailey et al. 2000;
Rogers et al. 2001), lower adaptive behaviour scores
(Bailey et al. 2001; Rogers et al. 2001; Hatton et al.
2003; Kau et al. 2004) and more problem behaviours (Hatton et al. 2002; Kau et al. 2004) than
males with FXS alone.
Identification of autism in young males with FXS

There is consensus that early identification and
intervention lead to improved outcomes for children
with autism (National Research Council 2001).
Even though FXS is a known risk factor for autism,
there is no simple screening instrument specific to
FXS that can be used to efficiently aid in the early
identification process. Because 90% of males with
FXS display at least one autistic behaviour (Hagerman 2002), and 67% eventually meet criteria for
autistic spectrum disorder (Clifford et al. 2007),
those who will later display the most severe autistic
behaviour and meet criteria for autism disorder may
be difficult to distinguish at a young age. Infants
who are later diagnosed with autism may more
closely resemble infants with developmental delays
than infants with idiopathic autism (Baranek et al.
2005).
To date, the only published study to use an
autism screening tool with a group of males with
FXS involved the Checklist for Autism in Toddlers
(CHAT; Baron-Cohen et al. 1992), a screening
tool with 14 yes/no items intended to be used by
general practitioners to screen for autism during
18-month developmental check-ups. The CHAT
only correctly flagged 50% of children who met
DSM-IV autism criteria (Scambler et al. 2007).
Although sensitivity was poor, the CHAT was
highly specific (100%). Scambler and his colleagues used a relatively small sample of children
with FXS (n = 17), so generalisation of their findings may be limited. Studies of males with idiopathic autism demonstrated a similarly high level
of specificity (98%) and an even lower level of
sensitivity (38%) (Baron-Cohen et al.). The CHAT
does not provide an ideal level of sensitivity for
screening males who already have a known risk
factor for autism, such as FXS.

896
Selecting screening items: features unique to

males with FXS/Autism
In sum, there is a body of evidence demonstrating
that males with comorbid FXS/autism are at risk
for markedly poorer outcomes than males with
FXS alone, and there is reason to believe that
early identification and intervention would improve
these outcomes. However, accurately screening
males with FXS at an early age for autism
remains a challenge, as current screening tools
may not be adequate. To advance understanding
of young males with FXS who will be later diagnosed with autism, a systematic review of the
research was used to identify the specific behaviours and risk factors that differentiate this group
by area of impairment: communication, social
behaviour, and repetitive behaviour and restricted
interests. Studies were included that had samples
of 15 or more young males with FXS, directly
compared a group of males with comorbid FXS/
autism with those with only FXS, demonstrated
significant differences between groups using
accepted statistical methods, and were published
after the year 2000. Most of the reviewed studies
compared individuals with FXS with and without
autistic disorder, although some use the broader
classification of ASD, and are explicitly noted.
Communication
To date, no factors relating to communication that
consistently predict autism status in males with
FXS across studies have been reported. Some
researchers have found that while most boys with
FXS score higher on receptive than expressive language measures, boys with FXS/autism lack this
relative strength in receptive language (Philofsky
et al. 2004; Lewis et al. 2006). However, other
researchers using the same or similar measures did
not find significant differences between receptive
and expressive language skills (Roberts et al. 2001;
Price et al. 2007). Even if findings were similar
across studies, most of the researchers described
broad findings detected only after administering
entire diagnostic batteries; closer examination of
sub-scales and individual items did not reveal any
specific features that were strong predictors of
autism.
Social behaviour
A number of research groups suggest that social
behaviour may predict autism status in males with
FXS (Kau et al. 2004; Kaufmann et al. 2004;
Budimirovic et al. 2006; Roberts et al. 2007). Adaptive socialisation and social withdrawal have been
identified as independent predictors of ASD in FXS
(Budimirovic et al. 2006), with the most predictive
items relating to recognising and responding to
emotions, and social avoidance (Kau et al. 2004;
Budimirovic et al. 2006). Roberts et al. (2007) theorised that while most males with FXS are socially
anxious and withdraw in novel situations, males
with FXS/autism are more likely to display social
withdrawal regardless of whether the situation is
novel or familiar. When Kaufmann et al. (2004)
analysed items from the Autism Diagnostic
Interview-Revised items reflecting imaginative play
and peer interaction were most predictive of autism
diagnosis (Kaufman et al.).
Repetitive behaviour and restricted interests
No studies meeting the criteria for this literature
review were found that suggest repetitive behaviours
and restricted interests predict autism status in
males with FXS. Although Baranek et al. (2005) did
not compare subgroups of males with FXS with
and without autism, their comparison of very young
males with FXS, idiopathic autism and other developmental disabilities yielded some unique distinguishing features. Using retrospective video analysis,
they found that 9- to 12-month-old males with FXS
were best distinguished from males with other
developmental disabilities by unusual object play
(including spinning objects) and unusual motor patterns (including repetitive leg movements). Because
these features relate to repetitive behaviours and
restricted interests, a defining category of autistic
behaviour, it is possible that they might distinguish
males with FXS/autism from males with only FXS.
Research aims
If males with FXS/autism are to achieve optimal
outcomes, early identification and intervention are
paramount. Current autism screening tools are not
sensitive enough to identify the majority of males
with FXS who will later be diagnosed with autism

897
(Scambler et al. 2007). A review of the literature

demonstrates that behavioural features may distinguish males with FXS/autism. The purpose of this
study was to identify parent questionnaire items
related to these distinguishing traits and to determine the extent to which they predict autism
symptom severity in males with FXS. Parent questionnaires have the potential to provide a quick and
efficient method for identifying males with FXS
who appear to be at risk for autism, leading to
earlier referrals for formal autism assessment. Based
on the most promising findings in the literature,
three major categories of behaviour were identified
as being most likely to distinguish males with FXS/
autism: (1) recognising emotions in self and
others (Kau et al. 2004; Kaufmann et al. 2004;
Budimirovic et al. 2006); (2) social anxiety, withdrawal and avoidance (Roberts et al. 2001; Kau
et al. 2004; Kaufmann et al. 2004; Budimirovic et al.
2006); and (3) imaginative play, social play and
object play (Kaufmann et al. 2004; Baranek et al.
2005).
Method
Participants
The participants were a subgroup from the Carolina Fragile X Project who recently participated in
a pilot study of the genetics of ASD in males with
FXS, for which informed consent had been
obtained. De-identified data were used for this
study. Participants included 60 boys with full
mutation FXS between 4 and 18 years old
(M = 11.9; SD = 4.4) who were of European
American (n = 53) and African American (n = 7)
descent. The process by which participants were
recruited and identified as having full mutation
FXS has been described elsewhere (Hatton et al.
2006). Three parent-report measures were collected for each participant: the Social Responsiveness Scale (SRS; Constantino & Gruber 2005),
the Social Communication Questionnaire (SCQ;
Rutter et al. 2003) and the Repetitive Behavior
Scale (RBS; Bodfish et al. 1999). In addition,
scores from the Childhood Autism Rating Scale
(CARS; Schopler et al. 1988) were used to
describe autism symptom severity.
Measures
The SRS (Constantino & Gruber 2005) is a
65-item parent-report questionnaire designed to
screen for autism, Aspergers, PDD-NOS and schizoid personality disorder of childhood. Items are
rated on a 5-point Likert scale and summed to yield
five sub-scale scores (social awareness, social cognition, social communication, social motivation and
autistic mannerisms) and a total score.
A companion screening tool for the Autism
Diagnostic Interview-Revised (Lord et al. 1994),
the SCQ (Rutter et al. 2003) is a 40-item parentreport questionnaire designed to provide a dimensional measure of autism symptoms, compare
overall levels of autism symptoms across samples,
and to approximate severity of autism symptoms.
Parents respond yes or no to each item, and items
are summed to yield three sub-scale scores (reciprocal social interaction; communication; and
restricted, repetitive and stereotyped patterns of
behaviour) and a total score.
The RBS (Bodfish et al. 1999) assesses the presence and severity of abnormal repetitive behaviours
and restricted interests associated with disorders
such as ASDs. Items are rated on a 5-point Likert
scale and summed to yield five sub-scale scores
(stereotypic behaviour, self-injurious behaviour,
compulsive behaviour, ritualistic/sameness behaviour and restricted interests) and a total score using
a revised scoring algorithm. The RBS is an experimental instrument, and its psychometric properties
have been assessed in an independent validation
study (Lam & Aman 2007). RBS sub-scales showed
high internal consistency (0.780.91) and fair to
good inter-rater reliability (0.570.73; Lam & Aman
2007).
On the CARS (Schopler et al. 1988), professionals rate 15 items on a scale from 1 (within normal
limits) to 4 (severely abnormal) to measure autism
symptom severity, with individuals scoring 30 or
above being considered autistic (Schopler et al.).
One of the advantages of the CARS is that it yields
a single score that represents a continuous measure
of autistic behaviour (Bailey et al. 2001; Hatton
et al. 2009). Multiple research groups have used the
CARS effectively as a measure of autism symptom
severity in subjects with FXS (Levitas et al. 1983;
Bailey et al. 2001; Demark et al. 2003; Hatton et al.

898
2009). Because the participants participated in the

Carolina Fragile X Project, a longitudinal study, the
CARS had been administered on multiple occasions. For this analysis, CARS scores were averaged
to yield a single score for each participant. Each
subject had been assessed with the CARS between
one and eight times (M = 3.5; SD = 1.5), with scores
ranging from 17.5 to 50.5 (n = 209; M = 27.2;
SD = 5.7). Other research groups have also used
average CARS scores for the purpose of statistical
analysis (Stone & Caro-Martinez 1990). The procedure by which CARS scores were obtained and the
rationale for using the CARS have been described
elsewhere (Hatton et al. 2006).
Table 1 Items that relate to recognising emotions in self and

others*
Test/item
number
SRS7
SRS12
SRS15
SRS26
SRS38
SRS60
SCQ27
Procedure
Variables
A systematic literature review identified three major
categories of behaviour that may distinguish boys
with FXS who have autism: (1) recognising emotions in self and others (Kau et al. 2004; Kaufmann
et al. 2004; Budimirovic et al. 2006); (2) social
anxiety, withdrawal and avoidance (Roberts et al.
2001; Kau et al. 2004; Kaufmann et al. 2004;
Budimirovic et al. 2006); and (3) imaginative play,
social play and object play (Kaufmann et al. 2004;
Baranek et al. 2005). The three parent-reported
questionnaires (SRS, SCQ and RBS) were examined, and items that related to one of the three
identified categories of behaviour were identified, as
summarised in Tables 13.
Statistical analysis
Predictors identified in Table 1 were entered as
independent variables into a regression model with
the CARS total score as the dependent variable,
and items with significant unique contributions
were identified using backward regression. This
process was repeated for predictor variables in
Tables 2 and 3, so that three groups of optimal predictors were identified from each of the three
respective categories. These three groups of predictors were entered into a regression model, and
backward regression again was used to find the
combination of predictor items that best explained
the variance in CARS scores.
SCQ31
Item
Is aware of what others are thinking or feeling
Is able to communicate his or her feelings to
others
Is able to understand the meaning of other
peoples tone of voice and facial expressions
Offers comfort to others when they are sad
Responds appropriately to mood changes in
others
Is emotionally distant, doesnt show his or her
feelings
When she/he was 4 to 5, did she/he smile back
if someone smiled at him/her?
When she/he was 4 to 5, did she/he ever try to
comfort you when you were sad or hurt?
* Budimirovic et al. 2006.
Kau et al. 2004.
Kaufmann et al. 2004.
In backward regression, a type of stepwise regression, all independent variables are initially included
in the model. The independent variable with the
smallest partial correlation coefficient is identified
and removed from the model if it fails to meet the
selection criterion (P < 0.10). Then a new model is
generated, and the process is repeated until the
variable with the smallest partial correlation coefficient meets the minimum criterion (Cohen et al.
2003). Backward regression was selected for this
analysis, as opposed to other types of stepwise
regression, because the elimination process effectively ranks the independent variables in order of
unique contribution to the model, aiding in the
interpretation of findings.
Regression models were generated for each of the
screening questionnaires (the SCQ and SRS), with
the total score the sole predictor, and the CARS
total score the dependent variable. An R2 statistic
was generated for each regression equation, representing how well each screening tools total score
predicted the variance in CARS scores. Then the
prediction ability of the identified combination of
items was compared with that of the SCQ and SRS
total scores, because the goal is for a specialised

899
Table 2 Items that relate to social anxiety*, withdrawal and

avoidance
Table 3 Items that relate to imaginative play*, social play* and object
play
Test/item
number
Item
Test/item
number
SRS1
SRS20
SRS3
SRS6
SRS16
SRS23
SRS27
SRS34
SRS45
SRS64
SCQ19
SCQ20
SCQ26
SCQ28
SCQ36
SCQ37
Seems much more fidgety in social situations than

when alone
Seems self-confident when interacting with others
Would rather be alone than with others
Avoids eye contact or has unusual eye contact
Does not join group activities unless told to do so
Avoids starting social interactions with peers or
adults
Avoids people who want to be emotionally close
to him or her
Focuses his or her attention to where others are
looking or listening
Is too tense in social settings
Does she/he have any particular friends or best
friend?
When she/he was 4 to 5, did she/he ever talk with
you just to be friendly (rather than to get
something)?
When she/he was 4 or 5, did she/he usually look
at you directly in the face when doing things
with you or talking with you?
When she/he was 4 to 5, did she/he ever show
you things that interested her/him to engage
your attention?
When she/he was 4 to 5, did she/he seem
interested in other children of approximately
the same age whom she/he did not know?
When she/he was 4 to 5, did she/he respond
positively when another child approached
her/him?
SRS22
SRS40
SCQ12
SCQ29
SCQ34
SCQ35
SCQ39
SCQ40
RBS5
RBS42
RBS43
* Roberts et al. 2001.
Budimirovic et al. 2006.
Kau et al. 2004.
Kaufmann et al. 2004.
screening tool to outperform existing screening

instruments.
Because the wide age range of the sample could
possibly confound interpretation of the results, the
backward regression analysis was replicated with
two subgroups sorted by age, to determine if the
results of the analysis would differ for younger and
older males with FXS. The younger group (n = 30)
ranged from 4 to 12 years old (M = 7.5; SD = 2.8),
while the older group (n = 30) ranged from 13 to 18
years old (M = 15.5; SD = 1.8).
Item
Shows unusual sensory interests or strange ways
of playing with toys
Plays appropriately with children his or her own
age
Is imaginative, good at pretending
Has she/he ever seemed more interested in parts
of a toy or an object (e.g. spinning the wheels
of a car), rather than using the object as it was
intended?
When she/he was 4 to 5, did she/he ever offer to
share things other than food with you?
When she/he was 4 to 5, did she/he ever
spontaneously join in and try to copy the
actions of social games, such as The Mulberry
Bush or London Bridge Is Falling Down?
When she/he was 4 to 5, did she/he play pretend
or make-believe games?
When she/he was 4 to 5, did she/he ever play
imaginative games with another child in a way
that you could tell that they each understood
what the other was doing?
When she/he was 4 to 5, did she/he play
cooperatively in games that required joining in
with a group of other children such as
hide-and-seek or ball games?
Object usage (spins or twirls objects, twiddles or
slaps or throws objects, lets objects fall out of
hands)
Preoccupation with part(s) of object rather than
the whole object
Fascination, preoccupation with movement/things
that move (e.g. fans, clocks)
* Kaufmann et al. 2004.
Baranek et al. 2005
Results
Phase I: regression analysis within categories
First, backward regression was used to identify the
optimal combination of predictors relating to recognising emotions in self and others (see Table 1).
The identified predictors, SRS12 (does not
communicate feelings) and SRS26 (does not offer
comfort), were somewhat predictive of CARS
scores (R2 = 0.28). When backward regression was
used to identify an optimal combination of predic-

900
Table 4 Phase II backward regression statistics
Model
1
CARS ~ SRS23 + SRS45 + SCQ19 + SRS20 + SRS34 + RBS5 + SRS12* + SRS26
CARS ~ SRS23 + SRS45 + SCQ19 + SRS20 + SRS34* + RBS5 + SRS26
CARS ~ SRS23 + SRS45 + SCQ19 + SRS20* + RBS5 + SRS26
CARS ~ SRS23 + SRS45 + SCQ19* + RBS5 + SRS26
CARS ~ SRS23 + SRS45 + RBS5 + SRS26
Item
SE
SRS23
SRS45
SCQ19
SRS20
SRS34
RBS5
SRS12*
SRS26
SRS23
SRS45
SCQ19
SRS20
SRS34*
RBS5
SRS26
SRS23
SRS45
SCQ19
SRS20*
RBS5
SRS26
SRS23
SRS45
SCQ19*
RBS5
SRS26
SRS23
SRS45
RBS5
SRS26
1.308
-1.435
1.449
0.744
0.433
1.260
0.114
-1.466
1.293
-1.423
1.458
0.729
0.459
1.251
-1.464
1.334
-1.422
1.345
0.586
1.288
-1.342
1.453
-1.542
1.536
1.638
-1.364
1.572
-1.717
1.719
-1.455
0.489
0.588
0.991
0.559
0.566
0.685
0.648
0.525
0.476
0.578
0.980
0.547
0.542
0.677
0.519
0.473
0.577
0.968
0.519
0.674
0.498
0.462
0.569
0.956
0.601
0.499
0.462
0.566
0.607
0.503
2.677
-2.441
1.463
1.332
0.765
1.838
0.177
-2.795
2.714
-2.460
1.488
1.334
0.848
1.848
-2.819
2.823
-2.465
1.389
1.131
1.911
-2.696
3.146
-2.713
1.606
2.727
-2.734
3.400
-3.033
2.832
-2.895
0.010
0.018
0.150
0.189
0.448
0.072
0.860
0.007
0.009
0.017
0.143
0.188
0.400
0.070
0.007
0.007
0.017
0.171
0.263
0.062
0.009
0.003
0.009
0.114
0.009
0.008
0.001
0.004
0.006
0.005
* Least significant predictor in model with P > 0.10; removed from subsequent models.
tors that relate to social anxiety, withdrawal and

avoidance (see Table 2), three items were identified:
SRS23 (avoids group activities), SRS45 (lack of
response to joint attention) and SCQ19 (lack of
best friend). This combination of items was moderately predictive of CARS scores (R2 = 0.44). Finally,
three items were identified that relate to imaginative
play, social play and object play (see Table 3):
SCQ34 (lack of participation in social games),
RBS5 (stereotypical object usage) and SRS20
(unusual sensory interests or object play). This
combination of three items was moderately predictive of CARS scores (R2 = 0.48).
Phase II: regression analysis among categories

The eight items identified from the three categories
were entered into a backward regression model with
CARS scores as the dependent variable, as summarised in Table 4. Four items were identified as an
optimal combination of predictors: SRS23 (avoids
group activities), SRS45 (lack of response to joint
attention), RBS5 (stereotypical object usage) and
SRS26 (does not offer comfort). These items were
highly intercorrelated (see Table 5) but represented
all three categories. This optimal combination of
predictors explained more variance in the CARS
scores (R2 = 0.59) than the SCQ (R2 = 0.24) or SRS
(R2 = 0.27) for this sample of boys with FXS.
Although the results from the backward regression with younger and older subgroups were not
identical to the findings for the larger group, they
were similar in that they both identified items that
relate to social impairments and repetitive behaviours. For the younger subgroup, the optimal com-

901
Table 5 Intercorrelations between items identified in Phase II analysis
Item
SRS23
SRS45
RBS5
SRS26
SRS23 Does not join group activities unless told to do so

SRS45 Focuses his or her attention to where others are looking or listening
RBS5 Object usage (spins or twirls objects, twiddles or slaps or throws objects, lets
objects fall out of hands)
SRS26 Offers comfort to others when they are sad
-0.104
0.252
-0.352**
-0.145
0.205
-0.443**
* P < 0.05,
** P < 0.01.
bination of predictors included items representing

increased repetitive behaviour (RBS5 stereotypical
object usage) and social impairments (SCQ19
lack of best friend, SCQ34 lack of participation in
social games). In the older subgroup, optimal predictors included items that related to repetitive
behaviour and restricted interests (SRS20 unusual
sensory interests or object play, RBS43 preoccupation with movement), social anxiety (SRS23
avoids group activities) and social avoidance
(SRS45 lack of response to joint attention). The
associations between CARS scores and items representing repetitive play and social anxiety were stronger in the older subgroup.
scores were analysed as predictors of autism classification. Different cut-off scores were considered by
calculating how many subjects would be correctly
sorted into each group (Autism or No Autism)
as a function of the cut-off score (see Table 6). To
identify the optimal cut-off score, the sensitivity
(percentage of subjects correctly identified in the
Autism group) and specificity (percentage of subjects correctly identified in the No Autism group)
were summed for each cut-off score. The optimal
cut-off score with the greatest sum, 6, was both
highly sensitive (83%) and specific (79%).
Discussion
Sensitivity and specificity
When the analysis was originally planned, computing composite scores and analysing sensitivity and
specificity were not included, as differences in
scaling between the SCQ (2-point yes/no scale) and
the SRS and RBS (5-points Likert scales) posed
methodological problems. However, the final backward regression analysis resulted in items from only
the equivalently scaled SRS and RBS questionnaires. Items were summed to compute a composite
score that could range from 0 to 16.
Composite scores were calculated for all subjects.
Then, subjects were classified into two groups
(Autism and No Autism) based on CARS scores.
Subjects with CARS scores of 30 or greater were
classified into the Autism group, while subjects
with CARS scores less than 30 were classified into
the No Autism group. This resulted in 17 subjects
(28%) in the Autism group, and 43 subjects (72%)
in the No Autism group. Then, the composite
The items identified in the final phase of analysis,

SRS23 (avoids group activities), SRS45 (lack of
response to joint attention), RBS5 (stereotypical
object usage) and SRS26 (does not offer comfort),
represent the combination of items that best predict
autism symptom severity as measured by the
CARS. This combination of items was highly predictive (R2 = 0.59), markedly superior to the ability
of the SCQ (R2 = 0.24) or SRS (R2 = 0.27) in predicting CARS scores in this sample of males with
FXS. As a screening tool, these four items achieve
high sensitivity (82%) and specificity (79%) as
shown in Table 6. However, these items were identified because of their unique prediction abilities with
this particular sample, and it is not clear whether
they would perform similarly in a different sample
of males with FXS. Despite this limitation, the
results of this study suggest that specific types of
parent-report questionnaire items may be useful as
part of a screening tool and provide evidence that

902
Table 6 Sensitivity and specificity of summed score of identified items (n = 60)
Cut-off
Subjects correctly
identified in Autism
Group
Subjects correctly
identified in No Autism
Group
Sensitivity
Specificity
11
10
9
8
7
6*
5
1
5
6
9
11
14
17
43
43
43
40
38
34
24
6%
29%
35%
53%
65%
82%
100%
100%
100%
100%
93%
88%
79%
56%
* Optimal cut-off score (yielded highest sum of sensitivity + specificity).
repetitive behaviours may play a surprisingly prominent role in distinguishing males with FXS who
have autism.
An item representing repetitive behaviour, RBS5
(stereotypical object usage) had the highest partial
correlation coefficient (b = 1.719) and was the most
powerful single predictor in the final model.
Although one research group (Baranek et al. 2005)
found that early object play may differentiate
infants with FXS from infants with other developmental disabilities, their analysis did not include
subgroups of FXS subjects with and without
autism. This study may be the first to support
repetitive behaviours as a key determinant of autism
in FXS. This finding may be extremely valuable, as
repetitive behaviours are observable at a young age
and could play an important role in early
identification.
The results of the regression analysis for older
and younger subgroups did not markedly differ
from the results of the analysis with the full sample.
Repetitive behaviours and restricted interests were
associated with increased CARS scores for both
subgroups, suggesting that repetitive behaviours distinguish males with FXS/autism regardless of age.
The types of items in the two subgroups differed
qualitatively; while the younger subgroup was distinguished by repetitive motor behaviours, the older
subgroup was distinguished by restricted interests
and sensory behaviours. Also, the association
between CARS scores and items representing social
anxiety was stronger in the older subgroup, similar
to previous findings that social withdrawal
(Budimirovic et al. 2006) and overall autism symptoms (Hatton et al. 2006) increase with age in males
with FXS/autism. These qualitative differences in
the types of repetitive behaviours, restricted interests and social behaviours that distinguished the
two subgroups suggests that a specialised screening
instrument would be most effective if it were specific to a certain age group.
While identification of repetitive behaviours as a
key determinant of autism in FXS is an emergent
finding, differences in social behaviours have been
established as a distinguishing feature of males with
FXS/autism (Kau et al. 2004; Kaufmann et al. 2004;
Budimirovic et al. 2006; Roberts et al. 2007). Interrelated social impairments and repetitive behaviours
distinguished males with more severe autism symptoms. Increased repetitive behaviour (RBS5
stereotypical object usage) was associated with
increased social impairment, specifically lack of
joint attention (SRS45 lack of response to joint
attention) and lack of empathy (SRS26 does not
offer comfort). The association between repetitive
behaviour and joint attention may stem from
impairments in imitation skills. Children with FXS/
autism tend to have more deficits in imitation skills,
and these deficits have been linked to impairments
in object play and joint attention (Rogers et al.
2003). Impaired joint attention skills likely result in
fewer opportunities to practice imitation skills. Deficits in imitation may decrease a childs ability to
acquire object play skills, resulting in an increase in
repetitive behaviours. The association between
repetitive behaviour and lack of empathy may stem

903
from impairments in theory of mind. Impairments

in theory of mind, or the ability to impute mental
states to oneself and to others (Baron-Cohen et al.
1985, p. 39), have been found to discriminate males
with FXS who do and do not have autism, even
after controlling for cognitive functioning (Lewis
et al. 2006). Clearly, empathy, which requires
understanding and reacting to someone elses
mental state, is closely tied to theory of mind.
Theory of mind may be indirectly tied to repetitive
behaviour; males without theory of mind may not
respond to social stimuli because it is not motivating, but instead engage in more motivating selfstimulating repetitive behaviours.
Limitations and directions for future research

Several limitations should be considered before generalising the findings of this study. First, candidate
questionnaire items were chosen based on research
reports from multiple groups. Ideally, a factor
analysis of all the questionnaire items would have
been conducted to aid in this process, but the relatively small sample size (n = 60) did not permit this
type of analysis. Next, when exploring sensitivity
and specificity, CARS scores of 30 or greater were
used to assign autism status, rather than a confirmed formal diagnosis from a team with extensive
experience in diagnosing autism. In addition, the
CARS is not considered a gold standard tool for
diagnosing autism; it is better established as a continuous measure of autism symptom severity (Bailey
et al. 2001; Demark et al. 2003; Hatton et al. 2009).
Also, while the combination of identified questionnaire items is highly predictive of both autism status
and symptom severity in this particular sample,
these findings should be replicated with a different
sample before being generalised to the FXS
population.
Conclusion
Both social behaviour (avoiding group activities,
lack of response to joint attention, lack of empathy)
and repetitive behaviours distinguish males with
FXS/autism from males with only FXS. Repetitive
behaviours related to stereotyped object play may
be the best single predictor of autism in males with
FXS. Healthcare workers and early interventionists

may be able to interview parents about a few key
behaviours to determine if young children with
FXS should be formally evaluated for autism.
Better identification of males with FXS/autism will
allow for increased implementation of evidencebased practices that have been demonstrated to
improve outcomes for individuals who have autism.
Acknowledgements
We would like to thank the families that participated in the Carolina Fragile X Project, along with
funding agencies, the National Institute of Child
Health and Human Development, the National
Institute of Health (HD0031110-35SI), and Autism
Speaks. In addition, we would like to thank Samuel
Odom, PhD, Harriet Boone, PhD and Patrick Sullivan, MD, for their contributions to this project.
References
Abrahams B. S. & Geschwind D. H. (2008) Advances in
autism genetics: on the threshold of a new neurobiology. Nature Reviews Genetics 9, 34155. DOI: 10.1038/
nrg2346.
American Psychiatric Association (APA) (2000) Diagnostic
and Statistical Manual of Mental Disorders DSM-IV-TR.
American Psychiatric Association, Washington, DC.
Bailey D. B., Hatton D. D., Mesibov G., Ament N. &
Skinner M. (2000) Early development, temperament,
and functional impairment in autism and fragile X syndrome. Journal of Autism and Developmental Disorders 30,
4960. DOI: 10.1023/A:1005412111706.
Bailey D. B., Hatton D. D., Skinner M. & Mesibov G.
(2001) Autistic behavior, FMR1 protein, and developmental trajectories in young males with fragile X syndrome. Journal of Autism and Developmental Disorders 31,
16574. DOI: 10.1023/A:1010747131386.
Baranek G. T., Danko C. D., Skinner M. L., Bailey D. B.,
Hatton D. D., Roberts J. E. et al. (2005) Video analysis
of sensory-motor features in infants with fragile X syndrome at 912 Months of Age. Journal of Autism and
Developmental Disorders 35, 64556. DOI: 10.1007/
s10803-005-0008-7.
Baron-Cohen S., Leslie A. & Frith U. (1985) Does the
autistic child have a theory of mind? Cognition 21,
3746. DOI: 10.1016/0010-0277(85)90022-8.
Baron-Cohen S., Allen J. & Gillberg C. (1992) Can
autism be detected at 18 months? The needle, the
904
haystack, and the CHAT. The British Journal of Psychiatry 161, 83943. DOI: 10.1192/bjp.161.6.839.
Bodfish J. W., Symons F. W. & Lewis M. H. (1999) The
Repetitive Behavior Scale. Western Carolina Center
Research Reports.
Budimirovic D. B., Bukelis I., Cox C., Gray R. M.,
Tierney E. & Kaufmann W. E. (2006) Autism spectrum
disorder in fragile X syndrome: differential contribution
of adaptive socialization and social withdrawal. American
Journal of Medical Genetics. Part A 140A, 181426. DOI:
10.1002/ajmg.a.31405.
Clifford S., Dissanayake C., Bui Q. M., Huggins R.,
Taylor A. K. & Loesch D. Z. (2007) Autism spectrum
phenotype in males and females with fragile X full
mutation and premutation. Journal of Autism and Developmental Disorders 37, 73847. DOI: 10.1007/s10803006-0205-z.
Cohen I. L., Vietze P. M., Sudhalter V., Jenkins E. C. &
Brown W. T. (1989) Parent-child dyadic gaze patterns in
fragile X males and in non-fragile X males with autistic
disorder. Journal of Child Psychology and Psychiatry 30,
84556. DOI: 10.1111/j.1469-7610.1989.tb00286.x.
Hatton D. D., Hooper S. R., Bailey D. B., Skinner M. L.,

Sullivan K. M. & Wheeler A. (2002) Problem behavior
in boys with fragile X syndrome. American Journal of
Medical Genetics. Part A 108, 10516. DOI: 10.1002/
ajmg.10216.
Hatton D. D., Wheeler A. C., Skinner M. L., Bailey D. B.,
Sullivan K. M., Roberts J. E. et al. (2003) Adaptive
behavior in children with fragile X syndrome. American
Journal on Mental Retardation 108, 37390. DOI:
10.1352/0895-8017(2003)108<373:ABICWF>2.0.CO;2.
Hatton D. D., Sideris J., Skinner M., Mankowski J., Bailey
D. B., Roberts J. et al. (2006) Autistic behavior in children with fragile X syndrome: prevalence, stability, and
the impact of FMRP. American Journal of Medical
Genetics. Part A 140A, 180413. DOI: 10.1002/
ajmg.a.31286.
Hatton D., Wheeler A., Sideris J., Sullivan K., Reichardt
A., Roberts J. et al. (2009) Developmental trajectories of
young girls with fragile X syndrome. American Journal
on Intellectual and Developmental Disabilities 114, 16171.
DOI: 10.1352/1944-7558-114.3.161.
Cohen J., Cohen P., West S. G. & Aiken L. S. (2003)

Applied Multiple Regression/Correlation Analysis for the
Behavioral Sciences. Lawrence Erlbaum Associates, Inc,
Mahwah, NJ.
Hessl D., Glaser B., Dryer-Friedman J. & Reiss A. (2006)

Social behavior and cortisol reactivity in children with
fragile X syndrome. Journal of Child Psychology and
Psychiatry 47, 60210. DOI: 10.1111/j.14697610.2005.01556.x.
Collacott R. A., Cooper S. & McGrother C. (1992) Differential rates of psychiatric disorders in adults with
downs syndrome compared with other mentally handicapped adults. The British Journal of Psychiatry 161,
6714. DOI: 10.1192/bjp.161.5.671.
Kau A., Tierney E., Bukelis I., Stump M. H., Kates

W. R., Trescher W. H. et al. (2004) Social behavior
profile in young males with fragile X syndrome: characteristics and specificity. American Journal of Medical
Genetics. Part A 126A, 917. DOI: 10.1002/ajmg.a.20218.
Constantino J. N. & Gruber C. P. (2005) Social Responsiveness Scale (SRS). Western Psychological Services,
Los Angeles.
Kaufmann W. E., Cortell R., Kau A., Bukelis I., Tierney

E., Gray R. et al. (2004) Autism spectrum disorder in
fragile X syndrome: communication, social interaction,
and specific behaviors. American Journal of Medical
Genetics. Part A 129A, 22534. DOI: 10.1002/
ajmg.a.30229.
Crawford D. (2001) FMR1 and Fragile X Syndrome. Available at: http://www.cdc.gov/genomics/hugenet/file/print/

factsheets/FS_FragileX.pdf (retrieved 1 August 2009).
Demark J., Feldman M. & Holden J. (2003) Behavioral
relationship between autism and fragile X syndrome.
American Journal of Mental Retardation 108, 31426.
DOI: 10.1352/0895-8017(2003)108<314:
BRBAAF>2.0.CO;2.
Hagerman R. J. (2002) Physical and behavioral
phenotype. In: Fragile X Syndrome: Diagnosis, Treatment, and Research, 3rd edn. (eds R. J. Hagerman & P. J.
Hagerman), pp. 387. The John Hopkins University
Press, Baltimore, MD.
Kent L., Evans J., Paul M. & Sharp M. (1999) Comorbidity of autistic spectrum disorders in children with down
syndrome. Developmental Medicine and Child Neurology
41, 1538. DOI: 10.1017/S001216229900033X.
Lam K. & Aman M. (2007) The repetitive behavior scalerevised: independent validation in individuals with
autism spectrum disorders. Journal of Autism and Developmental Disorders 37, 85566. DOI: 10.1007/s10803006-0213-z.
Hagerman R. J. (2006) Lessons from fragile X regarding

neurobiology, autism, and neurodegeneration. Journal of
Developmental and Behavioral Pediatrics 27, 6374. DOI:
10.1097/00004703-200602000-00012.
Levitas A., Hagerman R., Braden M., Rimland B.,

McBogg P. & Matus I. (1983) Autism and the fragile X
syndrome. Journal of Development and Behavioral Pediatrics 4, 1518. DOI: 10.1097/00004703-19830900000002.
Happ F., Ronald A. & Plomin R. (2006) Time to give up

on a single explanation for autism. Nature Neuroscience
9, 121820. DOI: 1218-1220. 10.1038/nn1770.
Lewis P., Abbeduto L., Murphy M., Richmond E., Giles

N., Bruno L. et al. (2006) Cognitive, language and
social-cognitive skills of individuals with fragile X syn-

905
drome with and without autism. Journal of Intellectual

Disability Research 50, 53245. DOI: 10.1111/j.13652788.2006.00803.x.
Lord C., Rutter M. & Le Coteur A. (1994) Autism diagnostic interview-revised: a revised version of a diagnostic
interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and
Developmental Disorders 24, 65985. DOI: 10.1007/
BF02172145.
National Research Council (2001) Educating Children with
Autism. National Academy Press, Washington, DC.
Philofsky A., Hepburn S. L., Hayes A., Hagerman R. &
Rogers S. J. (2004) Linguistic and cognitive functioning
and autism symptoms in young children with fragile X
syndrome. American Journal of Mental Retardation 109,
20818. DOI: 10.1352/0895-8017(2004)109<208:
LACFAA>2.0.CO;2.
Price J., Roberts J., Vandergrift N. & Martin G. (2007)
Language comprehension in boys with fragile X syndrome and boys with down syndrome. Journal of Intellectual Disability Research 51, 31826. DOI: 10.1111/
j.1365-2788.2006.00881.x.
Roberts J. E., Mirrett P. & Burchinal M. (2001) Receptive
and expressive communication development of young
males with fragile X syndrome. American Journal of
Mental Retardation 106, 21630. DOI: 10.1352/08958017(2001)106<0216:RAECDO>2.0.CO;2.
Roberts J., Weisenfeld L., Hatton D., Heath M. & Kaufmann W. (2007) Social approach and autistic behavior
in children with fragile X syndrome. Journal of Autism
and Developmental Disorders 37, 174860. DOI: 10.1007/
s11689-009-9028-5.
Rogers S. J., Wehner E. & Hagerman R. J. (2001) The

behavioral phenotype in fragile X: symptoms of autism
in very young children with fragile X syndrome, idiopathic autism, and other developmental disorders.
Journal of Developmental and Behavioral Pediatrics 22,
40917.
Rogers S., Hepburn S., Stackhouse T. & Wehner E.
(2003) Imitation performance in toddlers with autism
and those with other developmental disorders. Journal
of Child Psychology and Psychiatry 44, 76381. DOI:
10.1111/1469-7610.00162.
Rutter M., Bailey A. & Lord C. (2003) Social Communication Questionnaire (SCQ). Western Psychological Services, Los Angeles.
Scambler D. J., Hepburn S. L., Hagerman R. J. & Rogers
S. J. (2007) A preliminary study of screening for risk of
autism in males with fragile X syndrome: testing two
risk cut-offs for the checklist for autism in toddlers.
Journal of Intellectual Disability Research 51, 26976.
DOI: 10.1111/j.1365-2788.2006.00874.x.
Schopler E., Reichler R. J. & Renner B. R. (1988) The
Childhood Autism Rating Scale (CARS). Western Psychological Services, Los Angeles, CA.
Stone W. & Caro-Martinez L. (1990) Naturalistic observations of spontaneous communication in autistic children. Journal of Autism and Developmental Disorders 20,
43753. DOI: 10.1007/BF02216051.
Accepted 20 July 2010
This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.
Users should refer to the original published version of the material.

Distinguishing Features of Autism in Boys With Fragile

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Distinguishing Features of Autism in Boys With Fragile

Enviado por

Direitos autorais:

Formatos disponíveis

Journal of Intellectual Disability Research

volume 54 part 10 pp 894905 october 2010