Severe mood problems in adolescents with
Journal of Child Psychology and Psychiatry 53:11 (2012), pp 1157–1166
doi:10.1111/j.1469-7610.2012.02600.x
Severe mood problems in adolescents
with autism spectrum disorder
Emily Simonoff, 1 Catherine R.G. Jones, 2 Andrew Pickles, 3 Francesca Happé, 4
Gillian Baird, 5 and Tony Charman6
1
Department of Child and Adolescent Psychiatry, King’s College London, Institute of Psychiatry and NIHR Biomedical
Research Centre for Mental Health, De Crespigny Park, London, UK; 2Department of Psychology, University of Essex,
Wivenhoe Park, Colchester, Essex, UK; 3Department of Biostatistics, King’s College London, Institute of Psychiatry,
London, UK; 4MRC SDGP Research Centre, King’s College London, Institute of Psychiatry, London, UK; 5Guy’s & St
Thomas’ NHS Foundation Trust, Newcomen Centre, London, UK; 6Centre for Research in Autism and Education,
Institute of Education, London, UK
Introduction: Severe mood dysregulation and problems (SMP) in otherwise typically developing youth
are recognized as an important mental health problem with a distinct set of clinical features, family
history and neurocognitive characteristics. SMP in people with autism spectrum disorders (ASDs) have
not previously been explored. Method: We studied a longitudinal, population-based cohort of
adolescents with ASD in which we collected parent-reported symptoms of SMP that included rage, low
and labile mood and depressive thoughts. Ninety-one adolescents with ASD provided data at age
16 years, of whom 79 had additional data from age 12. We studied whether SMP have similar correlates
to those seen in typically developing youth. Results: Severe mood problems were associated with
current (parent-rated) and earlier (parent- and teacher-rated) emotional problems. The number of prior
psychiatric diagnoses increased the risk of subsequent SMP. Intellectual ability and adaptive
functioning did not predict to SMP. Maternal mental health problems rated at 12 and 16 years were
associated with SMP. Autism severity as rated by parents was associated with SMP, but the relationship
did not hold for clinician ratings of autistic symptoms or diagnosis. SMP were associated with difficulty
in identifying the facial expression of surprise, but not with performance recognizing other emotions.
Relationships between SMP and tests of executive function (card sort and trail making) were not
significant after controlling for IQ. Conclusions: This is the first study of the behavioural and cognitive
correlates of severe mood problems in ASD. As in typically developing youth, SMP in adolescents
with ASD are related to other affective symptoms and maternal mental health problems. Previously
reported links to deficits in emotion recognition and cognitive flexibility were not found in the
current sample. Further research is warranted using categorical and validated measures of
SMP. Keywords: Severe mood dysregulation, mood disorders, childhood autism, autism spectrum
disorder, SNAP.
Introduction
Severe mood problems (SMP) in children and adolescents include high levels of irritability, often
manifested by temper tantrums, as well as low and
labile mood; together, these have been identified as
an important cause of psychosocial impairment.
Debate has raged about the aetiology of mood dysregulation symptoms, most specifically the extent to
which these are best conceptualized as part of the
spectrum of juvenile bipolar disorder, attention deficit hyperactivity disorder (ADHD) or as a separate
syndrome (Leibenluft, 2011). Leibenluft Cohen,
Gorrindo, Brook, & Pine, (2006) argue persuasively
for a new diagnostic category, severe mood dysregulation (SMD), currently under consideration for
DSM-5 (www.dsm5.org). Under current proposals,
SMD would include severe and prominent mood
abnormalities, hyperarousal and increased reactivity
to negative emotional stimuli, with consequent
Conflict of interest statement: No conflicts declared.
functional impairment. In support of this new diagnosis, Leibenluft and colleagues provide evidence for
distinctive presenting and longitudinal clinical features, family history and neurocognitive profile (Leibenluft, 2011). They argue that, while features
aligned with irritability are included in several diagnostic categories, the syndrome of severe and
impairing irritability with predominant negative
mood includes features not adequately captured by
other diagnoses. Unlike classic juvenile bipolar disorder, manic episodes do not appear to be common
adult outcomes in SMD (Brotman et al., 2006),
whereas unipolar depression and anxiety are
(Stringaris et al., 2009). One small family study
failed to find elevated rates of bipolar disorder in
parents of children with SMD, in contrast with the
parents of children with juvenile bipolar disorder
(Brotman et al., 2007). Neurocognitive differences
exist in young people with SMD in relation to:
labelling facial emotions (Guyer et al., 2007);
response to frustration (Rich et al., 2011); and
performance on response reversal paradigms
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
1158
Emily Simonoff et al.
(Dickstein, Finger, Brotman, et al., 2010), with
neural circuitry differences on fMRI from juvenile
bipolar disorder patients in the first two tasks
(Brotman et al., 2010; Rich et al., 2011).
There has recently been an appreciation that other
psychiatric problems frequently occur in people with
autism spectrum disorders (ASDs), with rates as
high as 60–70%. These co-occurring disorders
include high rates of ADHD, anxiety and ODD in
children (Joshi et al., 2010; Leyfer et al., 2006;
Simonoff et al., 2008). The emergence and timing of
different psychiatric problems in ASD has not been
explicity studied, but comparison of cross-sectional
studies of different age groups suggests that
depressive and obsessive-compulsive disorder may
be more common in older adolescents and adults
(Bakken et al., 2010; Mazefsky et al., 2010) and one
longitudinal clinical study showed that affective
disorder was amongst the most common newly
emerging psychiatric disorders in adults with autism
(Hutton, Goode, Murphy, Le Couteur, & Rutter,
2008). Affective disorders in autism have included
bipolar disorder (Bradley & Bolton, 2006; Munesue
et al., 2008), although the ascertainment methods
and sample sizes in these studies do not provide a
conclusion on whether bipolar disorder is disproportionately increased in ASD.
Most of the research on co-occurring psychiatric
symptoms and disorders in people with ASD has
used standardized instruments to measure recognized symptom patterns and diagnoses. The syndrome of SMD has not, to our knowledge, been
previously explored. There are several reasons to
consider this a useful concept to explore in people
with ASD. First, several of the symptom domains
that are increased in people with ASD, including
ADHD and affective disorder, have been associated
with SMD. Second, people with ASD have high levels
of psychosocial impairment that are greater than
would be expected based on their level of intellectual
functioning. While this has often been attributed to
the core autistic deficits, it is an empirical question
whether co-occurring psychiatric problems, such as
SMD, contribute to this psychosocial impairment.
Third, the relationship between low mood and
‘challenging’ behaviour has long been recognized in
intellectual disability, where communication is significantly impaired (Hayes, McGuire, O’Neill, Oliver,
& Morrison, 2011). Challenging behaviour occurs in
10–20% of people with ASDs, affects the entire
intellectual ability spectrum (Emerson et al., 2001),
but its causes are less well-understood than in those
with intellectual disability without ASD. One possibility is that unrecognized mood problems partially
explain high rates of challenging behaviour in ASD.
In the present study, we use data collected from
the Special Needs and Autism Project (SNAP) cohort
(Baird et al., 2006) at age 16 years to create a
measure of SMP. We test whether the psychiatric,
family and neurocognitive correlates to this scale are
J Child Psychol Psychiatry 2012; 53(11): 1157–66
similar in our ASD sample to those seen in typically
developing populations.
Methods
Participants
The sample in the present analyses comprises
ninety-one 16-year olds with ASD from the 158
participants with ASD in the original SNAP cohort. In
addition, longitudinal data from 12 years were
available on 79 of the 91 individuals. As described
previously [see Baird et al., (2006) for details], SNAP
was drawn from a total population cohort of 56,946
children. All those with a current clinical diagnosis of
pervasive developmental disorder (PDD, N = 255) or
considered ‘at risk’ for being an undetected case by
virtue of having a statement of Special Educational
Needs (SEN; N = 1,515) were surveyed using the
Social Communication Questionnaire [SCQ (Rutter,
Bailey, & Lord, 2003)]. A diagnostic assessment of a
stratified sample at 12 years (also 255 individuals)
identified 158 young people with ASD. The follow-up
assessment at 16 years focussed on the cognitive
phenotype of ASD and therefore only those who had
estimated IQs of ‡50 at 12 years were invited to
participate (Charman et al., 2011). From the SNAP
database, 131 possible participants were identified
on the basis of IQ; of these, 19 indicated they were
not interested in participating, 11 could not be contracted and 1 indicated interest, but was not included before the end of the study leaving 100
adolescent participants, for whom 91 had data to
provide an SMP score for these analyses.
For this cohort, consensus clinical ICD-10 ASD
diagnoses at 12 years were made using the Autism
Diagnostic Interview-Revised [ADI-R (Le Couteur
et al., 1989)] and Autism Diagnostic Observation
Schedule-Generic [ADOS-G (Lord et al., 2000)] as well
as IQ, language and adaptive behaviour measures.
The 91 in the contemporaneous analyses included 83
male, 8 female; 48 met consensus criteria for childhood autism and 43 for another ASD. For the subset of
79 included in the longitudinal analyses, 73 were
male and 41 had a diagnosis of autism. The sample
had a mean age of 15 years 6 months (SD =5 months;
range 14 years 8 months–16 years, 9 months) with a
mean time interv al from the 12 year assessment of
4 years 0 months (SD =11 months, range 1 year
7 months–5 years 8 months).
The study was approved by the South East
Research Ethics Committee (05/MRE01/67) and
informed consent was obtained from all participants.
Measures
Questionnaires and interviews. A scale comprising
‘SMP’ was generated a priori from four items on the
parent-reported Profile of Neuropsychiatric Symptoms (PONS), completed at 16 years. The PONS is a
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
doi:10.1111/j.1469-7610.2012.02600.x
62-item questionnaire that assesses the severity and
impact of 31 symptoms commonly reported in children and young people with neurodevelopmental
disorders (Santosh, Baird, Pityaratstian, Tavare, &
Gringras, 2006). For each symptom, a brief definition
is given and the respondent is asked to endorse the
overall frequency and impact on everyday life. Each
component (frequency and impact) is each scored 0–5
(‘not at all’ to ‘all the time’/‘extremely’), with a combined score ranging from 0 to 10. Four items were
included in the SMP scale, taking into consideration,
the proposed DSM-5 criteria: ‘explosive range’, ‘low
mood’, ‘depressive thoughts’ and ‘labile mood’. A
description of the PONS, the presentation of the individual items and the means and ranges for the SNAP
ASD samples are described in the supplementary
online appendix and Supplementary Table S1. The
scale had good internal consistency with a Cronbach’s
a of .92. The raw scale was nonnormally distributed
with a mean of 7.8 (SD 8.0, range 0–36) and a squareroot transformation was applied to generate a more
normally distributed continuous measure with
skewness of 0.16 and kurtosis 2.78. A binary classification divided the top 25% of scores (13–36, N = 24)
from the rest of the distribution (0–12, N = 67). This
threshold was chosen pragmatically because (a) it was
likely to have reasonable power to detect mean differences and (b) it is conservative compared to the
rates of ADHD (28%) and anxiety disorders (42%) reported in this cohort and is therefore, a plausible
threshold to select.
The Strengths and Difficulties Questionnaire [SDQ
(Goodman, Ford, Simmons, Gatward, & Meltzer,
2000)], rated by parents at 12 and 16 years and
teachers at 12 years, was also used to measure
mental health symptoms. The SDQ is a widely used
screening instrument for child psychiatric problems
and its psychometric properties have been established in several samples, including UK (Goodman
et al., 2000) and US studies (Bourdon, Goodman,
Rae, Simpson, & Koretz, 2005). The present analyses
use the hyperactivity, conduct and emotional subscales.
At 12 years, the parent-reported Child and Adolescent Psychiatric Assessment [CAPA (Angold &
Costello, 2000)] was completed on 69 of the present
sample. The CAPA is a semistructured psychiatric
interview and the following diagnostic areas were
included: all anxiety and phobic disorders (including
obsessive-compulsive disorder); major depression
and dysthymic disorder; ODD and conduct disorder
(CD); ADHD; tics/Tourette/trichotillomania; enuresis and encopresis. The prevalence rates and diagnostic correlates have been reported previously
(Simonoff et al., 2008).
Autism severity was assessed in three ways. We
used the diagnostic dichotomy of childhood autism/
other PDD. Clinicians undertaking the review of
autism diagnostic information based on the ADI-R
and ADOS-G described above scored the 12 ICD-10
Mood problems in autism
1159
symptoms that comprise the autism spectrum disorder diagnoses. The Social Responsiveness Scale
[SRS (Constantino et al., 2003)] T scores were used
as a quantitative measure of autism severity, scored
at 12 years in 60 participants and at 16 years in 27,
where data were missing at 12 years.
Adaptive functioning at 12 years was measured
using the Vineland Adaptive Behaviour Scales composite score (Sparrow, Balla, & Cichetti, 1984). A
quantitative measure of the shortfall, or adaptive
‘under-function’, was generated by subtracting the
Vineland score from the full-scale IQ, both measured
at 12 years and standardized to mean of 100, SD 15
and.
Maternal self-reports on the General Health
Questionnaire [GHQ-30 (Goldberg & Muller, 1988)]
when the participants were 12 and 16 years provided a measure of maternal psychiatric symptoms
with particular emphasis on mood, anxiety and somatic difficulties. The Parenting Stress Index [PSI
Short Form; (Abidin, 1995)] measures difficulties in
the parent-child relationship on three subscales:
disturbed child, parental distress and parent-child
dysfunctional interaction. Parental distress was
used, herein, to index the parental component of
stress, as it attempts to measure parental characteristics rather than aspects of the parent-child
relationship, which may be affected by the presence
of an ASD in the child.
Neurocognitive measures. IQ was measured at
12 years with the Wechsler Intelligence Scales for
Children-Third Edition [WISC-IIIUK (Wechsler,
1992)] and at 16 years with the Wechsler Abbreviated Scales of Intelligence [WASI (Wechsler, 1999)].
Details of the neurocognitive tasks are given in the
Supplementary online appendix. All were administered at 16 years. The emotion recognition task has
been previously described (Jones et al., 2011). In the
present analysis, we used the Ekman-Friesen test of
affect recognition (Ekman & Friesen, 1976), as this
was most similar to tasks undertaken in typically
developing youths with SMD (Brotman et al., 2008).
We measured total number of correct responses.
The Card Sort was included as a measure of cognitive flexibility and response reversal (Tregay, Gilmour, & Charman, 2009). The task requires the
participant to correctly sort cards to one of three
alternative sets across three trials, with the correct
set varying in each trial. The key variable was the
number of sorts required to reach criterion. In the
present analyses, we included only those participants who demonstrated an understanding of the
rule in the first trial by reaching criterion before the
end. The number of sorts required in the second and
third trials was divided into four levels: top half
(scores 12–18, N = 42); third quartile (scores 19–24,
N = 22), bottom quartile (scores 25–40, N = 17) and
those who did not reach criterion by the end of both
trials (N = 8).
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
1160
Emily Simonoff et al.
Trail Making was included as a measure of attentional switching and response reversal. The task was
comprised of three separate trials (Reitan & Wolfson,
1985). The participant was asked to ‘join the dots’ in
numerical order, then, in a second trial, in alphabetic order, followed by a third trial switching between numbers and letters. The difference score
between the time taken on the first and the third trial
comprised a measure of switching ability. The mean
difference score was 57.8 (SD 40.7, range 10.5–
229.1). As the data were highly skewed, a squareroot transformed score was used in the present
analyses.
J Child Psychol Psychiatry 2012; 53(11): 1157–66
overall effect and those specific to an emotion.
Ordinal logistic regression was also used for ordinal
outcomes, such as number of diagnoses. For sets of
ordinal items, such as SDQ items, specificity of
association was tested using similar models estimated in gllamm (www.gllamm.org) using a generalized estimating equations approach with an
Independent Working Model. The models allowed
separate threshold parameters for each item and
estimated a common and an item-specific effect in
the manner of testing for differential item functioning. Significance of effects was determined from
Wald tests using the robust form of the parameter
covariance matrix.
Statistical analysis
Data reduction and statistical analysis were
undertaken in Stata version 11 (StataCorp, 2009).
Linear regression was used to examine the continuous outcome of the transformed SMP score and
logistic regression for the binary variable of high
versus low SMP scores. Ordinal logistic regression
was used for the Card Sort, where a 4-level scale
was generated, and for the analyses using the total
number of psychiatric diagnoses on the CAPA.
Multivariate regression, analogous to multiple
analysis of variance, was employed to analyse the
emotion recognition profile to allow for tests of an
Results
Participant characteristics, according to high/low
SMP are shown in Table 1.
Emotional and behavioural characteristics
associated with SMP
Examining the contemporaneous relationships between the three parent-rated mental health problems
domains of the SDQ (Table 2) revealed that hyperactivity, conduct and emotional problems all were
associated with SMP in bivariate analyses, but that
Table 1 Sample characteristics according to severe mood dysregulation and problems (SMP) classification [M (SD)]
High SMP (N = 24)
Raw PONS scores on individual items
Explosive rage
Low mood
Labile mood
Depressive thoughts
Other characteristics at 16 years
Full-scale IQ
SDQc Hyperactivity
SDQc Conduct problems
SDQc Emotional problems
Maternal GHQ score
Other characteristics at 12 years
Adaptive behaviour
Diagnosed childhood autism N (%)
ICD-10 symptom severity
SRSb
SDQc Hyperactivity (parent)
SDQc Conduct problems (parent)
SDQ3 Emotional problems (parent)
CAPA Any emotional problem N (%)
CAPA Oppositional defiant/conduct disorder N (%)
CAPA ADHD N (%)
Maternal GHQ score
Neurocognitive measures at 16
Ekman faces total score
Card sort errors to criterion
Trail making difference score
4.9
4.8
4.8
4.6
(2.4)
(2.3)
(2.8)
(2.8)
Low SMPa (N = 67)
1.2
1.2
0.8
0.6
(1.4)
(1.4)
(1.6)
(1.2)
80.0
6.6
2.6
5.3
8.0
(16.6)
(2.6)
(1.1)
(2.1)
(8.5)
85.8
5.7
1.5
2.9
4.1
(17.5)
(2.4)
(1.6)
(2.3)
(6.3)
50.6
13
8.4
101.3
7.4
3.5
6.3
11
8
8
7.3
(12.3)
(54)
(2.4)
(25.9)
(2.8)
(1.9)
(2.5)
(57.9)
(42.1)
(42.1)
(7.3)
52.1
35
8.0
90.3
7.5
2.9
3.8
13
9
9
4.9
(14.4)
(52)
(2.5)
(22.4)
(2.5)
(2.1)
(2.4)
(26.0)
(18.0)
(16.0)
(6.5)
41.8 (8.2)
24.5 (8.3)
69.1 (48.8)
42.8 (7.7)
21.8 (8.7)
61.9 (43.7)
ADHD, attention deficit hyperactivity disorder; CAPA, Child and Adolescent Psychiatric Assessment; GHQ, General Health
Questionnaire; SDQ, Strengths and Difficulties Questionnaire; SRS, Social Responsiveness Scale.
a
High SMP refers to those scoring in the top quartile, whereas low SMP is the rest of the distribution.
b
Measured at 12 years in 60, at 16 years in 27.
c
Parent-reported measures at 12 and 16 years.
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
doi:10.1111/j.1469-7610.2012.02600.x
Mood problems in autism
1161
Table 2 Personal characteristics under various scales associated with severe mood problems [95 per cent confidence intervals]
Unadjusted
Emotional and behavioural problems at 16 (parent-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Emotional and behavioural problems at 12 (parent-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Emotional and behavioural problems at 12 (teacher-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Other personal characteristics
Full-scale IQ (current)
Adaptive functioning (12 years)
IQ-adaptive functioning discrepancy (12 years)
Social Responsiveness Scale (SRS) Autism severity
ICD-10 total autism symptom score
Maternal distress
Maternal GHQ, 16 years
Maternal GHQ, 12 years
Adjusted
B (95% CIs)
p
b (95% CIs)
p
.15 (.03, .28)
.37 (.19, .56)
.35 (.22, .45)
.02
doi:10.1111/j.1469-7610.2012.02600.x
Severe mood problems in adolescents
with autism spectrum disorder
Emily Simonoff, 1 Catherine R.G. Jones, 2 Andrew Pickles, 3 Francesca Happé, 4
Gillian Baird, 5 and Tony Charman6
1
Department of Child and Adolescent Psychiatry, King’s College London, Institute of Psychiatry and NIHR Biomedical
Research Centre for Mental Health, De Crespigny Park, London, UK; 2Department of Psychology, University of Essex,
Wivenhoe Park, Colchester, Essex, UK; 3Department of Biostatistics, King’s College London, Institute of Psychiatry,
London, UK; 4MRC SDGP Research Centre, King’s College London, Institute of Psychiatry, London, UK; 5Guy’s & St
Thomas’ NHS Foundation Trust, Newcomen Centre, London, UK; 6Centre for Research in Autism and Education,
Institute of Education, London, UK
Introduction: Severe mood dysregulation and problems (SMP) in otherwise typically developing youth
are recognized as an important mental health problem with a distinct set of clinical features, family
history and neurocognitive characteristics. SMP in people with autism spectrum disorders (ASDs) have
not previously been explored. Method: We studied a longitudinal, population-based cohort of
adolescents with ASD in which we collected parent-reported symptoms of SMP that included rage, low
and labile mood and depressive thoughts. Ninety-one adolescents with ASD provided data at age
16 years, of whom 79 had additional data from age 12. We studied whether SMP have similar correlates
to those seen in typically developing youth. Results: Severe mood problems were associated with
current (parent-rated) and earlier (parent- and teacher-rated) emotional problems. The number of prior
psychiatric diagnoses increased the risk of subsequent SMP. Intellectual ability and adaptive
functioning did not predict to SMP. Maternal mental health problems rated at 12 and 16 years were
associated with SMP. Autism severity as rated by parents was associated with SMP, but the relationship
did not hold for clinician ratings of autistic symptoms or diagnosis. SMP were associated with difficulty
in identifying the facial expression of surprise, but not with performance recognizing other emotions.
Relationships between SMP and tests of executive function (card sort and trail making) were not
significant after controlling for IQ. Conclusions: This is the first study of the behavioural and cognitive
correlates of severe mood problems in ASD. As in typically developing youth, SMP in adolescents
with ASD are related to other affective symptoms and maternal mental health problems. Previously
reported links to deficits in emotion recognition and cognitive flexibility were not found in the
current sample. Further research is warranted using categorical and validated measures of
SMP. Keywords: Severe mood dysregulation, mood disorders, childhood autism, autism spectrum
disorder, SNAP.
Introduction
Severe mood problems (SMP) in children and adolescents include high levels of irritability, often
manifested by temper tantrums, as well as low and
labile mood; together, these have been identified as
an important cause of psychosocial impairment.
Debate has raged about the aetiology of mood dysregulation symptoms, most specifically the extent to
which these are best conceptualized as part of the
spectrum of juvenile bipolar disorder, attention deficit hyperactivity disorder (ADHD) or as a separate
syndrome (Leibenluft, 2011). Leibenluft Cohen,
Gorrindo, Brook, & Pine, (2006) argue persuasively
for a new diagnostic category, severe mood dysregulation (SMD), currently under consideration for
DSM-5 (www.dsm5.org). Under current proposals,
SMD would include severe and prominent mood
abnormalities, hyperarousal and increased reactivity
to negative emotional stimuli, with consequent
Conflict of interest statement: No conflicts declared.
functional impairment. In support of this new diagnosis, Leibenluft and colleagues provide evidence for
distinctive presenting and longitudinal clinical features, family history and neurocognitive profile (Leibenluft, 2011). They argue that, while features
aligned with irritability are included in several diagnostic categories, the syndrome of severe and
impairing irritability with predominant negative
mood includes features not adequately captured by
other diagnoses. Unlike classic juvenile bipolar disorder, manic episodes do not appear to be common
adult outcomes in SMD (Brotman et al., 2006),
whereas unipolar depression and anxiety are
(Stringaris et al., 2009). One small family study
failed to find elevated rates of bipolar disorder in
parents of children with SMD, in contrast with the
parents of children with juvenile bipolar disorder
(Brotman et al., 2007). Neurocognitive differences
exist in young people with SMD in relation to:
labelling facial emotions (Guyer et al., 2007);
response to frustration (Rich et al., 2011); and
performance on response reversal paradigms
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
1158
Emily Simonoff et al.
(Dickstein, Finger, Brotman, et al., 2010), with
neural circuitry differences on fMRI from juvenile
bipolar disorder patients in the first two tasks
(Brotman et al., 2010; Rich et al., 2011).
There has recently been an appreciation that other
psychiatric problems frequently occur in people with
autism spectrum disorders (ASDs), with rates as
high as 60–70%. These co-occurring disorders
include high rates of ADHD, anxiety and ODD in
children (Joshi et al., 2010; Leyfer et al., 2006;
Simonoff et al., 2008). The emergence and timing of
different psychiatric problems in ASD has not been
explicity studied, but comparison of cross-sectional
studies of different age groups suggests that
depressive and obsessive-compulsive disorder may
be more common in older adolescents and adults
(Bakken et al., 2010; Mazefsky et al., 2010) and one
longitudinal clinical study showed that affective
disorder was amongst the most common newly
emerging psychiatric disorders in adults with autism
(Hutton, Goode, Murphy, Le Couteur, & Rutter,
2008). Affective disorders in autism have included
bipolar disorder (Bradley & Bolton, 2006; Munesue
et al., 2008), although the ascertainment methods
and sample sizes in these studies do not provide a
conclusion on whether bipolar disorder is disproportionately increased in ASD.
Most of the research on co-occurring psychiatric
symptoms and disorders in people with ASD has
used standardized instruments to measure recognized symptom patterns and diagnoses. The syndrome of SMD has not, to our knowledge, been
previously explored. There are several reasons to
consider this a useful concept to explore in people
with ASD. First, several of the symptom domains
that are increased in people with ASD, including
ADHD and affective disorder, have been associated
with SMD. Second, people with ASD have high levels
of psychosocial impairment that are greater than
would be expected based on their level of intellectual
functioning. While this has often been attributed to
the core autistic deficits, it is an empirical question
whether co-occurring psychiatric problems, such as
SMD, contribute to this psychosocial impairment.
Third, the relationship between low mood and
‘challenging’ behaviour has long been recognized in
intellectual disability, where communication is significantly impaired (Hayes, McGuire, O’Neill, Oliver,
& Morrison, 2011). Challenging behaviour occurs in
10–20% of people with ASDs, affects the entire
intellectual ability spectrum (Emerson et al., 2001),
but its causes are less well-understood than in those
with intellectual disability without ASD. One possibility is that unrecognized mood problems partially
explain high rates of challenging behaviour in ASD.
In the present study, we use data collected from
the Special Needs and Autism Project (SNAP) cohort
(Baird et al., 2006) at age 16 years to create a
measure of SMP. We test whether the psychiatric,
family and neurocognitive correlates to this scale are
J Child Psychol Psychiatry 2012; 53(11): 1157–66
similar in our ASD sample to those seen in typically
developing populations.
Methods
Participants
The sample in the present analyses comprises
ninety-one 16-year olds with ASD from the 158
participants with ASD in the original SNAP cohort. In
addition, longitudinal data from 12 years were
available on 79 of the 91 individuals. As described
previously [see Baird et al., (2006) for details], SNAP
was drawn from a total population cohort of 56,946
children. All those with a current clinical diagnosis of
pervasive developmental disorder (PDD, N = 255) or
considered ‘at risk’ for being an undetected case by
virtue of having a statement of Special Educational
Needs (SEN; N = 1,515) were surveyed using the
Social Communication Questionnaire [SCQ (Rutter,
Bailey, & Lord, 2003)]. A diagnostic assessment of a
stratified sample at 12 years (also 255 individuals)
identified 158 young people with ASD. The follow-up
assessment at 16 years focussed on the cognitive
phenotype of ASD and therefore only those who had
estimated IQs of ‡50 at 12 years were invited to
participate (Charman et al., 2011). From the SNAP
database, 131 possible participants were identified
on the basis of IQ; of these, 19 indicated they were
not interested in participating, 11 could not be contracted and 1 indicated interest, but was not included before the end of the study leaving 100
adolescent participants, for whom 91 had data to
provide an SMP score for these analyses.
For this cohort, consensus clinical ICD-10 ASD
diagnoses at 12 years were made using the Autism
Diagnostic Interview-Revised [ADI-R (Le Couteur
et al., 1989)] and Autism Diagnostic Observation
Schedule-Generic [ADOS-G (Lord et al., 2000)] as well
as IQ, language and adaptive behaviour measures.
The 91 in the contemporaneous analyses included 83
male, 8 female; 48 met consensus criteria for childhood autism and 43 for another ASD. For the subset of
79 included in the longitudinal analyses, 73 were
male and 41 had a diagnosis of autism. The sample
had a mean age of 15 years 6 months (SD =5 months;
range 14 years 8 months–16 years, 9 months) with a
mean time interv al from the 12 year assessment of
4 years 0 months (SD =11 months, range 1 year
7 months–5 years 8 months).
The study was approved by the South East
Research Ethics Committee (05/MRE01/67) and
informed consent was obtained from all participants.
Measures
Questionnaires and interviews. A scale comprising
‘SMP’ was generated a priori from four items on the
parent-reported Profile of Neuropsychiatric Symptoms (PONS), completed at 16 years. The PONS is a
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
doi:10.1111/j.1469-7610.2012.02600.x
62-item questionnaire that assesses the severity and
impact of 31 symptoms commonly reported in children and young people with neurodevelopmental
disorders (Santosh, Baird, Pityaratstian, Tavare, &
Gringras, 2006). For each symptom, a brief definition
is given and the respondent is asked to endorse the
overall frequency and impact on everyday life. Each
component (frequency and impact) is each scored 0–5
(‘not at all’ to ‘all the time’/‘extremely’), with a combined score ranging from 0 to 10. Four items were
included in the SMP scale, taking into consideration,
the proposed DSM-5 criteria: ‘explosive range’, ‘low
mood’, ‘depressive thoughts’ and ‘labile mood’. A
description of the PONS, the presentation of the individual items and the means and ranges for the SNAP
ASD samples are described in the supplementary
online appendix and Supplementary Table S1. The
scale had good internal consistency with a Cronbach’s
a of .92. The raw scale was nonnormally distributed
with a mean of 7.8 (SD 8.0, range 0–36) and a squareroot transformation was applied to generate a more
normally distributed continuous measure with
skewness of 0.16 and kurtosis 2.78. A binary classification divided the top 25% of scores (13–36, N = 24)
from the rest of the distribution (0–12, N = 67). This
threshold was chosen pragmatically because (a) it was
likely to have reasonable power to detect mean differences and (b) it is conservative compared to the
rates of ADHD (28%) and anxiety disorders (42%) reported in this cohort and is therefore, a plausible
threshold to select.
The Strengths and Difficulties Questionnaire [SDQ
(Goodman, Ford, Simmons, Gatward, & Meltzer,
2000)], rated by parents at 12 and 16 years and
teachers at 12 years, was also used to measure
mental health symptoms. The SDQ is a widely used
screening instrument for child psychiatric problems
and its psychometric properties have been established in several samples, including UK (Goodman
et al., 2000) and US studies (Bourdon, Goodman,
Rae, Simpson, & Koretz, 2005). The present analyses
use the hyperactivity, conduct and emotional subscales.
At 12 years, the parent-reported Child and Adolescent Psychiatric Assessment [CAPA (Angold &
Costello, 2000)] was completed on 69 of the present
sample. The CAPA is a semistructured psychiatric
interview and the following diagnostic areas were
included: all anxiety and phobic disorders (including
obsessive-compulsive disorder); major depression
and dysthymic disorder; ODD and conduct disorder
(CD); ADHD; tics/Tourette/trichotillomania; enuresis and encopresis. The prevalence rates and diagnostic correlates have been reported previously
(Simonoff et al., 2008).
Autism severity was assessed in three ways. We
used the diagnostic dichotomy of childhood autism/
other PDD. Clinicians undertaking the review of
autism diagnostic information based on the ADI-R
and ADOS-G described above scored the 12 ICD-10
Mood problems in autism
1159
symptoms that comprise the autism spectrum disorder diagnoses. The Social Responsiveness Scale
[SRS (Constantino et al., 2003)] T scores were used
as a quantitative measure of autism severity, scored
at 12 years in 60 participants and at 16 years in 27,
where data were missing at 12 years.
Adaptive functioning at 12 years was measured
using the Vineland Adaptive Behaviour Scales composite score (Sparrow, Balla, & Cichetti, 1984). A
quantitative measure of the shortfall, or adaptive
‘under-function’, was generated by subtracting the
Vineland score from the full-scale IQ, both measured
at 12 years and standardized to mean of 100, SD 15
and.
Maternal self-reports on the General Health
Questionnaire [GHQ-30 (Goldberg & Muller, 1988)]
when the participants were 12 and 16 years provided a measure of maternal psychiatric symptoms
with particular emphasis on mood, anxiety and somatic difficulties. The Parenting Stress Index [PSI
Short Form; (Abidin, 1995)] measures difficulties in
the parent-child relationship on three subscales:
disturbed child, parental distress and parent-child
dysfunctional interaction. Parental distress was
used, herein, to index the parental component of
stress, as it attempts to measure parental characteristics rather than aspects of the parent-child
relationship, which may be affected by the presence
of an ASD in the child.
Neurocognitive measures. IQ was measured at
12 years with the Wechsler Intelligence Scales for
Children-Third Edition [WISC-IIIUK (Wechsler,
1992)] and at 16 years with the Wechsler Abbreviated Scales of Intelligence [WASI (Wechsler, 1999)].
Details of the neurocognitive tasks are given in the
Supplementary online appendix. All were administered at 16 years. The emotion recognition task has
been previously described (Jones et al., 2011). In the
present analysis, we used the Ekman-Friesen test of
affect recognition (Ekman & Friesen, 1976), as this
was most similar to tasks undertaken in typically
developing youths with SMD (Brotman et al., 2008).
We measured total number of correct responses.
The Card Sort was included as a measure of cognitive flexibility and response reversal (Tregay, Gilmour, & Charman, 2009). The task requires the
participant to correctly sort cards to one of three
alternative sets across three trials, with the correct
set varying in each trial. The key variable was the
number of sorts required to reach criterion. In the
present analyses, we included only those participants who demonstrated an understanding of the
rule in the first trial by reaching criterion before the
end. The number of sorts required in the second and
third trials was divided into four levels: top half
(scores 12–18, N = 42); third quartile (scores 19–24,
N = 22), bottom quartile (scores 25–40, N = 17) and
those who did not reach criterion by the end of both
trials (N = 8).
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
1160
Emily Simonoff et al.
Trail Making was included as a measure of attentional switching and response reversal. The task was
comprised of three separate trials (Reitan & Wolfson,
1985). The participant was asked to ‘join the dots’ in
numerical order, then, in a second trial, in alphabetic order, followed by a third trial switching between numbers and letters. The difference score
between the time taken on the first and the third trial
comprised a measure of switching ability. The mean
difference score was 57.8 (SD 40.7, range 10.5–
229.1). As the data were highly skewed, a squareroot transformed score was used in the present
analyses.
J Child Psychol Psychiatry 2012; 53(11): 1157–66
overall effect and those specific to an emotion.
Ordinal logistic regression was also used for ordinal
outcomes, such as number of diagnoses. For sets of
ordinal items, such as SDQ items, specificity of
association was tested using similar models estimated in gllamm (www.gllamm.org) using a generalized estimating equations approach with an
Independent Working Model. The models allowed
separate threshold parameters for each item and
estimated a common and an item-specific effect in
the manner of testing for differential item functioning. Significance of effects was determined from
Wald tests using the robust form of the parameter
covariance matrix.
Statistical analysis
Data reduction and statistical analysis were
undertaken in Stata version 11 (StataCorp, 2009).
Linear regression was used to examine the continuous outcome of the transformed SMP score and
logistic regression for the binary variable of high
versus low SMP scores. Ordinal logistic regression
was used for the Card Sort, where a 4-level scale
was generated, and for the analyses using the total
number of psychiatric diagnoses on the CAPA.
Multivariate regression, analogous to multiple
analysis of variance, was employed to analyse the
emotion recognition profile to allow for tests of an
Results
Participant characteristics, according to high/low
SMP are shown in Table 1.
Emotional and behavioural characteristics
associated with SMP
Examining the contemporaneous relationships between the three parent-rated mental health problems
domains of the SDQ (Table 2) revealed that hyperactivity, conduct and emotional problems all were
associated with SMP in bivariate analyses, but that
Table 1 Sample characteristics according to severe mood dysregulation and problems (SMP) classification [M (SD)]
High SMP (N = 24)
Raw PONS scores on individual items
Explosive rage
Low mood
Labile mood
Depressive thoughts
Other characteristics at 16 years
Full-scale IQ
SDQc Hyperactivity
SDQc Conduct problems
SDQc Emotional problems
Maternal GHQ score
Other characteristics at 12 years
Adaptive behaviour
Diagnosed childhood autism N (%)
ICD-10 symptom severity
SRSb
SDQc Hyperactivity (parent)
SDQc Conduct problems (parent)
SDQ3 Emotional problems (parent)
CAPA Any emotional problem N (%)
CAPA Oppositional defiant/conduct disorder N (%)
CAPA ADHD N (%)
Maternal GHQ score
Neurocognitive measures at 16
Ekman faces total score
Card sort errors to criterion
Trail making difference score
4.9
4.8
4.8
4.6
(2.4)
(2.3)
(2.8)
(2.8)
Low SMPa (N = 67)
1.2
1.2
0.8
0.6
(1.4)
(1.4)
(1.6)
(1.2)
80.0
6.6
2.6
5.3
8.0
(16.6)
(2.6)
(1.1)
(2.1)
(8.5)
85.8
5.7
1.5
2.9
4.1
(17.5)
(2.4)
(1.6)
(2.3)
(6.3)
50.6
13
8.4
101.3
7.4
3.5
6.3
11
8
8
7.3
(12.3)
(54)
(2.4)
(25.9)
(2.8)
(1.9)
(2.5)
(57.9)
(42.1)
(42.1)
(7.3)
52.1
35
8.0
90.3
7.5
2.9
3.8
13
9
9
4.9
(14.4)
(52)
(2.5)
(22.4)
(2.5)
(2.1)
(2.4)
(26.0)
(18.0)
(16.0)
(6.5)
41.8 (8.2)
24.5 (8.3)
69.1 (48.8)
42.8 (7.7)
21.8 (8.7)
61.9 (43.7)
ADHD, attention deficit hyperactivity disorder; CAPA, Child and Adolescent Psychiatric Assessment; GHQ, General Health
Questionnaire; SDQ, Strengths and Difficulties Questionnaire; SRS, Social Responsiveness Scale.
a
High SMP refers to those scoring in the top quartile, whereas low SMP is the rest of the distribution.
b
Measured at 12 years in 60, at 16 years in 27.
c
Parent-reported measures at 12 and 16 years.
2012 The Authors. Journal of Child Psychology and Psychiatry 2012 Association for Child and Adolescent Mental Health.
doi:10.1111/j.1469-7610.2012.02600.x
Mood problems in autism
1161
Table 2 Personal characteristics under various scales associated with severe mood problems [95 per cent confidence intervals]
Unadjusted
Emotional and behavioural problems at 16 (parent-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Emotional and behavioural problems at 12 (parent-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Emotional and behavioural problems at 12 (teacher-rated)
SDQ Hyperactivity
SDQ Conduct problems
SDQ Emotional problems
Other personal characteristics
Full-scale IQ (current)
Adaptive functioning (12 years)
IQ-adaptive functioning discrepancy (12 years)
Social Responsiveness Scale (SRS) Autism severity
ICD-10 total autism symptom score
Maternal distress
Maternal GHQ, 16 years
Maternal GHQ, 12 years
Adjusted
B (95% CIs)
p
b (95% CIs)
p
.15 (.03, .28)
.37 (.19, .56)
.35 (.22, .45)
.02