Unformatted text preview: 0005 80% OF CHILDREN WITH PDD-NOS HAVE SEVERAL CO-MORBID PSYCHIATRIC
E. I. de Bruin, R.F. Ferdinand, F. Verheij, P.F.A. de Nijs
Department of Child and Adolescent Psychiatry, Erasmus Medical Center Rotterdam/Sophia Children’s
Hospital, Rotterdam, Netherlands
A diagnosis of Pervasive Developmental Disorder-Not Otherwise Specified (PDD-NOS) applies when an
individual fails to meet specific criteria for autistic disorder or another explicitly defined pervasive
developmental disorder (PDD), but has similar difficulties in social interaction, reciprocal communication,
and/or stereotypical behavior (APA, 1994). These difficulties may be milder or of different quality than
those seen in autistic disorder (Towbin, 1997). Like children with autistic disorder, children with PDD-NOS
may have stereotyped interests, preoccupations, or limitations in imaginative play, although such features
may also be mild or even absent. And also in contrast with autistic disorder, PDD-NOS does not have to
be associated with a language deficit (Towbin, 1997).
DSM-IV (APA, 1994) provides 12 explicit criteria for autistic disorder, divided over the
domains of social interaction, communication, and stereotyped interests and repetitive behaviors. A child
must display at least six criteria for a diagnosis of autistic disorder to be assigned. For a DSM-IV diagnosis
of PDD-NOS, criteria are not explicit and are somewhat ambiguous. No specific items or scoring
algorithms are provided. As a consequence, children with PDD-NOS may have different combinations of
symptoms (Walker et al., 2004) and therefore constitute of a very heterogeneous group. Buitelaar, Van der
Gaag, Klin, and Volkmar (1999) developed more standardized research criteria for PDD-NOS that
differentiated reliably between PDD-NOS and non-PDD children. These explicit criteria are used in the
Studies that provide information on the prevalence rates of PDD-NOS often fall short of strict
diagnostic criteria, and assessment procedures differ per study and change over time, which makes
prevalence rates difficult to compare (Fombonne, 1999). As Wing and Potter (2002) pointed out,
prevalence rates of all types of PDDs in recent studies seem considerably higher than in older studies.
Recently, Chakrabarti and Fombonne (2001) suggested that PDD-NOS is at least twice as common as
autistic disorder in the general population and that this substantial group may have similar treatment needs
as the autistic group (Fombonne, 1999). The paradox is that, although PDD-NOS may be much more
common, the disorder is much less frequently studied than autistic disorder (Volkmar & Lord, 1998).
Despite the high prevalence, knowledge about psychiatric co-morbidity patterns
associated with PDD-NOS is hardly available. Knowledge about specific types of co-morbidity that occur
frequently with PDD-NOS would be useful to efficiently focus clinical assessment. Such additional
symptoms may cause considerable distress, interfere markedly with daily functioning, and may respond
favorably to treatment. For instance, pharmacotherapy can play a role in treatment of co-morbid attention
and hyperactivity problems, may be helpful in reducing anxiety, aggression and obsessive preoccupations
(e.g., Keen & Ward, 2004; Santosh & Baird, 2001), but has a limited effect on improving social
communication deficits in PDDs (Tanguay, 2000). Medication is also beneficial for reducing these
interfering co-morbid symptoms, and may subsequently facilitate effective application of treatments, such
as behavioral intervention (McDougle, Stigler, & Posey, 2003). Early use of behavioral interventions in
children with PDD may lead to reductions of aggression, tantrums, and self-injurious behavior up to 8090% (e.g., Horner, Carr, Strain, Todd, & Reed, 2002; Iwata et al., 1994).
Knowledge of co-morbidity patterns in PDD-NOS may also enhance further research
regarding subtypes of PDD-NOS. For instance, PDD-NOS that is associated with specific types of comorbidity, such as anxiety or mood disorders, may have different genetic or neurobiological correlates,
may differ with respect to prognosis, or may warrant different treatment approaches than, for instance,
PDD-NOS with co-morbid disruptive behavior disorders.
Some studies investigated co-morbidity between PDD-NOS and symptoms of AttentionDeficit/-Hyperactivity Disorder (ADHD) in school-aged children. However, exact data on rates of ADHD in
children with PDD-NOS are not available. This may be partly related to the priority rules of DSM-IV, which
only permit the use of a co-morbid ADHD classification when the symptoms do not occur during the course
of any PDD (APA, 1994). Therefore, it seems nearly impossible to simultaneously apply a classification of
both PDD-NOS and ADHD. Luteijn, Serra et al. (2000) compared 5 to 12 year old children with PDD-NOS
(n = 190) and with ADHD (n = 152), from an outpatients’ clinic for child and adolescent psychiatry.
Classifications of PDD-NOS and ADHD were based on DSM-IV criteria (APA, 1994). Parents filled out the Full Text Versions – Autism Safari 2006 – 2nd World Autism Congress – 30 Oct to 2 Nov, Cape Town, South Africa 1 Child Behavior Checklist 4-18 (CBCL; Achenbach, 1991). It was found that average scores of the PDDNOS group on the CBCL Attention Problems scale equaled those of the ADHD group. Hence, this study
indicated that co-morbid attention problems seemed to occur frequently in children with PDD-NOS.
To the present authors’ knowledge no data are available about the co-morbidity of PDD-NOS and
other externalizing disorders such as Oppositional-Defiant Disorder (ODD) or Conduct Disorder (CD).
Gilmour, Hill, Place, and Skuse (2004) showed a substantial co-morbidity of PDD in children with CD, but
no conclusions could be drawn about the reverse: rates of co-morbid CD in children with PDD-NOS.
Muris, Steerneman, Merkelbach, Holdrinet, and Meesters (1998) investigated the prevalence of
co-morbid anxiety disorders in children with autistic disorder (n = 15) and PDD-NOS (n = 29), aged
between 5 and 14 years. Classification of PDD-NOS or autistic disorder was based on DSM-III-R criteria
(APA, 1987). All children had undergone extensive psychological and psychiatric screening, and a
multidisciplinary team of professionals of the Center of Autism assigned the diagnoses. Co-morbid anxiety
disorders were investigated using the Diagnostic Interview Schedule for Children-Parent version 2.3
(DISC-P; Shaffer et al., 1996). Results indicated that 84.1% of the children met diagnostic criteria for at
least one anxiety disorder. The most common anxiety disorder was simple phobia (63.3%), and the least
frequent anxiety disorder was panic disorder (9.1%), although panic attacks occurred regularly (56.8%).
Furthermore, 11.4% of the children met diagnostic criteria for obsessive compulsive disorder (OCD). Some
anxiety disorders (e.g., simple phobia, separation anxiety disorder, avoidant disorder, and overanxious
disorder) were significantly more prevalent among children with PDD-NOS than among children with
autistic disorder. The sample size of this study was small, which limits its generalizability. Further,
diagnosis of PDD-NOS was not based on standardized assessment.
Further information regarding the co-morbidity of psychiatric disorders in children with PDD-NOS
would be very useful for clinical and research purposes. Therefore the aim of this study was to investigate
psychiatric co-morbidity patterns in school-aged children with PDD-NOS.
Participants were 94 children, 6 to 12 year old, who fulfilled PDD-NOS research criteria (age range 6; 5 –
12; 11 years, M = 8.5, SD = 1.9, 88.3% boys; n = 83, and 11.7% girls; n = 11). All patients visited the
outpatients’ department of child and adolescent psychiatry, Erasmus Medical Center Rotterdam, the
Netherlands. Co-morbidity was assessed with the DISC-IV-P (Shaffer, Fisher, & Lucas, 1998).
Subsequently, subgroups of PDD-NOS children with different co-morbid disorders were compared on
measures of severity of PDD related social contact and communication problems (i.e., Autism Diagnostic
Observation Schedule-Generic [ADOS-G]; Lord, Rutter, DiLavore, & Risi, 1999; Children’s Social Behavior
Questionnaire [CSBQ]; Luteijn, Jackson, Volkmar, & Minderaa, 1998; Luteijn, Minderaa, & Jackson, 2002).
All consecutive referrals between July 2002 and September 2004 (n = 503) were screened for the
presence or absence of a PDD-NOS research diagnosis. One hundred and eight children met sufficient
criteria for a research diagnosis of PDD-NOS and were therefore eligible to participate in the study. Two of
them were excluded due to parental language difficulties. A further nine parents refused to take part in the
DISC-IV-P assessment, and three children were excluded because of severe neurological or physical
problems (e.g., blindness). This yielded 94 children with a research diagnosis of PDD-NOS for whom
reliable DISC-IV-P data were available.
PDD-NOS research criteria
In the current study, a diagnosis of PDD-NOS was based on explicit research criteria. Buitelaar et al.
(1999) compared children with clinical diagnoses of autistic disorder (n = 205), PDD-NOS (n = 80), and
non-PDD diagnoses such as mental retardation and language disorders (n = 174) on the 12 criteria for
autistic disorder. Both ICD-10 (WHO, 1993) and DSM-IV (APA, 1994) classification systems were used.
They found that a short set of seven criteria, derived from the 12 original criteria for autistic disorder,
discriminated best between the PDD-NOS group and the group of non-PDD children. These seven items
were divided over the domains of social interaction (four items), communication (two items), and
stereotyped interests and repetitive behavior (one item). To diagnose PDD-NOS, at least three items had
to be present including at least one social interaction item, and the child should not meet criteria for autistic
disorder or other types of PDD.
Child psychiatrists, registrars, and psychologists supervised by child psychiatrists rated the
research criteria. Twenty-four different raters were involved. Rating was based on assessment of early
development through current level of social, communicative, and adaptive functioning, obtained from semistructured interviews carried out with the parent(s) or caretaker(s) as well as psychiatric observation of the
Full Text Versions – Autism Safari 2006 – 2nd World Autism Congress – 30 Oct to 2 Nov, Cape Town, South Africa 2 child in a one-to-one situation. School and relevant medical information was obtained, as well as
psychological assessment information. Immediately after all diagnostic procedures were finished, a
multidisciplinary team obtained consensus with regard to the final DSM-IV (APA, 1994) classification, and
PDD-NOS research criteria were rated subsequently.
We carried out an interrater reliability study on 30 randomly selected children (32%). Two
clinicians independently rated all the PDD-NOS research criteria. Agreement between the raters on the
presence or absence of a PDD-NOS diagnosis was moderate to good (κ = .62, 80.77% agreement).
Further, we computed a score for the total number of PDD-NOS criteria rated positive by each rater for
each child. The correlation between these scores by the two raters was high (Spearman’s rho = .79),
indicating excellent agreement (Cichetti & Sparrow, 1981).
The Dutch version of the DISC-IV (Ferdinand & Van der Ende, 1998; Shaffer et al., 1998) is a highly
structured respondent based interview to assess DSM-IV Axis I psychiatric disorders in the past year, in
children and adolescents. The DISC-IV has a parent version (DISC-IV-P) for parents of children aged 6 to
17, and a child version (DISC-IV-C) to be administered to children aged 11 to 17. In this study, the DISCIV-P was used to assess anxiety disorders, mood disorders, schizophrenia, and disruptive behavior
disorders. DISC-IV diagnoses are solely based on parent reports about the presence or absence of
symptoms. Clinical observations of the interviewer are not used.
The complete Dutch DISC-IV (Ferdinand & Van der Ende, 1998) contains just fewer than 3000
questions of which around 10% are considered ‘stem’ questions that are always asked. The stem
questions are concerned with broad symptom descriptions and are designed to lead to false-positive
answers. Subsequently, there are many ‘contingent’ questions that are asked if a stem question is
answered positively. Contingent questions assess whether symptoms meet the intensity, frequency, and
duration criteria as specified by DSM or ICD classification systems. Thus, the contingent questions
improve the accuracy of the stem questions (Shaffer, Fisher, Lucas, Dulcan, & Schwab-Stone, 2000). To
obtain a Dutch version of the DISC-IV, the original American version was translated into Dutch, and then
translated back by an independent translator. Subsequently, the original version and the back translation
were compared by the developers of the original version at Columbia University New York, as well as by
the Dutch translators, after which the translation was adapted.
Studies of earlier versions of the DISC-P have shown good test-retest and interrater reliability
(Schwab-Stone et al., 1993; Shaffer et al., 1993; Shaffer et al., 1996). The DISC-IV showed moderate to
good reliability (Shaffer et al., 2000). Psychometric properties of the Dutch DISC-IV are not available yet.
In this study, psychologists, research assistants, and psychology undergraduate students
(supervised by psychologists) had all been trained by the authors of the Dutch DISC-IV (Ferdinand & Van
der Ende, 1998) who, in turn, had been trained as trainers at Columbia University New York by the authors
of the original DISC. The interviewers were blind to any other diagnostic information about the child. In
69.1% (n = 65) of the cases the mother of the child has been interviewed, in 5.3% (n = 5) the father, in
24.5% (n = 23) both parents, and in 1.1% (n = 1) a different caretaker has been interviewed.
The Dutch version of the Wechsler Intelligence Scale for Children-Revised (WISC-R; Van Haasen et al.,
1986; Wechsler, 1974) was administered. As the original version, the Dutch version has shown sufficient
reliability and validity. The WISC-R generates a Full Scale Intelligence Quotient (FSIQ), a Verbal
Intelligence Quotient (VIQ), and a Performance Intelligence Quotient (PIQ) (M = 100, SD = 15). For 95.7%
(n = 90) of the children WISC-R data were available. FSIQ varied between 55 and 120 (M = 91.22, SD =
17.43). PIQ varied between 49 and 129 (M = 92.84, SD = 18.90), and VIQ varied between 51 and 122 (M
= 91.57, SD = 16.60). For four children WISC-R could not be administered due to mental retardation (n =
2), hearing problems (n = 1), and insufficient knowledge of the Dutch language (n = 1).
The ADOS-G (Lord et al., 1999) provides a standardized context to observe PDD related behaviors in the
domains of social interaction, communication, and stereotyped behavior. In this study, subgroups of PDDNOS children with different co-morbid disorders were compared on the subscales Reciprocal Social
Interaction (i.e., eye contact) and Communication (i.e., stereotyped language) of the ADOS-G. These two
subscales have shown high reliability and validity. The diagnostic algorithm of the ADOS-G allows
classification of participants as having autistic disorder or ASD. The distinction between the categories
depends on symptom severity. For 93.6% (n = 88) of the children in this study ADOS-G data were
available. An ADOS-G classification of autistic disorder was assigned in 10.2% (n = 9) of the cases, and a
further 47.7% (n = 42) had an ADOS-G classification of ASD. The remaining 42.1% (n = 37) received Full Text Versions – Autism Safari 2006 – 2nd World Autism Congress – 30 Oct to 2 Nov, Cape Town, South Africa 3 scores that were below the threshold for an ASD classification. Six children were unable (i.e., severe
communication deficits) or unwilling to take part in the ADOS-G assessment.
The CSBQ (Luteijn et al., 1998; Luteijn et al., 2002) is a 49-item parent questionnaire that covers a wide
range of PDD features of a child in the past two months. The items refer to problem behaviors seen in
children with milder variants of PDD (Luteijn, Luteijn, Jackson, Volkmar, & Minderaa, 2000). The score
format is ‘does not apply’ (score 0), ‘sometimes or somewhat applies’ (score 1), or ‘clearly or often applies’
(score 2). The items are divided over six subscales, referring to Behaviors not tuned to situation (e.g.,
Doesn’t know when to stop), Withdrawn (e.g., Acts as if others are not there), Orientation problems in time,
place, or activity (e.g., Easily gets lost), Difficulties understanding social information (e.g., Does not
understand jokes), Stereotyped behavior (e.g., Flaps arms/hands when excited), and Fear of and
resistance to changes (e.g., Opposes change).
Psychometric properties of the CSBQ have been studied in a large Dutch sample (n = 3407)
(Hartman, Luteijn, Serra, & Minderaa, in press). Three types of reliability were studied. Cronbach’s α
(based on n = 3407), that reflects internal consistency was good (α = .94 for total score, and between .76
and .90 for the subscales). Intraclass Correlation Coefficients (ICC), reflecting interrater reliability (mother
versus father information) was good (ICC = .86 for total score, and between .75 and .89 for the subscales).
Further, test-retest reliability (interval of approximately four weeks) was good as well (r = .90 for total
score, and between .80 and .88 for the subscales) (Hartman et al., in press).
For 97.9% (n = 92) of the children in this study, CSBQ data were available. For two children,
CSBQ data were not available due to refusal of the parent(s) to fill out the questionnaire. In the current
sample Crohnbach’s α was good; alphas were between .81 and .88 for the six subscales, and .94 for the
To determine which co-morbid psychiatric disorders were present in children with PDD-NOS, rates and
95%-confidence intervals were calculated for each co-morbid diagnosis.
To compare four different co-morbidity groups in the total sample of PDD-NOS children, on IQ,
ADOS-G, and CSBQ data, analysis of variance (ANOVA) and multivariate analysis of variance (MANOVA)
were used. To further compare the four groups on number of PDD-NOS criteria that were present, nonparametric statistics were used. For these additional analyses, mood and anxiety disorders were summed
as internalizing disorders.
Parent(s) or caretaker(s) of the children had all signed informed consent forms prior to participation in the
study. Children who were 12 years old signed the consent forms themselves too. The Medical Ethics
Committee of the Erasmus Medical Center has approved the study.
PDD-NOS research criteria
In Table 1 the percentage of children who were scored positively on each research criterion for PDD-NOS
is listed. Although all 94 children were assigned a research diagnosis of PDD-NOS, not necessarily every
criterion was present in each child. Criterion 1a (marked impairment in the use of multiple non-verbal
behaviors to regulate social interaction) and criterion 1b (failure to develop peer relationships appropriate
to developmental level) received the highest percentages of positive scores (77.7% and 92.5%
Insert Table 1 about here
Table 2 shows rates of co-morbid DISC-IV-P diagnoses. Overall, 80.9% (n = 76) of the children had at
least one co-morbid psychiatric disorder, and 54.3% (n = 51) had two or more co-morbid psychiatric
In 61.7% (n = 58) of the children with PDD-NOS, criteria for at least one disruptive behavior
disorder (ADHD, ODD, and/or CD) were met. ADHD was present in 44.7% (n = 42) of the children, ODD in
37.2% (n = 35), and CD in 9.6% (n = 9) of the children. I...
View Full Document