Download Follow-up of Maladaptive Behaviors in Youth with Autism Spectrum

Document related concepts

Conversion disorder wikipedia , lookup

Factitious disorder imposed on another wikipedia , lookup

Mental disorder wikipedia , lookup

Mental status examination wikipedia , lookup

Narcissistic personality disorder wikipedia , lookup

Antisocial personality disorder wikipedia , lookup

Political abuse of psychiatry in Russia wikipedia , lookup

History of psychiatry wikipedia , lookup

Generalized anxiety disorder wikipedia , lookup

Developmental disability wikipedia , lookup

Emergency psychiatry wikipedia , lookup

Separation anxiety disorder wikipedia , lookup

Dissociative identity disorder wikipedia , lookup

Conduct disorder wikipedia , lookup

History of mental disorders wikipedia , lookup

Spectrum disorder wikipedia , lookup

Classification of mental disorders wikipedia , lookup

Diagnostic and Statistical Manual of Mental Disorders wikipedia , lookup

Autism wikipedia , lookup

Controversy surrounding psychiatry wikipedia , lookup

Child psychopathology wikipedia , lookup

Pyotr Gannushkin wikipedia , lookup

Abnormal psychology wikipedia , lookup

Epidemiology of autism wikipedia , lookup

Autism therapies wikipedia , lookup

Autism spectrum wikipedia , lookup

Asperger syndrome wikipedia , lookup

Transcript
Follow-up of Maladaptive Behaviors in Youth with Autism Spectrum Disorders: Changes
and Predictors Over Two to Eight Years
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy
in the Graduate School of The Ohio State University
By
Monali Chowdhury
Graduate Program in Psychology
The Ohio State University
2012
Dissertation Committee:
Dr. Michael Aman, Advisor
Dr. Luc Lecavalier
Dr. Michael Edwards
Dr. Laura Wagner
Copyrighted by
Monali Chowdhury
2012
Abstract
Ample evidence suggests that young people with autism spectrum disorders (ASDs)
present with a wide range of maladaptive behaviors beyond the core symptoms that
define the disorder. Such maladaptive behaviors represent additional handicaps for these
individuals and are associated with negative outcomes including poor social competency,
academic and vocational under-achievement, among others. In spite of the prevalence
and significance of maladaptive behaviors, the natural course and predictors of such
behaviors are not delineated in the ASD literature. This study was conducted to add to the
scant body of knowledge on this topic. Participants were recruited using patient records at
Nisonger Center clinics, where potential participants were given an ASD diagnosis twoto-eight years previously. All individuals seen at the clinics had a measure of maladaptive
behavior on file, assessed using the Nisonger Child Behavior Rating Form (NCBRF;
Aman, Tassé, Rojahn, & Hammer, 1996). I conducted follow-ups on 342 potential
participants and collected follow-up data from 143 (41.8%) of these individuals. Followup data included current parent-rated NCBRF and supplementary demographic
information (about the individual with an ASD) from families via a structured phone
interview. This information included (a) comorbid psychiatric diagnoses, (b) current
educational placement, (c) current psychotropic medications, and (d) interventions
received. Results indicated substantial presence of maladaptive behaviors both at initial
ii
visit (T1) and follow-up (T2). Scores differed significantly between T1 and T2
assessments on all six NCBRF subscales indicating that maladaptive behaviors changed
considerably with age. Scores on three of the six NCBRF subscales (Conduct,
Hyperactivity, and Self-injury/Stereotypic) showed significant decline (improvement)
over time, while scores on the remaining three subscales (Insecure/Anxious, Selfisolated/Ritualistic, and Overly Sensitive) increased (deteriorated) at follow-up. Levels of
maladaptive behaviors were found to vary considerably based on gender, ASD subtype,
and language abilities of participants. As far as predictors of maladaptive behavior
change, the tested models accounted for 39% to 50% of the variance in T2 NCBRF
subscale scores. The T1 scores on the respective NCBRF subscales were the most
consistent predictors of all six NCBRF subscale scores at follow-up. This indicates that a
child’s individual levels of maladaptive behavior at T1 were the best predictor of
maladaptive behavior over time. Other variables that significantly predicted T2 scores on
one or more NCBRF subscales included T1 age, ASD subtype, and T1 language ability.
Parents reported high rates (68.5%) of comorbid psychiatric conditions in this community
sample unselected for psychiatric disorders and also a high rate of psychotropic
medication use (52.4%). The most common comorbid disorders were anxiety disorders
(37.8%) and ADHD (31.5%). At follow-up, the highest proportion of participants were
placed in developmentally handicapped classes (22.7%), followed by regular classes with
minimum accommodations (17.9%). The vast majority (79.7%) received at least one of
the following interventions: speech and language therapy, occupational/physical therapy,
or applied behavioral analytic therapy. Findings from this study add to the limited data on
iii
the natural course of maladaptive behavior in ASDs and have implications for clinicians,
parents, and service providers in anticipating change over time and planning
interventions.
iv
Acknowledgments
I wish to thank my advisor, Dr. Michael Aman, for his unparalleled expertise,
patience, and support. His guidance shaped this dissertation from planning to completion.
I wish to thank my Dissertation Committee members. I am grateful to Dr. Michael
Edwards for his statistical guidance throughout this study, and to Dr. Luc Lecavalier and
Dr. Laura Wagner for their thoughtful comments and suggestions.
I also wish to express my appreciation to Dr. Paula Rabidoux for her inputs
regarding this project.
I wish to thank the families of all study participants. This project would not be
possible without their cooperation.
I am grateful for the monetary support from The Ohio State University Alumni
Grants and Nisonger Center Research Fund.
Last, but not least, I wish to thank my husband for his support and humor which
energizes me in all my endeavors.
v
Vita
September 18, 1982 ....................................... Born, Kolkata, India
2003................................................................ B.A. Psychology,
Calcutta University, India
2004-2005 ...................................................... University Fellow,
The Ohio State University
2005-2006 ..................................................... Graduate Teaching Associate, Department
of Psychology, The Ohio State University
2006-2009 ..................................................... Behavior Support Specialist,
Adult Behavior Support Services,
Nisonger Center, The Ohio State University
2009-2012 ...................................................... Graduate Teaching Associate, Department
of Psychology, The Ohio State University
Publications
Chowdhury, M., & Benson, B.A. (2011). Deinstitutionalization and quality of life of
individuals with intellectual disability: A review of the international literature.
Journal of Policy and Practice in Intellectual Disabilities, 8 (4), 256-265.
Chowdhury, M., & Benson, B.A. (2011). Use of differential reinforcement to reduce
behavior problems in adults with intellectual disabilities: A methodological
review. Research in Developmental Disabilities, 324(2), 383-394.
Chowdhury, M., Aman, M.G., Scahill, L., Swiezy, N., Arnold, L.E., Lecavalier, L., et al.
(2010). The Home Situations Questionnaire-PDD version: Factor structure and
psychometric properties. Journal of Intellectual Disability Research, 54(3), 281291.
vi
Chowdhury, M., Benson, B.A., & Hillier, A. (2010). Changes in restricted repetitive
behaviors with age: A study of high-functioning adults with autism spectrum
disorders. Research in Autism Spectrum Disorders, 4(2), 210-216.
Fields of Study
Major Field: Psychology
Specialization: Intellectual and Developmental Disabilities; Developmental Psychology
Minor
vii
Table of Contents
Abstract ............................................................................................................................... ii
Acknowledgments............................................................................................................... v
Vita..................................................................................................................................... vi
Publications ........................................................................................................................ vi
Fields of Study .................................................................................................................. vii
Table of Contents ............................................................................................................. viii
List of Tables ...................................................................................................................... x
List of Figures .................................................................................................................. xiii
Chapter 1: Introduction ...................................................................................................... 1
Chapter 2: Methods ........................................................................................................... 51
Chapter 3: Results ............................................................................................................. 76
Chapter 4: Discussion ....................................................................................................... 99
References ....................................................................................................................... 126
Appendix A: Tables and Figures .................................................................................... 148
Appendix B: Initial Letter to Potential Participants ........................................................ 180
Appendix C: Demographics Form .................................................................................. 182
viii
Appendix D: Cover Letter in Study Packets to Potential Participants............................ 186
Appendix E: Combination of NCBRF Raters at T1 and T2 ........................................... 189
Appendix F: Ancillary HMLR Analyses with IQ as Predictor ....................................... 191
Appendix G: Supplementary Logistic Regression Analyses .......................................... 205
Appendix H: Binary Logistic Regression Predicting Disruptive Behavior Disorder at T2
......................................................................................................................................... 211
ix
List of Tables
Table 1. Comparative Analysis Between Participants and Non-participants ................. 149
Table 2. Comparative Analysis Between Participants from the ASD Clinic and FDD
Clinic ............................................................................................................................... 150
Table 3. Sample Characteristics at Time 1 ..................................................................... 151
Table 4. Associations Between Categorical Variables and NCBRF Problem Behavior
Subscale Scores at T1 ..................................................................................................... 152
Table 5. Correlations Between Continuous Variables and NCBRF Problem Behavior
Subscale Scores at T1 ..................................................................................................... 153
Table 6. Difference in Demographic and Assessment Variables based on ASD Subtypes
at Time 1 ......................................................................................................................... 154
Table 7. Sample Characteristics at Time 2 ..................................................................... 156
Table 8. Stability of NCBRF Problem Behavior Subscale Scores Between Time 1 and
Time 2 ............................................................................................................................. 158
Table 9. Correlations Between NCBRF Problem Behavior Subscale Scores Compared
Based on Median Split of Intervening Time ................................................................... 159
Table 10. Improvement, Worsening, and No Change in NCBRF Problem Behavior
Subscale Scores at T2 Relative to T1 ............................................................................. 160
Table 11. Difference in NCBRF Change Scores Based on ASD Subtype ..................... 161
x
Table 12. Overview of Hierarchical Multiple Linear Regression (HMLR) Analyses
Predicting NCBRF Problem Behavior Subscale Scores at T2........................................ 162
Table 13. HMLR Analysis Predicting NCBRF Conduct at T2 ...................................... 163
Table 14. HMLR Analysis Predicting NCBRF Insecure/Anxious at T2 ........................ 164
Table 15. HMLR Analysis Predicting NCBRF Hyperactive at T2 ................................ 165
Table 16. HMLR Analysis Predicting NCBRF Self-injury/Stereotypic at T2 ............... 166
Table 17. HMLR Analysis Predicting NCBRF Self-isolated/Ritualistic at T2 .............. 167
Table 18. HMLR Analysis Predicting NCBRF Overly Sensitive at T2 ......................... 168
Table 19. Associations between T2 NCBRF Subscale Scores and Psychotropic
Medication Use at T2 ...................................................................................................... 171
Table 20. Association between T2 NCBRF Subscale Scores and Educational Placements
at T2 ................................................................................................................................ 172
Table 21. Logistic Regression Predicting Comorbid Anxiety Disorder at T2 ............... 173
Table 22. Logistic Regression Predicting Comorbid ADHD at T2 ................................ 174
Table 23. Logistic Regression Predicting Comorbid Depressive Disorder at T2 ........... 175
Table 24. Logistic Regression Predicting Restrictive Educational Placement at T2 ...... 176
Table 25. Summary of Significant Predictors of T2 NCBRF Outcomes ........................ 177
Table 26. Combination of NCBRF Raters at T1 and T2 ................................................ 190
Table 27. Summary of Ancillary HMLR Analyses Predicting T2 NCBRF Outcomes with
IQ as Additional Predictor in Block 2............................................................................. 192
Table 28. Ancillary HMLR Analysis Predicting T2 NCBRF Conduct with IQ as
Additional Predictor in Block 2 ...................................................................................... 193
xi
Table 29. Ancillary HMLR Analysis Predicting T2 NCBRF Insecure/Anxious with IQ as
Additional Predictor in Block 2 ...................................................................................... 195
Table 30. Ancillary HMLR Analyses Predicting T2 NCBRF Hyperactivity with IQ as
Additional Predictor in Block 2 ...................................................................................... 197
Table 31. Ancillary HMLR Analyses Predicting T2 NCBRF Self-injury/Stereotypic with
IQ as Additional Predictor in Block 2............................................................................. 199
Table 32. Ancillary HMLR Analyses Predicting T2 NCBRF Self-isolated/Ritualistic with
IQ as Additional Predictor in Block 2............................................................................. 201
Table 33. Ancillary HMLR Analyses Predicting T2 NCBRF Overly Sensitive with IQ as
Additional Predictor in Block 2 ...................................................................................... 203
Table 34. Supplementary Logistic Regression Predicting T2 Comorbid Anxiety Disorder
with ADI–R Domain Scores as Additional Predictors ................................................... 206
Table 35. Supplementary Logistic Regression Predicting T2 Comorbid ADHD with
ADI–R Domain Scores as Additional Predictors............................................................ 207
Table 36. Supplementary Logistic Regression Predicting T2 Comorbid DBD with ADI–R
Domain Scores as Additional Predictors ........................................................................ 208
Table 37. Supplementary Logistic Regression Predicting T2 Comorbid Depressive
Disorder with ADI–R Domain Scores as Additional Predictors .................................... 209
Table 38. Supplementary Logistic Regression Predicting T2 Educational Placement with
ADI–R Domain Scores as Additional Predictors............................................................ 210
Table 39. Binary Logistic Regression Predicting T2 Comorbid DBD ........................... 212
xii
List of Figures
Figure 1. Overlap Between Psychiatric Comorbid Conditions at T2 ............................. 178
Figure 2. Dot plot of bivariate relation between T1 NCBRF Hyperactivity scores and T2
ADHD diagnosis ............................................................................................................. 179
xiii
Chapter 1: Introduction
Autism spectrum disorders (ASDs; Autistic Disorder, Asperger’s Disorder,
Pervasive Developmental Disorder–Not Otherwise Specified or PDD–NOS ) are
clinically heterogeneous developmental disorders, with a wide range of severity in the
core features, seen in individuals of all cognitive levels (American Psychiatric
Association, 2000). Although the most critical phase and symptom manifestation is in
childhood, ASDs have been documented to be life-long conditions typically continuing
into adulthood (Seltzer, Shattuck, Abbeduto, & Greenberg, 2004; Szatmari, Bryson,
Boyle, Streiner, & Duku, 2003). The question of outcome and prognosis has been raised
since the very first studies on autism (Kanner, 1971). Accurate information about the
developmental trajectory of the disorder is crucial for all of those involved with the
individual with an ASD, whether in the role of parent, clinician, researcher, or service
provider (Tsatsanis, 2002). Outcome studies provide useful findings with regard to
positive prognosticators and long-term adaptation to everyday life. This information may
assist the family of the individual with ASD in forming realistic expectations, availing
appropriate intervention, and in planning for the future. However, predicting long-term
outcome in autism is complicated by the very wide spectrum of cognitive, linguistic,
social, and behavioral functioning associated with the condition (Howlin, Goode, Hutton,
& Rutter, 2004).
1
Currently, there is a growing body of literature that provides insight into
adolescent and adult outcomes in ASDs. Most studies describe outcome on the basis of
social adaptation, IQ, and speech, but results are difficult to compare because different
samples, models, and variables are used (Darrou et al., 2010). Longitudinal studies show
a wide range of outcomes in adaptive behavior domains (Bolte & Poustka, 2002), degree
of autism (Jonsdottir et al., 2006; Szatmari et al., 2003), speech (Eaves & Ho, 1996;
Turner, Stone, Pozdol, & Coonrod, 2006), and IQ (Eaves & Ho, 1996; Szatmari et al.,
2003).
From a public health perspective, ASDs are an important cause of morbidity and
high service utilization (Jarbrink, Fombonne, & Knapp, 2003) because of their early
onset, lifelong persistence, high level of associated impairment, and absence of
established treatment for the core problems (Simonoff et al., 2008). Impairment due to
the social and communicative deficits that constitute the core features of autism is well
demonstrated in ASDs. A further, less well-investigated cause of impairment may be cooccurring maladaptive behaviors. Maladaptive behaviors are behaviors that interfere
with everyday activities, include behaviors that are not socially acceptable, can physically
harm someone, or affect education or living placement (Shattuck et al., 2007).
Maladaptive behaviors have been referred to using a variety of terminology including
behavior problems, problem behaviors, challenging behaviors, and aberrant behaviors,
among others (McClintock, Hall, and Oliver, 2003). In clinical practice it has long been
recognized that children and adolescents with ASDs often present with maladaptive
behaviors or comorbid psychopathology which further contribute to the broad variability
2
in the clinical presentation of ASD (Gjevik, Eldevik, Fjaeran-Granum, & Sponheim,
2011). A growing number of studies have confirmed clinical observations and have
converged in finding that in addition to the core features of ASDs, children and
adolescents on the autism spectrum are at risk of developing various maladaptive
behaviors and difficulties beyond those defining the disorder (Anderson, Maye, & Lord,
2011). Comorbid maladaptive behaviors commonly include symptoms of hyperactivity,
irritability, aggression, oppositional conduct, self-injury, depression, anxiety, and other
socially unacceptable behaviors (e.g., Howlin, 2007; Lecavalier, 2006; Shattuck et al.,
2007; Simonoff et al., 2008). Only recently have systematic studies started to explore
how often these additional difficulties are due to, or associated with, comorbid DSM–IV
psychiatric disorders (Leyfer et al., 2006).
Historically, maladaptive behaviors have received less attention in the scientific
literature than core features of ASDs (Anderson et al., 2011). Though the extant
literature has explored developmental outcomes of ASDs as mentioned above, there is a
significant gap in our knowledge when it comes to follow-up studies reporting on the
outcome of comorbid maladaptive behaviors in adolescence and adulthood.
In the sections to follow, I present a broad overview of recent findings from
studies assessing short-term developmental outcomes in childhood and adolescence. I
then discuss overall long-term outcomes in adulthood (including change in autistic
symptomatology). The subsequent sections hone in on maladaptive behaviors in ASDs–
the central theme of this dissertation. I summarize findings from studies reporting on
3
maladaptive behaviors, and I follow this with a discussion of the scant literature reporting
follow-up data on maladaptive behavior outcomes in individuals with ASDs.
Short term Outcomes in Childhood and Adolescence in ASDs
A large number of follow-up studies of short-term outcomes have examined
diagnostic stability, cognitive stability, and/or language development for young children
with ASDs (Freeman et al., 1985; 1991; Gonzalez et al., 1993; Harris & Handleman,
2000; Harris et al., 1991; Lord & Schopler, 1989a; Piven et al., 1996; Sigman, 1998;
Sigman & Ruskin, 1999; Venter et al., 1992). Findings have revealed good diagnostic
stability with about 90-100% of preschool-aged children diagnosed with autistic disorder
remaining on the autism spectrum, and 75-100% retaining the specific diagnosis of
autistic disorder into school age and adolescence (Gonzalez et al., 1993; Piven et al.,
1996; Sigman and Ruskin, 1999; Turner et al., 2006; Venter et al., 1992). Stability for a
specific diagnosis has been reported to be higher for autistic disorder than for PDD–NOS
(Turner et al., 2006).
Cognitive scores for children with ASDs also remain relatively stable from
preschool age into school age, adolescence, and early adulthood. At the group level, IQ or
developmental quotient (DQ) scores have been found to be quite stable over time, with
mean scores generally varying by less than 10 points, and correlations across
measurement periods ranging from 0.50 to above 0.80 (Lord and Schopler, 1988; 1989a,
1989b; Sigman and Ruskin, 1999). In addition, cognitive status appears to be relatively
stable over time when considered within the context of broad categories of functioning
(e.g. high, medium, or low IQ) (Freeman et al., 1985, 1991; Gonzalez et al., 1993; Lord
4
and Schopler, 1989a). However, for individual children, scores can vary considerably,
with average changes ranging from 10 to over 20 points, most of which are in a positive
direction (Lord & Schopler, 1989a; Sigman & Ruskin, 1999; Turner et al., 2006).
Improvements in cognitive scores in preschool-aged children have been found to be
associated with younger initial age, high initial cognitive skills, and participation in early
intervention (Harris & Handleman, 2000).
In a series of two comparative studies, Szatmari and colleagues (Szatmari et al.,
2000, 2003) reported on a 2- and 6-year follow-up of 66 children (46 with autism and 20
with Asperger’s disorder) with ages between 4 to 6 years to assess if outcomes differed
for these two diagnostic subtypes. As compared to the autism group, children with
Asperger’s disorder had higher socialization scores on the Vineland Adaptive Behavior
Scales (Sparrow, Balla, & Cicchetti, 1984), and fewer total autistic symptoms as
measured by the Autism Behavior Checklist (Krug, Arik, & Almond, 1980; Volkmar et
al., 1988) at the time of follow-up. These differences were substantial and continued to
exist even after initial differences in nonverbal IQ, expressive language, and verbal
reasoning were statistically controlled. These outcome variations between the children
with autism and Asperger’s were in fact well explained by the observed changes in the
level of language fluency, as measured by an oral vocabulary test. There were large
differences between the groups with Asperger’s and autism on oral vocabulary at both the
beginning of the study and at follow-up. However, autistic children who were fluent at
follow-up were not statistically different from the children with Asperger’s at the
beginning of the study. Based on these findings, the authors suggested that the
5
differences between Asperger’s and autism may largely be a matter of timing. In other
words, the groups seemed to be on different but parallel developmental trajectories
initially, but there is the possibility that some children with autism may join the trajectory
of children with Asperger’s, once they develop a certain level of fluent language. The
important implication of this result is that the clinical pathway that a child follows largely
depends on if, and when, the child develops fluent language. The authors cautioned that
this model, though promising, must be seen as exploratory at this point.
More recently, Baghdadli et al. (2007) described the psychological development
of 219 preschoolers with autism over a period of 3 years. Outcome was assessed in terms
of object-related cognition (manipulation of objects and the quality of eye and hand
coordination, cognitive shifting, and perceptual organization), person-related cognition
(social interaction, joint attention, and behavior adjustment), and adaptive behavior. High
variability was found in the short-term outcome of the preschoolers.
In a follow-up to Baghdadli et al. (2007), the same research team (Darrou et al.,
2010) explored if the amount of intervention influenced developmental trajectories in 208
children with ASDs, assessed first at age 5 years and reassessed three years later. The
participants’ clinical characteristics and total number of hours of intervention provided
per week were recorded. The intervention programs were eclectic and consisted of a
mixture of special education, rehabilitation, and psychotherapy. A high-level (HL) group
and low-level (LL) group of participants were identified using latent class analysis at
Time 1, and their outcome and associated risk and protective factors examined at Time 2.
Children in the HL group had a lower degree of autism, higher level of speech, and
6
higher developmental age than children in the LL group. As found in Baghdadli et al.
(2007), autism severity and lack of speech at Time 1 emerged as risk factors of belonging
to the LL group at follow-up. Contrary to the authors’ expectations, the number of hours
of eclectic intervention was not linked at all to the children’s developmental trajectories.
Predictors of Outcome
Several investigators have examined factors that predicted short term outcome in
ASDs. A good prognosis in middle childhood to adolescence was found to be dependent
on a number of variables. Most consistent among these were initial severity of autism
symptoms, initial speech and cognitive level, early diagnosis, and interventions.
The initial level of intellectual functioning appeared to be an important predictive
factor of psychological development; children who functioned at high levels made the
most developmental progress in socio-adaptive and cognitive domains (Fein et al., 1999;
Liss et al., 2001; Starr et al., 2003). Baghdadli et al. (2007) pointed out that it is difficult
to determine if autism severity is independent of intellectual functioning or if the two
variables have a common underlying dimension.
The presence or absence of speech is another important variable that has long
been considered a robust predictor of overall outcome (Baghdadli et al., 2007; Sutera et
al., 2007) with the absence of speech at age 5 years being associated with poor
development (Mawhood, Howlin, & Rutter, 2000; Nordin & Gillberg, 1998). Being able
to express oneself–even with only a few words and shaky syntax–appears to correspond
to a critical moment in development and seems to function as an important protective
factor (Darrou et al., 2010). Turner et al. (2006) also reported that higher language skills
7
at age 2 years were associated with better outcomes in their 7-year follow-up of
developmental outcomes of children diagnosed with ASDs at age 2 years. Moreover,
Charman et al. (2004) contended that children’s initial communication level is a predictor
of changes in their daily living skills.
Overall, the ability of early language and cognitive level to predict later outcome
were found to be stable over time, at least until the pre-adolescent years (Szatmari et al.,
2003). Further, Szatmari et al. noted that outcome should be measured as a
multidimensional construct among children with ASDs. For example, in their study, the
explanatory power of models (that included early language and nonverbal skills as
predictors) was found to be much higher for communication and social skills than for
autistic symptoms. Consequently, the authors noted that, while we can predict outcome
fairly well in terms of communication skills, our ability to do so with respect to social
skills is weaker, and even poorer for autistic symptoms. Additionally, the strength of
prediction differed among ASD subtypes. Language was a better predictor of outcome for
the autism group than for the group with Asperger’s disorder thus indicating that
determinants of outcome in children with autism are different than in children with
Asperger’s.
In line with findings discussed above, Baghdadli et al. (2007) reported that (a)
cognitive functioning, (b) socio-adaptive skills, (c) verbal expression, and (d) severity of
autism were factors that predicted developmental outcomes in mid-childhood. However,
in their study, IQ, speech level, and adaptive functioning taken together, only explained
up to 30% of the developmental variance observed in the sample. The authors pointed out
8
that these results indicate that variability in psychological development is possibly
captured by variables other than those mentioned above. They suggested that impact of
intervention was a variable of interest.
A large number of studies have reported that early intervention services are
associated with significant improvements in language, cognitive ability, and social skills
for young children with autism (Bondy and Frost, 1995; Harris and Handleman, 2000;
Harris et al., 1991; McEachin et al., 1993; Rogers and Lewis, 1989; Strain et al., 1985).
Some evidence also suggests that children with autism who start intervention at younger
ages have better outcomes than those who enroll at older ages (Harris and Handleman,
2000). One possible explanation for this finding is that increased brain plasticity enables
younger children to reap more benefits from early intervention than older children
(Dawson et al., 2000; Mundy and Neal, 2001). However, participation in intervention is
not always predictive of better outcome. For example, Darrou et al. (2010) reported lack
of any link between hours of intervention received and outcome. This finding is in line
with results obtained by Gabriels et al. (2001) and Eaves and Ho (2004), and it suggests
that the impact of the sheer amount of intervention might not be as influential to outcome
as the type of intervention. Results from a few studies suggested that intensive, eclectic
intervention approaches are not as effective as intensive behavioral approaches (Gabriels
et al., 2001; Jonsdottir et al., 2006). Darrou et al. reasoned that, since children with
autism need routines and regularity in their environment, eclectic intervention approaches
could be a source of excessive change and transition between activities resulting in
problem behavior and anxiety. They noted that the failure of eclectic intervention
9
intensity and diversity to have an impact in their study raises the question of whether the
specificities of an intervention method are what leads to gains and improved outcomes for
the child (Eldevik et al., 2006; Howard et al., 2005). Interestingly, Turner et al. (2006)
found only speech-language therapy hours to be predictive of better outcomes but not
educational therapy hours. In explaining this finding, they suggested that speechlanguage therapy is more often provided within the context of one-on-one sessions,
and/or that the content of the sessions is more likely to focus on remediation of core
social-communicative deficits. Overall, it seems that, while there is a trend that
intervention–especially early intervention–is predictive of better outcome, the elements
that lead to successful therapy are still not well established.
Adult Outcomes in ASDs
The first systematic follow-up studies of adult outcome in ASDs were conducted
around the 1970s and 1980s and indicated that, while adult outcome was variable, it was
on average psychosocially poor (Lotter, 1974; Lockyer & Rutter, 1969, 1970; Nordin &
Gillberg, 1998; Rutter, Greenfeld, & Lockyer, 1967; Rutter & Lockyer, 1967). According
to these studies, there was no indication of independence (in terms of work, education,
and living skills) in early adult life for about two thirds of people with autism. High rates
of epilepsy were reported in early childhood and adolescence (Olsson, Gillberg, &
Steffenburg, 1988; Volkmar & Nelson 1990) and possibly a higher rate of poor outcome
in the epilepsy subgroup (Gillberg & Steffenburg 1987; Kobayashi & Murata, 1998a).
Females with autism appeared to have worse outcomes than males. Those with “classic”
autism were reported to have poorer outcomes than those with autistic-like conditions
10
(Nordin & Gillberg, 1998). In terms of follow-up of individuals with comparatively
higher levels of cognitive functioning, Venter et al. (1992) described outcome for 22
individuals aged 18 years or over with an IQ score of over 60. Outcomes were generally
poor across the board. Although a third were competitively employed, jobs were
generally at a very low level, and the majority was in sheltered employment or special
training programs; three were unemployed. Only four individuals lived more or less
independently.
In contrast, Szatmari and colleagues (Szatmari, Bartolucci, Bremner,Bond, &
Rich, 1989) reported more positive findings for their group of 12 males and four females
(all 17 years or over; mean IQ > 90). Half had attended college or university, and over a
third were in regular, fulltime employment. Half were described as being completely
independent, although some of them still lived at home. Although over half the group
never formed close relationships, a quarter dated regularly or had long-term relationships
and one was married. Mawhood and colleagues (Howlin, Mawhood, & Rutter, 2000;
Mawhood, Howlin, & Rutter, 2000) followed up 19 men with autism (mean performace
IQ 83) who had initially been diagnosed between 4 and 9 years of age. Although the
majority improved over time, all showed continuing problems in communication, social
relationships, and independence. Almost half remained socially isolated, only three lived
independently, and over two-thirds had significant difficulties associated with obsessional
or ritualistic tendencies. Only three individuals (16%) were considered to have a good
outcome; two (10%) remained moderately impaired and 14 (74%) continued to show
substantial impairments.
11
More recently, Howlin et al. (2004) reported on adult outcomes of 68 individuals
with autism with performance IQ of 50 or above in childhood. Mean age at follow-up
was 29 years (range 21-48 years). Although a minority of adults achieved relatively high
levels of independence, most remained very dependent on families or other support
services. Few lived alone, had close friends, or had permanent employment. Stereotyped
behaviors and interests, as well as communication difficulties, frequently persisted into
adulthood. Overall, only 12% were rated as having a “very good” outcome; 10% as
“good,” and 19% as “fair.” The majority was rated as having “poor” (46%) or “very
poor” (12%) outcome.
Direct comparison between studies discussed above should be avoided because of
marked differences in initial sample characteristics and measures used. Overall judgment
of whether outcome was “good,” “fair,” or “poor” also tended to be based on variable
criteria. Outcomes for higher-functioning individuals varied; on average, even
cognitively able adults with autism were reported to live at home or in a supervised
facility, and occupy lower level jobs or were unable to gain competitive employment
(Tsatsanis, 2003). As pointed out by Tsatsanis, better prognosis in more recent samples
(e.g., Venter et al., 1992) versus those from two decades before (e.g, Bartak & Rutter,
1976) may signify advanced outcome from the availability of continuous structured
educational and therapeutic programs. Conversely, it is also possible that the earlier
studies included more severely impaired individuals.
12
Predictors of Outcome
Not surprisingly, the same variables identified as strong predictors of good shortterm outcome (discussed in previous sections) were found to be most consistently
associated with good outcome in the adult outcome literature. Overall, presence of
communicative speech before 5 years of age, and IQ were the most robust predictors of
achievement and adaptive functioning (Venter et al., 1992). Howlin et al. (2000) also
found early language skills to be significantly related to social competence. Compared
with those who had no speech at age 5 years, those with speech had better outcomes in
overall social rating and living status, but there were no significant differences for
educational level, social or abnormal use of language, and only a marginal difference for
work level (Howlin et al., 2004).
Significantly better adult outcomes were reported for individuals with a childhood
performance IQ of at least 70 with a strong positive association between current verbal
IQ and friendship and social ratings in adulthood (Howlin et al., 2000, 2004).
Nevertheless, findings from Howlin et al. (2004) also showed that outcomes can be
extremely variable for individuals with performance IQ above 70. Few differences in
adulthood were noted between those with a childhood IQ of 100 or more, compared with
those with an IQ between 70 and 99, and even those in the highest IQ group experienced
persistent problems as adults. Compared with those who had no speech at age 5 years,
those with speech had better outcomes in overall social rating and living status. However,
there were no significant differences for educational level, social, or abnormal use of
language, and only a marginal difference for work level. The authors concluded that
13
while initial level of nonverbal IQ is a relatively good predictor of adult outcome, it is by
no means perfectly reliable. They agreed with Lord and Bailey (2002) who suggested that
the presence of useful speech by age 2 years might be a far more reliable indicator of
later outcome.
Change in Core Symptoms of Autism in Adolescence and Adulthood
Studies addressing changes in core symptoms of ASDs have spanned four
decades and therefore differ in terms of diagnostic practices in effect across time (Seltzer
et al., 2004). However, despite differences in diagnostic practices, design, sample, and
measures used, the accumulated evidence summarized in a review by Seltzer et al. (2004)
indicate that the core symptoms of autism abate to some degree during adolescence and
young adulthood. Among recent studies, Boelte and Poustka (2000) administered the
Autism Diagnostic Interview–Revised (ADI–R; Lord, Rutter, & LeCouteur, 1994) to 93
individuals (age 15-37 years; mean age= 22.3 years) and found that current symptoms
had decreased in comparison to lifetime ratings. Although the overall trend was that of
improvement for many, the trajectory of symptom development was far from
homogenous. Plateaus or periods of symptom aggravation were reported over the life
course, and for some individuals symptoms did not abate and even worsened for a few
(Gillberg & Steffenberg, 1987). The findings of some studies also suggested that
symptom development may be “splintered” over time, with improvement in only some
behaviors that define autism, with other behaviors improving at different times. Such
differential degrees of abatement in ASD symptoms have been reported by Szatmari et al.
(2003), who studied 68 subjects with ASDs over a six-year period. They found that
14
socialization levels diminished, but communication skills remained stable. Starr,
Szatmari, Bryson, and Zwaigenbaum (2003) compared children with high-functioning
autism and children with Asperger’s disorder and found a major decrease in symptom
severity in communication in the autism group, increased symptom severity in the social
domain in both groups, and no significant change in repetitive activities.
There is some evidence that age-related improvement may be more limited in the
domain of restricted repetitive behaviors (RRB) compared to the domains of reciprocal
social interaction and communication. For example, Piven et al. (1996) in their
retrospective study of 38 high-IQ adolescents and adults with ASDs found that only half
showed improvement in RRB symptoms, while more than 80% improved in both social
and communication domains. A similar trend of lesser age-related improvement in the
RRB domain was observed by Fecteau et al. (2003) in their retrospective study of 28
individuals with autism. Seltzer et al. (2003) found that, based on current ratings on the
ADI–R, 87.7% of their sample continued to score above diagnostic cut-offs in the RRB
domain, compared to 67.9% for the Communication domain and 85.4% for the
Reciprocal Social Interaction domain.
In summary, improvement seems to be a “dominant, although not universal,
pattern of change shown by persons with autism, existing alongside persistent
impairments in multiple areas of functioning” (Seltzer et al., 2004, p.567). Importantly,
improvements were not seen for all individuals and, even in those who did improve,
changes were seldom substantial enough to move the individual into the normal range of
functioning. Nevertheless, there was a suggestion in the literature that a very small
15
proportion of individuals who received an ASD diagnosis in childhood showed sufficient
improvement during adolescence and adult years to outgrow the diagnosis (Boelte &
Poustka, 2000; Mesibov et al., 1989). These tend to be individuals who, as children,
manifested the least severe symptoms–often those diagnosed as having Asperger’s or
PDD–NOS (Seltzer et al, 2003).
Prevalence of Psychopathology in ASDs
Growing evidence in the literature suggests that individuals with ASDs frequently
present with a wide range of emotional and behavior problems. These include symptoms
of obsessive-compulsive disorders (McDougle et al., 1995), mood disorders (Bradley,
Summers, Wood, & Bryson, 2004; Ghaziuddin, Ghaziuddin, & Greden, 2002; Lainhart &
Folstein, 1994), high anxiety or fears (Fombonne, 1997; Gillot, Furniss, & Walter, 2001;
Kim, Szatmari, Bryson, Streiner, & Wilson, 2000), low frustration tolerance, temper
outbursts, mood lability, and symptoms of Attention Deficit Hyperactivity Disorder
(ADHD) such as inattention, impulsivity, hyperactivity (Goldstein & Schwebach, 2004;
Kim et al., 2000; Loveland & Tunali-Kotowski, 1997). Other behavior problems
commonly reported in the literature include verbal aggression, property destruction,
physical aggression, tantrum behavior, self-injurious behavior, and stereotypies
(Lecavalier, 2006; Rojahn, Aman, Matson, & Mayville, 2003). Up to 94.3% of children
and adolescents with ASDs have been found to display at least one challenging behavior
(Matson, Wilkins, & Macken, 2009). Young people with ASDs have been found to be
more likely to engage in problem behaviors than children with Fragile X syndrome,
Williams syndrome, Prader-Willi syndrome, or Down syndrome (Tonge & Einfeld,
16
2003). Similar statements have been made with respect to children with intellectual
disability (ID) alone (Holden & Gitlesen, 2006; Murphy et al., 2005) and typically
developing children (Nicholas et al., 2008). A meta-analysis of research on maladaptive
behaviors in children and adults with different types of IDs found that aggression,
disruptive behavior, and self-injury were significantly more prevalent among those with
ASDs than other types of IDs (McClintock et al., 2003).
Studies assessing psychopathology in ASDs can be classified as belonging to one
of two groups: (1) studies reporting on prevalence rates of comorbid DSM–IV psychiatric
disorders and psychiatric symptoms, and (2) studies reporting on levels and patterns of
individual maladaptive behaviors. Findings from these two groups of studies are
discussed below.
Comorbid Psychiatric Disorders in ASDs
A growing number of systematic studies have reported comorbid psychiatric
disorders to be common in individuals with ASDs. These were first reported in a number
of studies using unstandardized assessments (Chung, Luk, & Lee, 1990; Goldstein &
Schwebach, 2004), and subsequently confirmed by studies using questionnaires in
samples of both children (Herring et al., 2006; Steinhausen & Metzke, 2004) and adults
(Kobayashi & Murata, 1998b). Studies assessing prevalence of psychiatric disorders have
most commonly reported high rates of ADHD and anxiety (Lee & Ousley, 2006; Leyfer
et al., 2006; Muris et al., 1998, Rosenberg, Kaufmann, Law, & Law, 2011, Simonoff et
al., 2008). Findings on prevalence of depression and oppositional defiant
disorder/conduct disorder (ODD/CD) appear more inconsistent (de Bruin et al., 2007;
17
Ghaziuddin et al., 2002; Leyfer et al., 2006; Simonoff et al., 2008). Other disorders
reported at low rates include tic disorders (Gjevick et al., 2011; Simonoff et al., 2008),
enuresis, and encopresis (Mattilla et al., 2010). Schizophrenia, related psychotic
disorders, and eating disorders are rarely, if at all, reported in ASDs (Leyfer et al., 2006;
Gjevick et al, 2011).
Importance of Diagnosing Comorbid Psychiatric Disorders in ASDs
Accurate, reliable diagnosis of comorbid psychiatric conditions in individuals
with ASDs is of major importance for the following reasons (Leyfer et al., 2006). First, as
outlined by Leyfer et al., a comorbid disorder may result in significant clinical
impairment and represent additional burden of illness on the individual with ASD and on
the family. More specific treatment paths can be adopted when maladaptive behaviors are
classified as a comorbid psychiatric disorder rather than just random, isolated behaviors.
Second, from a public health and service perspective, additional clarification of
psychiatric comorbidity in ASDs can help in planning for service provision. Diagnosis of
a comorbid psychiatric disorder may qualify a child with ASD for coverage by medical
insurance. Third, from a point of view of research, identification of psychiatric
comorbidities may help delineate etiological and neurobiological heterogeneity in ASDs.
It may also aid genetic studies of ASDs by accurately subgrouping children with autism
according to comorbidity.
Problems with Diagnosing Comorbid Psychiatric Disorders in Presence of ASD
Diagnosing comorbid psychiatric disorders in presence of an ASD is considerably
challenging (Gjevik et al., 2011). First, as outlined by Gjevik et al., the hierarchical and
18
categorical structure of the DSM–IV implies that psychiatric symptoms should not be
diagnosed by clinicians if better accounted for by a more severe disorder (American
Psychiatric Association, 2000). This approach has the advantage of parsimony (Simonoff
et al., 2008). In clinical service catering to individuals with ASDs, there might be a
tendency of “diagnostic overshadowing,” i.e., attributing all presenting symptoms to the
ASD–a trend also found in ID, where psychiatric symptoms might end up being
attributed to the ID. In addition, there are some diagnoses that DSM–IV explicitly
mentions when it comes to being diagnosed concurrent with an ASD. For example, DSM
guidelines recommend against diagnosing ADHD if exclusively occurring during the
course of an ASD. DSM–IV also does not allow a diagnosis of separation anxiety
disorder, generalized anxiety disorder, or social phobia in individuals with ASDs
(American Psychiatric Association, 2000; see also Grondhuis & Aman, 2012). However,
in clinical practice it has long been recognized that children and adolescents with ASDs
often have comorbid psychopathology (Gjevik et al., 2011), and hence it is increasingly
argued that comorbid psychiatric disorders should be assessed and diagnosed when
present with an ASD (Simonoff et al., 2008). Grondhuis and Aman (2012) argued that
failure to recognize separation anxiety, generalized anxiety, and social phobia may impair
professional communication and provision of appropriate care among such individuals.
Second, in addition to issues related to DSM guidelines, assessing comorbid
psychiatric disorders in children and adolescents with ASDs presents other diagnostic
challenges related to instrumentation. Instruments and standardized diagnostic interviews
developed for the general population have been historically used to measure maladaptive
19
behaviors and aspects of comorbidity in individuals with developmental disabilities
(DDs), including ASDs. The majority of these instruments, however, have not yet been
tested for reliability and validity in individuals with ASDs (see Grondhuis & Aman,
2012; Leyfer et al., 2006). Such instruments, as outlined by Leyfer et al., include (a) the
Child Behavior Checklist (Achenbach, Howell, Quay, & Conners, 1991; Dekker, Koot,
van der Ende & Verhulst, 2002; Dykens, 2000; Masi, Brovedani, Mucci, & Favilla,
2002), (b) the Diagnostic Interview Schedule for Children-IV (Dekker et al., 2002;
Shaffer, Fisher, Lucas, Dulcan, & Schwab-Stone, 2000); (c) the Kiddie Schedule for
Affective Disorders and Schizophrenia (Masi, Favilla, & Mucci, 2000; Masi, Mucci,
Favilla, & Poli, 1999), and (d) the Schedule for Affective Disorders and Schizophrenia–
Lifetime Version (Meadows et al.,1991). Next, there are some other instruments which
have been specifically designed for use in individuals with DDs and have contributed to
the study of comorbidity in autism. These include (a) the Aberrant Behavior Checklist
(Aman, Singh, Stewart, & Field, 1985; Aman & Singh, 1994; Rojahn et al., 2003), (b) the
Developmental Behaviour Checklist (Clarke, Tonge, Einfeld, & Mackinnon, 2003;
Dekker, Nunn, Einfeld, Tonge, & Koot, 2002; Einfeld, Tonge, Turner, Parmenter, &
Smith, 1999; Tonge & Einfeld, 2000), (c) the Behavior Problems Inventory (Rojahn,
Matson, Lott, Esbensen, & Smalls, 2001), (d) the Anxiety, Depression, and Mood Scale
(Esbensen, Rojahn, Aman, & Ruedrich, 2003), (e) the Nisonger Child Behavior Rating
Form (Aman et al., 2002; Lecavalier et al., 2006; Witwer & Lecavalier, 2010), and
others. As Leyfer et al. (2006) pointed out, several of these instruments were developed
for the purpose of assessing maladaptive behaviors and their reliability and validity are
20
yet to be tested for the purpose of diagnosing comorbid psychiatric disorders in
individuals with ASDs. Hence, very few specialized instruments are available for
assessing psychiatric conditions in this population (Bradley & Bolton 2006; Leyfer et al.
2006).
Third, limited cognitive and language abilities make it difficult for individuals
with ASDs to communicate their thoughts and feelings aptly, thus further complicating
the evaluation of psychiatric symptoms. In addition to communication problems, children
and adults with autism often have impairments in ‘‘theory of mind,” complex information
processing, and executive functioning (Baron-Cohen, 1991; Baron-Cohen, Leslie, &
Frith, 1985; Minshew, Goldstein, & Siegel, 1997; Ozonoff, Strayer, McMahon, &
Filloux, 1994). These overarching problems with social cognition can make it difficult for
many individuals with ASDs to describe their mental states, mental experiences, and even
daily life experiences. Thus, it can be challenging to determine if a person’s difficulties
are due to effects of the core features of autism, life experiences, or a comorbid
psychiatric disorder superimposed on the autism and the life experiences of the child
(Leyfer et al., 2006).
Finally, in addition to problems with communicating with the patient about
symptoms, symptoms of certain comorbid psychiatric disorders can be overlapping with
core features of ASD, making the process of diagnosis even more difficult. As outlined
by Gjevik et al. (2011), symptoms of obsessive compulsive disorder (OCD) might be
difficult to distinguish from repetitive and ritualistic behaviors, tic disorders from
stereotyped and repetitive movements, social phobia from impairments in social
21
interaction, and psychotic symptoms from stereotyped and repetitive thinking (Dossetor,
2007; Zandt et al., 2007).
Overall, when it comes to diagnosing psychiatric comorbidity in ASDs, there
continues to be a lack of consensus as to whether and how symptom definitions should be
varied to take account of specific characteristics of people with ASDs. For example,
Leyfer et al. (2006) modified an existing instrument to form the Autism Comorbidity
Interview–Present and Lifetime Version (ACI–PL) as discussed in a later section. It is not
established whether this is an appropriate methodology as previous incidences of altering
diagnostic criteria for specific groups (such as depression in children), or altering
diagnostic criteria for people with significant ID, have usually led to uncertainty about
the validity of classification (McBrien, 2003; Taylor & Rutter, 2000). This further
indicates the need for a comparison of instruments in developing clinical and research
tools for assessing psychiatric comorbidities in ASDs (Simonoff et al., 2006).
Studies Assessing Comorbid DSM Psychiatric Disorders and Symptoms
Studies from both North America and Europe provide evidence of high rates of
psychiatric comorbidity among individuals with ASDs with varied cognitive levels.
Samples and assessment instruments from such studies are described followed by a
comparative analysis of rates of psychiatric comorbid conditions reported in these
studies.
In a study from the United States, Leyfer et al. (2006) modified the K-SADS
(Chambers et al., 1985; Kaufman et al., 1997; Ambrosini, 2000), a semi-structured
interview based on DSM–IV criteria for use with children and adolescents with ASDs.
22
The authors developed additional screening questions and coding options that reflect the
presentation of psychiatric disorders in ASDs. The modified instrument, the Autism
Comorbidity Interview–Present and Lifetime Version (ACI–PL) was used to assess 109
children with ASDs ranging from 5-to-17 years of age. In a British study, Simonoff et al.
(2008) used the Child and Adolescent Psychiatric Assessment–Parent Version (CAPA) to
assess comorbid psychiatric disorders in an epidemiological, population-derived sample
consisting of 112 children ranging from 10 to 13.9 years of age.
In a study from Finland, Mattila et al. (2010) focused solely on individuals with
ASDs with higher cognitive levels. These authors used the K-SADS–PL (Kaufman et al.,
1997) to identify prevalence of comorbid psychiatric disorders particularly associated
with Asperger’s disorder/high-functioning autism (HFA) in a combined community- and
clinic-based sample of fifty 9- to-16-year-old participants. Gjevik et al. (2011), in a study
from Norway, also used the K-SADS to assess the prevalence of current comorbid
psychiatric disorders in a special school population of 71 children and adolescents with
ASDs who were 6 to 17.9 years of age. In all studies, DSM–IV exclusionary criteria for
ASD were not applied and all comorbid disorders were recorded when diagnostic criteria
were met.
In a series of two studies from the United States, Gadow, DeVincent, Pomeroy,
Azizian (2004, 2005) used the Child Symptom Inventory–4 (both parent and teacher
ratings) (CSI-4; Gadow & Sprafkin, 1994, 2002) to compare DSM–IV symptom
prevalence and severity in preschool children (ages 3 to 5 years, inclusive) and
elementary school-age children (ages 6 to 12 years, inclusive) with PDDs (autistic
23
disorder, Asperger’s disorder, and PDD–NOS). In both studies, children with PDDs were
recruited from a developmental disabilities specialty clinic. Comparison groups in both
studies included non-PDD psychiatric clinic referrals, and pupils in regular and special
education classes.
Also in a study from North America, Witwer and Lecavalier (2010) examined
prevalence of DSM psychiatric symptoms in ASDs in a clinically-referred sample of 61
children and adolescents with a mean age of 11.2 years (range 6–17 years). The authors
used the Children’s Interview for Psychiatric Symptoms–Parent version (P–ChIPS;
Weller et al., 1999), a structured interview designed to assess 20 disorders according to
DSM–IV criteria in children and adolescents.
Rosenberg et al. (2011) explored rates of psychiatric comorbidity in a large,
perhaps the largest, community sample of 4,343 participants. This study differed from the
ones mentioned above in that authors used parent-report of comorbid psychiatric
conditions from an online registry. The Interactive Autism Network (IAN) is an USbased online autism registry of parent reported data which provides information on
community trends in ASD among over 14,000 affected children and more than 22,000
family members. The database was begun in 2007, is continually updated, and all data are
voluntarily submitted by families.
Any Psychiatric Comorbidity
The rate of at least one comorbid psychiatric disorder was comparable across
most of the studies: 72% in both Gjevik et al. (2011) and Simonoff et al. (2008), and 70.8
% in Leyfer et al. (2006). This exceeded the rate of psychiatric disorders both in the
24
general child and adolescent populations (7.0-13.3%) (Costello et al., 2003; Heiervang et
al., 2007), and in children with ID (38.6%) (Dekker & Koot, 2003). As with ASD groups
with broader cognitive levels assessed in studies above, Mattila et al. (2010) reported
psychiatric conditions to be common and often with multiple occurrences even in
individuals with Asperger’s disorder/HFA. In contrast, Rosenberg et al. (2011) reported a
lower proportion of participants (26.9%) with at least one psychiatric comorbidity in their
community sample.
Anxiety Disorders
One of the most prevalent diagnostic groups reported was anxiety disorders,
diagnosed in 42% of the sample in both Gjevik et al. (2011) and Mattila et al. (2010)
studies. Comparable rates were reported by Simonoff et al. (2008) and Leyfer et al.
(2006). On the other hand, Rosenberg et al. (2011) reported a comparatively lower rate
(26.2%) of anxiety disorders. Specific phobia was the most prevalent anxiety disorder
(31% in Gjevik et al.; 32% in Leyfer et al.), congruent with previous reports by other
researchers (de Bruin et al., 2007; Muris et al., 1998). Witwer and Lecavalier (2010)
reported a higher rate for phobia (67.2%). Much lower rates for phobia (8.5%) were
reported by Simonoff et al.
Simonoff et al. (2008) found agoraphobia and panic disorder in about 8% of their
sample. On the other hand, these conditions were not diagnosed in both the Gjevik et al.
(2011) and the Leyfer et al. (2006) sample. Social phobia was diagnosed in 7% of the
Gjevik et al. sample and in 7.5% of the Leyfer et al. sample. In contrast, higher rates of
social phobia were reported by Witwer and Lecavalier (2010) (16.4%), Simonoff et al.
25
(29.2%) and Muris et al. (20.5%) (whose study included a clinical sample comprising
mostly individuals with PDD–NOS). Rate differences may be explained in part by
different ways of interpreting and defining symptoms of anxiety, such as whether
avoidance in social interaction is to be interpreted as a symptom of anxiety or a feature of
ASD. Gjevik et al. interpreted DSM–IV criteria of social phobia as requiring observable
and expressed symptoms of anxiety and not merely avoidance of social interactions.
As regards diagnosis of OCD vs. ASD, the DSM–IV indicates a qualitative
distinction: repetitive behavior is seen as a source of pleasure in ASDs, whereas it is
associated with anxiety and distress in OCD. However, as indicated by Gjevik et al.
(2011), anxiety and distress related to repetitive and stereotyped behavior might easily be
overlooked in children and adolescents with ASDs. Diagnostic rates of OCD were
reported as 4.9% (Witwer & Lecavalier, 2010), 8.5 % (Simonoff et al., 2008), 10%
(Gjevik et al.), and 11.4% (Muris et al.1998). Leyfer et al. (2006) reported a considerably
higher prevalence (37%) which might be explained by their use of a modified criterion
for OCD which was widened to include signs and symptoms that could be observed by
others. Leyfer et al. reasoned that only a small minority of the sample would have met
OCD criteria had they only allowed for reports of subjective mental distress.
Generalized anxiety disorder (GAD) was not diagnosed in the Gjevik et al. (2011)
sample and was diagnosed at a rate of 2.4% by Leyfer et al. (2006). In contrast, Witwer
and Lecavalier (2010), Muris et al., (1998), and Simonoff et al., (2008) reported higher
rates of GAD (24.6%, 22.7%, and 13.4%, respectively). Differences in rates of subtypes
for anxiety disorders across studies may be explained by varying proportions of
26
participants with autistic disorders, varying age groups, and different cognitive and
language levels. In their comparative analysis, Gadow et al. (2004, 2005) found that the
PDD sample subgroup received higher severity ratings on DSM symptoms of specific
phobia, obsessions, compulsions, and social phobia, compared to the non-PDD subgroup.
Attention Deficit Hyperactivity Disorder
A high rate of ADHD was consistently reported in all studies: 38% in Matilla et
al. (2010), 38.1% in Rosenberg et al. (2011), 31% in both Gjevik et al. (2011) and Leyfer
et al. (2006), and 28.1% in Simonoff et al (2008). In the Leyfer et al. study, the rate was
increased to nearly 55% when subsyndromal cases were included. Subsyndromal
disorders were diagnosed when an individual had significant impairment due to
psychiatric symptoms but just fell short of meeting full DSM criteria. Rates as high as
53% to 78% have been reported in clinically referred samples (Witwer & Lecavalier,
2010; Sinzig et al., 2009; Lee & Ousley, 2006). ADHD, Predominantly Inattentive Type
was the most prevalent subtype reported by Gjevik et al., in line with findings by Leyfer
et al. Gadow et al. (2004, 2005) found severity of ADHD symptoms to be comparable
between the PDD sample subgroup and the subgroup referred for outpatient psychiatric
evaluation (but not diagnosed as having PDD).
As mentioned before, the DSM–IV precludes the diagnosis of ADHD in the
presence of ASD, and concurrent diagnosis of ADHD with ASD can be particularly
problematic. Deficits in attention are persistently reported in individuals with ASDs
(Allen & Courchesne, 2001), and such deficits can also be seen as fulfilling the criteria
for ADHD. On the other hand, it is well established that some children with ASDs can
27
attend for long periods to a stimulus they find interesting. Impaired attention becomes
apparent when such children are confronted with school work and other cognitively
demanding activities (Leyfer et al, 2006). This unusual and idiosyncratic attentioninattention pattern found in ASDs further complicates the diagnosis of ADHD in
individuals with ASDs (Dawson & Lewy, 1989). Overall, there exists an ongoing debate
whether ADHD is best conceptualized as a distinct comorbid disorder or as a part of the
ASD diathesis (Sinzig et al., 2009). It is noteworthy that proposed changes in ADHD
diagnostic criteria in the DSM–5 include removing PDD from the exclusion criteria
(http://www.dsm5.org/ProposedRevision/Pages/proposedrevision.aspx?rid=383#).
Oppositional Defiant Disorder/Conduct Disorder
Gjevik et al. (2011) and Leyfer et al. (2006) both reported low levels of ODD/CD
(7%). The fairly low rate of these disorders could indicate that cognitive and other factors
associated with oppositionality in children with ASDs may be different than the factors
reported in children without an ASD. Children with autism might not fully understand the
concepts of spitefulness, vindictiveness, and intentionality (e.g., blaming others for one’s
own mistakes) inherent in the description of ODD/CD (Leyfer et al., 2006). Lending
support to that line of thought, Gjevik et al. found ODD/CD to be more prevalent in
children with PDD–NOS in their sample who had relatively high cognitive level.
Similarly, Gadow and colleagues (2004, 2005) also reported that participants with
Asperger’s disorder were rated as more oppositional than the autism subgroup. Further,
Matilla et al. (2010), in their sample of individuals with Asperger’s disorder/HFA
reported higher rates of ODD (16%), though rates of CD remained low (2%). A similar
28
trend was noted by Simonoff et al. (2008) who reported high rates of ODD (28.1%), but
low rates of CD (3.2%).
Green et al. (2000) reported high rates of both CD and ODD (25% for each);
however, this could very well reflect referral bias in their study which compared
adolescents with CD and Asperger’s disorder. Witwer and Lecavalier (2010) reported the
highest rate for both ODD (75.4%) and CD (50%) in their clinically referred sample. As
DSM–IV guidelines indicate that CD supersedes ODD, it is likely that actual proportions
for ODD and CD were around 25% and 50%, respectively.
Mood Disorder
In the Leyfer et al. study (2006), 10% of the sample was diagnosed with major
depressive disorder (an additional 14% were subsybdromal). Leyfer et al. reported that
these rates were “quite striking” given the mean age of 9 years of their sample and the
fact that none of the children were pre-selected for known psychiatric comorbdity. Gjevik
et al. (2011) reported 10% of their sample to be diagnosed with any depressive disorder
(depressive disorder NOS being the most prevalent subtype); similar rates (11%) were
reported by Rosenberg et al. (2011). Matilla et al. (2010) found lower rates (6%), as did
Simonoff et al. (2008) (1%), who added that 10.9% of the sample just fell short of full
diagnostic criteria. Other investigators have found increased rates of depression in
individuals with ASDs (Abramson et al., 1992; Chung, Luk, & Lee, 1990; Ghaziuddin,
Weidmer-Mikhail, & Ghaziuddin, 1998; Tantam, 1991).
Manic episode or bipolar disorder were either absent (Gjevick et al., 2011) or
reported at low rates (<2%), consistent with rates (3.2%) reported by de Bruin et al.
29
(2007) in a referred sample of children and adolescents with PDD–NOS. Rates of mania
reported by Witwer and Lecavalier (2010) were much higher (14.8%).Wozniak et al.
(1997) reported a 20% prevalence rate of bipolar disorder in a clinically referred sample,
as opposed to 5.2% reported by Rosenberg et al. (2011) in their community sample.
Patterns of Comorbidity
Simonoff et al. (2008) found that for any psychiatric disorder, the majority of
children with at least one disorder had multiple diagnoses (41% of the 71% with at least
one psychiatric comorbidity), and one-third of these (24% of 71%) had three or more
disorders in addition to the ASD. Similar rates have been reported by Gjevick et al.
(2011) (41% of the 72% with any one psychiatric comorbidity) and Rosenberg et al.
(2011) (45.2% of the 26.9% with any one comorbidity). Within the main psychiatric
disorders in the Simonoff et al. study, 80% of those with ADHD had an emotional or
behavioral disorder or both; the same applied to roughly 60% of those with a behavioral
disorder and 40% of those with an emotional disorder. In the Leyfer et al. (2006) study,
the median number of diagnoses per child was 3. The authors suggested that the
frequency of multiple comorbid diagnoses that they reported was likely to be an
underestimate, as parent informants were less likely to complete lengthy interviews when
multiple types of psychopathology were present in their children.
Associations Between Psychiatric Disorders and Participant Characteristics
Studies discussed above also examined subject characteristics that could
potentially function as risk factors for psychiatric comorbidity. Autism severity was not a
significant predictor of comorbid psychiatric disorders in the Simonoff et al. (2008)
30
study. However, these authors did find presence of epilepsy to be a significant predictor
of psychiatric comorbidities. Neither IQ nor Vineland Adaptive Behavior Scale scores
was associated with any other psychiatric disorder category, which is consistent with the
report of Brereton et al. (2006) who failed to find an association between IQ and scores
on the Developmental Behavior Checklist (Einfeld & Tonge, 2002). Similarly, Gjevick et
al. (2011) found no associations between comorbid psychiatric disorders and age, gender,
intellectual level, receptive language, and ASD subtypes. Rosenberg et al (2011) also
found no gender difference in proportion of overall comorbidity. Neither was presence of
ID associated with any comorbidity except as a negative correlate of depression (i.e.,
higher IQ was protective for depression).
Contrasting findings regarding IQ were reported by Witwer and Lecavalier (2010)
who found a significant association between IQ and endorsement rates of psychiatric
symptoms on the P–ChIPS. Across disorder categories, children with an IQ<70 had fewer
reported symptoms than those with IQ≥70. This trend was particularly prominent for
symptoms of ADHD and GAD. These authors also found absence/presence of
conversational language significantly to affect symptom endorsement for several of the
disorder categories (ADHD, ODD, Phobia, GAD). Not surprisingly, the majority of
significant differences were observed in symptoms which require language to be
endorsed (e.g., arguments, blaming others, talking too much). Further, nonverbal
individuals were more likely to be subsyndromal for ODD. As far as differential
association of psychiatric comorbidity with ASD subtypes, Rosenberg et al. (2011)
31
reported a diagnosis of PDD–NOS and Asperger’s disorder to be significantly correlated
with increased odds of each comorbid condition and for overall comorbidity.
Gadow, De Vincent, and Schneider (2008) used parent- and teacher-ratings on the
CSI-4 for 238 children with ASDs aged 6- to 12-years to assess predictors of psychiatric
symptoms. All participants were referrals to a DD-specialty clinic. For mother-rated
symptoms, results indicated that psychotropic medication use was a unique predictor of
symptom severity of ADHD (both Inattention and Hyperactivity-Impulsivity) and ODD.
Pregnancy complications uniquely predicted separation anxiety disorder and specific
phobia. Family history of psychiatric problems uniquely predicted most symptom
categories (except ADHD Inattentive symptoms and specific phobia). For teacher-rated
symptoms, early childhood special education was uniquely predictive of less severe
ADHD Inattentive symptoms and specific phobia. Interestingly, unlike findings based on
mother’s ratings, family history of psychopathology was not associated with the severity
of most teacher-rated psychiatric symptoms (except ODD).
Overall, high rates of psychiatric comorbid conditions have been found in both
community- and clinic-based samples of young people with ASDs. Rosenberg et al.
(2011) commented that differences in rates between research- and clinician-confirmed
diagnosis and their community-based findings suggest that recognition of psychiatric
comorbidity is highly variable, and they recommended that future research consistently
define psychiatric comorbidity within ASDs. The common consensus across all studies
pointed towards an increased need for future research to understand better the
manifestation of DSM psychiatric conditions alongside ASD symptoms.
32
Studies Assessing Maladaptive Behaviors
The second type of study assessing psychopathology in ASDs described levels of
individual maladaptive behaviors. The following is a brief discussion of recent studies
using standardized instruments to characterize maladaptive behaviors in the ASD
population. In an epidemiological study of children with ID, Tonge and Einfeld (2003)
reported on 118 children with autism (mean age of 8.5 years). They used the
Developmental Behaviour Checklist–Parent version (DBC–P; Einfeld & Tonge, 1992,
1995) to rate all participating children at three different intervals across an eight-year
span and reported that 73.5% of children with autism scored above the cutoff considered
clinically significant for caseness. Scores remained fairly stable across time. Overall, the
authors indicated that youngsters with autism are at high risk of suffering from chronic
behavioral and emotional disturbance, over and beyond the symptoms establishing their
diagnosis of an ASD.
Brereton, Tonge, and Einfeld (2006) also used the Developmental Behaviour
Checklist to assess a broad range of emotional and behavioral disturbance in 381 young
people with autism, and a representative group of 581 young people with ID aged 4-18
years. Overall, participants with autism were found to present with significantly higher
levels of maladaptive behaviors than young people with ID. The youth with autism were
found to be highly disruptive and at risk of suffering from problems with impulse control
such as deficits in attention, impulsivity, hyperactivity, and disorganized behavior.
Participants with autism were also highly self-absorbed (displaying a range of
stereotypic, repetitive, and preoccupied behaviors) and anxious. Symptoms of anxious
33
behavior included fear of separation from familiar people, resistance to change, crying
easily over small upsets, tenseness, shyness, and irritability. The autism cohort had
relatively lower scores on the Antisocial subscale mirroring low rates of OD/CDD in
previously discussed studies (Gjevik et al., 2011; Simonoff et al., 2008).
The Brereton et al. (2006) study also found that young people with autism were at
risk of suffering symptoms of depression such as irritability, sleep and appetite
disturbance, psychomotor retardation, and might have had thoughts of suicide or suicidal
behavior. Increased symptoms of depression in adolescents aged 13 and older and those
in the normal IQ range, were consistent with earlier findings of chronic feelings of
unhappiness in verbal adolescents with autism (e.g., Rutter, 1970; Wing, 1982). There
was also a suggestion that some individuals with autism were at risk of major depression
during adolescence and adulthood because of a family history of affective disorders
(Gillberg & Coleman, 1992).
Lecavalier (2006) used parent- and teacher-rated NCBRFs to study the relative
prevalence of behavioral and emotional problems in a sample of 487 youngsters (mean
age 9.6 years) with PDDs. Data were collected in 37 school districts across Ohio. A
cluster analysis of parent-rated NCBRFs (n = 353) revealed that 49% of the sample was
characterized by some form of behavioral difficulties. Identified clusters included
Ritualistic and Hyperactive (13% of the sample), Hyperactive with Conduct problems
(14%), Anxious (13%), and Undifferentiated Behavior Disturbance (9% of the sample).
34
Association Between Maladaptive Behavior and Participant Characteristics
The level of overall maladaptive behavior was not affected by age, sex, or IQ in
the Brereton et al. (2006) study. However, DBC subscale scores differed as a function of
age and gender. For example, children with autism, aged 13 years and older, had more
problems with social relating (including aloofness, poor eye contact, preferring to be
alone) than the younger subjects. Also females had more problems with social relating
than males. Disruptive and ADHD-like behaviors and symptoms of depression and
anxiety were equally common for males and females with ASDs. This is in contrast to the
general population where ADHD is more common in males, reducing in prevalence
during adolescence, but where anxiety and depression are more likely in adolescent
females (Rutter et al., 2003). In explaining these intriguing differences in gender
distribution in their study, Brereton et al. suggested that the equal vulnerability for
ADHD in females and males with ASDs might reflect a shared neurobiological
impairment which overrides the usual relative neurodevelopmental resilience of females.
Similarly, the overriding effects of organic brain dysfunction and genetic risk of
psychopathology (Rutter et al., 2003) might explain the equivalence of depressive
symptoms in both adolescent males and females.
Language ability seemed to have a robust influence on maladaptive behaviors. In
the Brereton et al. (2006) study, verbal children in the autism cohort had significantly
more problems with communication disturbance (including social use of language) and
antisocial behavior, whereas non-verbal children had more problems with self-absorbed
and social relating behavior.
35
Overlapping findings were reported by Lecavalier (2006). Membership in
different clusters of behavioral difficulties did not differ based on age and gender.
However, clusters did differ as a function of adaptive behavior, with members of the
Anxious cluster having significantly higher adaptive behavior scores than others. The
Ritualistic and Hyperactive cluster members had significantly lower adaptive behavior
scores than others, except for members of the Hyperactive with Conduct problems
cluster.
Review of Studies Reporting on Changes in Maladaptive Behaviors
Though there is now an appreciable body of literature describing the high
prevalence of psychopathology in ASDs, the progression of maladaptive behaviors over
time in this population remains an understudied area. This section summarizes findings
from a handful of studies assessing change in maladaptive behaviors in ASDs.
Ballaban-Gil, Rapin, Tuchman, and Shinnar (1996) reported on problem behavior
outcomes in their longitudinal examination of behavioral, language and social changes in
102 individuals with ASDs. Follow-up data were collected over a structured telephone
interview with a parent and included information on problem behaviors in general, selfinjurious behaviors, and stereotypies. Results indicated that behavioral difficulties
continued to be a problem for 69% of adolescents and adults. In all, 35% of adolescents
and 49% of adults engaged in self-injurious behavior, and slightly more than 50% of
adolescents and adults exhibited some stereotypic behavior. Compared to behaviors at
initial evaluation, about half of the participants had worsened behavior at follow-up,
36
slightly less than 20% improved, and one third did not change. No association was found
between behavior change and IQ.
Shattuck et al. (2007) prospectively examined change in autism symptoms and
maladaptive behaviors during a 4.5 year period in a community sample of 241
adolescents and adults with ASDs. Participants were 10 to 52 years old (mean = 22.0)
when the study began. Assessments included data collected from the individual with
ASD and in-home interviews with mothers at various time points in the follow-up.
Autism symptoms were assessed using the ADI–R. Maladaptive behaviors were assessed
using the Problem Behavior scale of the Scales of Independent Behavior–Revised (SIBR; Bruininks et al., 1996), which measures behaviors grouped in three domains:
internalized, externalized, and asocial. Since cognitive testing results were not available
for most participants, ID status was determined using a variety of sources of information
including scores on the Wide Range Intelligence Test (50% of sample), Vineland
Screener, and reviews of available records (historical standardized assessments, parent
reports of prior diagnoses, clinical and school records). Overall level of language was
deduced from the single ADI–R item indicating whether the individual was nonverbal or
verbal (defined as using at least 3-word phrases). At follow-up, all four SIB–R scales
decreased significantly over time indicating an overall decline in maladaptive behaviors.
The most robust predictor of change in maladaptive behavior score was Time 1 measure
of the dependent variable. Other significant predictors included age cohort and ID status.
Being in the oldest age cohort was associated with significantly greater decline
(compared to those aged 10 through 21 years), thus indicating that reduction in
37
maladaptive behaviors continues well into, and may even accelerate, in midlife. Those
who were assigned an ID status by the authors improved less over time. The change in
proportion of variance explained attributable to adding other independent variables to the
Time 1 score ranged from 3% to 7% and was significant for three of the four models
(internalized, asocial, and maladaptive behavior total).
Lounds Taylor and Seltzer (2010) further built on the above mentioned dataset by
Shattuck et al. (2007) and focused on the important transitional event of exiting the high
school system to examine if this turning point was associated with rates of change in
autism symptoms and maladaptive behaviors in youth with ASDs. The sample consisted
of 242 youth with ASDs who recently exited the school system and were part of the
larger longitudinal study described above. This study extended the previous work by
Shattuck et al. by collecting sufficient data to examine change over a nearly 10-year
period (opposed to a 4.5 years follow-up). As in Shattuck et al., maladaptive behaviors
were assessed using the Behavior Problems subscale of the SIB–R (three domains of
externalized, internalized, and asocial behaviors). Results indicated that, on average,
maladaptive behaviors prior to high school exit improved over time for all three domains.
However, after high school exit, there was an overall slowing of improvement.
Improvement in internalized behaviors slowed significantly after high school exit from a
rate of 0.82 points/year prior to exit to 0.31 points/year after exit. This slowing down of
improvement was evident for both youth with and without comorbid ID. A different
pattern was observed in the domains of externalizing and asocial behaviors, where the
most pronounced slowing was observed for young adults with ASD who did not have ID.
38
For externalizing behaviors, after high school exit, change was reduced to 0.11
points/year (from 1.03 points/year) for youth without ID, whereas improvement became
more pronounced for those with ID (from 0.29 points/year to 0.53 points/year after exit).
A similar pattern was observed for asocial behaviors.
Overall, for youth without comorbid ID, high school exit appeared to have a more
pronounced influence on maladaptive behaviors over time. The authors suggested that
this may be related to difficulties finding appropriate educational and occupational
activities for individuals without ID on the autism spectrum. For example, Taylor and
Seltzer (2010) found that nearly 74% of young adults with ASD and comorbid ID
received adult day services in the years immediately following high school exit compared
to only 6% of those without ID. Family income also exerted a significant influence on
maladaptive behavior change after high school exit. For those youth whose families had
higher incomes (income at the 75th percentile or over $70,000 per year), internalized and
externalized behaviors after exit improved more over time relative to those whose
families had lower incomes. Interestingly enough, family income only impacted change
in maladaptive behaviors after high school exit, but not levels of behaviors when the
individual was still in the school system. This finding, the authors reasoned, may indicate
some degree of success of the Individuals with Disabilities Education Act (IDEA), which
mandates appropriate educational services for all school-aged children with disabilities. If
behavioral improvement is assumed to be related to access to services, the lack of
association between family income and behavioral improvement while still in high school
may reflect a positive influence of autism services through the school system that differ
39
little based on parental SES. This line of reasoning would also suggest that family income
may greatly influence access to services after youth with ASD exit the school system.
Matson, Mahan, Hess, Fodstad, and Neal (2010) examined the effect of age on
maladaptive behaviors in individuals with ASDs. This was not a follow-up study and
instead used a cross sectional design to study the effect of different age cohorts on
maladaptive behaviors in a sample of 167 individuals with ASDs, ages 3 to 14 years
(mean = 8.13). Maladaptive behaviors were assessed using the informant-rated Autism
Spectrum Disorders–Behavior Problem Child scale (ASD–BPC; Matson, Gonzalez, &
Rivet, 2008). The ASD–BPC has two dimensions–an internalizing scale (comprising
items assessing SIB, stereotypies, inappropriate sexual behaviors, and other odd
behavior), and an externalizing scale (comprising items assessing physical and verbal
aggression, property destruction, and tantrum-like behaviors) (Matson et al., 2008).
Results indicated lack of any significant differences in maladaptive behaviors among the
three age cohorts, i.e., 3-6-year-olds, 7-10-year-olds, and 11-14-year-olds. Furthermore, a
curve estimation to explore linear trend of maladaptive behaviors throughout the
childhood years indicated that behaviors remained stable over time. However, the authors
noted that the trend did approach significance suggesting that challenging behaviors may
decrease in older children compared to younger children which would be congruent with
findings reported earlier (Shattuck et al., 2007).
Kelley, Naigles, and Fein (2009) reported on a 8-year follow-up on multiple
aspects of functioning, including problem behaviors, in a group of “optimal outcome”
(OO) children with a history of ASDs. These children were matched on age, gender, and
40
nonverbal IQ to a group of typically developing children (TD) and a group of highfunctioning children with ASD (HFA) who still retained a diagnosis on the autism
spectrum. OO children were those who were mainstreamed into regular classrooms and
had, according to their records, a full-scale IQ within the average range (i.e. greater than
70). In this study, the OO children were no longer receiving extra help in the classroom,
had lost their diagnosis according to the school system, and no longer met criteria for a
diagnosis of ASD according to their Autism Diagnostic Observation Schedule (ADOS-G;
Lord et al, 2000). The HFA group also had a full-scale IQ within the average range
according to their records but, unlike the OO group, retained their diagnosis in the school
system and continued to meet criteria for a diagnosis of ASD according to the ADOS-G.
The Behavior Assessment System for Children (BASC; Reynolds & Kamphaus,
1992) was used to assess problem behaviors for all three groups (OO group N = 9, HFA
group N = 11, TD group N= 12). At follow-up, the HFA group had mean scores in the atrisk range on the Adaptability, Atypicality, Withdrawal, Social Skills, and Leadership
subscales, indicating that they continued to experience behavioral difficulties in these
areas. In contrast, the OO group scored in the normal range on all of the subscales of the
BASC, although they were not significantly different from the HFA group on Conduct
Problems, Depression, Atypicality, and Withdrawal. Further, they were found to be
borderline at-risk on the Attention subscale. The borderline at-risk Attention scores are
consistent with the existence of a subgroup of OO children who present with ADHD as
they lose their autism symptoms (Fein, Dixon, Paul, & Levin, 2005). Because some of
the OO children’s scores were elevated in the current study at 8 to 13 years of age, it is
41
quite possible that problems with internalizing or externalizing behaviors might emerge
as they enter adolescence.
Most recently, Anderson, Maye, and Lord (2011) examined trajectories of change
in maladaptive behaviors, as well as predictors of such behaviors, across ages 9 to 18
years for youth with ASDs and a comparison group with nonspectrum developmental
delays. Participants were 116 youth who were diagnosed with an ASD prior to 37 months
of age. This study focused on the transitional period spanning mid-childhood through late
adolescence. Assessment consisted of a battery of diagnostic and psychometric
instruments that were administered in person at three points during the follow-up–at ages
2 years, 5 years (for part of the sample), and 9 years. In addition to these instruments, a
log of early childhood educational and intervention treatments and records of any
seizures were obtained via caregiver interviews and diaries. Caregivers were also asked
to provide information on maladaptive behaviors, development during puberty,
medication use, and seizure activity. This was done through parent-rated instruments
when the youth were 9 years old and then every 4 months through mailed questionnaires
when the adolescents were between 13 and 18 years of age. Diagnostic instruments
included the ADI–R and the ADOS. Maladaptive behavior was assessed using three
subscales of the Aberrant Behavior Checklist (ABC; Aman et al, 1994). These were: the
Lethargy/Social Withdrawal, Hyperactivity, and Irritability subscales. Pubertal onset was
assessed using the parent-rated Pubertal Development Scale (PDS; Petersen, Crockett,
Richards, & Boxer, 1988).
42
Consistent with the researchers’ hypothesis, the children with autism scored
higher on externalizing problem behaviors related to irritability and hyperactivity at age 9
years than the comparison groups. With age, the overall trend was one of decreasing
irritability and hyperactivity problems, particularly for youth with an autism diagnosis.
This trend of improved behaviors over time is congruent with the general pattern of
improvement in core symptoms of ASDs (e.g., Seltzer et al., 2004) and general social and
cognitive skills (Anderson et al., 2007, 2009). However, in this particular study, the
picture was different for behaviors related to social withdrawal, which increased with age
for a substantial minority in both youth with autism and broader ASD (BASD), but not in
the nonspectrum comparison group. This increase was found to be unrelated to IQ
differences, and the increase was more pronounced for youth with BASD (a 10-point
increase for one-third of the children), despite a less severe presentation of core social
symptoms. The authors suggested that this pattern of increasing withdrawal over time
suggests the presence of vulnerability factors in late childhood and adolescence that may
exacerbate social difficulties associated with ASDs. These findings appear contrary to
those reported by Lounds Taylor and Seltzer (2010), who found that internalizing
symptoms decreased over age in high school-aged youth with ASDs. However, as the
authors cautioned, results from these two studies might not be directly comparable
because Lounds Taylor et al. used a more general measure of withdrawal that
encompassed social withdrawal, self-injury, and inattention.
As far as predictors of change in maladaptive behaviors in the Anderson et al.
(2011) study, onset of puberty was associated with more problems related to irritability
43
and social withdrawal. Seizure onset also predicted higher levels of social withdrawal,
though the effects of seizure diminished with age. Gender, race, mother’s education, and
hours of individual treatment received were consistently non-significant as predictors of
outcome.
Summary
In summary, although growing evidence indicates wide prevalence of problem
behaviors in ASDs, there has been relatively little research documenting their course as
youngsters with ASDs grow older. Even fewer studies have examined the predictors
associated with change in maladaptive behaviors over time. Findings from the scant
literature suggested that pattern of change in maladaptive behaviors was highly
heterogeneous. Although studies were too few in number to reach firm conclusions about
predictors of change, initial levels of maladaptive behavior and older age appeared to be
significantly associated with change over time.
Why is it Important to Study Progression of and Predictors of Maladaptive
Behaviors
Understanding the natural history of maladaptive behaviors and determining
predictors of change in maladaptive behaviors are important issues to consider for a
number of reasons.
First, chronic comorbid maladaptive behaviors represent an added handicap for
the individual with ASD and they require considerable attention and can become a focus
for intervention and medical treatment (Gadow et al, 2008; Holtmann et al., 2007;
Weisbrot et al., 2005). If not managed properly, maladaptive behaviors can have
44
pervasive and long-lasting negative effects on the lives of individuals and their families.
These include negative outcomes associated with academic achievement, social
competency outcomes (such as school/vocational placement), and adult psychiatric
outcomes for the individual with ASD (Chadwick, Piroth, Walker, Bernard, & Taylor,
2000; de Bildt, Sytema, Kraijer, Sparrow, & Minderaa, 2005; Kim, Szatmari, Bryson,
Streiner, & Wilson, 2000). Persisting maladaptive behaviors have also been found to be
related to (a) poor societal attitude toward these individuals (Morton & Campbell, 2008),
(b) increased risk of injury (Lee, Harrington, Chang, & Connors, 2008), (c) poor
adaptation to one’s environment (Vieillevoya & Nader-Grosbois, 2008), and (d)
increased and long-term use of psychotropic medications and their accompanying side
effects (Advokat, Mayville, & Matson, 2000; Singh, Matson, Cooper, Dixon, & Sturmey,
2005). Overall, maladaptive behaviors are associated with a lower quality of life for
individuals with ASDs with respect to social integration, leisure activities, health and
security, and self-determination (Gerber, Baud, Giroud, & Caminati, 2008). As is evident
from the findings mentioned above, the effects of chronic maladaptive behaviors are
indeed multi-faceted and significant for individuals with ASDs.
Second, as underscored by Anderson et al. (2011), maladaptive behaviors may not
necessarily follow the same trajectory over the life span as core symptoms of ASDs. In
fact, comorbid behaviors such as irritability, hyperactivity, and social withdrawal, show
low-to-moderate correlations with core features of ASDs (Lecavalier, 2006). Although,
as discussed before, a general tendency of abatement of core ASD symptoms in
adolescence and adulthood have been reported in the literature (Seltzer et al.,2004), some
45
internalizing maladaptive behaviors, such as depression and anxiety, appear to be more
common in adolescence and adulthood (Lecavalier, 2006; Tonge & Einfeld, 2003). This
is particularly true for youth with comparatively greater language ability such as those
with Asperger’s disorder (Ghaziuddin, 2002; Weisbrot et al., 2005, Gadow et al., 2005),
higher IQ scores (Estes et al., 2007), and/or those with fewer core symptoms of ASDs
(Gadow et al., 2005; Kanai et al., 2004).
Third, interventions for young people with ASDs may be rendered ineffective or
less effective if secondary maladaptive behaviors are not addressed (Anderson et al.,
2011). Additional clarification on the presentation and patterns of change in various
maladaptive behaviors for young people with a broad range of core ASD symptoms and
cognitive abilities can help in the refinement of treatment strategies for maximum
efficacy.
Given the significance that persisting maladaptive behaviors can have on the lives
of individuals with ASDs, their families and service providers, the recognition of the
nature of these behaviors over time is clearly important (Downs, Downs, & Rau, 2008;
O’Reilly et al., 2008; Ringdahl, Call, Mews, Boelter, & Christensen, 2008). The current
inadequacy of information in the scientific literature delving into the course of these
behaviors over time is indeed a point of concern. Additional longitudinal studies
examining maladaptive behaviors in ASDs are needed to extend our understanding on the
prevalence of, and predictors of, changes in such behaviors. Murphy et al. (2005)
commented that it would be preferable for information about the chronicity of
maladaptive behavior to be gained not just from longitudinal follow-up studies of
46
individuals selected because they showed challenging behaviors at the first time point but
from studies of individuals in whom maladaptive behavior (or its absence) can be logged
at two distinct time points, so that both old and new behaviors can be detected. Ideally,
such studies would examine outcomes and predictors of maladaptive behaviors using a
standardized instrument that is appropriate for use in the ASD population.
This dissertation was such a longitudinal study of a community sample of children
and adolescents with ASDs with the overall purpose of adding to knowledge in the
scientific literature on the course and predictors of maladaptive behaviors in ASDs.
Current Study: Aims and Hypotheses
Participants for the current study were recruited using case records of children and
adolescents who received an ASD diagnosis from clinics at the Nisonger Center at the
Ohio State University two-to-eight years previously. Follow-up data were collected from
families of the youth using a mail survey and supplemented by telephone interview.
Aims
The primary aims of the study were as follows:
(1) To characterize the sample of children and adolescents with ASDs in terms of
parent-rated maladaptive behaviors on a standardized rating scale at initial contact and
follow-up;
(2) To examine the stability of and changes in maladaptive behaviors from initial
contact to follow-up assessment; and
(3) To assess the predictive power of a host of variables in forecasting levels of
maladaptive behavior at follow-up.
47
These aims were considered primary because data on maladaptive behaviors
were collected both at the initial time point and follow-up using a standardized rating
scale designed for use in the DD population. These were considered relatively “hard
data” in this study.
Additionally, the study had the following exploratory aims:
(1) To characterize the sample, at follow-up, in terms of parent report on (a)
comorbid psychiatric diagnoses received , (b) current educational placement, (d)
psychotropic medications used, and (d) intervention history;
(2) To examine the role of variables collected at initial contact in predicting DSM
psychiatric diagnosis at follow-up; and
(3) To determine if these same variables had any role in predicting educational
placement at follow-up.
The above aims were categorized as exploratory because they utilized “soft
data” collected at follow-up via parent report during a telephone interview. This
methodology might be seen as less desirable than in-person structured clinical
interviews and/or additional direct observation of the individual with ASD.
Hypotheses
Based on reviews of aforementioned published studies, I developed the following
primary hypotheses for this study:
(1) Sample characterization: I hypothesized that maladaptive behaviors will be
present at substantial rates at both initial contact and follow up (Brereton et al., 2006;
Gadow et al., 2004, 2005; Lecavalier, 2006; Tonge and Einfeld, 2003).
48
(2) Stability and change in maladaptive behaviors: Given the finding of a general
pattern of improvement in autism symptoms (Piven et al., 1996; Seltzer et al., 2003), and
a similar pattern of improvement in maladaptive behaviors (Shattuck et al., 2007; Lounds
Taylor & Seltzer, 2010; Anderson et al., 2011), I expected to find a general decline in
maladaptive behaviors at follow-up. However, based on aforementioned findings
reported by Anderson et al. (2011), Ballaban-Gil et al. (1996), and Kelley et al. (2009), I
also predicted that decline might not be universal for all types of maladaptive behavior.
Overall, I hypothesized that rates of maladaptive behaviors would be moderately
correlated between initial assessment and follow-up.
(3) Predictors of maladaptive behaviors at follow-up: Regarding predictors, based
on my review of prior research (Shattuck et al., 2007, Murphy et al., 2005), I expected
that levels of maladaptive behavior at initial contact would be a significant predictor of
maladaptive behaviors at follow-up. I also expected that language abilities would be a
significant predictor–that those with better language abilities at initial contact would
show greater declines in maladaptive behaviors at follow-up (e.g., Lord & Bailey, 2002;
Szatmari, 2000). However, this hypothesis also had a caveat. Based on findings by
Brereton et al. (2006) and Witwer and Lecavalier (2011), I expected that individuals who
had better language capacity at initial contact would do worse at follow-up on
maladaptive behaviors which involved spoken language (such as arguments with adults
and peers, talking too much or too loud, threatening people, talking back). Among other
hypotheses for this exercise, I did not expect amount of intervention (i.e., intensity or
hours of intervention) to emerge as a significant predictor (Darrou et al., Anderson et al.,
49
2011; Gabriels et al., 2010; Eaves & Ho, 2004). In fact, as suggested by Darrou et al.
(2010), I planned on assessing the predictive power of type of intervention and expected
this to be predictive of outcome on maladaptive behavior at follow-up.
For the exploratory aims, I had the following hypotheses:
(1) Characterization of sample at follow-up: I predicted a substantial rate of
psychiatric comorbidity at follow-up, with comorbid diagnoses of anxiety disorder and
ADHD most commonly reported (Gjevik et al., 2011; Leyfer et al., 2006; Matilla et al.,
2010; Simonoff et al., 2008; Rosenberg et al., 2011). Based on broad findings in the ASD
literature, I hypothesized that at follow-up the majority of individuals would be in
classes/programs that provide special accommodations (White, Scahill, Klin, Koenig,
& Volkmar, 2007), that most would received some form of intervention (Turner et al.,
2006), and most would be taking at least one psychotropic medication (Aman,
Lam, & Van Bourgondien, 2005).
(2) Predictors of DSM diagnosis at T2: I predicted that high initial levels of
maladaptive behaviors associated with anxiety and hyperactivity would predict comorbid
diagnoses of anxiety disorder and ADHD, respectively, at follow-up.
(3) Predictors of educational placement at T2: I hypothesized that low levels of
externalizing maladaptive behaviors and high language ability would predict placement
in more normative classes.
50
Chapter 2: Methods
Participants and Procedure
Eligible participants were consecutive cases who were referred for a possible
diagnosis of autism spectrum [autistic disorder (referred to as “autism” throughout this
document), Asperger’s disorder, and PDD–NOS]. These individuals were seen at the
Autism Spectrum Disorder (ASD) Clinic or Family Directed Developmental (FDD)
Clinic at the Nisonger Center University Center for Excellence in Developmental
Disabilities (UCEDD) at the Ohio State University. The ASD Clinic provides services to
both children and adolescents, while the FDD clinic typically sees younger children
(younger than four years). These are collectively referred to as “clinics” throughout this
document, unless specification is necessary. Families seen at the clinics came from all
over Ohio, from varied ethnic and socio-economic backgrounds, providing good
representation of Ohio demographics.
Assessments at the clinics involved an interdisciplinary evaluation to determine
whether one or more diagnoses were appropriate. Members of the interdisciplinary team
routinely included a clinical psychologist, pediatrician, and speech and language
pathologist. The interdisciplinary team reviewed the child’s previous medical,
educational, and psychological records, and conducted behavioral and medical
assessments and standardized testing of cognitive and language skills, as relevant, with
51
the child. The Nisonger Child Behavior Rating Form (NCBRF; Aman, Tassé, Rojahn, &
Hammer, 1996; Tassé, Aman, Hammer, & Rojahn, 1996), a measure of problem behavior
in individuals with DDs, was part of the required intake packet. The NCBRF was
completed by all parents (or, other significant caregivers) before the initial visit. Routine
assessments included a structured interview with parents using the Autism Diagnostic
Interview—Revised (ADI—R; Lord, Rutter, & LeCouteur, 1994) (for all cases seen at
the ASD Clinic), and some form of standardized assessment of language of the child.
Other evaluations, determined on a case-by-case basis by the team, included IQ
assessment, completion of adaptive behavior scales, and various ASD-specific rating
scales by parents/caregivers. After the evaluation, the family was given a complete report
containing the diagnosis (as applicable), results from psychometric testing, and
recommendations for educational and behavioral interventions.
During the planning stage of this study, I conducted a feasibility check and
reviewed a sample of current and archived patient records from the clinics to assess the
range of data that were available for all cases seen. Information available in patient
records included (a) coverage of maladaptive behavior assessment, as measured by the
parent-rated NCBRF for all children assessed, (b) universal coverage of language
capabilities by some form of standardized assessment using tests such as the
Comprehensive Assessment of Spoken Language (CASL; Carrow-Woolfolk, 1999), the
Preschool Language Scale–4th ed. (PLS-4; Zimmerman, Steiner & Pond, 2002), and the
Bayley Scales of Infant Development–3rd ed. (Bayley, 1993), among others, (c) domain
scores on the ADI–R (for all children seen at the ASD Clinic), and (d) intelligence
52
testing and adaptive behavior evaluation on at least half of the cases seen. Following
approval from the OSU Social and Behavioral Sciences Institutional Review Board, the
study was carried out in two stages. All tables and figures appear in Appendix A.
Stage 1
I conducted a detailed review of all archived patient charts and entered relevant
data into an electronic database for the purposes of this study. Cases were included as
eligible participants in this follow-up study if they (a) had received an ASD diagnosis
from the clinics, and (b) had a parent-rated (or caregiver-rated) NCBRF on file. Other
information harvested from patient charts included (a) demographic information, (b)
contact information for the family, (c) ASD diagnosis, (d) scores on parent-rated
NCBRF, (e) scores on language assessment, (f) domain scores on the ADI-R, and (g)
scores on intelligence tests and adaptive behavior, if available. The final pool of eligible
participants comprised 342 young people who were seen at the clinics between 2003 and
early 2010 (data in patient records were increasingly inconsistent for cases seen before
2003 and were not included). Over 75% of these individuals were seen at the ASD Clinic.
Participants in this study included the children with an ASD and their parents or legal
guardians (referred to as “family”).
Stage 2
In Stage 2, I conducted a follow-up on the identified cases and collected (a) a
current parent-rated NCBRF via a mail survey and (b) current demographic information
for the child via a phone interview with the family. I used a 5-step follow-up process to
contact eligible participants as described below.
53
Step 1. An initial letter was sent out to all eligible families alerting them to the study
which was referred to as “The OSU Autism Spectrum Follow-up” (Appendix B). The
letter was signed by Sherry Feinstein and me. Ms. Feinstein was the Clinical Program
Manager at the clinics, she conducted all intake interviews before families were
scheduled for assessment, and she had an ongoing rapport with families seen at the
clinics. The letter described the goals and importance of the study, and asked families to
expect a phone call from me in the next ten days to explain more about the study and see
if they might be interested in participating. Finally, the letter indicated that families could
contact us within ten days of receiving the letter if they did not want to participate in the
study.
Step 2. If I did not receive phone calls from families in the next two weeks, I contacted
them via telephone. If the call was directed to an answering service, I explained my
reason for calling and left my contact information. If someone answered the call, I asked
to speak to the parent or legal guardian (“family”) of the child. When speaking to the
caregiver, I first asked if he or she received the initial letter. Then I explained that his or
her participation in the study would entail a 15-20 minute phone interview with me to
answer questions about the child and completing the NCBRF that I would mail out. I
further indicated that a small compensation ($15) was available for participation in this
study. I then sought verbal consent to participate.
If the family declined participation, I thanked them and the phone call was ended.
If the family provided verbal consent, I conducted a brief interview to complete a
demographic form (described in detail under Instruments and presented in Appendix C).
54
On some occasions, the family and I scheduled a future time to complete the phone
interview. I also asked families if, optionally, they would give me permission to contact
their child’s school for the purpose of confirming the child’s class placement. This was
done to verify the information regarding the child’s educational placement that families
provided during my phone interview. Providing permission to contact schools was
optional, and families could still participate in this study (and receive compensation)
without providing permission to contact the child’s school. If families for whom
telephone voice messages were left did not get in touch with me within ten days, I called
them a second time and followed the same procedure as described above.
Step 3. A packet containing study forms was mailed to families who provided verbal
consent to participate. The packet contained a cover letter (Appendix D), the NCBRF,
appropriate consent documentation including an assent form for the child’s signature,
and a self-addressed stamped envelope for returning documents. As soon as I received
the completed NCBRF from families, I mailed back the payment of $15 along with
copies of appropriate consent documentation. If data were missing on the NCBRF, I
tried calling the family for further clarification on the items missed. If families provided
written permission to contact the child’s school (via a Release of Information form), I
tried to confirm the child’s class placement by calling the school, providing the signed
Release of Information form, and talking to the appropriate personnel (most often the
class teacher or special education teacher).
Step 4. If I did not receive the completed NCBRF in the mail within three weeks of
initial mailing, I sent out a reminder packet via mail (same content as initial packets) and
55
followed this up with a reminder phone call to the family. A voice message was left if
the call went to an answering service.
Step 5. If no mail reply was received within two weeks of the first reminder packet, I
sent out a second reminder packet via mail (same content as initial packets).
The final sample consisted of 143 children and adolescents with an ASD (41.8%
of the target pool of 342 eligible participants). For all individuals in the final sample, I
had current demographic information (completed via phone interview with family) and
completed parent-rated NCBRF at follow-up. Sample characteristics at initial visit or
Time 1 (T1) and follow-up Time 2 (T2) are presented in Tables 3 and 7, and discussed in
the Results section.
I received funding for the study from two sources: $1995 through the Alumni
Grants for Graduate Research and Scholarship, a competition open to all Ohio State
graduate students, and $600 through the Nisonger Center Research Fund. I used these
funds to provide compensation to participants and to purchase postage and mailing
materials.
Instruments
Measure of Maladaptive Behavior
The principal instrument used in this study was the Nisonger Child Behavior
Rating Form (NCBRF; Aman et al., 1996; Tasse´ et al., 1996) that was used to assess
maladaptive behaviors at T1 and T2. The NCBRF is an informant-rated instrument
designed to assess behavior and emotional problems in individuals with DDs. The
NCBRF also includes a brief section on positive social behavior, which consists of 10
56
items scored on two subscales. Since this dissertation is concerned with changes in and
predictors of maladaptive behaviors, only ratings on the NCBRF Problem Behavior
subscales were included in my analyses. There are two forms of the NCBRF: a parentrated and a teacher-rated version. The current study used only the parent-rated form.
The 60 problem behavior items on the parent-rated NCBRF are distributed across
six subscales: Conduct Problem (16 items), Insecure/Anxious (15 items), Hyperactive (9
items), Self-Injury/Stereotypic (7 items), Self-Isolated/Ritualistic (8 items), and Overly
Sensitive (5 items). Parents were instructed to describe the child’s behavior as it was at
home over the previous month using a four-point Likert scale, ranging from “behavior
did not occur or was not a problem” (0), to “behavior occured a lot or was a serious
problem” (3). The 10 positive social behavior items are scored onto two subscales:
Compliant/Calm (6 items) and Adaptive/Social (4 items). These are also rated on a fourpoint Likert scale ranging from “not true” (0) to “completely or always true” (3).
The NCBRF has been found to show good psychometric properties in several
studies (Aman et al., 1996; Lecavalier, Aman, Hammer, Stoica, & Mathews, 2004;
Tasse´, Morin, & Girouard, 2000). In the original psychometric study, Aman et al.
reported a median Cronbach alpha value of 0.85 for the Problem Behavior subscales of
the parent-rated NCBRF, suggesting good internal consistency for the subscales. The
authors also established good convergent validity of the NCBRF with corresponding
subscales of the Aberrant Behavior Checklist (ABC, Aman, Singh, Stewart, & Field,
1985). Median correlations between analogous subscales of the parent-rated NCBRF and
the ABC were 0.72 indicating that clinically related subscales of the two instruments
57
assessed similar constructs. Lecavalier et al. (2004) reported on the factor structure of the
NCBRF is a sample of 330 children and adolescents with ASDs using both parent and
teacher ratings. Problem behavior items were found to be distributed across a “simpler”
five-factor solution for both rating forms (as opposed to six factors in the original
validation study of Aman et al.). In line with findings from Aman et al., factor structures
from the parent and teacher versions were overall similar, but with some differences. For
example, self-injurious and stereotypic items loaded on two distinct subscales for the
teacher form, but not on the parent form. Social competence items showed more
similarity with the original solutions than did problem behavior items. Factor loadings
and internal consistencies were generally lower than those reported for the original
versions but were still within the acceptable range. Overall, the NCBRF was found to be
a psychometrically valid and useful instrument for use in children and adolescents with
ASDs.
ASD Measure
The Autism Diagnostic Interview—Revised (ADI—R; Lord, Rutter, &
LeCouteur, 1994) is a standardized, semi-structured interview conducted with a primary
caregiver. It is based on the DSM–IV (American Psychiatric Association, 1994) and
International Classification of Diseases (World Health Organization, 1992) criteria for
autism. The ADI–R operationalizes the DSM–IV definition of autism by establishing
thresholds that indicate qualitative impairments in the three domains of reciprocal social
interaction, communication, and restricted, repetitive behavior and interests (RRBs).
Behavioral descriptions given by the caregiver are coded by the interviewer as 0 (no
58
abnormality), 1 (possible abnormality), 2 (definite autistic-type abnormality), and 3
(severe autistic-type abnormality). Scores of 3 are recoded to 2 for scoring purposes. For
each domain, two scores are assigned: one reflecting “current” levels of impairment, and
another score reflecting “lifetime” impairment, defined as impairment at any time in the
individual’s life (“ever”).
Language Measures
A variety of language assessment measures were used at the clinics depending on
the needs of the specific child. Often there were results from multiple tests on file. The
most commonly used tests at the clinics included the following.
(1) Comprehensive Assessment of Spoken Language (CASL; Carrow-Woolfolk, 1999).
The CASL is an oral language assessment battery of tests designed for use with
individuals between the ages of 3 and 21 years. It includes subtests in four different
categories of language: lexical/semantic, syntactic, supralinguistic, and pragmatic. In
addition to subtest scores, the CASL also produces an overall Core Composite score
which is a global measure of oral language. The Core Composite provides an index of
performance on a group of selected subtests that are representative of all the categories
for each age band. It is based on combinations of four subtests for children 6 years and
under and five subtests for individuals between 7 and 21 years.
(2) Preschool Language Scale –4th ed. (PLS-4; Zimmerman, Steiner & Pond, 2002). The
PLS-4 is a standardized tool intended for use with children from birth to six years. The
PLS-4 yields scores on two subscales (Auditory Comprehension and Expressive
Communication), a composite Total Language score, and three supplemental assessments
59
(the Language Sample Checklist, the Articulation Screener, and the Caregiver
Questionnaire). The Total Language score is the sum of the Expressive Communication
and Auditory Comprehension standard scores.
(3) Bayley Scales of Infant and Toddler Development–3rd ed. (Bayley-III; Bayley, 1993).
The Bayley-III is an instrument that assesses the developmental functioning of infants
and young children 1 month to 42 months of age. It consists of three administered
subscales with standard scores: the Cognitive Scale, the Language Scale (including the
Receptive Communication and Expresisve Communication subtests), and the Motor
Scale. Additionally, the Social-Emotional Scale and the Adaptive Behavior Scale form
the Social-Emotional and Adaptive Behavior Questionnaire, which is completed by the
parent or primary caregiver.
(4) Test of Pragmatic Language –2nd ed. (TOPL-2; Phelps-Terasaki & Phelps-Gunn,
1992, 2007). The TOPL-2 is a standardized comprehensive measure of pragmatic
language ability designed for use with children and adolescents between the ages of 6 and
18 years. It is primarily used to identify individuals who are evidencing difficulty in
social communication. The TOPL-2 provides one standard score, the Pragmatic
Language Usage Index. It also provides information on the seven core subcomponents of
pragmatic language: physical context, audience, topic, purpose, visual-gestural cues,
abstractions, and pragmatic evaluation.
Dr. Paula Rabidoux, Director of Leadership Education in Neurodevelopmental
and Related Disabilities (LEND) training at Nisonger Center and lead speech pathologist
on the interdisciplinary evaluation team at the Nisonger clinics, was consulted to decide
60
how best to use information from language measures available in patient records. Based
on feedback from Dr. Rabidoux, two composite language scores were used in this study.
The first was a general language composite which included any one of the following: (a)
the Core Composite of the CASL, (b) the Total Language score of the PLS-4, and (c) the
Language Composite of the Bayley-III. All composites are standard scores (M = 100, SD
= 15) and, therefore, are comparable. The second language score used in this study was a
pragmatic language composite and included either the Pragmatic Composite of the CASL,
or the Pragmatic Language Usage Index of the TOPL-2. Based on the consultation with
Dr. Rabidoux, standard composite scores were used as continuous variables for analyses.
Percentage above and below a standard composite of 70 (comparable to the widely
accepted 70 split for IQ scores) was also used for sample characterization.
Cognitive Measures
As with language, a wide variety of cognitive testing measures was used in the
clinics. In this study, IQ information included one of the following: composite scores
from the Wechsler Preschool and Primary Scale of Intelligenc–3rd ed.(WPPSI;
Wechsler, 2002), the Wechsler Intelligence Scales for Children–4th ed.(WISC; Wechsler,
2004), the Stanford-Binet Intelligence Scale, 5th ed.(SB5; Roid, 2003), the Leiter
International Performance Scale–Revised (Leiter–R; Roid & Miller, 1997), and less
commonly, the Early Learning Composite of the Mullen Scales of Early Learning
(Mullen, 1995), and the Cognitive Composite of the BSID-III.
61
Adaptive Behavior Measures
Adaptive behavior information included the General Adaptive Composite of the Adaptive
Behavior Assessment System–2nd ed. (ABAS-II; Harrison & Oakland, 2003), the General
Adaptive Composite of the BSID-III, or the Adaptive Behavior Composite of the Vineland
Adaptive Behavior Scales–2nd ed. (VABS; Sparrow, Cicchetti, & Baila, 2005).
Demographic Form
A demographic form (Appendix C) was used during the semi-structured telephone
interview with families at T2. The form collected information on (a) presence of ID, (b)
current occurrence of seizures, (c) comorbid psychiatric diagnoses received, (d) current
educational placement, (e) intervention history, and (f) currently prescribed psychotropic
medications. A series of 3- or 4-part questions were asked to gather information on
comborbid psychiatric conditions. Psychiatric disorders were endorsed only if families
affirmed that a doctor assigned that particular diagnosis to the child and/or the child was
taking prescribed medication for that diagnosis. The demographic form also collected
information on the informant including relationship with the individual with an ASD,
age, and highest level of education.
Statistical Analyses
Primary Analyses
The Statistical Package for the Social Sciences–Version 19 was used for all
analyses.
62
Aim 1: Sample Characterization
Measures of central tendency were used to characterize the sample in terms of
demographic and assessment variables (including the NCBRF) at T1 and T2. Analysis of
variance (ANOVA) was used to examine differences between NCBRF subscale scores at
T1 based on gender, ASD subtypes, and T1 language composite scores. Tukey’s
Honestly Significant Difference tests were used for post-hoc comparisons. Associations
among the six subscales of the NCBRF, and between NCBRF subscale scores and other
continuous variables (T1 age in months, ADI–R domain scores) were explored using
Pearson correlations. Further, ANOVAs, and chi-square statistics were used to examine
difference in demographic and assessment variables at T1 based on ASD subtype.
Fisher’s exact test was used for analyses where cell counts were less than 5.
Aim 2: Stability of NCBRF Problem Behavior Scores Between T1 and T2
Differences between T1 and T2 NCBRF subscale scores were examined using
paired t-tests. Effect size of change between T1 and T2 was examined using Cohen’s d.
The widely accepted guidelines of Cohen (1988) were used for interpreting the
magnitude of effect sizes where 0.20 to 0.49 is small, 0.50 to 0.79 is medium, and ≥ 0.8 is
large. Associations between NCBRF subscale scores at T1 and T2 were further explored
with Pearson correlations.
To characterize the pattern of change for each NCBRF subscale, participants
were classified into three categories based on the magnitude of individual change relative
to T1 standard deviation. Participants whose T2 scores were within half the T1 standard
deviation above or below their T1 scores were classified as showing “no change.”
63
Participants who changed more than half a standard deviation above or below their T1
scores were classified as having worsened or improved, respectively. A similar approach
has been used by Shattuck et al. (2007) to assess change in repeated measures over time.
As outlined by Shattuck et al., the half standard deviation unit of change has long been
considered a guideline for what represents a “medium” effect size in behavioral research
(Cohen, 1988; Kline, 2004), and it is also an unit of change found to represent clinically
visible change for a variety of behavioral measures (Norman, Sloan, & Wyrwich, 2003).
Change scores were calculated for each NCBRF subscale by subtracting T1
scores from T2 scores. ANOVAs were used to see if NCBRF change scores differed
based on ASD subtypes.
Aim 3: Assessing Predictor Variables for T2 NCBRF Scores
A series of hierarchical multiple linear regressions were run to assess the
contribution of different variables in predicting NCBRF Problem Behavior subscale
scores at T2. Hierarchical regression builds successive linear regression models, each
adding new predictors of interest to see if they predict the outcome above and beyond the
effect of previously entered variables. In this type of regression analysis, we enter
variables into the regression model in “blocks,” each block representing one step in the
hierarchy. The change in R2 resulting from the inclusion of a new predictor (or block of
predictors) is assessed at each step. This procedure enables us to examine the unique
contribution of a new block of predictors in explaining variance in the outcome.
The following predictors were included in each block of the hierarchical
regressions in the current analyses.
64
(a) Block 1. In the first block, the demographic variables were entered to control for these
when testing the other predictor variables of interest in subsequent blocks. Variables in
Block 1 included T1 age in months, gender, and ethnicity. Ethnicity was dichotomously
coded (1 = Caucasian, 0 = minority). I chose the dichotomous split because individual
minority categories did not have high enough frequencies to be considered separately.
(b) Block 2. Block 2 was conceptualized as including all the information that was
available on the participants at T1. These included T1 NCBRF subscale scores, ADI–R
domain scores, general language composite scores, pragmatic language composite scores,
and ASD subtype. Dummy coding was used for ASD subtype, with “autism” as the
reference category.
(c) Block 3. Block 3 was conceptualized as including all information occurring after T1
(or in other words, information collected at T2). Block 3 included the time lag between
T1 and T2 (in months), and types of interventions that were received between T1 and T2.
Types of interventions included were Speech and Language therapy, Occupational and
Physical therapy (OT/PT), and Applied Behavior Analytic Therapy (ABA). These were
coded dichotomously (1 = received, 0 = not received).
Assumptions of Multiple Linear Regression (MLR)
A series of steps were taken both before and during analysis to ensure that the
assumptions of MLR were met for the hierarchical multiple regressions (HMLR)
conducted.
Linearity. MLR assumes that the relationship between the predictors and the
outcome variable (criterion) is linear. Scatter plots were used to check the linearity
65
assumption and establish that the relation between predictors and the outcome were
indeed linear.
Normally distributed errors (residuals). Technically, normality is necessary only
for inference purposes and underlies the tests of significance of the MLR and individual
regression coefficients. Histograms of the HMLR regression residual values (overlaid
with normal probability curves) were generated to check for normality of residuals.
Further, normal P-P plots of residual values were also generated for a visual check on
normality.
Homogeneity of variance (homoscedasticity). This assumption states that the error
variance is approximately equal for all predicted outcome values. This was visually
inspected by plotting residual values against predicted values and was found to be
acceptable.
Independent errors. This states that the errors associated with one observation are
not correlated with the errors of any other observation. Given the nature of this study
(follow-up), temporal autocorrelation could have been a point of concern. The DurbinWatson statistic, used to test for the presence of correlation among residuals, was
requested for each MLR analysis. The Durbin-Watson statistic lies in the range 0-4. A
value of 2 or nearly 2 indicates that there is no autocorrelation. An acceptable range falls
between 1.50 and 2.50. Values of Durbin Watson were found to be within acceptable
ranges for all analyses, indicating there was no autocorrelation.
Collinearity. If predictors are highly correlated, then they can be redundant with
one another in the regression analysis. In such a case, the new predictor does not add any
66
predictive value over the other and can cause problems in estimating the regression
coefficients. Initially, a matrix of bivariate correlations among all predictors was
generated. Significant correlation coefficients did not exceed 0.67 (moderate correlation).
Additionally, to check for intercorrelations between predictor variables (i.e., problem of
multi-collinearity), collinearity statistics (tolerance and variance inflation factor) were
requested during the MLR analyses. Very small tolerance values indicate that a predictor
is redundant, and values that are less than .10 may merit further investigation. Variance
Inflation Factor (VIF) is 1/Tolerance. While there is no formal VIF value for determining
presence of multicollinearity, values of VIF that exceed 10 are often regarded as
indicating multicollinearity. Tolerance and VIF values were found to be within
acceptable ranges for all analyses.
Regression diagnostics to check for unusual and influential observations.
Regression diagnostics were generated to ensure that the data set did not include errant
data points that can influence the regression coefficient estimates and lead to inaccurate
conclusions about the relationship of predictors to the outcome variable. Errant data
points can be characterized in the three following ways.
(1) Leverage or h (Hoaglin & Welsh, 1978): Leverage concerns values on
predictor variables only and represents the extent to which a case deviates from the rest
of the cases in terms of values on the predictor. A case with an extreme value on a
predictor variable is called a point with high leverage, where leverage is defined as the
potential of the point for changing the position of the regression plane. It is commonly
suggested that cases with h > 2 p`/n should be considered a leverage point (where p
67
stands for the number of predictors and p` = p + 1 to indicate the number of predictors,
including the intercept). Netter et al. (1991) suggested that a quick reference cutoff of 0.2
to 0.5 represents moderate leverage and values greater than 0.5 represent high leverage.
Leverage (h) was assessed for all analyses in this study and found to be within the
acceptable range.
(2) Distance. In MLR, distance is a way of assessing outliers, i.e., cases whose
scores on the outcome variable is extreme or unexpected, given the values on the set of
predictors. All measures of distance are functions of the distance of an observed score
from the regression plane and are based on residuals from the regression analysis.
Studentized deleted residual values were used to assess distance. It is commonly
suggested that absolute values of studentized deleted residual between 2.5 and 3.0 can be
used as cut-offs. Residual values were found to be under or within this range in all
analyses, except for a few cases which had marginally higher values and were flagged.
(3) Influence. Influence, a function of both leverage and distance, is the extent to
which a single data point can change the outcome of the regression. It is noteworthy that
a point may have high leverage or high distance and not affect the position of the
regression plane. A point that is both an outlier and has high leverage potentially can
move the regression plane, but not all points do so. Cook’s D, a global measure of
distance, was used for all analyses. D values substantially larger than 1.0 should be
further investigated (Howell, 2007). D values were under 1.0 in all analyses. Since
influence indicates the extent to which a point actually moves the regression plane, and
since all D values were within acceptable ranges, I concluded that the data set did not
68
include any errant data points that could influence estimates of regression coefficients
and distort the relationship of predictors to the outcome.
Ancillary Analysis: Assessing IQ as a Predictor of T2 NCBRF Scores
As noted before, IQ testing at the clinics was done on a case-by-case basis and
consequently, T1 IQ information was not available for all cases seen at the clinics. Full
scale IQ scores at T1 were available for 83 participants (57.3% of the sample). This
included results from IQ testing done at the clinics and, much less commonly, historical
IQ information archived in the child’s previous education or psychological records.
Verbal and/or performance IQ scores were available for 71 participants (48.9% of the
sample). On some occasions, when IQ information came from previous records, only Full
Scale IQ scores were available which led to a difference in availability of full scale and
verbal/performance IQ scores.
Given that previous studies have indicated that initial IQ is often a useful
predictor of overall outcome in ASDs (Howlin et al., 2000, 2004; Billstedt et al., 2005,
McGovern & Sigman, 2005; Shea & Mesibov, 2005), I ran ancillary analyses to explore
the contribution of IQ in predicting T2 NCBRF problem behavior scores in the current
sample. In doing so, I used two methodological approaches.
(1) Past research has suggested considerable congruence between language and
IQ scores; children with better language skills also tend to have higher IQ scores
(Szatmari et al., 2003). Given universal availability of language scores for this sample, I
explored using language scores as an indicator for how IQ scores would perform in the
regression analyses. The advantage of this approach was that participants were not lost
69
for lack of T1 IQ information, and sample size was maintained at N = 143. In order to
consider language as a proxy for IQ, I computed the correlation between general
language composite standard scores and IQ scores for the 83 participants with full scale
IQ scores on file. Pearson correlations were used to measure the correspondence between
the two.
(2) As an alternative approach, I conducted a series of regression analyses using
the sample subset for whom T1 full scale IQ scores were available. For these analyses,
sample size was reduced to n = 83. I conducted HMLR analyses using the same blocks of
predictors as outlined before. This was to see if the predictive power of individual
variables was any different for this smaller sample of 83 participants. Following this, I
entered full scale IQ scores as a predictor in Block 2 of HMLRs and ran all analyses.
As a side note, adaptive behavior scores at T1 were not included in any analysis
given limited availability (n = 59; 40.7%) in patient charts for the current sample.
Exploratory Analyses
Aim 1: Characterization of the Sample at T2
Measures of central tendency were used to characterize the sample in terms of
information collected at follow-up. This included parent-reported psychiatric
comorbidity, current psychotropic medications, level of language compared to peers,
current educational placement, and interventions received between T1 and T2. Parentreported educational placements were verified by school personnel for 54 participants
(39.7% of 136 participants for whom current education placements were reported).
70
Associations between participant characteristics (age, gender, ASD diagnostic
subtype, T1 language scores) and parent-reported comorbid psychiatric disorders were
explored for any psychiatric disorder and the main DSM diagnostic groups reported using
ANOVAs, chi square statistics, and Fisher’s Exact Tests. Association of subject
characteristics with psychotropic medications used and T2 educational placements were
also examined using similar methods.
Aim 2: Predicting DSM Psychiatric Comorbidities at T2
Logistic regression is the method of choice when the outcome variable is
dichotomous, and predictor variables are a mix of continuous and categorical.
Associations between potential predictor variables and parent-reported comorbid
psychiatric disorders were tested in a series of bivariate logistic regressions using DSM
disorder categories with sufficient numbers of affected children to identify associations.
These were anxiety disorder, ADHD, disruptive behavior disorders (DBDs), and
depressive disorders.
Assumptions of Logistic Regression
Unlike MLR, logistic regression does not assume a linear relationship between the
dependent and independent variables. As noted above, the dependent variable must be a
dichotomy (two categories). The independent variables need not be interval, or normally
distributed, or linearly related, or of equal variance within each group. The categories
(groups) must be mutually exclusive and exhaustive; a given case can only be in one
group and every case must be a member of one of the groups.
71
The regression technique is based on maximum likelihood (ML) estimation. One
ramification of the ML estimation is that logistic regression requires larger samples than
for linear regression, because maximum likelihood coefficients are large sample
estimates.
Recommended minimum observation-to-predictor ratio. In terms of the adequacy
of sample sizes, the literature has not offered specific rules applicable to logistic
regression (Peng et al., 2002). However, authors on multivariate statistics (Lawley &
Maxwell, 1971; Marascuilo & Levin, 1983; Tabachnick & Fidell, 1996, 2001) have
recommended minimum sample sizes of 50 or 100, plus a variable number that is a
function of the number of predictors (ranging from a minimum ratio of 5-to-1 to 50-to-1).
Peduzzi et al. (1995, 1996) have published simulation studies suggesting that logistic
regression models will produce reasonably stable estimates if the sample size allows a
ratio of approximately 10 to 15 observations per predictor.
Choosing Final Model for Predicting DSM Psychiatric Comorbidities at T2
In the current analyses, the sample subset for each comorbid diagnostic category
was small (largest was n = 54 for anxiety disorder, followed by n = 45 for ADHD, n = 26
for DBD, and n = 17 for depressive disorders). The HMLR analyses described in
previous sections included 20 predictors. Retaining all 20 predictors for the logistic
regression analyses would make the observation to predictor ratio much higher than the
suggested ratio of 10 cases per predictor (Peduzzi et al., 1995, 1996). This could lead to
the problem of overfitting, i.e. capitalizing on the idiosyncrasies of the sample at hand.
72
Simply put, the notion of overfitting can be thought of as asking too much from the
available data (Babyak, 2004).
To avoid overfitting the model and improve the observation to predictor ratio, I
followed guidelines given by Babyak (2004). First, in an attempt to retain only those
variables that were of primary importance to the research question, I removed “ethinicty,”
“intervening time,” and the three types of “interventions received” as predictors. Next, I
sought to eliminate related variables and consequently removed the predictor “pragmatic
language score.” I retained “general language score,” as it represented an aggregate of the
individual’s overall language capacities. After these two steps, I was left with the
following predictors: T1 age, gender, ASD subtypes, T1 NCBRF problem behavior
subscale scores, general language composite scores, and scores on ADI–R domains. I ran
a series of binary logistic regressions predicting T2 comorbid psychiatric diagnoses with
this set of predictors. I also ran two additional sets of analyses: (1) excluding the three
ADI–R domains from the above mentioned set of predictors, and (2) using only the six
NCBRF subscales as predictors.
Results from the above analyses and findings from the final model are discussed
under the Results section.
Aim 3: Predicting Educational Placement at T2
Educational placements were dichotomously coded in an attempt to have
educational placement categories achieve sufficient size to detect associations. Placement
in regular education class with no accommodations, and regular education class with
minimum accommodation (e.g., aide services for part of day/tutoring services) were
73
collapsed into one category “Regular” that was coded as 0. The other category
“Restrictive,” coded as 1, represented more restrictive placements with more intensive
accommodations. This included placement in a developmental handicapped (DH) class,
multihandicapped (MH) class, a combination of classroom environments (partial
inclusion), and placement in a specialized academy designed to meet educational needs
of students with ASDs and other DDs (for this sample, such academies included the
Haugland Learning Center, Step by Step Academy, and Helping Hands Center, among
others). A similar approach for categorizing education placements has been used by
Williams White, Scahill, Klin, Koenig, and Volkmar (2007).
I used similar strategies, as described above for logistic regressions predicting T2
psychiatric comorbidity, in an attempt to present the most stable estimates while using the
smallest number of predictors. The final model for predicting educational placements at
T2 are discussed under the Results section.
Missing Data
At T1, ADI–R domain scores were missing for 24 participants. Language
composite scores were missing for 6 participants at T1. At T2, NCBRF data was missing
for 9 participants. Among these, the maximum number of missing algorithm items was 5,
out of 60 algorithm items. Missing algorithm items were spread across all six subscales.
As a default, the SPSS statistical analysis software uses the list wise deletion (LD)
method when missing data are encountered in regression analysis. With LD, all cases that
have even one value missing are deleted from the analyses. Consequently, if using LD in
the current analyses, the total sample size of N = 143 would be further reduced. I
74
explored the method of mean substitution as an alternative to LD. I replaced missing
ADI–R domain scores and language composite scores with the mean scores on that
variable. Missing T2 NCBRF data were replaced with the mean item score on that
subscale. Regression analyses outputs were compared using both LD and mean
substitution method and found to be remarkably similar in terms of the pattern of
statistical significance, or lack thereof, of predictor blocks and individual predictor
variables. Hence, in order to utilize information from the full data set of N = 143, mean
substitution method was use to handle missing data.
75
Chapter 3: Results
From a target pool of 342 eligible participants, the final sample for the present
study included 143 individuals whose families participated in a phone interview and
returned completed NCBRFs at follow-up (T2). This represents a response rate of 41.8%.
Of the remaining families who were not involved in the study (n = 199), 115 were lost
due to unreachable status (including possible geographical relocation) and 84 declined to
participate. Participants were compared with non-participants and no significant
differences were found between the two groups in terms of age and gender of the sample,
ASD subtype, NCBRF problem behavior scores at T1, and language composite scores at
T1 (Table 1). Hence, the individuals who were not involved in this study did not differ in
any obvious way from individuals who were involved. For the final sample, the time lag
between the initial visit (T1) and T2 ranged from 2.2-8.9 years, with a mean lag of 5.2
(SD = 1.91) years.
Over 75% of cases in the original target pool of 342 individuals came from the
ASD Clinic. The final sample (N =143) comprised 122 (84.2%) individuals who were
seen at the ASD Clinic and 21 (14.8%) who were seen at the FDD Clinic. Participant
subgroups from these two clinics were compared to see if they differed on age, gender,
ASD subtype, and T1 NCBRF problem behavior scores. Results (Table 2) indicated that
the sample subgroup seen at the FDD clinic was significantly younger (M = 41.8, SD =
4.10) than the other participants (M = 66.2, SD = 17.76) (F = 32.21, p < .001). This is not
76
surprising because, as described in the Methods section, the FDD clinic typically sees
children younger than 4 years of age, while the ASD Clinics sees children older than 4
years and through adolescence. Also, the sample subgroup from the FDD clinic had a
significantly higher proportion of participants diagnosed with autism (p = .002, Fisher’s
Exact Test), than those from the ASD Clinic. No other significant differences between
the sample subgroups were found.
Informants
The vast majority of informants were mothers (n = 124, 85.6%) of whom three
were step-mothers, followed by fathers (n = 14, 9.7%), and legal guardians including one
aunt and four grandmothers (n = 5, 3.4%). A large proportion had graduated from college
(n = 83, 57.3%). NCBRFs at T1 and T2 were completed by the same rater for the
majority of the sample (n = 99, 74.2% of 134 participants for whom rater information
was available at T1). Details on raters at T1 and T2 appear in Appendix E.
Primary Analyses
Aim 1: Sample Characterization
T1 Sample Characterization
The sample characteristics at T1 are presented in Table 3. The sample members
ranged in age from 3 to 11 years [36 months to 139 months (M = 62.65, SD = 18.61)].
Approximately three-fourths were male (n = 109, 76.2%), and the vast majority were
Caucasians (n = 112, 78.3%). More than half had a diagnosis of autistic disorder (n = 78,
54.5%), followed by Asperger’s disorder (n = 44, 30.8%), and finally PDD–NOS (n = 21,
14.7%).
77
Assessments at T1 included the NCBRF, ADI–R, and language measures. Table 3
shows mean scores on ADI–R domains, mean scores on general language and pragmatic
language composites, and percentage of sample scoring above and below 70 on both
language composites. Turning to the NCBRF, the highest mean score was found on the
Conduct subscale (M = 17.90, SD = 6.87), followed by Hyperactive (M = 15.94, SD =
5.44). The lowest mean score was found on Self-injury/Stereotypic (M = 3.48, SD =
2.07). Mean item scores on each subscale were also examined, as the subscales have
differing number of items. Conduct and Insecure/Anxious have the highest number of
items (16 and 15 items, respectively) and the remaining subscales have item counts
ranging from 5 (Overly Sensitive) to 9 (Hyperactive). The Hyperactive subscale had the
highest mean item score (M = 1.77), followed by Overly Sensitive (M = 1.25), Conduct
(M = 1.12), Self-isolated/Ritualistic (M = 1.02), Insecure/Anxious (M = 0.72), and finally,
Self-injury/Stereotypic with the lowest item mean (M = 0.49). Scores were widely
distributed for most of the subscales with the widest range of 36 points (2–38) for
Conduct, followed by Hyperactive with a range of 25 (2–27). The Self-injury/Stereotypic
subscale had the narrowest range of 9 points (0–9).
I used ANOVAs to examine differences in T1 NCBRF subscale scores based on
gender, ASD subtypes, and T1 language composite scores (split at 70). Results are
presented in Table 4. Gender differences were found on two subscale scores, with girls
scoring significantly higher on Insecure/Anxious (F = 9.18, p = .003), and boys scoring
significantly higher on the Self-injury/Stereotypic subscale (F = 8.94, p = .003). As far as
differences based on ASD subtypes, scores on Conduct and Hyperactive subscales were
78
found to be significantly different for all three subtypes. Individuals with an autism
diagnosis scored the highest and those with a PDD–NOS diagnosis scored the lowest on
both subscales. There were no differences between ASD subtypes for other subscale
scores. Turning to differences based on language skills, those with a higher general
language composite (>70) had significantly lower scores on the Conduct
(F = 9.68, p = .002) and Hyperactive (F = 18.62, p < .001) subscales. Interestingly, those
with a higher general language composite had significantly higher scores on the
Insecure/Anxious subscale (F = 19.39, p < .001). The same findings were noted for
pragmatic language composite.
Pearson correlations of T1 NCBRF subscale scores with each other and between
NCBRF subscale scores and other continuous variables (T1 age in months, ADI-R
domain scores) are presented in Table 5. At T1, scores on Hyperactive and Conduct were
significantly correlated (r = 0.669). Other significant associations included low positive
correlations between Insecure/Anxious and Overly Sensitive (r = 0.166), and between
Self-injury/Stereotypic and Self-isolated/Ritualistic (r = 0.201). T1 age was significantly
negatively correlated with scores on the Conduct (r = -0.302) and Hyperactive
(r = -0.475) subscales, indicating that younger age was associated with higher scores on
these subscales. Scores on both ADI–R Reciprocal Social Interaction and
Communication domains had medium positive correlations with the Conduct and
Hyperactive subscales.
There were quite a few differences in demographic and assessment variables at T1
based on ASD subtypes (Table 6). First, those with a diagnosis of Asperger’s disorder
79
and PDD–NOS were significantly older at T1 than those with a diagnosis of autism (F =
44.71, p < .001). T1 age for those diagnosed with autism ranged from 36–82 months (M
= 51.37, SD = 11.19), Asperger’s disorder ranged from 52–139 months (M = 75.59, SD =
19.56), and PDD-NOS ranged from 49–106 months (M = 73.59, SD = 15.52). Second,
ADI–R scores for those with autism were significantly higher than those with Asperger’s
disorder and PDD–NOS for the Reciprocal Social Interaction (F = 44.64, p < .001) and
Communication domains (F = 47.42, p < .001). The PDD–NOS group had significantly
lower scores than the other two subtypes for the ADI–R RRB domain (F = 11.91, p <
.001). Third, those with a diagnosis of Asperger’s disorder were more likely to have a
general language composite greater than 70 (χ2 = 34.59, p < .001). A similar outcome was
present for pragmatic language composite (χ2 = 23.94, p < .001). The gender and
ethnicity distributions did not differ significantly among the three ASD subtypes.
Time 2 Sample Characterization
Table 7 presents characteristics of the sample at follow-up (T2). Age ranged from
5 to 17 years [65 to 204 months (M = 124.64, SD = 31.98)], with 69.2% (n = 99) falling
between ages 5 to 11 years (categorized as children), and 30.8% (n = 44) falling between
ages 12 to 17 years (categorized as adolescents).
Turning to NCBRF subscale scores at T2, the pattern was similar to T1 with
Conduct showing the highest mean subscale score (M = 16.2, SD = 7.84), followed by
Hyperactive (M= 13.9, SD = 8.30). Mean item scores, however, indicated that the highest
mean item score occurred for the Hyperactive (M = 1.54) subscale, followed by Self-
80
isolated/Ritualistic and Overly Sensitive (both M = 1.46). The Conduct subscale came in
third with a mean item score of 1.01.
Other sample characteristics presented on Table 7 (parent-reported DSM
psychiatric comorbidities, psychotropic medications used, current educational placement,
interventions received) are discussed later in the section on Exploratory Analyses.
Aim 2: NCBRF Stability
The results from paired t-tests indicated that scores on all six NCBRF subscales
changed significantly from T1 to T2 (Table 8). Mean change scores and range of change
scores for each subscale are also presented in Table 8. In terms of mean change scores,
the greatest decrease was found for the Hyperactive subscale, with a mean change of
-1.97 at T2, followed by Conduct with a mean change of -1.71. The greatest increase was
found for Self-isolated/Ritualistic, with a mean increase of 3.54 at T2, followed by
Insecure/Anxious with a mean increase of 2.13. The range of change was also quite wide
for most subscales. Effect sizes for the changes (Cohen’s d) were “small” for the most
part, according to guidelines given by Cohen (1988), and they ranged from d = 0.23 for
Conduct to d = 0.44 for Insecure/Anxious. One exception was change on the Selfisolated/Ritualistic subscale, which was accompanied by a moderate-to- high effect size
of d = 0.71. Pearson correlations provided an additional estimate of association between
T1 and T2 NCBRF scores. All T1 scores had low-to-moderate positive significant
correlations with T2 scores, with correlation coefficients ranging from r = .376 for
Insecure/Anxious to r = 0.650 for Self-injury/Stereotypic. Table 9 presents correlations
between T1 and T2 NCBRF subscale scores as a function of intervening time (split across
81
median lag of 5 years; i.e., scores were examined for 2-4 years follow-up and for 5-8
years follow-up). As shown in Table 9, correlations were stronger between T1 and T2
scores over the shorter time lag (2-4 years) than over the longer time lag (5-8 years)
(except for the Conduct subscale for which the correlation was marginally lower for the
shorter time lag). Correlations continued to be significant even for the longer time lag
(except for scores on the Insecure/Anxious subscale).
The analyses mentioned above provided a broad overview of change from T1 to
T2. To allow a closer examination of change over time, the percentages of sample
members who improved, did not change, or worsened on each NCBRF subscale, relative
to their T1 SD, are presented in Table 10. As described in the Methods section, units of
0.50 SD were used to denote improvement, no change, and worsening (Shattuck et al.,
2007). The percent who improved at T2 ranged from a low of 14.7% for Selfisolated/Ritualistic to a high of 61.5% for Hyperactive. The percent who worsened at T2
ranged from a low of 9.1% for Self-injury/Stereotypic to a high of 66.4% for Selfisolated/Ritualistic. The percent with no change at T2 was the highest for Selfinjury/Stereotypic (67.8%) indicating that scores on this subscale were the most stable
over time. This seems consistent with correlation estimates between T1 and T2 scores
reported in Table 8, with Self-injury/Stereotypic showing the highest correlation
compared to the other five subscales (r = 0.65). The proportion that improved was larger
than the proportion that worsened for three of the six subscales (Conduct, Hyperactive,
Self-injury/Stereotypic). The Self-isolated/Ritualistic subscale stood out, with the highest
percent that worsened and with the lowest percent that improved. These extremes in
82
change between T1 and T2 likely contributed to the highest effect size of change (d =
- 0.71) for this subscale, as reported in Table 8.
Examination of differences in NCBRF change scores based on ASD subtype
yielded some interesting findings as presented in Table 11. Overall, degree of change
differed significantly based on ASD subtypes for three of the six subscales. Interestingly,
these were the very three subscales where mean scores had increased over time
(Insecure/Anxious, Self-isolated/Ritualistic, and Overly Sensitive). Those with Asperger’s
disorder and PDD–NOS deteriorated significantly (i.e., had higher scores) on the
Insecure/Anxious and the Overly Sensitive subscales, relative to those with autism (in
fact, those with autism improved slightly). There was no significant difference in
deterioration on Insecure/Anxious and Overly Sensitive between those with Asperger’s
disorder and PDD–NOS. For Self-isolated/Ritualistic, those with Asperger’s disorder (but
not PDD–NOS) significantly deteriorated at T2 compared to those with autism.
Aim 3: Predicting NCBRF Subscale Scores at T2
The third primary aim was to assess the predictive power of a host of variables in
forecasting parent-rated maladaptive behavior on the NCBRF at follow-up. An overview
of results from hierarchical regression models predicting NCBRF subscale scores at T2
are presented in Table 12. Table 12 presents, for all six NCBRF subscales, the variance
explained (R2) by each block of predictors and change in R2 (with associated statistical
significance values) with introduction of each successive block (unique contribution of
each block in explaining variance in the outcome). A series of tables (Tables 13-18)
provide in detail, for each NCBRF subscale, coefficients and significance values
83
associated with individual predictors in each block. The results for each NCBRF subscale
at T2 are discussed in the following paragraphs. Coefficients are interpreted for the last
block which significantly predicted the outcome. Significant predictors are highlighted
using bold font in Tables 13-18.
Conduct. As shown in Table 12, the linear combination of predictors in Block 2
significantly predicted T2 NCBRF Conduct (F= 5.462, p <.001). The R2 value was
0.418, indicating that 41.8% of the variance in T2 Conduct scores was explained by this
combination of predictors. Block 3 also approached (but did not reach) significance (F =
2.357, p = .058) explaining an additional 4.4% of the variance. B and beta coefficients
and p values for Block 2 presented in Table 13 indicate three significant predictors of T2
Conduct. First, higher T1 age predicted higher T2 Conduct scores, with one month
increase in age predicting .087 increase in T2 Conduct scores. Second, a diagnosis of
Asperger’s disorder or PDD–NOS , as compared to autism, predicted substantially lower
T2 Conduct scores (-7.23 and -8.60 units, respectively). This was close to one standard
deviation difference in T2 Conduct scores (SD = 7.84 for T2 Conduct). Third, higher
Conduct scores at T1 predicted higher Conduct scores at T2 (each 1.0 unit increase in T1
scores predicting a 0.650 increase at T2).
Insecure/Anxious. Initially Block 1, containing demographic variables,
significantly predicted the outcome (F = 7.882, p < .001) (see Table 12). However,
introduction of predictors in Block 2 resulted in a significant R2 change of 0.283 (F =
4.231, p < .001). This indicated that Block 2 accounted for significant additional variance
beyond what was explained by predictors in Block 1. As such, Block 2 accounted for
84
43.5% of the variance in T2 Insecure/Anxious scores. Introduction of Block 3 did not
result in a significant R2 change. Thus, the linear combination of predictors in Block 2
best predicted the outcome. As seen on Table 14, ASD diagnostic subtypes were
significant predictors: a diagnosis of Asperger’s disorder or PDD–NOS, as compared to
autism, predicted substantially higher Insecure/Anxious scores at T2 (+6.906 and +6.265
units, respectively). This was more than one standard deviation difference in T2
Insecure/Anxious scores (SD = 5.17 for T2 Insecure/Anxious). Other significant
predictors included T1 scores on Insecure/Anxious, with higher T1 values predicting
higher T2 values (each 1.0 unit increase at T1 predicted a .480 increase at T2). Finally,
T1 Self-injury/Stereotypic was also a significant predictor, though not as strong as T1
Insecure/Anxious (as suggested by standardized beta coefficients).
Hyperactive. Though Block 1 significantly predicted the outcome (F = 13.226, p
< .001), R2 change after introducing Block 2 continued to be significant (F = 2.361, p =
.006). Block 2 explained 39.9% of the variance in the outcome (see Table 12). The
introduction of Block 3 did not result in significant R2 change. Table 15 show three
individual predictors in Block 2 which significantly predicted the outcome. Higher T1
age predicted lower Hyperactive scores at T2 (1 month increase in T1 age predicting a
.09 decrease in T2 scores). As with other subscales discussed above, higher T1
Hyperactive scores predicted higher T2 scores. Finally, higher general language
composite scores at T1 predicted lower Hyperactive scores at T2.
Self-injury/Stereotypic. The linear combination of predictors in Block 2 accounted
for 50.3% of the variance in T2 Self-injury/Stereotypic scores (F = 8.011, p < .001). The
85
introduction of Block 3 did not result in significant R2 change. Table 16 presents
coefficients and p values for the Self-injury/Stereotypic subscale. Inspection of
standardized beta coefficients indicated that the strongest predictor was T1 Selfinjury/Stereotypic, with higher scores predicting higher scores at T2 (1 unit increase in T1
scores predicting .633 increase at T2). The other significant predictor was T1
Insecure/Anxious (1 unit increase in T1 values predicting .08 increase in T2 values of the
outcome). No other variables were significant predictors of T2
Self-injury/Stereotypic subscale scores.
Self-isolated/Ritualistic. For this subscale, initially Block 1 significantly predicted
the outcome (F = 10.016, p <.001), but as with the other subscales discussed above,
introduction of Block 2 led to significant R2 change (F = 4.845, p < .001). Block 2
accounted for 48.3% of variance in the outcome. Introduction of Block 3 did not lead to
significant change in R2 (see Table 12). Examination of coefficients and p values
presented in Table 17 indicated only one significant predictor of the outcome. This was
T1 Self-isolated/Ritualistic: 1 unit increase in T1 scores predicted a .566 increase in T2
scores.
Overly Sensitive. For this subscale, Block 1 and Block 2 both significantly
predicted the outcome (see Table 12). The introduction of Block 3 continued to result in
significant R2 change (F= 4.301, p = .003). The linear combination of variables in Block
3 accounted for 45% of the variance in the outcome. Table 18 presents coefficients and p
values of individual predictors for this analysis. Examination of Block 3 predictors
indicated that T1 values on Overly Sensitive was a significant predictor of T2 values (1
86
unit increase in T1 values predicted .593 increases in T2 values). The only other
significant predictor was time lag between T1 and T2, with 1 month increase in time lag
predicting an increase of .048 in T2 scores.
Ancilliary Analysis: IQ as a Predictor Variable
In the current sample, T1 general language composite scores were strongly
correlated with T1 full scale IQ scores (r = 0.802, p < .01). Given a correlation of this
magnitude, we can reasonably use general language composite scores to infer the
predictive capacity of IQ in forecasting NCBRF problem behavior scores at T2. As
discussed above, general language composite score emerged as a significant predictor
only for one of the six NCBRF subscales: higher general language composite scores at T1
predicted lower scores on Hyperactive at T2.
To explore the direct contribution of IQ in predicting NCBRF maladaptive
behaviors at T2, I ran a series of hierarchical regression analyses (Appendix F, Tables 2733) using the sample subset for whom full scale IQ scores were available at T1 (n = 83,
57.3%). Full scale IQ was entered as a predictor in Block 2 for all regression analyses.
For all six subscales, Block 2 was the last block that significantly predicted the outcome
(R2 ranging from 0.35 for Hyperactive to 0.46 for Self-injury/Stereotypic). Examination
of regression coefficients indicated that full scale IQ was not among the significant
predictors of T2 NCBRF scores for any of the subscales. Detailed results appear in
Appendix F.
87
Exploratory Analyses
Aim 1: Sample Characterization at T2
The first exploratory aim was to characterize the sample of children and
adolescents, at follow-up, in terms of (a) parent-reported comorbid psychiatric diagnoses,
(b) psychotropic medications used, (c) current educational placement, and (d)
intervention history.
Parent-reported Comorbid Psychiatric Diagnoses at T2
A high rate of psychiatric comorbidities was reported by parents, with 98
participants (68.5%) reported as having at least one comorbid psychiatric disorder. The
highest rate was reported for anxiety disorder (n = 54, 37.8%) including 2 participants
with Obsessive Compulsive Disorder, followed by ADHD (n = 45, 31.5%), disruptive
behavior disorders (DBD) (n = 26, 18.2%), and depressive disorder (n = 17, 11.9%).
Other parent-reported comorbidities reported included Tourette syndrome (n = 2), tic
disorder (n = 1), bipolar disorder (n = 1), and enuresis (n = 1).
Multiple Psychiatric Comorbidities
A total of 36 participants (36.7% of those with any comorbid psychiatric disorder,
and 25.2% of the total sample) were reported as having more than one psychiatric
comorbid condition. Of these, 30 participants had two psychiatric comorbidities, and 6
participants were reported as having three. Of the 54 participants reported as having an
anxiety disorder, 23 had a second comorbid disorder, most frequently ADHD (n = 13),
followed by depressive disorder (n = 7) and DBD (n = 3). Overlap between ADHD and
88
DBD was reported for 5 participants. The overlap between main diagnostic groups is
illustrated using a Venn diagram (Fig.1).
Association of Psychiatric Comorbidities with Participant Characteristics
The associations between participant characteristics and comorbid psychiatric
disorders were tested for “any psychiatric disorder” and for the main diagnostic groups
(anxiety disorder, ADHD, DBD, and depressive disorder). As far as age at T2,
participants with comorbid ADHD were significantly younger (M = 9.6 years, SD = 2.4)
than those without ADHD (M = 10.7, SD =2.7) (F = 4.657, p = 0.033).
Examination of gender differences in the rate of reported comorbid conditions
yielded some interesting findings. First, boys were more likely than girls to have any
psychiatric comorbid condition (60.4% versus 39.6%, respectively, p = .019, Fisher’s
exact test). Further, boys were also more likely than girls to have an anxiety disorder
(64.8% versus 35.2%, respectively, p = .016, Fisher’s exact test). No other associations
were found between parent-reported psychiatric comorbidity and gender.
As far as language capabilities were concerned, those reported to have a comorbid
anxiety disorder had significantly higher general language composite scores (M = 84.98,
SD = 14.66) than those who were not (M = 77.21, SD = 15.73) (F = 8.618, p = .004).
Conversely, those reported to have comorbid ADHD had significantly lower general
language composite scores (M = 73.73, SD = 14.09) than those without ADHD (M =
83.09, SD = 15.65) (F = 11.715, p = .001). No significant differences were found in
comorbid psychiatric diagnoses based on ASD diagnostic subtype, which was surprising
given the differences based on language capabilities mentioned above.
89
Psychotropic Medications
A little more than half of the sample (n = 76, 52.4%) were reported as taking at
least one psychotropic medication at follow-up. The most frequently reported drug
classes were antidepressants (n = 34, 23.5%), stimulants (n = 32; 22.1%), antipsychotics
(n = 25, 17.2%), anxiolytics (n = 19, 13.1%), alpha agonists (n = 12, 8.3%), and mood
stabilizers/anticonvulsants (n = 9; 6.2%). Of those taking mood
stabilizers/anticonvulsants, seizures were reported as being currently present for 3
participants. The most common drugs within each class were as follows:
(a) antidepressants: fluoxetine (n = 13), sertraline (n = 14), paroxetine (n = 5), and
citalopram (n = 2); (b) stimulants: methylphenidate (n = 21), dextroamphetamine (n = 7),
and amphetamine salts (n = 4); (c) antipsychotics: risperidone (n = 13), olanzapine (n =
5), thioridazine (n = 5), and ziprasidone (n = 2); (d) anxiolytics: buspirone (n = 12),
lorazepam (n = 6), and diazepam (n = 1); (e) alpha agonists: clonidine (n = 8), and
guanfacine (n = 4); (f) mood stabilizers/anticonvulsants: valproic acid (n = 4), lithium (n
= 3), and carbamazepine (n = 2).
Participants were also reported as taking different types of complementary and
alternative medications (CAM) (n = 21; 14.5%). The rates of CAM treatments might
have been underestimated, because the Demographic Form asked for psychotropic
medications, and it was not clear whether all informants reported CAM/supplement use
(see Table 7).
90
Association of Psychotropic Medication Use with Participant Characteristics
The associations between participant characteristics and taking “any psychotropic
medication” were examined. Overall, participants taking psychotropic medications had
significantly higher T2 scores on three NCBRF subscales (Table 19). These were
Conduct, Insecure/Anxious, and Hyperactive. Individuals with a diagnosis of autism were
more likely to be taking psychotropic medications as compared to those with Asperger’s
disorder and PDD–NOS (60.6% versus 32.4% and 7%, respectively) (χ2 = 6.67, p =
.036). Association between participant age and psychotropic medication use was first
explored using the two age categories mentioned in T2 sample characterization (5 to 11
years and 12 to 17 years). No significant differences in psychotropic medication use were
found based on this dichotomization. Given that greater age has been consistently found
to be associated with use of psychotropic medication (Langworthy-Lam, 2002; Witwer &
Lecavalier, 2005; Aman et al., 2005), I further explored this relationship in the present
sample. I split T2 age such that the two categories were 5 to 9 years, inclusive and 10 to
17 years, inclusive. Using this dichotomization yielded significant differences in
psychotropic medication use such that those with ages 10 years and above were more
likely to use psychotropic medications than those with ages 9 years and below (59.1%
versus 40.9%, χ2 = 4.37, p = .037). Gender and language capabilities were not
significantly associated with psychotropic medication use at T2.
Current Educational Placement
Parent-reported class placement was available for 136 participants (seven parents
responded “don’t know” to the question on education placement during telephone
91
interview). Parent-reported class placement was verified by school personnel for 54
participants (39.7%). Agreement between parent-report and school personnel report was
calculated using Cohen’s kappa coefficient. A high degree of agreement was found with
k = 0.82.
The largest number of participants were reported to be in a developmentally
handicapped (DH) class (n = 33, 22.8%). DH classes typically have students with IQ
scores of 80 or less and established deficits in adaptive behavior (Brown, Aman, &
Havercamp, 2002; Marshburn & Aman, 1992). The second most commonly reported
placement was regular class with minimum accommodations such as aide and/or tutoring
services (n = 26. 17.9%). An appreciable number of participants were also reported to be
in regular class with no accommodations (n = 23, 15.9%). Other educational placements
reported included academies exclusively for students with DDs including ASDs (n = 21,
14.5%), combined class environments (e.g., in a regular classroom for part of the day)
(n = 15, 10.3%), and multihandicapped (MH) class (n = 9, 6.2%). MH classes are
typically for students having developmental handicaps and at least one other handicap,
such as severe behavior disorder, hearing, orthopedic, or speech impairment (Brown et
al., 2002; Marshburn & Aman, 1992). A total of 9 participants in the current sample were
home-schooled.
Association of Educational Placements and Participant Characteristics
The association between T2 educational placements and T2 NCBRF subscales
scores yielded some significant results (Table 20). Overall, participants in “restrictive”
educational placements (as defined in the Methods section) had significantly higher
92
scores on indices of externalizing maladaptive behaviors (Conduct and Hyperactive
subscales). On the other hand, participants in “regular” educational placements had
significantly higher scores on indices of internalizing maladaptive behaviors
(Insecure/Anxious and Overly Sensitive subscales). These findings corresponded nicely
with findings based on ASD subtypes. Those found in “restrictive” educational
placements at T2 were more likely to have autism, as opposed to Asperger’s disorder or
PDD–NOS (71.2% versus 19.7% and 9.1%, respectively; χ2 = 13.75, p = .001). T2
educational placements did not differ significantly based on gender or T2 age.
Intervention History
As per parent-report, the vast majority of participants (n = 114; 79.7%) received
at least one of the following interventions: speech and language therapy, occupational
and/or physical therapy (OT/PT), and applied behavior analytic therapy (ABA) (see
Table 7). Amount of speech therapy ranged from 20 minutes/week to four hours/week,
with an average of 1.2 hours/week. Amount of OT/PT ranged from 15 minutes/week to
two hours/week, with an average of 53.4 minutes/week. Amount of ABA ranged from
one hour/week to 40 hours/week, with an average of 12.3 hours/week. Other
interventions received included social skills training, sensory integration, and Floortime.
Aim 2: Predicting DSM Psychiatric Comorbidities at T2
The second exploratory aim was to examine the role of variables collected at
initial contact (T1) in predicting DSM psychiatric comorbid conditions at follow-up (T2).
To reduce the risk of overfitting the model, the number of predictor variables was
narrowed as described under the Methods section. Initially, a series of logistic regressions
93
were run using the following predictors: T1 age, gender, ASD subtypes, T1 NCBRF
problem behavior subscale scores, general language composite scores, and scores on
ADI–R domains.
Scores on the three ADI–R domains were non-significant predictors in all
analyses. Subsequently, I ran all analyses without the three ADI–R domain scores and
found resulting outputs from these models to be remarkably similar to the ones including
ADI–R domain scores as predictors. To present the most stable estimates while also using
the smallest number of predictors, the models excluding ADI–R domains were selected
as the final models and interpreted. To reiterate, the final set of predictors included T1
age, gender, ASD subtype, T1 NCBRF subscale scores, and T1 general language
composite scores. I retained non-significant predictors (age, gender, ASD subtype,
general language composite) in the final model because these were variables of primary
interest in these analyses (from a theoretical perspective) and their non-significant status
as a predictor was meaningful in terms of interpretation. Previous research suggests that
these participant characteristics might be differentially associated with psychiatric
symptoms and comorbidity. For example, internalizing maladaptive behaviors related to
anxiety and depressions have been found to be more common in older children
(Lecavalier, 2006; Tonge & Einfeld, 2003). Witwer and Lecavalier (2011) found
absence/presence of conversational language significantly to affect endorsement rates of
DSM disorder categories. Further, ASD subtype has been suggested to be differentially
associated with psychiatric comorbidity (Rosenberg et al., 2011; Gadow et al., 2005).
94
Given interest, the supplementary models including ADI–R domain scores as
predictors (in addition to the set of predictors mentioned above) are presented in
Appendix G (Tables 34-37). To explore other possible models, I also ran a series of
analyses with only the six NCBRF subscales as predictors (thus eliminating all nonsignificant predictors) and found the pattern of significance to remain similar to the
selected final model.
Results from Final Logistic Regression Models
Results from the final models are presented in Tables 21-23 and discussed in the
following paragraphs. Since number of observations for each diagnostic category was
small, the ratio of observation to predictor was not optimum. Consequently, findings
discussed in the following paragraphs are considered tentative.
Anxiety disorder. The final model was statistically significant, indicating that the
set of predictors reliably distinguished between those who had an anxiety disorder at T2
and those who did not (χ2 = 33.12, p = .001). In logistic regression, the statistical
significance of individual regression coefficients is tested using the Wald chi-square
statistic (Table 21). Significant predictors are indicated using bold font. As shown in
Table 21, T1 scores on NCBRF Conduct was a significant predictor of being diagnosed
with anxiety disorder at T2. Examination of the corresponding odds ratio indicated that 1
unit increase in T1 Conduct decreased the odds of an anxiety disorder diagnosis at T2 by
0.920 times (or by 8%). T1 Insecure/Anxious was also a significant predictor of an
anxiety disorder diagnosis at T2. The corresponding odds ratio indicated that 1 unit
increase in T1 Insecure/Anxious increased the odds of an anxiety disorder diagnosis at T2
95
by 1.206 times (or by 20%). No other variables were significant predictors of anxiety
disorder diagnosis at T2.
Next, the Hosmer-Lemeshow (H–L) test, a goodness-of-fit statistic was examined.
This goodness-of-fit statistic assesses the fit of the logistic model against actual outcomes
(i.e., whether a participant is diagnosed with anxiety disorder at T2). Well-fitting models
show non-significance on the H–L goodness-of-fit test. This non-significant result is
desirable because it indicates that outcomes predicted by the model do not significantly
differ from what is observed. For this particular analysis, the H–L test yielded χ2 = 12.034
and was non-significant (p = .150), indicating that the model’s estimates fit the data at an
acceptable level.
ADHD. A similar binary logistic regression analysis was run to predict a
diagnosis of ADHD at T2. A test of the final model was statistically significant (χ2 =
11.46, p < .001). As shown in Table 22, the only significant predictor was T1 scores on
the NCBRF Hyperactive subscale. Corresponding odds ratios indicated that one unit
increase in T1 Hyperactivity increased the odds of an ADHD diagnosis at T2 by 1.672
times or by 67%. To gain a simpler understanding of this seemingly strong association
between T1 Hyperactivity score and T2 ADHD diagnosis, the bivariate relation between
T1 Hyperactivity scores and T2 ADHD diagnosis was explored using a dot plot (Fig. 2).
As seen in Fig 2, the vast majority of individuals with a T2 ADHD diagnosis scored
above the median value of T1 Hyperactivity subscale scores. Conversely, most nondiagnosed subjects had T1 Hyperactivity scores below the median for T1.
96
The H–L goodness of fit of this regression model yielded χ2 = 8.695 and was nonsignificant (p = .369), indicating acceptable fit of the logistic model.
Disruptive Behavior Disorder. A test of the final model against a constant only
model was statistically non-significant (χ2 = 19.16, p = .074). Consequently, significance
values and odds ratios associated with individual predictors were not interpreted
(Appendix H). As a side note, the supplementary model including ADI–R domain scores
as predictors (Appendix G, Table 36) was also non-significant.
Depressive Disorder. A test of the final model was statistically significant (χ2 =
24.11, p = .012). As shown in Table 23, three significant predictors emerged. Of these,
two were T1 scores on the NCBRF Insecure/Anxious and Overly Sensitive subscales.
Corresponding odds ratios indicated that one unit increase on T1 Insecure/Anxious and
T1 Overly Sensitive increased the odds of a depressive disorder diagnosis at T2 by 1.145
and 1.166 times, respectively. The last significant predictor was T1 Hyperactive. The
odds ratio indicated that one unit increase on the T1 Hyperactive subscale decreased the
odds of a depressive disorder diagnosis at T2 by .807 times. The H–L goodness-of-fit test
yielded χ2 = 6.70 and was non-significant (p = .569).
Aim 3: Predicting Education Placement at T2
The third exploratory aim was to examine the role of variables collected at
initial contact in predicting educational placement at follow-up. After a similar
exploratory exercise, as described above for T2 psychiatric comorbidity analyses, the
same set of predictor variables were retained for the final model predicting educational
placements at T2. These predictors were T1 age, gender, ASD subtype, T1 NCBRF
97
subscale scores, and T1 general language composite scores. Findings are presented in
Table 24 and discussed below. The supplementary model including ADI–R domain
scores as predictors (in addition to the predictors mentioned above) is presented in
Appendix G, Table 38.
A test of the final model was statistically significant (χ2 = 45.28, p < .001). As
shown in Table 24, T1 scores on NCBRF Conduct was a significant predictor, such that
one unit increase on the T1 Conduct subscale increased the odds of a “Restrictive”
educational placement by 1.148 times (or by 14%). General language composite score
was also a significant predictor: higher general language composite scores at T1
decreased the odds of “Restrictive” educational placement at T2 by .941 times (or by
6%). The H–L goodness of fit yielded χ2 = 5.875 and was non-significant (p = .661),
indicating acceptable fit of the logistic model.
98
Chapter 4: Discussion
The primary aim of this dissertation was to examine changes in, and predictors of,
maladaptive behaviors in ASDs. Exploratory aims included examination of sample
characteristics at follow-up which included (a) parent-reported DSM psychiatric
comorbidities, (b) psychotropic medications used, (c) current educational placements, and
(d) interventions received. Of the target pool of 342 potential participants, 143 (41.8%)
participated in the study. The final sample did not differ significantly from nonparticipants in terms of age, gender, ASD subtype, and T1 scores on NCBRF problem
behavior subscales. So the final sample can be regarded as fairly representative of cases
typically seen at Nisonger Center clinics.
Primary Aims
Sample Characterization at Time 1
The mean age of the sample was 5.2 years at initial assessment. For the large
majority of children this was probably the time when an ASD diagnosis was first
assigned since cases were referred to Nisonger Center clinics for possible diagnosis on
the autism spectrum. Initial diagnosis at around 5 years of age is consistent with findings
from a 2011 survey on “Pathways to Diagnosis and Services” among children aged 6-17
years with ASDs, conducted by the National Center of Health Statistics (NCHS) of the
Centers for Disease Control and Prevention (CDC)
99
(http://www.cdc.gov/nchs/data/databriefs/db97.htm). The survey found that one-half of
school-aged children with ASD were 5 years and over when they were first identified as
having an ASD. In the current sample, the mean age was significantly higher for those
receiving a diagnosis of either Aperger’s disorder (6.3 years) or PDD-NOS (6.1 years)
than for those receiving a diagnosis of autism (4.3 years). This echoes similar findings in
the literature which suggest that on average, children with Asperger’s disorder receive a
diagnosis and receive help much later in their development than do children with autism
(Eisenmajer et al., 1996; Green, Gilchrist, Burton, & Cox, 2000; Howlin & Asgharian,
1999). Consistent with the male preponderance typically seen in ASDs, the final sample
comprised a higher number of boys with a male-to-female ratio of approximately 3:1
(76.2% vs. 23.8%) which deviated slightly from the generally accepted male-to-female
ratio of 4:1 in the ASDs (Fombonne, 2003).
Maladaptive Behaviors at T1: Association with Subject Characteristics
The wide distribution of NCBRF scores at T1 indicated a varied prevalence of
maladaptive behaviors in the sample. The highest mean scores were noted for Conduct
and Hyperactive (which also had the highest mean item score); the lowest mean scores, as
well as lowest mean item scores, were found on the Self-injurious/Stereotypic subscale.
Further, maladaptive behaviors at T1 were found to vary considerably based on
age, gender, ASD subtype, and language abilities of participants. Overall, young age,
diagnosis of autism (as opposed to the other two ASD subtypes), and lower language
ability (scores lower than 70) were significantly associated with higher scores on the
NCBRF Conduct and Hyperactive subscales. It is interesting to note some parallels
100
between these results and findings from the ID population where lower IQ (which is more
likely in youngsters with autism as opposed to those with Asperger’s disorder/PDD–
NOS), and deficits in communication have been associated with higher levels of
disruptive behavior (McClintock et al., 2003). Further, in the typically developing
population, disruptive behavior has been associated with younger age (National Institute
of Child Health and Human Development Early Child Care Research Network; NICHD,
2004). On the other hand, female gender and higher language ability (scores greater than
70) were significantly associated with higher scores on the NCBRF Insecure/Anxious
subscale. A similar trend is observed in the general population, where maladaptive
behaviors associated with affective disorders are more common in girls than boys by
about 30% (Costello et al., 2003; Rutter et al., 2003). An association between higher
language ability and higher maladaptive behaviors associated with anxiety has also been
previously suggested in the ASD literature (Howlin, 2007; Sukhodolsky et al., 2008).
Change in Maladaptive Behaviors at Follow-up
Examination of change in NCBRF maladaptive behavior scores at follow-up
yielded a number of key findings.
(1) First of all, there were significant differences between T1 and T2 NCBRF
subscale scores for all six subscales indicating that maladaptive behaviors change
considerably as children with ASDs grow older.
(2) Scores on three of the six NCBRF subscales (Conduct, Hyperactivity, and
Self-injury/Stereotypic) showed significant decline (i.e., improvement) over time.
However, effect sizes of change were quite small (in the 0.23–0.34 range), as per the
101
widely accepted guidelines by Cohen (1988). Correlations between T1 and T2 scores on
these three subscales were moderate in magnitude (albeit statistically significant). To
allow direct comparison in change among subscales, percentages of participants who
improved, worsened, or showed no change were examined based on whether their T2
scores changed more than ±½ SD of T1 scores or remained within ½ SD of T1 scores.
Based on this categorization, the greatest proportion of participants improved on
Hyperactivity (61.5%), followed by Conduct (43.3%) Interestingly, research on behavior
problems in typically developing children suggest a similar trend in that disruptive
behavior problems (such as tantrums) improve with age. Some of these changes with age
have been thought to be maturational and some have been related to improving language
skills (NICHD, 2004; Stevenson & Richman, 1978).
(3) Scores on the remaining three NCBRF subscales (Insecure/Anxious, Selfisolated/Ritualistic, and Overly Sensitive) increased at follow-up, and those with
Asperger’s disorder and PDD–NOS (as compared to autism) deteriorated (had
significantly higher scores). Effect sizes of change (in the 0.34–0.71 range) were
comparatively higher in magnitude than change in the subscales discussed above.
Correlations between T1 and T2 scores were significant, and low-to-moderate in
magnitude. The highest proportion of subjects worsened in the Self-isolated/Ritualistic
subscale (66.4%), followed by worsening on the Insecure/Anxious subscale (58%). The
increase in scores on the Self-isolated/Ritualistic subscale (which contains items such as
“isolated self from others” and “withdrawn, uninvolved with others”) overlaps with
102
findings reported by Anderson et al. (2011). These authors observed increases in socially
withdrawn behavior over time for a substantial proportion of youth with ASDs.
As mentioned before, a significantly greater proportion of those with Asperger’s
and PDD–NOS deteriorated on Insecure/Anxious, Self-isolated/Ritualistic, and Overly
Sensitive, at follow-up. Such a trend is being reported with increasing consistency in the
ASD literature. For example, internalizing behaviors (particularly those related to anxiety
and depression) have been found to be more common in older children and adolescents
who have higher IQ (Estes et al., 2007) and/or fewer core symptoms of ASDs (Gadow et
al., 2005; Kanai et al., 2004). Both descriptions are likely applicable for those diagnosed
with Asperger’s disorder and PDD–NOS, as compared to autism. Further, some research
has shown that with increasing age young people with Asperger’s disorder perceive
themselves as more dissimilar to their peers and this, in turn, might be associated with
higher levels of anxiety and social withdrawal (Butzer & Konstantareas, 2003; Hedley &
Young, 2006). Anderson et al. (2011) noted that youth with Asperger’s disorder and
PDD–NOS may be more likely to find themselves among typically developing peers in
and outside of the classroom setting. They suggested that inability to keep up with the
social demands of these situations might contribute to higher anxiety and social
withdrawal in these youth compared to those with autism who may experience fewer
social demands.
(4) Maladaptive behaviors associated with the Self-injury/Stereotypic subscale
were the most stable over time with (a) the highest correlation between T1 and T2 scores
(r = 0.65) among all NCBRF subscales and (b) the highest proportion of participants
103
(67.8%) who showed no change at follow-up (i.e., their T2 scores remained within ½ SD
of their T1 scores). This subscale had the lowest mean item score at both time points.
This suggests that while behaviors such as “rocking body or head,” “hitting or slapping
own body parts,” “gouging self,” or other similar behaviors related to “physically hurting
self on purpose” had low prevalence in this sample, those who did exhibit such behaviors
were likely to continue displaying them over time.
(5) Findings associated with the Self-isolated/Ritualistic subscale were also
noteworthy. The highest proportion of participants (66.4%) deteriorated on these types of
maladaptive behaviors, and the smallest proportion improved (14.7%). In other words,
these types of maladaptive behaviors showed the most extreme change over time. Not
surprisingly, change in this subscale was accompanied with the highest effect size (d =
- 0.71) among all NCBRF subscales. It is worth mentioning that of the 8 items on this
subscale, the majority are concerned with self-isolation-type behaviors and only two have
a ritualistic flavor [“has rituals such as head rolling or floor pacing,” and “odd repetitive
behaviors (e.g., stares, grimaces, rigid postures”)].
Overall, keeping with the trend of declining maladaptive behaviors over time that is
reported in previous research (Anderson et al., 2011; Lounds Taylor & Seltzer, 2010;
Shattuck et al., 2007), decline was observed in the Conduct, Hyperactive, and Selfinjury/Stereotypic subscales in the current sample. Further, the hypothesis that decline
might not be universal for all types of maladaptive behavior was supported as shown by
increase in Insecure/Anxious, Self-isolated/Ritualistic, and Overly Sensitive. Also, the
104
prediction that rates of maladaptive behaviors will be moderately correlated between
initial assessment and follow-up was supported for all six NCBRF subscales.
Predicting Maladaptive Behaviors at Follow-up
Our study yielded several important findings regarding predictors of NCBRF
subscale scores at follow-up. Results from a series of hierarchical regression analyses
indicated that the tested models accounted for 39% to 50% of the variance in T2
maladaptive scores. The ability to predict up to 50% of the variance in maladaptive
behaviors over time is indeed significant. This high percent of variance explained by the
models can largely be attributed to the inclusion of T1 scores on the same respective
NCBRF subscales as predictors. T1 scores on the respective NCBRF subscales were the
single most consistent predictors of all six NCBRF subscale scores at follow-up. This
indicates that a child’s individual levels of maladaptive behavior are the best predictor of
maladaptive behavior levels over time. A similar trend has been observed in previous
research (Murphy et al., 2005; Shattuck et al., 2007).
In an effort to systematize the findings from the hierarchical regression analyses,
T2 NCBRF outcomes and their associated significant predictors are summarized in Table
25 and discussed below.
Conduct. Three significant predictors emerged for T2 scores on the Conduct
subscale. (a) Higher Conduct scores at T1 predicted higher Conduct scores at T2. (b)
Higher T1 age predicted higher T2 Conduct scores. Though this might seem counterintuitive at first glance, it might not actually be so for two reasons. First, such a finding is
not unique to this study; studies in children with ID have identified older age as a
105
predictor of disruptive behaviors such as aggression (Borthwick-Duffy, 1994; Kiernan &
Kiernan, 1994). Second, closer examination of items on the NCBRF Conduct subscale
might shed some light on why increasing age might predict higher Conduct scores at
follow-up. The Conduct subscale contains items which require language ability on the
part of the child in order to be endorsed (e.g., “argues with parents, teachers, peers, and
other children,” “threatening people,” “talking back”). It is possible that growing
language capabilities with increasing age contributed to greater levels of this type of
Conduct problems for older children. (c) A diagnosis of Asperger’s disorder or PDD–
NOS, as compared to autism, predicted substantially lower (close to one SD) T2 Conduct
scores.
Insecure/Anxious. Three variables significantly predicted T2 scores on the
Insecure/Anxious subscale. (a) Higher T1 scores on Insecure/Anxious predicted higher
T2 scores. (b) Higher T1 scores on Self-injury/Stereotypic predicted higher T2 scores. (c)
A diagnosis of Asperger’s disorder or PDD–NOS, as compared to autism, predicted
substantially higher (more than one SD) Insecure/Anxious scores at T2. This finding is
consistent with the suggestion in the ASD literature that internalizing maladaptive
behaviors, such as anxiety, are found to be more common in older children and
adolescents who have higher IQ (Estes et al., 2007) and/or fewer core symptoms of ASDs
(Gadow et al., 2005; Kanai et al., 2004).
Hyperactive. T2 Hyperactive scores also had three significant predictors.
(a) As with other subscales discussed above, higher T1 Hyperactive scores predicted
higher T2 scores. (b) Higher T1 age predicted lower Hyperactive scores at T2. This is
106
consistent with findings from the typically developing population discussed previously.
Increased maturity with increasing age might be one of the reasons for decreasing
hyperactivity over time. (c) Higher general language composite scores at T1 predicted
lower Hyperactive scores at T2. A negative association between language skills and
externalizing maladaptive behaviors has been reported in previous studies in the ASD
literature (Estes, Dawson, Sterling, & Munson, 2007; Hartley, Sikora, & McCoy, 2008).
Self-injury/Stereotypic. T2 scores on the Self-injury/Stereotypic subscale had two
significant predictors. (a) Higher T1 scores on Self-injury/Stereotypic predicted higher T2
scores. (b) Higher scores on T1 Insecure/Anxious also predicted higher T2 scores.
Self-isolated/Ritualistic. Only one significant predictor emerged for this subscale.
Higher T1 scores on the same subscale at T1 predicted higher scores at T2.
Overly Sensitive. T2 scores on this subscale had two significant predictors.
(a) Higher scores on Overly Sensitive at T1 predicted higher scores at T2. (b) Increasing
time lag between T1 and T2 predicted higher T2 Overly Sensitive scores. This was the
first occurrence of “time lag” significantly predicting an NCBRF outcome.
Overall, the hypothesis that levels of maladaptive behavior at initial contact will
be significant predictors of maladaptive behaviors at follow-up was supported for all six
NCBRF subscales. The prediction that greater language abilities at T1 would predict
greater declines in T2 maladaptive behaviors was supported for only one of the six
NCBRF subscales (Hyperactivity). We made a related prediction that better language
ability at T1 would predict higher levels of maladaptive behaviors at T2 which involved
spoken language (such as arguments with adults and peers, talking too much or too loud,
107
threatening people, talking back). This hypothesis was not supported. However,
increasing age at T1 was found to be a predictor of higher T2 scores on the Conduct
subscale which incidentally contains all but one of the items mentioned above. This
might suggest some indirect support for the last hypothesis, as better language ability is
likely associated with increasing age. However, this is highly speculative and certainly
requires independent support to be taken seriously. Finally, the hypothesis that type of
intervention received would predict T2 maladaptive behavior scores was not supported in
these findings. Previous research has also found a lack of relation between intensity or
amount of intervention and changes in maladaptive behavior (Anderson et al., 2011) and
overall developmental trajectories (Darrou et al., 2010). Lack of association between
maladaptive behavior change and both type and amount of intervention is a little
surprising. Perhaps this points towards the methodological complexity of assessing
“interventions received” as a predictor variable. Type, duration, and quality of
intervention vary considerably over the years for most children with ASDs, and it is
problematic to capture this complex conglomeration of information in a way that can be
used in statistical analyses to denote the overall contribution of interventions received.
Ancillary Analyses Examining IQ as a Predictor
In the scant literature available on the changes in maladaptive behaviors over
time, reports on the role of IQ in predicting changes have been inconsistent. Shattuck et
al. (2007) found presence of ID to be a predictor of change (less improvement over time
for those with ID). However, other studies failed to report significant associations
between IQ and change in maladaptive behaviors (Ballaban-Gil et al., 1996; Murphy et
108
al., 2005). Further, it is not known whether language and IQ are independent predictors of
outcome; children who are more verbal also tend to have higher IQs so the prognostic
ability of one variable may be largely accounted for by variation in the other (Szatmari et
al., 2003). If that is the case, predictive performance of IQ may be strongly related to the
performance of language as a predictor.
In the current study, I also conducted ancillary regression analyses with the
sample subset for whom IQ information was available (n = 83; 57.3% of the sample). IQ
scores at T1 were not found to be significant predictors of maladaptive behavior scores at
follow-up. Even in this smaller sample, T1 scores on a particular subscale remained the
most consistent predictor of scores at T2, providing reassurance that the regression
models were performing consistently.
Exploratory Aims
Sample Characterization at Follow-up: Psychiatric Comorbidity
One of the most notable findings from sample characterization at follow-up was a
high rate of comorbid psychiatric disorders reported by parents. This was particularly
striking given that ours was a community sample (unselected for psychiatric disorders).
Overall, findings did confirm previous research suggesting that psychiatric comorbidity is
increased in children and adolescents with ASD compared with general population
estimates, including risk of multiple comorbidities (Gjevik et al., 2011; Leyfer et al.,
2006; Simonoff et al., 2008). Prevalence of at least one psychiatric comorbidity in the
general child and adolescent populations is 7% to 13.3% (Costello et al., 2003; Heiervang
et al., 2007), and around 38% in children with ID (Dekker & Koot, 2003). Among the
109
current sample with ASDs, the figure was 68.5% (n = 98). This is comparable to other
previously reported rates of psychiatric comorbodity in ASDs [Leyfer et al. (70.8%),
Simonoff et al. (72%), and Gjevik et al. (72%)]. A total of 36.7% (n = 36) was reported to
have more than one psychiatric comorbid condition.
Association to Participant Characteristics
In the general population, affective disorders are more common among girls,
whereas ADHD is more common in boys (Rutter et al., 2003). Historically, studies of
psychiatric comorbidity in ASDs have failed to detect these gender differences (Brereton
et al., 2006; Rosenberg et al., 2011; Simonoff et al., 2008). One explanation might be that
presence of an ASD surpasses other risk factors such as the gender differences seen in
typically developing children. In the current sample, boys were more likely than girls to
be reported to have any psychiatric comorbid condition. Further, completely opposite to
the trend in the typically developing population, boys were also more likely than girls to
have an anxiety disorder. It is worth mentioning that the small proportion of girls (23%)
could have caused lack of power in the current sample. As suggested by Simonoff et al.,
similar studies in the future might need to oversample girls, but that entails its own set of
challenges.
Other significant associations between psychiatric comorbidities and participant
characteristics included language abilities. Those reported to have comorbid ADHD at
T2 had significantly lower language ability at T1. On the other hand, those reported to
have a comorbid anxiety disorder at T2 had significantly higher language ability at T1.
This link between better language ability and higher maladaptive behaviors associated
110
with anxiety was a consistent finding from different analyses in this study and has also
been reported in past research (Howlin, 2007; Sukhodolsky et al., 2008). One explanation
might be that anxiety is triggered by greater cognizance of one’s disability (as higher
language ability is often found in individuals with higher IQ) and greater social
expectations from adults and peers. In the current sample, given that those with a
comorbid anxiety disorder had significantly higher language scores, I also expected to see
a diagnosis of Asperger’s disorder to be associated with a comorbid anxiety disorder.
However this was not the case. Though the sample subset with Asperger’s disorder (and
PDD–NOS) had a higher percentage of individuals with a comorbid anxiety disorder
(than in the subset diagnosed with autism), these differences were not statistically
significant.
Point Prevalence of Psychiatric Comorbid Conditions
In the current sample, anxiety disorder was the most frequently reported
diagnostic group (n = 54; 37.8%). As outlined by Leyfer et al., various types of anxiety
are believed to be so common in autism that symptoms of anxiety disorders have been
thought by some clinicians and investigators to be aspects of autism, rather than
comorbid features (also see Grondhuis & Aman, 2012). However, as Leyfer et al. pointed
out, impairing anxiety is neither a defining characteristic nor a universal occurrence in
autism. Reported rates of at least one anxiety disorder in individuals with ASDs have
varied from 17% to 84% (Ando & Yoshimura, 1979; Muris et al., 1998; Rumsey,
Rapoport, & Sceery, 1985). More recent systematic studies have consistently reported
high rates of anxiety disorder. The 37.8% rate of anxiety disorders in our study was
111
slightly lower than rates reported by Gjevik et al. (42%), Simonoff et al (41.9%), and
Leyfer et al. (44%), and higher than the rate (26%) reported by Rosenberg et al. (2011).
The second-most commonly reported psychiatric comorbidity in the current
sample was ADHD (n = 45, 31.5%). This is consistent with rates reported by Leyfer et al.
(2006) (28.1%), Simonoff et al. (2008) (31%) and Gjevik et al. (2011) (also 31%). The
high rate of comorbid ADHD reported in ASDs is particularly interesting given that
DSM–IV–TR prohibits the diagnosis of ADHD if a PDD exists (APA, 2000). Clearly,
clinicians are identifying and diagnosing ADHD despite guidance not to do so in the
diagnostic manual. Results from the current and previous studies suggest that a
significant minority of individuals with ASDs present with ADHD symptoms that cannot
all be attributed to the ASD itself. This strengthens the argument for making a separate
diagnosis of both conditions when appropriate, as proposed by many in the DD field
(Frazier et al., 2001; Gillberg & Billstedt, 2000; Goldstein & Schwebach, 2004). It is
noteworthy that proposed changes in ADHD diagnostic criteria in the DSM-5 include
removing ASD from the exclusion criteria
(http://www.dsm5.org/ProposedRevision/Pages/proposedrevision.aspx?rid=383#).
We found a comparatively lower rate of disruptive behavior disorders (n = 26,
18.2%), than other conditions. This lower rate in our sample, and other studies (Gjevik et
al., 2011; Leyfer et al., 2006) could indicate that DSM criteria of the OD/CDD diagnosis
do not tap into the type of behavioral problems in many children with ASDs. Features of
ODD/CD, such as lying and blaming others, might require a fairly sophisticated level of
cognitive maturity and ability of think abstractly that may be uncommon for many
112
individuals with ASDs (Leyfer et al., 2006). On that note, oppositional behaviors in
ASDs (refusal to follow directions, being uncooperative) might very well be related to
intrinsic cognitive deficits in autism, such as lack of appreciation of mental states of
others (i.e., deficits in theory of mind), and problems with executive function, such as
rigid, inflexible thinking and behavior (Baron-Cohen, 1985, 19913; Ozonoff, Pennignton,
& Rogers, 1991).
Overall, the rates of comorbid psychiatric disorders at follow-up seem quite
striking, especially given the relatively young age of the sample (mean age of 10 years).
Advances in autism genetics, neuroimaging, and epidemiology may help determine why
individuals with ASDs have features of so many other psychiatric conditions. Given the
phenotypical overlap between comorbid disorders and ASD, it seems that reporting the
true prevalence of psychiatric conditions and symptoms in ASDs is complicated by
differences in interpreting and defining psychiatric symptoms. In any case, psychiatric
comorbidity is a particularly salient feature that is likely to account for differences in
overall developmental outcomes of children and adolescents on the autism spectrum.
Further research is needed to examine the extent and manner in which comorbid
conditions or symptomatology influence the course and outcome in ASDs.
Other Sample Characteristics at Follow-up
Psychotropic Medication Use
The treatment of ASDs often includes psychotropic medications to address
aggression, self-injurious behavior, hyperactivity, anxiety, sleep problems, repetitive
behaviors, and other symptoms common in children with these disorders (Bryson et al.,
113
2003; Filipek et al., 2000; Myers et al., 2007; Witwer & Lecavalier, 2005). Previous
studies estimated that 30% to 60% of children with ASDs use at least one psychotropic
medication, predominantly antidepressants, stimulants, or antipsychotics (Aman et al.,
2005; Green et al., 2006; Langworthy-Lam et al., 2002; Mandell et al., 2008; Martin et
al., 1999; Oswald & Sonenklar, 2007; Witwer & Lecavalier 2005). Consistent with these
findings in the literature, we found a high rate of psychotropic medication use (as
reported by parents) in the current sample. A little more than half of the sample (n = 76,
52.4%) were reported as using any one psychotropic medication, similar to rates reported
by Aman et al. (55.4%), and most recently, in the CDC NCHS survey (2011) (56%). The
rate found in this sample was slightly higher than rates reported by Langworthy-Lam et
al. (45.7%) and Witwer and Lecavalier (46.7%). This higher rate corroborates findings
from Aman et al. indicating that use of psychotropic medications in ASDs is increasing
over the years. In contrast, a study by Rosenberg et al. (2010) using data from the
Interactive Autism Network (IAN), a national online registry of children with ASD
younger than 18 years of age, reported a much lower rate (35.3%) of psychotropic
medication use. One reason for this might be the relatively low rate of comorbid
psychiatric conditions (39%) reported in the IAN sample. In the current sample, the
highest rate was reported for antidepressant use (n = 34, 23.5%), followed by stimulant
use (n = 32, 22.1%) (see Table 7).
Educational Placement at Follow-up
Through the last decade, there has been a growing emphasis on the inclusive
model of education with wide-scale adoption of the Individuals with Disabilities
114
Education Act (IDEA, 1997) which (a) promotes placement of children with DDs in
regular education curriculum to the maximum extent possible and, (b) mandates
appropriate educational services for all school-aged children with disabilities. In surveys
assessing individuals with DDs in the special education population in central and
northern Ohio, about 60% of participants were found to be placed in DH classes and
around 20% in MH classes (Marshburn & Aman, 1992; Brown et al. 2002). Unlike these
studies, the current sample was not restricted to youth in special education programs. The
largest number of participants were reported to be in a DH class (n = 33, 22.7%). This
was followed by regular class with minimum accommodations (e.g., aide services) (n =
26, 17.9%) and regular classes with no accommodations (n = 23, 15.9%). A small
proportion was reported to be in a MH class (n = 9; 6.2%).
Interventions Received
As far as interventions are concerned, the vast majority of the current sample
received at least one of the following: speech and language therapy, occupation
therapy/physical therapy, or applied behavioral analytic therapy (n = 114, 79.7%). This
indicates some degree of success of the adoption of IDEA. An even higher rate of service
use was reported in the 2011 CDC NCHS survey
(http://www.cdc.gov/nchs/data/databriefs/db97.htm), but this is likely because the survey
covered a much broader range of eight possible services. The survey reported that 91.1%
of youngsters in the 6-11 years age group and 84.2% of those in the 12-17 years age
group received at least one of eight services (social skills training, speech or language
therapy, occupational therapy, physical therapy, behavioral intervention or modification,
115
cognitive based therapy, sensory integration therapy, and complementary and alternative
medicine). In the current sample, speech and language therapy was reported to be the
most commonly received service, which is consistent with findings from the CDC NCHS
survey.
Prediction of Comorbid Psychiatric Disorders at Follow-up
Results from a series of logistic regression analyses indicated that only T1
NCBRF subscale scores were significant predictors of psychiatric comorbidity at followup. No other variables were predictive of T2 psychiatric comorbidity in our analyses. The
following findings, although interesting and potentially very useful, are considered
tentative due to the small subset of observations in each psychiatric diagnostic category.
The correspondence between indices of maladaptive behavior at T1 and psychiatric
comorbidity at T2 was relatively clear cut.
Anxiety Disorder. Each unit increase on T1 Insecure/Anxious increased the odds
of an anxiety disorder diagnosis at T2 by 1.206 times. Additionally, each unit increase on
T1 Conduct decreased the odds of an anxiety disorder diagnosis at T2 by 0.920 times.
ADHD. As far as a T2 diagnosis of ADHD was concerned, each unit increase on
T1 Hyperactive increased the odds of an ADHD diagnosis by 1.672 times.
Depressive Disorder. Finally, each unit increase on the Insecure/Anxious and
Overly Sensitive subscales at T1 increased the odds of a Depressive Disorder diagnosis at
T2 by 1.145 and 1.166 times, respectively. In contrast, each unit increase on the T1
Hyperactivity decreased the odds of a Depressive Disorder Diagnosis by 0.807 times.
116
The finding that T1 levels of certain maladaptive behaviors predict certain
comorbid psychiatric conditions later in childhood and adolescence is of considerable
importance. For example, such information may help clinicians and families to recognize
maladaptive behaviors as symptoms of a potentially treatable and perhaps undiagnosed
comorbidity rather than attributing those behaviors to the ASD alone. It may also enable
the field to contemplate the use of preventative measures early in life and to plan future
mental health needs.
Predicting Education Placements at Follow-up
Results from logistic regression analyses indicated two significant predictors of
educational placement at T2.
(1) T1 Maladaptive Behavior. We found that higher T1 scores on NCBRF
Conduct increased the odds of placement in “restrictive” classes (which included a DH or
MH placement) by 1.148 times. Disruptive behavior has been found to be directly related
to a loss of productive learning time, and consequently, this tends to reduce chances of
being placed in an integrated class setting (Einfeld & Tonge, 1992; King, Ollendick, &
Tonge, 1995). This makes intuitive sense, as higher frequency and intensity of
externalizing maladaptive behaviors likely deter placement in more normative “regular”
classes.
(2) T1 language ability. Higher language ability at T1 decreased the odds of a
“restrictive” placement by 0.941 times. This finding also makes intuitive sense and is
consistent with results of previous studies (Eaves & Ho, 1997; Oti et al., 2004, Williams
White et al., 2007).
117
Summary and Significance of Findings
ASDs are usually lifelong disorders (Seltzer et al., 2004). Most children with
ASDs have persisting difficulties in multiple areas including language and socialization
(Ballaban-Gil et al., 1996). In addition to impairments caused by core ASD symptoms,
presence of maladaptive behaviors further complicates outcomes for youth with ASDs
and their families. Given this big picture, findings from this study are significant because
they show that it is possible to predict up to 50% of the variance in change in maladaptive
behavior using information collected at initial assessment. These findings have important
implications for parents, clinicians, and service providers.
First, findings about the natural course of maladaptive behaviors in ASDs can
help service providers to address these behaviors in treatment plans. The clinical
implication of these findings is that clinicians can reassure parents that certain
maladaptive behaviors (e.g., conduct and hyperactivity problems) likely decline with age,
especially in those with higher language abilities. However, clinicians can also educate
parents that higher language ability has been associated with increases in social
withdrawal and anxiety with increasing age.
In our sample of children and adolescents who were unselected for psychiatric
disorders, high rates of comorbid psychiatric conditions were reported by parents. Timely
recognition of psychiatric comorbidity is important because this allows for more specific
treatment plans to be adopted. Treatments for psychiatric disorders are available but a
child likely will not benefit from these unless maladaptive behaviors are appropriately
classified as a comorbid psychiatric disorder rather than being treated as random, isolated
118
behaviors. We reported preliminary findings about predictors of comorbid psychiatric
disorders in ASDs. Service providers and families can benefit from such information
predicting which children are more likely to receive comorbid psychiatric diagnoses.
Such information can guide them in anticipating these conditions and planning
interventions targeted at minimizing maladaptive behaviors associated with certain
comorbid psychiatric disorders.
Findings associated with predictors of educational placement also have obvious
implications for parents, service providers, and administrators.
Study Limitations
The findings from this study should be interpreted in the light of a number of
considerations.
(1) The first set of limitations concern the sample of this study. (a) First, of 342
potential participants, follow-up data could be collected from 143 (41.8%). Some amount
of attrition is inherent in the nature of a follow-up study. Though comparative analysis
between final sample and non-participants did not reveal any significant differences,
nevertheless additional information might have been gained if follow-up data could be
collected from the complete target pool of 342 individuals. (b) Second, the sample size of
N = 143, though comparable to similar follow-up studies in the literature (Anderson et
al., 2007), is moderately small. (c) Third, at follow-up, the oldest participant in this
sample was 17 years of age and less than one-third (30.8%) were in the adolescence age
range (12-17 years). In other words, this study examined changes in maladaptive
behaviors only until adolescence; it is possible that additional changes in sample
119
characteristics and patterns of maladaptive behaviors continue into adulthood that could
not be captured in the current sample.
(2) A second potential limitation was our reliance on parent report for collecting
information on sample characterization at follow-up, including prevalence of psychiatric
comorbidty. Parent report on presence or absence of psychiatric comorbidity could not be
validated with an independent written diagnosis. During the phone interview with
parents, psychiatric disorders were endorsed only if parents affirmed that a doctor
assigned that particular diagnosis to the child and/or the child was taking prescribed
medication for that diagnosis (interview questions about psychiatric comorbidity are
found on the demographic form in Appendix C). These steps might increase confidence
in the parent-reported data on psychiatric comorbidity. Further, parent report is a
relatively common method of collecting follow-up data that has been used in studies in
the past (Ballaban-Gil et al., 1996, Rosenberg et al., 2011). Questions about DSM
psychiatric comorbidity used in this study were very similar to IAN questionnaires used
by Rosenberg et al. Additionally, cumulative prevalence and point prevalence rates of
psychiatric disorders reported by parents in this study were fairly consistent with rates of
research- and clinician-confirmed diagnoses reported in previous studies (Gjevik et al.,
2011; Leyfer et al., 2006, Simonoff et al., 2008).
(3) We found T1 measures of maladaptive behavior (T1 NCBRF scores) to be the
most consistent predictor of maladaptive behavior at follow-up (T2 NCBRF scores).
Since ratings on the NCBRF were collected from parent informants at both time points, it
is possible that findings were affected by single source bias (SSB). A form of common
120
method variance, SSB arises when overlapping variability is due to data collected from a
single source (Campbell & Fiske, 1959)–in this case, parents. However, T2 NCBRF
ratings showed a heterogeneous pattern of change, with ratings on some subscale items
improving significantly (and to varying degrees), others remaining stable, and others
deteriorating to varying degrees. If change was solely an artifact of SSB, we would
expect more uniformity than was observed.
(4) Finally, given the nature of data available in patient records at the Nisonger
Center clinics, IQ information was not available on all participants. Previous studies have
found IQ to be significantly associated with overall outcome in ASDs (Billstedt et al.,
2005; Howlin et al., 2000, 2004; Shea & Mesibov, 2005). However, association of IQ
with change in maladaptive behaviors over time has been inconsistent (Ballaban-Gil et
al., 1996; Murphy et al., 2005; Shattuck et al., 2007). A similar pattern of inconsistency is
seen in the association between IQ and psychiatric comorbidity in ASDs. While some
reports have found psychiatric comorbidity to be unrelated to differences in IQ (Gjevick
et al., 2011; Simonoff et al., 2008), association between IQ and psychiatric symptoms
have also been reported (Witwer & Lecavalier, 2011). Ancillary analyses in this study did
not indicate IQ to be a significant predictor of change in maladaptive behaviors.
Nevertheless, broader coverage of IQ information on all participants might have revealed
relationships not captured in this study.
121
Study Strengths
Findings from the current study extend our understanding of maladaptive
behaviors in the ASD population–an area which has received comparatively less attention
in past research. The following strengths of the study seem noteworthy.
(1) Although previously described as a limitation, the sample used in the current
study also had some advantages. (a) First, the sample included children and adolescents
representing all three ASD subgroups and participants of varying language level. (b) It
was a community sample, not referred for psychiatric assessment or treatment or
attending special schools or programs. This reduced the likelihood of artificially high
rates of maladaptive behavior or psychiatric comorbidity due to referral bias. Overall, the
sample could be regarded as fairly representative of a broader population of youth with
ASD. This increases confidence in the generalizability of findings from this study.
(2) The ASD diagnoses in the current sample were received from a single reputed
source–via evaluation by an interdisciplinary team at Nisonger Center clinics where
standardized diagnostic instruments (e.g., the ADI–R) were routinely used in
assessments. This allowed increased confidence that ASD diagnoses in the current
sample were indeed valid.
(3) Data from standardized testing of language skills were available for over 94%
of the sample at T1. This allowed an examination of outcomes associated with a broad
range of language abilities as opposed to being limited to a single qualitative label
(verbal or non verbal) as done in previous studies (Howlin et al., 2000; Mawhood et al.,
2000; Shattuck et al., 2007).
122
(4) A standardized rating scale (the NCBRF) designed for use in the DD
population was used to assess maladaptive behavior at both time points in all participants.
This provided confidence in how well maladaptive behaviors were assessed in the current
sample.
(5) Although stated as a limitation in the previous section, DSM psychiatric
comorbidity gathered via parent report had some associated advantages. As opposed to
research-assigned diagnosis, parent reports of psychiatric comorbidity provided
information on community diagnostic trends, which in turn may indicate variation in
application and interpretation of current DSM guidelines out in the community. A similar
advantage is also applicable to other parent-reported information collected at follow-up
(current education placements, psychotropic medications used, and interventions
received). Such parent-reported information reflects “real-world” situations and may be
useful for public policy considerations.
Future Directions
Currently, only a handful of studies report on changes in maladaptive behaviors in
ASDs. Given that chronic maladaptive behaviors have significant negative impacts on
the life of the person with an ASD, additional studies are definitely needed to clarify
further the developmental trajectory of maladaptive behaviors in this population. Such
studies should ideally involve individuals across a wide age range, including young and
older adults to trace possible changes in maladaptive behaviors into adulthood. Future
studies would also benefit from conducting in-person follow-up assessments to collect
information on comorbid psychiatric conditions via clinical interviews and additional
123
observations of the person with ASD. Ratings of maladaptive behaviors from multiple
sources (e.g., parents and teachers) would be helpful in examining situational differences
in maladaptive behaviors as the individual grows older.
The current sample had limited availability of IQ and adaptive behavior
information for study participants. Broader availability of these assessments at both T1
and follow-up would help to clarify the association between these variables and
maladaptive behavior change. The current study reported preliminary findings on
predictors of comorbid psychiatric disorders in ASDs. Independent support for such
findings from studies including a greater number of affected individuals in each
diagnostic category would be useful.
There is also a need to understand better the contribution of interventions on
maladaptive behavior change. Future studies could explore more optimum methods of
capturing information related to interventions. Information on predictive power of
particular type/amount/timing of interventions in reducing maladaptive behaviors or risk
of psychiatric comorbidity would be significant in the ASD field.
Conclusion
In addition to the core features of ASD, children and adolescents on the autism
spectrum present with various maladaptive behaviors beyond those defining the disorder.
Chronic maladaptive behaviors frequently result in additional impairment for the
individual and may ultimately lead to a lower quality of life. Such behaviors are also
associated with increased caregiver stress. A well established body of research suggests
that maladaptive behaviors are among the most difficult aspects of caring for a child with
124
ASD (Hastings, 2003; Hastings and Brown, 2002; Tomanik et al., 2004). Despite the
prevalence and significance of maladaptive behaviors in ASDs, the natural course of
these behaviors over time is not well documented. I hope that this study contributes to
the ASD literature by providing additional clarification on (1) patterns of stability and
change in different types of maladaptive behaviors from childhood to adolescence, (2)
association between subject characteristics and improvement/deterioration in maladaptive
behaviors, and (3) specifying variables that predict changes in these behaviors over time.
These and other exploratory findings from this study have important implications for
everyone involved with the individual with an ASD–parents, clinicians, researchers, and
service providers.
125
References
Abramson, R. K., Wright, H. H., Cuccaro, M. L., Lawrence, L. G., Babb, S., &
Pencarinha, D., et al. (1992). Biological liability in families with autism. Journal
of American Academy of Child Adolescent Psychiatry, 31, 370–371.
Achenbach, T. M., Howell, C. T., Quay, H. C., & Conners, C. K. (1991). National survey
of problems and competencies among four- to sixteen-year-olds: Parents’ reports
for normative and clinical samples. Monographs of the Society for Research in
Child Development, 56, 1-131.
Advokat, C. D., Mayville, E. A., & Matson, J. L. (2000). Side effect profiles of atypical
antipsychotics, typical antipsychotics, or no psychotic medications in persons
with mental retardation. Research in Developmental Disabilities, 21, 75–84.
Allen, G., & Courchesne, E. (2001). Attention function and dysfunction in autism.
Frontiers in Bioscience, 6, 105-119.
Aman, M. G., Singh, N. N., Stewart, A. W., & Field, C. J. (1985). The aberrant behavior
checklist: A behavior rating scale for the assessment of treatment effects.
American Journal of Mental Deficiency, 89, 485–491.
Aman, M. G., Tasse´, M. J., Rojahn, J., & Hammer, D. (1996). The Nisonger CBRF: A
child behavior rating form for children with developmental disabilities. Research
in Developmental Disabilities, 17, 41–57.
Aman, M.G., Lam, K.S.L., & Van Bourgondien, M.E. (2005). Medication patterns in
patients with autism: Temporal, regional, and demographic influences. Journal of
Child and Adolescent Psychopharmacology, 15, 116-126.
Ambrosini, P. J. (2000). Historical development and present status of the schedule for
affective disorders and schizophrenia for schoolage children (K-SADS). Journal
of American Academy of Child Adolescent Psychiatry, 39, 49–58.
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders, fourth edition. Washington, DC: American Psychiatric Association
Press.
126
American Psychiatric Association. (2002). Diagnostic and statistical manual of mental
disorders, fourth edition, text revision. Washington, DC: American Psychiatric
Association Press.
Anderson, D.K., Maye, M., & Lord, C. (2011). Changes in maladaptive behaviors from
midchildhood to young adulthood in autism spectrum disorder. American
Journal of Intellectual and Developmental Disabilities, 116 (5), 381-397.
Ando, H. & Yoshimura, I. (1979). Effects of age on communication skill levels and
prevalence of maladaptive behaviors in autistic and mentally retarded children.
Journal of Autism and Developmental Disorders, 9(1), 83-93.
Babyak, M. (2004). What you see may not be what you get: A brief, nontechnical
introduction to overfitting in regression-type models. Psychosomatic Medicine,
66, 411-421.
Baghdadli, A., Picot, M., Michelon, C., Bodet, J., Pernon, E., Burstezjn, C., Hochmann,
J., Lazartigues, A., Prey, R., Aussilloux, C. (2007). What happens to children with
PDD when they grow up? Prospective follow-up of 219 children from preschool
age to mid-childhood. Acta Psychiatrica Scandinavica, 115(5), 403-412.
Ballaban-Gil, K., Rapin, I., Tuchman, R., & Shinnar, S. (1996). Longitudinal
examination of the behavioral, language, and social changes in a population of
adolescents and young adults with autistic disorder. Pediatric Neurology, 15(3),
217-223.
Baron-Cohen, S. (1991). The theory of mind deficit in autism: How specific is it? British
Journal of Developmental Psychology, 9, 301–314.
Baron-Cohen, S., Leslie, A., & Frith, U. (1985). Does the autistic child have a ‘theory of
mind’? Cognition, 21, 37–46.
Bartak L, Rutter M. (1976). Differences between mentally retarded and normally
intelligent autistic children. Journal of Autism and Childhood Schizophrenia ,6,
109- 20.
Bayley, N. (1993). Bayley Scales of Infant Development: Second Edition. San Antonio,
TX: The Psychological Corporation.
Billstedt, E., Gillberg, I. C., & Gillberg, C. (2007). Autism after adolescence: Populationbased 13- to 22-year follow-up study of 120 individuals with autism diagnosed in
childhood. Journal of Autism and Developmental Disorders, 37(9), 1822.
127
Boelte, S., & Poustka, F. (2000). Diagnosis of autism: The connection between current
and historical information. Autism, 4, 382-390.
Bolte, S. & Poustka, F. (2002). The relation between general cognitive level and
adaptive behavior domains in individuals with autism with and without comorbid mental retardation. Child Psychiatry and Human Development, 33(2),
165–172.
Bondy, A. S. & Frost, L .A. (1995) ‘Educational Approaches in Preschool: Behavior
Techniques in a Public School Setting’, in E . Schopler & G. B. Mesibov (eds).
Learning and Cognition in Autism, pp. 311–33. New York: Plenum.
Borthwick-Duffy, S. (1994). Prevalence of destructive behaviors: A study of aggression,
self-injury, and property destruction. In T. Thompson and D. Gray (Eds.),
Destructive Behavior in Developmental Disabilities (pp. 3-21). Thousand Oaks,
CA: Sage Publications, Inc.
Bradley, E., & Bolton, P. (2006). Episodic psychiatric disorders in teenagers with
learning disabilities with and without autism. British Journal of Psychiatry, 189,
361-366.
Bradley, E., Summers, J., Wood, H., & Bryson, S. (2004). Comparing rates of psychiatric
and behavior disorders in adolescents and young adults with severe intellectual
disability with and without autism. Journal of Autism and Developmental
Disorders, 34, 151–161.
Brereton, A. V., Tonge, B. J., & Einfeld, S. L. (2006). Psychopathology in Children and
Adolescents with Autism Compared to Young People with Intellectual Disability.
Journal of Autism and Developmental Disorders, 36(7), 863-870.
Brown, E. C., Aman, M. G., & Havercamp, S. M. (2002). Factor analysis and norms for
parent ratings on the Aberrant Behavior Checklist-Community for young people
in special education. Research in Developmental Disabilities, 23(1), 45-60.
Bruininks, R. H., Woodcock, R. W., Weatherman, R. F., & Hill, B. K. (1996). Scales of
Independent Behavior—Revised: Boston: Riverside Publishing Company.
Bryson, S. E., Rogers, S. J., & Fombonne, E. (2003). Autism spectrum disorders: Early
detection, intervention, education, and psychopharmacological management.
Canadian Journal of Psychiatry (Revue Canadienne De Psychiatrie), 48(8), 506–
516.
128
Butzer, B., & Konstantareas, M. M. (2003). Depression, temperament and their
relationship to other characteristics in children with Asperger's disorder. Journal
on Developmental Disabilities, 10(1), 67-72.
Campbell, D., & Fiske, D., (1959). Convergent and discriminant validation by the
multitrait-multimethod matrix. Psychological Bulletin, 56(2), 81-105.
Carrow-Woolfolk, E. (1999) Comprehensive Assessment of Spoken Language. American
Guidance Service, Circle Pines, MN, USA.
Chadwick, O., Piroth, N., Walker, J., Bernard, S., & Taylor, E. (2000). Factors affecting
the risk of behaviour problems in children with severe intellectual disability.
Journal of Intellectual Disability Research, 44, 108-123.
Chambers, W. J., Puig-Antich, J., Hirsch, M., Paez, P., Ambrosini, P. J., & Tabrizi, M.
A., et al. (1985). The assessment of affective disorders in children and adolescents
by semistructured interview. Test–retest reliability of the schedule for affective
disorders and schizophrenia for school-age children, present episode version.
Archives of General Psychiatry, 42, 696–702.
Charman,T., Howlin, P., Berry, B. & Prince, E. (2004). Measuring Developmental
Progress of Children with Autism Spectrum Disorder on School Entry Using
Parent Report, Autism, 8(1), 89-100.
Chung, S. Y., Luk, S. L., & Lee, P. W. (1990). A follow-up study of infantile autism in
Hong Kong. Journal of Autism Developmental and Disorders, 20, 221–232.
Clarke, A. R., Tonge, B. J., Einfeld, S. L., & Mackinnon, A. (2003).Assessment of
change with the developmental behaviour checklist. Journal of Intellectual
Disability Research, 47, 210–212.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd edn.). New
York: Academic Press.
Costello, E. J., Mustillo, S., Erkanli, A., Keeler, G., & Angold, A. (2003). Prevalence
and development of psychiatric disorders in childhood and adolescence. Archives
of General Psychiatry, 60 (8), 837-844.
Darrou, C., Pry, R., Pernon, E., Michelon, C., Aussilloux, C., & Baghdadli, A. (2010).
Outcome of young children with autism: Does the amount of intervention
influence developmental trajectories? Autism, 14(6), 663-677.
129
Dawson, G., & Lewy, A. (1989). Arousal, attention, and the socioemotional impairments
of individuals with autism. In G. Dawson (Ed.), Autism: Nature, diagnoses, and
treatment (pp. 49–74). New York: Guilford Press.
Dawson, G., Ashman, S . B. & Carver, L . J. (2000).The role of early experience in
shaping behavioral and brain development and its implications for social policy.
Development and Psychopathology, 12, 695–712.
de Bildt, A., Sytema, S., Kraijer, D., Sparrow, S., & Minderaa, R. (2005). Adaptive
functioning and behaviour problems in relation to level of education in children
and adolescents with intellectual disability. Journal of Intellectual Disability
Research, 49, 672-681.
de Bruin, E. I., Ferdinand, R. F., Meester, S., de Nijs, P. F., &Verheij, F. (2007). High
rates of psychiatric co-morbidity in PDD-NOS. Journal of Autism and
Developmental Disorders, 37, 877–886.
Dekker, M. C., & Koot, H. M. (2003). DSM-IV disorders in children with borderline to
moderate intellectual disability. I: Prevalence and impact. Journal of American
Academy of Child Adolescent Psychiatry, 42, 915-922.
Dekker, M. C., Koot, H. M., van der Ende, J., & Verhulst, F. C. (2002). Emotional and
behavioral problems in children and adolescents with and without intellectual
disability. Journal of Child Psychology & Psychiatry & Allied Disciplines, 43,
1087-1098.
Dekker, M. C., Nunn, R. J., Einfeld, S. E., Tonge, B. J., & Koot, H. M. (2002). Assessing
emotional and behavioral problems in children with intellectual disability:
Revisiting the factor structure of the Developmental Behavior Checklist. Journal
of Autism & Developmental Disorders, 32, 601–610.
Disabilities, 29, 235–246.
Dossetor, D. R. (2007). ‘All that glitters is not gold’: Misdiagnosis of psychosis in
pervasive developmental disorders—A case series. Clinical Child Psychology and
Psychiatry, 12, 537–548.
Downs, A., Downs, R. C., & Rau, K. (2008). Effects of training and feedback on discrete
trial teaching skills and student performance. Research in Developmental
Disabilities, 29, 247–255.
Dykens, E. M. (2000). Psychopathology in children with intellectual disability. Journal of
Child Psychology & Psychiatry & Allied Disciplines, 41, 407–417.
130
Eaves, L. C., & Ho, H. H. (2004). The Very Early Identification of Autism: Outcome to
Age 4 1/2-5. Journal of Autism and Developmental Disorders, 34(4), 367-378.
Eaves, L.C. & Ho, H.H. (1996). Brief Report: Stability and Change in Cognitive and
Behavioral Characteristics of Autism through Childhood, Journal of Autism and
Developmental Disorders, 26(5), 557–569.
Einfeld, S. L., & Tonge, B. J. (1992). Manual for the developmental behaviour checklist.
Clayton, Melbourne and Sydney: Monash University Centre for Developmental
Psychiatry and School of Psychiatry, University of NSW.
Einfeld, S. L., & Tonge, B. J. (1995). The developmental behavior checklist: The
development and validation of an instrument to assess behavioural and emotional
disturbance in children and adolescents with mental retardation. Journal of Autism
and Developmental Disorders, 25, 81–104.
Einfeld, S., Tonge, B., Turner, G., Parmenter, T., & Smith, A. (1999). Longitudinal
course of behavioural and emotional problems of young persons with PraderWilli, Fragile X, Williams and Down syndromes. Journal of Intellectual &
Developmental Disability, 24, 349–354.
Eisenmajer, R., Prior, M., Leekam, S., Wing, L., Gould, J., Welham, M., et al. (1996).
Comparison of clinical symptoms in autism and Asperger’s disorder. Journal of
American Academy of Child and Adolescent Psychiatry, 35(11), 1523– 31.
Eldevik, S., Eikeseth, S., Jahr, E. & Smith,T. (2006). Effects of Low-Intensity Behavioral
Treatment for Children with Autism and Mental Retardation. Journal of Autism
and Developmental Disorders, 36(2), 211–224.
Esbensen, A. J., Rojahn, J., Aman, M. G., & Ruedrich, S. (2003). Reliability and validity
of an assessment instrument for anxiety, depression, and mood among individuals
with mental retardation. Journal of Autism & Developmental Disorders, 33, 617629.
Estes, A. M., Dawson, G., Sterling, L., & Munson, J. (2007). Level of intellectual
functioning predicts patterns of associated symptoms in school-age children with
autism spectrum disorder. American Journal on Mental Retardation, 112(6), 439449.
Fecteau, S., Mottron, L., Berthiaume, C., & Burack, J. A. (2003). Developmental changes
of autistic symptoms. Autism, 7, 255-268.
131
Fein, D., Dixon, P., Paul, J., & Levin, H. (2005). Brief report: Pervasive developmental
disorder can resolve into ADHD: Case illustrations. Journal of Autism and
Developmental Disorders, 35, 525–534.
Fein, D., Stevens, M., Dunn, M., Waterhouse, L., Allen, D., Rapin, I., & Feinstein, C.
(1999). Subtypes of pervasive developmental disorder: Clinical characteristics.
Child Neuropsychology, 5(1), 1-23.
Filipek, P. A., Accardo, P. J., Ashwal, S., Baranek, G. T., Cook, E. H., Jr, Dawson, G., et
al. (2000). Practice parameter: Screening and diagnosis of autism: Report of the
quality standards subcommittee of the American academy of neurology and the
child neurology society. Neurology, 55(4), 468–479.
Fombonne, E. (1997). Autism: Recent research findings. Current Opinion in Psychiatry,
10, 373–377.
Fombonne, E. (2003). Epidemiological surveys of autism and other pervasive
developmental disorders: An update. Journal of Autism and Developmental
Disorders, 33, 365–38.
Frazier, J. A., Biederman, J., Bellordre, C. A., Garfield, S. B., Geller, D. A., Coffey, B. J.,
& Faraone, S. V. (2001). Should the diagnosis of attention-deficit/hyperactivity
disorder be considered in children with pervasive developmental
disorder? Journal of Attention Disorders, 4(4), 203-211.
Freeman, B. J., Rahbar, B., Ritvo, E .R., Bice, T.L., Yokota, A., & Ritvo, R. (1991). The
stability of cognitive and behavioral parameters in autism: A twelve-year
prospective study, Journal of the American Academy of Child and Adolescent
Psychiatry, 30, 479–82.
Freeman, B. J., Ritvo, E .R., Needleman, R. & Yokota, A. (1985). The stability of
cognitive linguistic parameters in autism: A five-year prospective study. Journal
of the American Academy of Child Psychiatry, 24, 459–64.
Gabriels, R. L., Hill, D. E., Pierce, R.A., Rogers, S. J., & Wehner, B. (2001). Predictors
of Treatment Outcome in Young Children with Autism, Autism (5), 407–29.
Gadow, K. D., DeVincent, C. J., & Azizian, A. (2004). Psychiatric symptoms in
preschool children with PDD and clinic and comparison samples. Journal of
Autism and Developmental Disorders, 34, 379–393.
Gadow, K. D., DeVincent, C. J., Pomeroy, J., & Azizian, A. (2005). Comparison of
DSM-IV symptoms in elementary school-aged children with PDD versus clinic
and community samples. Autism, 9, 392–415.
132
Gadow, K. D., DeVincent, C., & Schneider, J. (2008). Predictors of psychiatric
symptoms in children with an autism spectrum disorder. Journal of Autism and
Developmental Disorders, 38(9), 1710-1720.
Gadow, K. D., & Sprafkin, J. (1994). Child symptom inventories manual. Stony Brook,
NY: Checkmate Plus.
Gadow, K. D., & Sprafkin, J. (2002). Child symptom inventory-4 screening and norms
manual. Stony Brook, NY: Checkmate Plus.
Gerber, F., Baud, M. A., Giroud, M., & Carminati, G. G. (2008). Quality of life of adults
with Pervasive Developmental Disorders and intellectual disabilities. Journal of
Autism and Developmental Disorders, 38(9), 1654-1665.
Ghaziuddin, M. (2002). Asperger syndrome: Associated psychiatric and medical
conditions. Focus on Autism and Other Developmental Disabilities, 17(3), 138144.
Ghaziuddin, M., Ghaziuddin, N., & Greden, J. (2002). Depression in persons with autism:
Implications for research and clinical care. Journal of Autism and Developmental
Disorders, 32, 299–306.
Ghaziuddin, M., Weidmer-Mikhail, E., & Ghaziuddin, N. (1998). Comorbidity of
Asperger syndrome: A preliminary report. Journal of Intellectual Disability
Research, 42, 279–283.
Gillberg, C., & Billstedt, E. (2000). Autism and Asperger syndrome: Coexistence with
other clinical disorders. Acta Psychiatrica Scandinavica, 102(5), 321-330.
Gillberg, C., & Coleman, M. (1992). The biology of the autistic syndromes. London:
Mackeith Press.
Gillberg, C., & Steffenburg, S. (1987). Outcome and prognostic factors in infantile autism
and similar conditions: A population-based study of 46 cases followed through
puberty. Journal of Autism and Developmental Disorders, 17, 272–288.
Gillot, A., Furniss, F., & Walter, A. (2001). Anxiety in high-functioning children with
autism. Autism, 4, 117–132.
Gjevik, E., Eldevik, S., Fjran-Granum, T., & Sponheim, E. (2011). Kiddie-SADS reveals
high rates of DSM-IV disorders in children and adolescents with autism
spectrum disorders. Journal of Autism and Developmental Disorders, 41(6), 761769.
133
Goldstein, S., Schwebach, A. (2004). The comorbidity of pervasive developmental
disorders and attention deficit hyperactivity disorder: results of a retrospective
chart review. Journal of Autism and Developmental Disorders, 34, 329-339.
Gonzalez, N.M., Murray, A., Shay, J., Campbell, M. & Small, A.M. (1993). Autistic
children on followup: Change of diagnosis. Psychopharmacology Bulletin, 29,
353–358.
Green, J., Gilchrist, A., Burton, D., & Cox. A. (2000). Social and psychiatric functioning
in adolescents with Asperger syndrome compared with conduct disorder. Journal
of Autism and Developmental Disorders, 30, 279-293.
Green, V. A., Pituch, K. A., Itchon, J., Choi, A., O’Reilly, M., & Sigafoos, J. (2006).
Internet survey of treatments used by parents of children with autism. Research in
Developmental Disabilities, 27(1), 70–84.
Grondhuis, S. & Aman, M. G. (2012). Assessment of anxiety in children and adolescents
with autism spectrum disorders. Research in Autism Spectrum Disorders, 6, 13451365.
Harris, S. L. & Handleman, J. S. (2000). Age and IQ at intake as predictors of placement
for young children with autism: A four-to six-year follow-up. Journal of Autism
& Developmental Disorders, 30, 137–42.
Harris, S. L., Handleman, J. S ., Gordon, R., Kristoff, B. & Fuentes, F. (1991). Changes
in cognitive and language functioning of preschool children with autism. Journal
of Autism and Developmental Disorders, 21, 281–90.
Harrison, P. L., & Oakland, T. (2003). Adaptive Behavior Assessment System manual
(2nd ed.). Los Angeles: Western Psychological Services.
Hartley, S., Sikora, D., & McCoy, R. (2008). Prevalence and risk factors of maladaptive
behaviour in young children with autistic disorder. Journal of Intellectual
Disability Research, 52(10), 819-829.
Hedley, D., & Young, R. (2006). Social comparison processes and depressive symptoms
in children and adolescents with Asperger syndrome. Autism, 10(2), 139-153.
Heiervang, E., Stormark, K. M., Lundervold, A. J., Heimann, M., Goodman, R.,
Posserud, M. B., et al. (2007). Psychiatric disorders in Norwegian 8–10-year-olds:
An epidemiological survey of prevalence, risk factors, and service use. Journal of
the American Academy of Child and Adolescent Psychiatry, 46, 438–447.
134
Herring, S., Gray, K., Taffe, J., Tonge, B., Sweeney, D., Einfeld, S. (2006). Behaviour
and emotional problems in toddlers with pervasive developmental disorders and
developmental delay: associations with parental mental health and family
functioning. Journal of Intellectual Disability Research, 50, 874-882.
Hoaglin, D. C., & Welsh, R.E. (1978). The hat matrix in regression and ANOVA. The
American Statistician, 32, 17-22.
Holden, B., & Gitlesen, J. P. (2006). A total population study of challenging behavior in
the country of Hedmark, Norway: Prevalence, and risk markers. Research in
Developmental Disabilities, 27, 456–465.
Holtmann, M., Bolte, S., & Poustka, F. (2007). Autism spectrum disorders: Sex
differences in autistic behaviour domains and coexisting psychopathology.
Developmental Medicine & Child Neurology, 49(5), 361-366.
Howard, J.S., Sparkman, C.R., Cohen, H.G., Green, G. & Stanislaw, H. (2005). A
Comparison of Intensive Behavior Analytic and Eclectic Treatments for Young
Children with Autism. Research in Developmental Disabilities, 26(4), 359–383.
Howell, D. (2007). Correlation and Regression. Statistical Methods for Psychology (6
th Ed). Australia: Thomson.
Howlin, P. (2007). The outcome in adult life for people with ASD. In F. R. Volkmar
(Ed.). Autism and pervasive developmental disorders (2nd edn. pp. 269-306).
Cambridge: Child and Adolescent Psychiatry.
Howlin, P., & Asgharian, A. (1999). The diagnosis of autism and Asperger syndrome:
findings from a survey of 770 families. Developmental Medicine and Child
Neurology, 41(12), 834–839.
Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2004). Adult outcome for children with
autism. Journal of Child Psychology and Psychiatry, 45(2), 212-229.
Howlin, P., Mawhood, L., & Rutter, M. (2000). Autism and developmental receptive
language disorder – a follow-up comparison in early adult life. II: Social,
behavioural, and psychiatric outcomes. Journal of Child Psychology and
Psychiatry, 41, 561–578.
Jarbrink, K., Fombonne, E., & Knapp, M. (2003). Measuring the Parental, Service and
Cost Impacts of Children with Autistic Spectrum Disorder: A Pilot Study.
Journal of Autism and Developmental Disorders, 33(4), 395-402.
135
Jonsdottir, S.L., Saemundsen, E., Asmundsdottir, G., Hjartardottir, S., Asgeirsdottir,
B.B., Smaradottir, H.H., et al. (2006). Follow-Up of Children Diagnosed with
Pervasive Developmental Disorders: Stability and Change during the Preschool
Years. Journal of Autism and Developmental Disorders, 37(7), 1361–1374.
Kanai, C., Koyama, T., Kato, S., Miyamoto, Y., & Osada, H. (2004). Comparison of
high-functioning atypical autism and childhood autism by Childhood Autism
Rating Scale-Tokyo version. Psychiatry and Clinical Neurosciences, 58(2), 217221.
Kanner, L. (1971). Follow-up study of eleven autistic children originally reported in
1943. Journal of Autism & Childhood Schizophrenia, 1(2), 119-145.
Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., & Moreci, P., et al. (1997).
Schedule for affective disorders and schizophrenia for school-age children-present
and lifetime version (K-SADS-PL): Initial reliability and validity data. Journal of
American Academy of Child Adolescent Psychiatry, 36, 980–988.
Kelley, E., Naigles, L., & Fein, D. (2010). An in-depth examination of optimal outcome
children with a history of autism spectrum disorders. Research in Autism
Spectrum Disorders, 4(3), 526-538.
Kiernan, C. & Kiernan, D. (1994). Challenging behaviour in schools for pupils with
severe learning difficulties. Mental Handicap Research, 7(3), 177-201.
Kim, J. A., Szatmari, P., Bryson, S. E., Streiner, D. L., & Wilson, F. J. (2000). The
prevalence of anxiety and mood problems among children with autism and
Asperger syndrome. Autism, 4(2), 117-132.
King, N. J., Olendick, T. H., & Tonge, B. J. (1995). School refusal: assessment and
treatment. Boston: Allyn & Bacon.
Kline, R. B. (2004). Beyond significance testing: Reforming data analysis methods in
behavioral research. Washington, DC: American Psychological Association.
Kobayashi, R, & Murata, T. (1998a). Setback phenomenon in autism and long-term
prognosis. Acta Psychiatrica Scandinavica, 4, 296–303.
Kobayashi, R., & Murata. T. (1998b). Behavioral characteristics of 187 young adults with
autism. Psychiatry and Clinical Neuroscience, 52, 383-390.
Krug, D.A., Arik, J., Almond, P. (1980). Behavior checklist for identifying severely
handicapped individuals with high levels of autistic behavior. Journal of Child
Psychology and Psychiatry, 21, 221–229.
136
Lainhart, J. E., & Folstein, S. E. (1994). Affective disorders in people with autism: A
review of published cases. Journal of Autism and Developmental Disorders, 24,
587–601.
Langworthy-Lam, K. S., Aman, M. G., & Van Bourgondien, M. E. (2002). Prevalence
and patterns of use of psychoactive medicines in individuals with autism in the
Autism Society of North Carolina. Journal of Child and Adolescent
Psychopharmacology, 12(4), 311-321.
Lawley, D. N., & Maxwell, A. E. (1971). Factor analysis as a statistical method.
London: Butterworth & Co.
Lecavalier, L. (2006). Behavioral and emotional problems in young people with
pervasive developmental disorders: Relative prevalence, effects of subject
characteristics, and empirical classification. Journal of Autism and Developmental
Disorders, 36(8), 1101-1114.
Lecavalier, L., Aman, M. G., Hammer, D., Stoica, W., & Matthews, G. L. (2004). Factor
analysis of Nisonger Child Behavior Rating Form in children with autism
spectrum disorders. Journal of Autism and Developmental Disorders, 34, 709–
721.
Lee, D. O., & Ousley, O. Y. (2006). Attention-deficit hyperactivity disorder symptoms in
a clinic sample of children and adolescents with pervasive developmental
disorders. Journal of Child and Adolescent Psychopharmacology, 16, 737-746.
Lee, L. C., Harrington, R. A., Chang, J. J., & Connors, S. L. (2008). Increased risk of
injury in children with developmental disabilities. Research in Developmental
Disabilities, 29, 351–362
Leyfer, O. T., Folstein, S. E., Bacalman, S., Davis, N. O., Dinh, E., Morgan, J., TagerFlusberg, H., Lainhart, J. E. (2006). Comorbid Psychiatric Disorders in Children
with Autism: Interview Development and Rates of Disorders. Journal of Autism
and Developmental Disorders, 36(7), 849-861.
Liss, M., Harel, B., Fein, D., Allen, D., Dunn, M., Feinstein, C., Robin. M., Waterhouse,
L., Rapin, I. (2001). Predictors and correlates of adaptive functioning in children
with developmental disorders. Journal of Autism and Developmental Disorders,
31(2), 219-230.
Lockyer, L., & Rutter, M. (1969). A five-to-fifteen year follow-up study of infantile
psychosis: III. Psychological aspects. British Journal of Psychiatry, 115,865–882.
137
Lockyer, L., & Rutter, M. (1970). A five-to-fifteen year follow-up study of infantile
psychosis: IV. Patterns of cognitive ability. British Journal of Social and Clinical
Psychology, 9, 152–163.
Lord, C., & Schopler, E. (1988). Intellectual and Developmental Assessment of Autistic
Children from Preschool to School Age: Clinical Implications of Two Follow-Up
Studies’, in E . Schopler & G. Mesibov (Eds.) Diagnosis and Assessment in
Autism (pp. 167–81). New York: Plenum.
Lord, C. & Schopler, E. (1989a). The role of age at assessment, developmental level, and
test in the stability of intelligence scores in young autistic children. Journal of
Autism Developmental Disorders, 19, 483–99.
Lord, C. & Schopler, E. (1989b). Stability of assessment results of autistic and
nonautistic language-impaired children from preschool years to early school age,
Journal of Child Psychology and Psychiatry, 30, 575–90.
Lord, C., & Bailey, A. (2002). Autism spectrum disorders. In M. Rutter & E. Taylor
(Eds.), Child and adolescent psychiatry (4th edn, pp. 664–681). Oxford:
Blackwell Scientific.
Lord, C., Risi, S., Lambrecht, L., Cook, E. H., Leventhal, B. L., DiLavore, P. C., et al.
(2000). The autism diagnostic observation schedule-generic: A standard measure
of social and communicative deficits associated with the autism spectrum.
Journal of Autism and Developmental Disorders, 30, 205–223.
Lord, C., Rutter, M., & LeCouteur, 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 & Developmental
Disorders, 24, 659-685.
Lotter, V. (1974). Factors related to outcome in autistic children. Journal of Autism and
Childhood Schizophrenia, 4, 263–277.
Lounds, T. J., & Seltzer, M. M. (2010). Changes in the Autism Behavioral Phenotype
During the Transition to Adulthood. Journal of Autism and Developmental
Disorders, 40(12), 1431-1446.
Loveland, K. A., & Tunali-Kotoski, B. (1997). The school age child with autism. In D. J.
Cohen, & F. R. Volkmar (Eds.), Handbook of autism and pervasive
developmental disorders (2nd ed. pp. 283–308). New York: Wiley.
138
Mandell, D. S., Morales, K. H., Marcus, S. C., Stahmer, A. C., Doshi, J., & Polsky, D. E.
(2008). Psychotropic medication use among medicaid-enrolled children with
autism spectrum disorders. Pediatrics, 121(3), 441–448.
Marascuilo, L. A., & Levin, J. R. (1983). Multivariate statistics in the social sciences: A
researcher’s guide. Monterey, CA: Brooks/Cole.
Marshburn, E. C., & Aman, M. G. (1992). Factor validity and norms for the Aberrant
Behavior Checklist in a community sample of children with mental retardation.
Journal of Autism and Developmental Disorders, 22(3), 357-373.
Martin, A., Scahill, L., Klin, A., & Volkmar, F. R. (1999). Higher functioning pervasive
developmental disorders: Rates and patterns of psychotropic drug use. Journal of
the American Academy of Child and Adolescent Psychiatry, 38(7), 923–931.
Masi, G., Brovedani, P., Mucci, M., & Favilla, L. (2002). Assessment of anxiety and
depression in adolescents with mental retardation. Child Psychiatry & Human
Development, 32, 227–237.
Masi, G., Favilla, L., & Mucci, M. (2000). Generalized anxiety disorder in adolescents
and young adults with mild mental retardation. Psychiatry, 63, 54–64.
Masi, G., Mucci, M., Favilla, L., & Poli, P. (1999). Dysthymic disorder in adolescents
with intellectual disability. Journal of Intellectual Disability Research, 43, 80–87.
Matson, J. L., Gonzalez, M. L., & Rivet, T. T. (2008). Reliability of the Autism Spectrum
Disorder-Behavior Problems for Children (ASD-BPC). Research in Autism
Spectrum Disorders, 2(4), 696-706.
Matson, J. L., Wilkins, J., & Macken, J. (2009). The relationship of challenging
behaviors to severity and symptoms of autism spectrum disorders. Journal of
Mental Health Research in Intellectual Disabilities, 2, 29–44.
Matson, J.L., Mahan, S., Hess, J. A., Fodstad, J. C., & Neal, D. (2010). Progression of
challenging behaviors in children and adolescents with Autism Spectrum
Disorders as measured by the Autism Spectrum Disorders-Problem Behaviors for
Children (ASD-PBC). Research in Autism Spectrum Disorders, 4, 400-404.
Mattila, M., Hurtig, T., Haapsamo, H., Jussila, K., Kuusikko-Gauffin, S., Kielinen, M.,
Linna, S., Ebeling, H., Bloigu, R., Joskitt, L., Pauls, D., Moilanen, I. (2010).
Comorbid psychiatric disorders associated with Asperger syndrome/highfunctioning autism: A community- and clinic-based study. Journal of Autism and
Developmental Disorders, 40(9), 1080-1093.
139
Mawhood, L.M., Howlin, P., & Rutter, M. (2000). Autism and developmental receptive
language disorder: A follow-up comparison in early adult life: I. Cognitive and
language outcomes. Journal of Child Psychology and Psychiatry, 41, 547–559.
McBrien, J. A. (2003). Assessment and diagnosis of depression in people with
intellectual disability. Journal of Intellectual Disability Research, 47, 1–13.
McClintock, K., Hall, S., & Oliver, C. (2003). Risk markers associated with challenging
behaviours in people with intellectual disability: A meta-analytic study. Journal
of Intellectual Disability Research, 47(6), 405-416.
McDougle, C. J., Kresch, L. E., Goodman, W. K., Naylor, S. T., Volkmar, F. R., Cohen,
D. J., & Price, L. H. (1995). A case controlled study of repetitive thoughts and
behaviour in adults with autistic disorder and obsessive-compulsive disorder.
American Journal of Psychiatry, 152, 772–777.
Mceachin, J. J., Smith, T. & Lovaas, O. I . (1993). Long-Term Outcome for Children
with Autism Who Received Early Intensive Behavioral Treatment, American
Journal on Mental Retardation , 97, 359–72.
McGovern, C. W., & Sigman, M. (2005). Continuity and change from early childhood to
adolescence in autism. Journal of Child Psychology and Psychiatry, 46(4), 401408.
Meadows, G., Turner, T., Campbell, L., & Lewis, S. W., et al. (1991). Assessing
schizophrenia in adults with mental retardation: A comparative study. British
Journal of Psychiatry, 158, 103–105.
Mesibov, G. B., Schopler, E., Schaffer, B., & Michal, N. (1989). Use of the Childhood
Autism Rating Scale with autistic adolescents and adults. Journal of the American
Academy of Child & Adolescent Psychiatry, 28(4), 538-541.
Minshew, N. J., Goldstein, G., & Siegel, D. J. (1997). Neuropsychologic functioning in
autism: Profile of a complex information processing disorder. Journal of the
International Neuropsychological Society, 3, 303–316.
Morton, J. F., & Campbell, J. M. (2008). Information source affects peers’ initial
attitudes toward autism. Research in Developmental Disabilities, 29, 189–201.
Mullen, E. M. (1995). Mullen Scales of Early Learning. American Guidance Service.
Circle Pines: Minn.
Mundy, P. & Neal, A.R. (2001). Neural Plasticity, Joint Attention, and a Transactional
Social-Orienting Model of Autism, in L .M. Glidden (ed.) International Review of
140
Research in Mental Retardation, (Vol. 23, pp. 139–68). San Diego, CA:
Academic Press.
Muris, P., Steerneman, P., Merckelbach, H., Holdrinet, I., & Meesters, C. (1998).
Comorbid anxiety symptoms in children with pervasive developmental disorders.
Journal of Anxiety Disorders, 12, 387-393.
Murphy, G. H., Beadle-Brown, J., Wing, L., Gould, J., Shah, A., & Holmes, N. (2005).
Chronicity of Challenging Behaviours in People with Severe Intellectual
Disabilities and/or Autism: A Total Population Sample. Journal of Autism and
Developmental Disorders, 35(4), 405-418.
Myers, S. M., Johnson, C. P., & American Academy of Pediatrics Council on Children
With Disabilities. (2007). Management of children with autism spectrum
disorders. Pediatrics, 120(5), 1162–1182.
NICHD (2004). Trajectories of physical aggression from toddlerhood to middle
childhood. Monographs of the Society for Research in Child Development, 69(4).
Nicholas, J. S., Charles, J. M., Carpenter, L. A., King, L. B., Jenner, W., & Spratt, E. G.
(2008). Prevalence and characteristics of children with autism spectrum disorders.
Annals of Epidemiology, 18, 130–136.
Nordin, V., & Gillberg, C. (1998). The long-term course of autistic disorders: Update on
follow-up studies. Acta Psychiatrica Scandinavica, 97, 99–108.
Norman, G. R., Sloan, J. A., & Wyrwich, K. W. (2003). Interpretation of changes in
health-related quality of life: The remarkable universality of half a standard
deviation. Medical Care, 41(5), 582–592.
Olsson, I., Gillberg, C., & Steffenburg, S. (1988). Epilepsy in autism and autistic-like
conditions: A population-based study. Archives of Neurology, 45, 666–668.
O’Reilly, M. F., Sigafoos, J., Lancioni, G., Rispoli, M., Lang, R., Chan, J., et al. (2008).
Manipulating the behavior-altering effect of motivating operation: Examination
of the influence on challenging behavior during leisure activities. Research in
Developmental Disabilities, 29, 333–340.
Oswald, D. P., & Sonenklar, N. A. (2007). Medication use among children with autism
spectrum disorders. Journal of Child and Adolescent Psychopharmacology, 17(3),
348–355.
Oti, R., Lord, C., Risi, S., & Carlson, C. (2004). Educational placement of children with
autism: Results from a longitudinal study. Poster presented at the International
141
meeting for autism research, Sacramento, CA.
Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991). Executive function deficits in
high-functioning autistic individuals: Relationship to theory of mind. Journal of
Child Psychology and Psychiatry, 32(7), 1081-1105.
Ozonoff, S., Strayer, D. L., McMahon, W. M., & Filloux, F. (1994). Executive function
abilities in autism and Tourette syndrome: An information processing approach.
Journal of Child Psychology & Psychiatry & Allied Disciplines, 35, 1015–1032.
Peduzzi, P.N., Concato, J., Holford, T.R., Feinstein, A.R. (1995). The importance of
events per independent variable in multivariable analysis, II: accuracy and
precision of regression estimates. Journal of Clinical Epidemiology, 48, 1503–
1510.
Peduzzi, P.N., Concato, J., Kemper, E., Holford, T.R., Feinstein, A.R. (1996). A
simulation study of the number of events per variable in logistic regression
analysis. Journal of Clinical Epidemiology, 49, 1373–1379.
Peng, C. J., Lee, K. L., & Ingersoll, G. M. (2002). An introduction to logistic regression
analysis and reporting. Journal of Educational Research, 96 (1), 3-14.
Petersen, A. C., Crockett, L., Richards, M., & Boxer, A. (1988). A Self-report measure of
pubertal status – reliability, validity, and initial norms. Journal of Youth and
Adolescence, 17(2), 117-133.
Phelps-Terasaki, D., & Phelps-Gunn, T. (2007). Test of Pragmatic Language, Second
Edition. East Moline, IL: Linguisystems.
Piven, J., Harper, J., Palmer, P., & Arndt, S . (1996). Course of behavioral change in
autism: A retrospective study of high-IQ adolescents and adults. Journal of the
American Academy of Child and Adolescent Psychiatry, 35, 523–529.
Pringle, B. A., Colpe, L. J., Blumberg, S. J., Avila, R. M., & Kogan, M. D.(2012).
Diagnostic history and treatment of school-aged children with autism spectrum
disorder and special health care needs. NCHS data brief, no 97. Hyattsville, MD:
National Center for Health Statistics. Retrieved June 1, 2012, from
http://www.cdc.gov/nchs/data/databriefs/db97.htm
Reynolds, C. R., & Kamphaus, R. W. (1992). Behavior assessment system for children.
American Guidance Service. Circle Pines: Minn.
142
Ringdahl, J. E., Call, N. A., Mews, J. B., Boelter, E. W., & Christensen, T. J. (2008).
Assessment and treatment of aggressive behavior without a clear social function.
Research in Developmental Disabilities, 29(4), 351-362.
Roid, G. H. (2003). Stanford-Binet Intelligence Scales, Fifth Edition, Examiner's Manual.
Itasca, IL: Riverside Publishing.
Roid, G., & Miller, L. (1997). Leiter International Performance Scale-Revised:
Examiner’s Manual. In G.H. Roid & L.J. Miller, Leiter International
PerformanceScale-Revised. Wood Dale, IL: Stoelting Co.
Rojahn, J., Aman, M. G., Matson, J. L., & Mayville, E. A. (2003). The Aberrant Behavior
Checklist and the Behavior Problems Inventory: Convergent and divergent
validity. Research in Developmental Disabilities, 24, 391–404.
Rojahn, J., Matson, J. L., Lott, D., Esbensen, A. J., & Smalls, Y. (2001). The behavior
problems inventory: An instrument for the assessment of self-injury, stereotyped
behavior, and aggression/destruction in individuals with developmental
disabilities. Journal of Autism & Developmental Disorders, 31, 577–588.
Rosenberg, R. E., Kaufmann, W. E., Law, J.K., & Law, P. A. (2011). Parent report of
community psychiatric comorbid disgnoses in autism spectrum disorders. Autism
Research and Treatment, Article 405849. Retrieved June 2, 2012, from
http://www.hindawi.com/journals/aurt/2011/405849/
Rumsey, J. M., Rapoport, J. L., & Sceery, W. R. (1985). Autistic children as adults:
Psychiatric, social and behavioral outcomes. Journal of the American Academy of
Child & Adolescent Psychiatry, 24, 465-473.
Rutter, M., & Lockyer, L. (1967). A five to fifteen year follow-up study of infantile
psychosis: I. Description of the sample. British Journal of Psychiatry, 113, 1169–
1182.
Rutter, M., Caspi, A., & Moffitt, T. (2003). Using sex differences in psychopathology to
study causal mechanisms: Unifying issues and research strategies. Journal of
Child Psychology and Psychiatry, 44, 1092–1115.
Rutter, M., Greenfeld, D., & Lockyer, L. (1967). A five to fifteen year follow-up of
infantile psychosis: II. Social and behavioural outcome. British Journal of
Psychiatry, 113, 1183–1189.
Rutter. M. (1970). Autistic children: Infancy to adulthood. Seminars in Psychiatry, 2,
435–450.
143
Seltzer, M. M., Krauss, M. W., Shattuck, P. T., Orsmond, G., Swe, A., & Lord, C.
(2003). The symptoms of autism spectrum disorders in adolescence and
adulthood. Journal of Autism & Developmental Disorders, 33, 565-581.
Seltzer, M. M., Shattuck, P., Abbeduto, L., , & Greenberg, J. S. (2004). Trajectory of
Development in Adolescents and Adults with Autism. Mental Retardation and
Developmental Disabilities Research Reviews, 10(4), 234-247.
Shaffer, D., Fisher, P., Lucas, C. P., Dulcan, M. K., Schwab-Stone, M. E. (2000). NIMH
diagnostic interview schedule for children version IV (NIMH DISC-IV):
Description, differences from previous versions, and reliability of some common
diagnoses. Journal of the American Academy of Child & Adolescent Psychiatry,
39, 28–38.
Shattuck, P. T., Seltzer, M. M., Greenberg, J. S., Orsmond, G. I., Bolt, D., Kring, S.,
Lounds, J.,Lord, C. (2007). Change in autism symptoms and maladaptive
behaviors in adolescents and adults with an autism spectrum disorder. Journal of
Autism and Developmental Disorders, 37(9), 1735-1747.
Sigman, M. & Ruskin, E. (1999). Continuity and change in the social competence of
children with autism, down syndrome, and developmental delays. Monographs of
the Society for Research in Child Development, 64, 1–114.
Sigman, M. (1998). The Emanuel Miller Memorial Lecture 1997: Change and continuity
in the development of children with autism. Journal of Child Psychology and
Psychiatry, 39, 817–827.
Simonoff, E., Pickles, A., Charman, T., Chandler, S., Loucas, T., & Baird, G. (2008).
Psychiatric disorders in children with autism spectrum disorders: Prevalence,
comorbidity, and associated factors in a population-derived sample. Journal of
the American Academy of Child & Adolescent Psychiatry, 47(8), 921-929.
Singh, A. N., Matson, J. L., Cooper, C. L., Dixon, D., & Sturmey, P. (2005). The use of
risperidone among individuals with mental retardation: Clinically supported or
not? Research in Developmental Disabilities, 26, 203–218.
Sinzig, J., Walter, D., & Doepfner, M. (2009). Attention deficit/hyperactivity disorder in
children and adolescents with autism spectrum disorder: Symptom or syndrome?
Journal of Attention Disorders, 13, 117-126.
Sparrow, S., Balla, D., Cicchetti, D. (1984) Vineland Adaptive Behavior Scales (Survey
Form). American Guidance Service. Circle Pines: Minn.
144
Starr, E., Szatmari, P., Bryson, S., Zwaigenbaum, L. (2003) Stability and change among
high-functioning children with pervasive developmental disorders: a 2-year
outcome study. Journal of Autism & Developmental Disorders, 33, 15–22.
Steinhausen, H. C., Metzke, C. (2004). Differentiating the behavioural profile in autism
and mental retardation and testing of a screener. European Child and Adolescent
Psychiatry, 13, 214-220.
Stevenson, J., & Richman, N. (1978). Behaviour, language and development in three year
old children. Journal of Autism and Childhood Schizophrenia, 8, 299–313.
Strain, P. S., Hoyson, M., & Jamieson, B. (1985). Normally developing preschoolers as
intervention agents for autistic-like children: Effects on class deportment and
social interaction. Journal of the Division for Early Childhood, Spring, 105–15.
Sutera, S., Pandey, J., Esser, E. L., Rosenthal, M. A., Wilson, L. B., Barton, M., Green.
J., Hodgson. S., Robins, D.L., Dumont-Mathieu, T., Fein, D. (2007). Predictors of
optimal outcome in toddlers diagnosed with autism spectrum disorders. Journal of
Autism and Developmental Disorders, 37(1), 98-107.
Szatmari, P., Barolucci, G., Bremmer, R., Bond, S., & Rich, S. (1989). A follow-up study
of high-functioning autistic children. Journal of Autism & Developmental
Disorders, 19, 213-225.
Szatmari, P., Bryson, S. E., Boyle, M. H., Streiner, D. L., & Duku, E. (2003). Predictors
of outcome among high functioning children with autism and Asperger
syndrome. Journal of Child Psychology and Psychiatry, 44(4), 520-528.
Szatmari, P., Bryson, S. E., Streiner, D. L., Wilson, F., Archer, L., & Ryerse, C. (2000).
Two-year outcome of preschool children with autism or Asperger's syndrome.
The American Journal of Psychiatry, 157(12), 1980-1987.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New
York: Harper Collins.
Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Needham
Heights, MA: Allyn & Bacon.
Tantam, D. (1991). Asperger syndrome in adulthood. In U. Frith (Ed.), Autism and
Asperger syndrome (pp. 147–183). Cambridge: Cambridge University Press.
Tasse´, M. J., Aman, M. G., Hammer, D., & Rojahn, J. (1996). The Nisonger Child
Behavior Rating Form: Age and gender effects and norms. Research in
Developmental Disabilities, 17, 59–75.
145
Tasse´, M. J., Morin, I. N., & Girouard, N. (2000). French Canadian translation and
validation of the Nisonger child behavior rating form. Canadian Psychology, 41,
116–123.
Taylor, E., & Rutter, M. (2000). Classification: conceptual issues and substantive
findings. In M Rutter & E, Taylor (Eds.), Child Psychiatry: Modern Approaches
(4th ed. pp. 3-17), Oxford: Blackwell Scientific.
Taylor, J. L., & Seltzer, M. M. (2010). The transition to adulthood for individuals with
ASD and their families. In P. Howlin (Chair), What really matters: Measuring
outcome and addressing the needs of adolescents and adults with ASD. Invited
Educational Symposium conducted at the International Meetings for Autism
Research, Philadelphia, PA.
Tonge, B. J., & Einfeld, S. L. (2003). Psychopathology and intellectual disability: The
Australian child to adult longitudinal study. In L. M. Glidden (Eds.), International
review of research in mental retardation (vol. 26, pp. 61–91). San Diego, CA:
Academic Press.
Tonge, B., & Einfeld, S. (2000). The trajectory of psychiatric disorders in young people
with intellectual disabilities. Australian & New Zealand Journal of Psychiatry,
34, 80–84.
Tsatsanis, K. D. (2003). Outcome research in Asperger syndrome and autism. Child and
Adolescent Psychiatric Clinics of North America, 12(1), 47-63.
Turner, L.M., Stone, W. L., Pozdol, S.L. & Coonrod, E.E. (2006). Follow-Up of Children
with Autism Spectrum Disorders from Age 2 to Age 9. Autism 10(3), 243–265.
Venter, A., Lord, C. & Schopler, E . (1992). A follow-up study of high functioning
autistic children. Journal of Child Psychology and Psychiatry,33, 489–507.
Vieillevoya, S., & Nader-Grosbois, N. (2008). Self-regulation during pretend play in
children with intellectual disability and in normally developing children. Research
in Developmental Disabilities, 29, 247–255.
Volkmar, F., Cicchetti, D. V., Dykens, E., Sparrow, S. S., Leckman, J. F., Cohen, D. J.
(1988). An evaluation of the Autism Behavior Checklist. Journal of Autism and
Developmental Disorders, 18, 81–98.
Volkmar, F. R., & Nelson, D. S. (1990). Seizure disorders in autism. Journal of American
Academy of Child and Adolescent Psychiatry, 29, 127–129.
146
Weisbrot, D. M., Gadow, K. D., DeVincent, C. J., & Pomeroy, J. (2005). The
presentation of anxiety in children with pervasive developmental disorders.
Journal of Child and Adolescent Psychopharmacology, 15, 477–496.
Weller, E. B., Weller, R. A., Teare, M., & Fristad, M. A. (1999). Parent Version
Children’s Interview for Psychiatric Syndromes (P-ChIPS). Washington:
American Psychiatric Press.
Wechsler, D. (2002). Wechsler Preschool and Primary Scale of Intelligence – third
edition. San Antonio, TX: Psychological Corp.
Wechsler, D. (2004). The Wechsler Intelligence Scale for Children—Fourth Edition.
London: Pearson Assessment.
White, S. W., Scahill, L., Klin, A., Koenig, K., & Volkmar, F. R. (2007). Educational
placements and service use patterns of individuals with autism spectrum
disorders. Journal of Autism and Developmental Disorders, 37(8), 1403-1412.
Wing, L. (1982). Language, social, and cognitive impairments in autism and severe
mental retardation. Journal of Autism and Developmental Disorders, 11, 31–44.
Witwer, A., & Lecavalier, L. (2005). Treatment Incidence and Patterns in Children and
Adolescents with Autism Spectrum Disorders. Journal of Child and Adolescent
Psychopharmacology, 15(4), 671-681.
Witwer, A., & Lecavalier, L. (2010). Validity of Comorbid Psychiatric Disorders in
Youngsters with Autism Spectrum Disorders. Journal of Developmental and
Physical Disabilities, 22(4), 367-380.
World Health Organization. (1992). The ICD-10 classification of mental and behavioral
disorders: Clinical descriptions and diagnostic guidelines. Geneva: World Health
Organization.
Wozniak, J., Biederman, J., Faraone, S. V., Frazier, J., Kim, J., Millstein, R., et al.
(1997). Mania in children with pervasive developmental disorder revisited.
Journal of American Academy of Child Adolescent Psychiatry, 36, 1552–1560.
Zandt, F., Prior, M., & Kyrios, M. (2007). Repetitive behaviour in children with high
functioning autism and obsessive compulsive disorder. Journal of Autism and
Developmental Disorders, 37, 251–259.
Zimmerman, I. L., Steiner, V. G., & Pond, R. E. (2002). Preschool Language Scale (4th
ed.). San Antonio, TX: The Psychological Corporation.
147
Appendix A: Tables and Figures
148
Table 1. Comparative Analysis Between Participants and Non-participants
Participants
(n = 143)
Age (months), M (SD)
62.09 (18.89)
Non-participants
(n = 199)
χ2/F
p
59.99 (19.53)
0.986
0.322
0.224
0.636
3.320
0.190
149
Gender, n (%)
Male
Female
109 (76.2)
34 (23.8)
156 (78.4)
43 (21.6)
ASD subtype, n (%)
Autism
Asperger’s Disorder
PDD-NOS
78 (54.5)
44 (30.8)
21 (14.7)
128 (64.3)
48 (24.1)
23 (11.6)
17.90 (6.86)
10.74 (4.30)
15.92 (5.44)
3.48 (2.07)
8.20 (4.37)
6.22 (3.23)
19.07 (6.02)
9.17 (3.39)
16.88 (4.86)
2.96 (3.88)
9.09 (6.24)
5.74 (2.05)
2.755
4.396
2.178
2.164
2.162
3.323
0.098
0.067
0.141
0.142
0.153
0.069
80.15 (15.75)
78.07 (14.65)
2.355
0.126
69.95 (9.92)
72.66 (23.86)
1.455
0.229
NCBRF, M (SD)
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly sensitive
General language
composite, M (SD)
Pragmatic language
composite, M (SD)
149
Table 2. Comparative Analysis Between Participants from the ASD Clinic and FDD Clinic
150
ASD Clinic
(n = 122)
FDD Clinic
(n = 21)
χ2/F
Age (months), M (SD)
66.24 (17.76)
41.79 (4.10)
32.21
< .001
Gender, n (%)
Male
Female
–
93 (76.2)
29 (23.8)
–
17 (81.0)
4 (19.0)
–
–
–
0.783 a
–
–
ASD subtype, n (%)
Autism
Asperger’s Disorder/
PDD-NOS
–
60 (49.2)
62 (50.8)
–
18 (85.7)
3 (14.3)
–
–
0.002 a
–
17.61 (6.86)
10.61 (4.32)
15.87 (5.70)
3.45 (2.05)
8.22 (4.59)
6.18 (3.30)
19.57 (5.83)
11.52 (4.21)
17.09 (3.67)
3.66 (2.19)
8.09 (2.81)
6.76 (2.81)
1.46
0.79
0.89
0.17
0.01
0.57
0.229
0.373
0.347
0.672
0.903
0.448
NCBRF, M (SD)
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly sensitive
General language
composite, M (SD)
Pragmatic language
composite, M (SD)
a
p
81.54 (16.47)
75.14 (12.34)
2.88
0.092
69.72 (10.11)
71.23 (8.85)
0.41
0.522
p value given by Fisher’s Exact Test
150
Table 3. Sample Characteristics at Time 1
Gender
Male
Female
Age (months)
Range
Ethnicity
Caucasian
African American
Other
ASD subtype
Autistic disorder
Asperger’s disorder
PDD–NOS
NCBRF
Conduct
Insecure/Anxious
Hyperactivity
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
ADI–R
Reciprocal Social Interaction
Communication
Restricted, Repetitive
Behaviors and Interests (RRB)
N (%)
Mean (SD)
109 (76.2%)
34 (23.8%)
—
—
—
—
62.65 (18.61)
(36 – 139)
112 (78.3%)
21 (14.7%)
10 (7.0%)
—
—
—
78 (54.5%)
44 (30.8%)
21 (14.7%)
—
—
—
—
—
—
—
—
—
17.90 (6.87)
10.75 (4.30)
15.94 (5.44)
3.48 (2.07)
8.20 (4.37)
6.26 (3.23)
—
—
—
16.41 (4.41)
13.0 (3.9)
4.76 (1.7)
General Language Composite
> 70
< 70
—
89 (62.2%)
54 (37.8%)
80.60 (16.05)
—
—
Pragmatic Language Composite
> 70
< 70
—
74 (51.7%)
69 (48.3%)
69.95 (9.92)
—
—
151
Table 4. Associations Between Categorical Variables and NCBRF Problem Behavior Subscale Scores at T1
Conduct
Insecure/
Anxious
Hyperactive
Self-injury
Stereotypic
Self-isolated
Ritualistic
Overly
Sensitive
18.03 (6.7)
17.50 (7.3)
.15 (.69)
10.16 (4.3)
12.65 (5.5)
9.18 (.003)
15.93 (5.4)
15.94 (5.5)
.00 (.99)
3.77 (2.1)
2.58 (1.9)
8.94 (.003)
8.40 (4.3)
7.56 (4.5)
.96 (.33)
6.19 (3.2)
6.50 (3.3)
.23 (.63)
10.69 (4.4)
10.54 (4.7)
11.38 (2.6)
0.28 (0.76)
19.50 (3.7)
12.72 (3.1)
9.42 (4.4)
29.27 (<.001)a
3.66 (2.2)
3.13 (1.9)
3.57 (1.8)
0.94 (0.39)
8.12 (4.4)
7.82 (3.9)
9.33 (4.9)
.89 (0.42)
6.46 (3.2)
6.02 (3.3)
6.26 (3.2)
0.31 (0.73)
8.6 (4.2)
7.5 (4.6)
2.26 (.58)
6.5 (3.3)
5.8 (2.9)
1.69 (.19)
n
M (SD)
Gender
Male
Female
F (p)
152
ASD subtype
Autism
Asperger’s
PDD-NOS
F (p)
109
34
78
44
21
20.92 (6.2)
15.70 (6.1)
11.29 (3.7)
26.58 (<.001) a
General Language
> 70
89
< 70
54
F (p)
16.55(6.3)
20.13(7.2)
9.68(.002)
11.91 (4.4)
8.83 (3.4)
19.39 (< .001)
14.49 (5.3)
18.31 (4.8)
18.62 (<.001)
3.42 (1.9)
3.61 (2.2)
.29(.58)
Pragmatic Language
> 70
74
< 70
69
F (p)
16.59(6.4)
19.30(7.1)
5.74(.018)
11.78 (4.3)
9.63 (4.1)
9.40 (.003)
14.72 (5.4)
17.24 (5.2)
8.11 (.005)
3.51(1.9)
3.46(2.2)
.02 (.89)
a
8.6 (3.7)
7.7 (4.9)
1.51 (.22)
Post hoc Tukey tests all groups to be significantly different from one another on Conduct and Hyperactive subscales
152
6.4 (3.2)
6.1 (3.4)
.71 (.40)
Table 5. Correlations Between Continuous Variables and NCBRF Problem Behavior Subscale Scores at T1
Conduct
NCBRF subscales
Conduct
1.00
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
153
Age (months)
ADI-R
RSI
Communication
RRB
- 0.302**
0.403**
0.479**
0.137
Insecure/
Anxious
- 0.001
1.00
0.070
- 0.086
- 0.035
0.081
Hyperactive
Self-injury/
Stereotypic
Self-isolated/
Ritualistic
Overly
Sensitive
0.669**
0.023
1.00
0.073
0.013
0.038
1.00
0.041
0.151
- 0.006
0.201*
1.00
0.037
0.166*
0.144
- 0.049
0.078
1.00
- 0.475**
-0.075
0.077
0.094
0.550**
0.559**
0.136
0.054
- 0.032
0.092
0.039
- 0.013
0.067
0.038
0.045
0.000
** significant at 0.01 level, * significant at 0.05 level
RSI = Reciprocal Social Interaction; RRB = Restricted, Repetitive Behaviors and Interests
153
Table 6. Difference in Demographic and Assessment Variables based on ASD Subtypes at Time 1
154
Asperger’s
n = 44
Age (months)
Mean (SD)
Range
51.37 (11.19)
36 – 82
75.59 (19.56)
52 – 139
73.59 (15.52)
49 – 106
F = 44.71
< .001a
Gender, n (%)
Male
Female
59 (75.6)
19 (24.4)
33 (75.0)
11 (25.0)
17 (81.0)
4 (19.0)
χ2 = 0.31
0.856
Ethnicity, n (%)
Caucasian
Minority1
65(83.3)
13(16.7)
34 (77.3)
10 (22.7)
13 (61.9)
8 (38.1)
χ2 = 4.52
0.105
ADI-R, Mean (SD)
RSI
Communication
RRB
18.94 (3.92)
14.14 (3.02)
4.80 (1.51)
12.42 (2.46)
7.75 (1.46)
5.34 (1.79)
11.10 (1.97)
7.10 (1.58)
3.28 (1.42)
F = 44.64
F = 47.42
F = 11.91
< .001a
< .001a
< .001b
41(93.2)
3 (6.8)
16 (76.2)
5 (23.8)
χ2 = 34.59
General Language, n (%)
> 70
32(41.0)
< 70
46(59.0)
PDD-NOS
n = 21
F/χ2
Autism
n = 78
p
< .001
continued
Minority includes individuals in “African American” and “Other” categories
a
Autism significantly different from Asperger’s and PDD-NOS; b PDD-NOS significantly different from Autism and Asperger’s
RSI = Reciprocal Social Interaction; RRB = Restricted, Repetitive Behaviors and Interests
1
154
Table 6, Continued
Autism
n = 78
Pragmatic Language, n (%)
> 70
26 (33.3)
< 70
52 (66.7)
1
a
Asperger’s
n = 44
PDD-NOS
n = 21
34 (77.3)
10 (22.7)
14 (66.7)
7 (33.3)
F/χ2
χ2 = 23.94
p
< .001
Minority includes individuals in “African American” and “Other” categories
Autism significantly different from Asperger’s and PDD-NOS; b PDD-NOS significantly different from Autism and Asperger’s
RSI = Reciprocal Social Interaction; RRB = Restricted, Repetitive Behaviors and Interests
155
155
Table 7. Sample Characteristics at Time 2
N (%)
Age (months)
Range
M (SD)
—
—
124.64 (31.98)
(65 – 204)
99 (69.2)
44 (30.8)
10.38 (2.66)
(5 – 17)
—
—
NCBRF
Conduct (16 items)
Insecure/Anxious (15 items)
Hyperactivity (9 items)
Self-injury/Stereotypic (7 items)
Self-isolated/Ritualistic (8 items)
Overly Sensitive (5 items)
—
—
—
—
—
—
16.2 (7.84)
12.8 (5.17)
13.9 (8.30)
2.8 (2.09)
11.7 (5.35)
7.3 (3.21)
Psychiatric comorbidity
Anxiety Disorder (including OCD)
ADHD
Disruptive Behavior Disorder
Depressive Disorder
Other a
Any psychiatric comorbidity
54 (37.8)
45 (31.5)
26 (18.2)
17 (11.9)
5 (3.45)
98 (68.5)
—
—
—
—
—
—
Psychotropic medications
Antidepressants
Stimulants
Antipsychotics
Anxiolytics
Alpha agonists
Anticonvulsant/mood stabilizer b
Any psychotropic
34 (23.5)
32 (22.1)
25 (17.2)
19 (13.1)
12 (8.3)
9 (6.2)
76 (52.4)
—
—
—
—
—
—
—
Age (years)
Range
5 – 11 (children)
12 – 17 (adolescents)
Complementary and Alternative Medication (CAM) c
Melatonin
11 (7.6%)
Super Nu-Thera
6 (4.1%)
Fish oil
4 (2.8%)
Vitamin B6
2 (1.4%)
St. John’s wort
1
Any CAM
21 (14.5%)
—
—
—
—
—
—
continued
156
Table 7, continued
N (%)
Parent-reported current language level
compared to peers d
Very much behind
Moderately behind
Mildly behind
Not behind
Current educational placement
Regular class without
accommodations
Regular class with minimal
Accommodations
DH class
MH class
Combined class environments
ASD specific academies
Home schooled
Not available
Interventions
Speech and language therapy
Occupational and Physical Therapy
Applied Behavior Analytic (ABA)
Therapy
Any of these services
Other e
4 (2.76)
32 (22.1)
38 (26.2)
69 (47.6)
M (SD)
—
—
—
—
23(15.87)
—
26(17.94)
—
33(22.77)
9 (6.21)
15(10.35)
21(14.49)
9 (6.21)
7 (4.83)
—
—
—
—
—
—
84 (58.7)
77 (53.8)
62 (43.4)
—
—
—
114 (79.7)
28 (19.3)
—
—
a
Other psychiatric comorbidities include Tourette syndrome (n = 2), tic disorder (n = 1),
bipolar disorder (n = 1), and enuresis (n = 1).
b
Seizures currently present for 3 participants
c
The rates of CAM treatments may be underestimated, because the demographic form asked
for psychotropic medications, and it was not clear whether all informants reported
CAM/supplement use.
d
Full description of anchors:
Very much behind; uses simple phrases only, and/or relies mostly on gestures/sign language
Moderately behind; limited vocabulary, uses basic sentences, at times struggles for
understanding
Mildly behind; Use of language OK, occasionally struggles with some language concepts
Not behind; fluent with language, has acquired speech skills comparable to or better than
other people of the same age
e
Other interventions include social skills training, sensory integration, and Floortime
157
Table 8. Stability of NCBRF Problem Behavior Subscale Scores Between Time 1 and Time 2
Subscale
(number of items)
Mean (SD)
158
Time 1
Time2
Conduct (16)
17.90 (6.87)
16.21 (7.84)
Insecure/Anxious (15)
10.75 (4.30)
12.83 (5.17)
Hyperactive (9)
15.94 (5.44)
13.91 (8.30)
Self-injury/Stereotypic (7)
3.48 (2.07)
2.84 (2.09)
Self-isolated/Ritualistic (8)
8.20 (4.37)
11.70 (5.35)
Overly Sensitive (5)
6.26 (3.23)
7.32 (3.21)
Mean Change
Score
(Range)
-1.71
(-23 – +18)
2.13
(-10 – +20)
-1.97
(-16 – +19)
-0.59
(-6 – +8)
3.54
(-10 – +15)
0.98
(-9 – +11)
Change score = T2 score – T1 score
**p = .004; *** p < .001; * significant at 0.01 level
158
t
Effect size
(d)
r
2.94 **
0.23
0.564*
- 4.75 ***
- 0.44
0.376*
3.43 ***
0.28
0.538*
4.08 ***
0.34
0.650*
- 8.27 ***
- 0.71
0.463*
- 0.34
0.443*
- 3.44 ***
Table 9. Correlations Between NCBRF Problem Behavior Subscale Scores Compared Based on Median Split of Intervening Time
Subscale
2-4 years (n = 68)
r
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
0.538**
0.516**
0.643**
0.765**
0.499**
0.648**
5-8 years (n = 75)
r
0.597**
0.219
0.391**
0.585**
0.443**
0.386**
159
** significant at .01 level
Note. Median number of years passed between T1 and T2 was 5.
159
Table 10. Improvement, Worsening, and No Change in NCBRF Problem Behavior Subscale Scores at T2 Relative to T1
Subscale
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
Improved
No change
Worsened
n (%) a
n (%) b
n (%) c
49 (34.3)
28 (19.6)
25 (17.5)
97 (67.8)
27 (18.9)
56 (39.2)
32 (22.4)
82 (58.0)
30 (21.0)
13 (9.1)
95 (66.4)
58 (40.6)
62 (43.4)
32 (22.4)
88 (61.5)
33 (23.1)
21 (14.7)
29 (20.3)
160
Note. NCBRF subscales are scored such that higher scores indicate more problem behaviors; a decrease over time indicates a decline
in problem behavior
a
T2 score was lower than T1 score by more than ½ SD of T1 mean
b
T2 score was within ± ½ SD of T1 mean
c
T2 score was higher than T1 score by more than ½ SD of T1 mean
160
Table 11. Difference in NCBRF Change Scores Based on ASD Subtype
Autism (1)
n = 78
Asperger’s (2)
n = 44
PDD-NOS (3)
n = 21
F
p
Source of
Significance
NCBRF subscales
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
- 2.00
- 0.09
-1.61
-0.38
1.70
-0.18
-1.8 1
5.16
-2.75
-0.86
6.40
2.59
-0.38
4.00
-1.66
-0.81
4.33
1.90
0.458
18.890
0.404
1.257
14.522
11.783
.634
<.001
.668
.288
<.001
<.001
—
1:2, 1:3
—
—
1:2
1:2, 1:3
161
Note. NCBRF subscales are scored such that higher scores indicate more problem behaviors; negative change scores indicates a
decline in problem behavior
161
Table 12. Overview of Hierarchical Multiple Linear Regression (HMLR) Analyses Predicting NCBRF Problem Behavior
Subscale Scores at T2
162
Predictors
Conduct
Block 1
R2
∆ R2
F (p)
.041
—
1.869 (.138)
Block 2
R2
∆ R2
F (p)
.418
.377
5.462 (<.001)
Block 3
R2
∆ R2
F (p)
.462
.044
2.357 (.058)
Insecure/
Anxious
Hyperactive
Self-injury/
Stereotypic
.152
—
7.882 (<.001)
.231
—
13.226 (<.001)
.030
—
1.379 (.252)
.435
.283
4.321 (<.001)
.399
.168
2.361 (.006)
.503
.473
8.011 (<.001)
.414
.015
.729 (.574)
.516
.013
.750 (.560)
.449
.014
.722 (.579)
162
Self-isolated/
Ritualistic
.185
—
10.016 (<.001)
.483
.297
4.845 (<.001)
.500
.018
1.007 (.407)
Overly
Sensitive
.083
—
3.999 (.009)
.367
.284
3.784 (<.001)
.450
.083
4.301 (.003)
Table 13. HMLR Analysis Predicting NCBRF Conduct at T2
B
SE(B)
p
      
163
Block 1
Age (months)
Gender (Male)
Ethnicity
-.067
1.815
-.261
.031
1.581
1.624
.034
.253
.873
-.184
.098
-.014
Block 2
Age (months)
Gender (Male)
Ethnicity
.087
1.353
-1.521
.040
1.441
1.478
.030
.350
.305
.240
.073
-.080
ASD subtype
Asperger’s
PDD-NOS
-7.231
-8.605
3.084
3.157
.021
.007
-.420
-.389
.650
-.055
-.033
.000
.054
.262
.112
.160
.181
.293
.139
.185
<.001
.734
.857
.999
.702
.158
.567
-.030
-.023
.000
.030
-.109
ADI-R
Reciprocal Social Interaction
-.176
Communication
-.243
Restricted Repetitive Behaviors -.356
.225
.253
.377
.436
.340
.347
-.090
-.115
-.078
Language composite
General Language
Pragmatic Language
.063
.073
.792
.365
-.033
.085
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.017
.066
163
Table 14. HMLR Analysis Predicting NCBRF Insecure/Anxious at T2
B

SE(B)
p
  
164
Block 1
Age (months)
Gender (Male)
Ethnicity
.095
-.384
.102
.020
.992
1.019
.000
.699
.920
.391
-.031
.008
Block 2
Age (months)
Gender (Male)
Ethnicity
.002
.527
-.518
.026
.946
.971
.947
.578
.595
.007
.043
-.041
ASD subtype
Asperger’s
PDD-NOS
6.906
6.265
2.026
2.074
.001
.003
.602
.424
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.126
.480
-.055
.380
-.070
.111
.074
.105
.119
.192
.092
.121
.089
<.001
.647
.050
.449
.361
.165
.397
-.057
.152
-.058
.069
ADI-R
Reciprocal Social Interests
Communication
Restricted Repetitive Behaviors
.086
-.007
-.082
.148
.166
.248
.563
.967
.741
.066
-.005
-.027
Language composite
General Language
Pragmatic Language
-.037
.036
.041
.048
.369
.448
-.111
.070
164
Table 15. HMLR Analysis Predicting NCBRF Hyperactive at T2

B
SE(B)
p
   
165
Block 1
Age (months)
Gender (Male)
Ethnicity
-.185
.454
.663
.030
1.516
1.557
.000
.765
.671
-.477
.023
033
Block 2
Age (months)
Gender (Male)
Ethnicity
-.090
.884
-.210
.043
1.567
1.607
.039
.574
.896
-.231
.045
-.010
ASD subtype
Asperger’s
PDD-NOS
2.226
.701
3.354
3.433
.508
.838
.121
.030
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.158
.187
.406
-.094
-.005
-.019
.122
.175
.197
.318
.152
.201
.197
.287
.041
.768
.972
.926
.129
.096
.264
-.023
-.003
-.007
ADI-R
Reciprocal Social Interaction
Communication
Restricted Repetitive Behaviors
.187
.006
-.207
.245
.276
.410
.446
.982
.615
.089
.003
-.042
Language composite
General Language
Pragmatic Language
-.140
.068
.068
.079
.042
.393
-.262
.082
165
Table 16. HMLR Analysis Predicting NCBRF Self-injury/Stereotypic at T2
B

SE(B)
p
   
Age (months)
Gender (Male)
Ethnicity
-.008
.776
.074
.008
.428
.440
.332
.072
.866
-.084
.156
.015
Block 2
Age (months)
Gender (Male)
Ethnicity
.011
.218
-.406
.010
.358
.368
.265
.545
.272
.113
.044
-.080
ASD subtype
Asperger’s
PDD-NOS
-1.115
-.874
.767
.785
.149
.268
-.241
-.147
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
. 029
.080
-.046
.633
.011
.028
.028
.040
.045
.073
.035
.046
.302
.046
.310
<.001
.754
.543
.094
.165
-.119
.628
.023
-.043
ADI-R
Reciprocal Social Interaction
Communication
Restricted Repetitive Behavior
.019
.012
.046
.056
.063
.094
.729
.846
.626
.037
.022
.037
Language composite
General Language
Pragmatic Language
.004
.002
.016
.018
.800
.979
.029
.002
166
Block 1
166
Table 17. HMLR Analysis Predicting NCBRF Self-isolated/Ritualistic at T2
B

SE(B)
p
 
Age (months)
Gender (Male)
Ethnicity
.097
1.740
-.665
.020
.999
1.026
.000
.084
.518
.392
.137
-.051
Block 2
Age (months)
Gender (Male)
Ethnicity
.036
.592
.029
.026
.931
.955
.163
.526
.976
.144
.047
.002
ASD subtype
Asperger’s
PDD-NOS
3.535
2.623
1.993
2.040
.079
.201
.300
.173
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.001
-.189
.081
.370
.566
.209
.072
.104
.117
.189
.090
.119
.986
.071
.488
.053
< .001
.082
.002
-.152
.083
.144
.462
.127
ADI-R
Reciprocal Social Interaction
Communication
RRB
.083
-.092
.122
.145
.164
.244
.568
.574
.617
-.062
-.064
.039
Language composite
General Language
Pragmatic Language
-.010
.018
.040
.047
.806
.697
-.029
.034
167
Block 1
167
Table 18. HMLR Analysis Predicting NCBRF Overly Sensitive at T2
Age (months)
Gender (Male)
Ethnicity
Block 2
Age (months)
Gender (Male)
Ethnicity
-.001
.216
.297
.017
.606
.622
.945
.722
634
-.008
.029
.039
ASD subtype
Asperger’s
PDD-NOS
1.683
1.641
1.298
1.329
.197
.219
.242
.184
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/ Ritualistic
Overly Sensitive
.002
-.056
-.093
-.073
-.044
.508
.047
.068
.076
.123
.059
.078
.960
.410
.225
.556
.451
<.001
.005
-.076
-.160
-.048
-.062
.522
ADI-R
Reciprocal Social Interaction
Communication
Restricted Repetitive Behaviors
-.038
.103
.239
.095
.107
.159
.686
.337
.134
-.048
.120
.130
Language composite
General Language
Pragmatic Language
.022
-.039
.026
.031
.407
.208
.109
-.124
continued
168
Block 1
168
SE(B)
.012
.624
.641
p
.001
.881
.748
 
B
.043
-.094
.206
.291
-.013
.027
Table 18, Continued
B

Block 3
SE(B)
p
  
169
Age (months)
Gender (Male)
Ethnicity
-.008
.225
-.138
.016
.578
.603
.627
.698
.819
-.053
.030
-.018
ASD subtype
Asperger’s
PDD-NOS
1.865
1.201
1.246
1.282
.137
.351
.268
.134
T1 NCBRF
Conduct
Insecure Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.003
-.058
-.073
.028
.088
.593
.045
.065
.073
.120
.057
.078
.940
.370
.317
.816
.127
< .001
.007
-.080
-.126
.018
-.122
.609
.021
.018
.165
.091
.105
.153
.817
.865
.282
.027
.021
.089
.020
-.006
.026
.031
.450
.841
.099
-.020
.048
.012
<.001
.340
ADI-R
Reciprocal Social Interaction
Communication
Restricted Repetitive Behaviors
Language composite
General Language
Pragmatic Language
Time lag between T1 and T2
continued
169
Table 18, Continued
B

Interventions
Speech and Language
OT/PT
ABA
SE(B)
-.073
.172
.367
.570
.536
.470
OT/PT = Occupational Therapy/Physical Therapy; ABA = Applied Behavioral Analytic Therapy
170
170
p
.899
.749
.437
 
-.011
.027
.057
Table 19. Associations between T2 NCBRF Subscale Scores and Psychotropic Medication Use at T2
Psychotropic
Medication
Use
Conduct
n
M (SD)
Insecure/
Anxious
M (SD)
Hyperactive
Self-injury/
Stereotypic
Self-isolated/
Ritualistic
M (SD)
M (SD)
M (SD)
Overly
Sensitive
M (SD)
Yes
76
18.16 (7.65)
13.87 (5.57)
15.83 (8.82)
3.05 (2.31)
12.30 (5.30)
7.23 (3.22)
No
67
14.25 (7.58)
11.88 (4.57)
12.12 (7.37)
2.73 (1.84)
11.16 (5.39)
7.25 (3.23)
0.85 (.358)
1.61 (.206)
0.25 (.974)
F (p)
9.45 (.003)
5.42 (.021)
7.43 (.007)
171
171
Table 20. Association between T2 NCBRF Subscale Scores and Educational Placements at T2
Educational
Placement
n
Conduct
Insecure/
Anxious
M (SD)
M (SD)
Hyperactive
M (SD)
Self-injury/
Stereotypic
M (SD)
Self-isolated/
Ritualistic
M (SD)
Overly
Sensitive
M (SD)
Regular
49
14.96 (8.13)
14.37 (5.22)
11.34 (6.84)
2.86 (1.91)
12.54 (5.31)
7.74 (3.24)
Restrictive
78
17.65 (7.40)
11.13 (4.62)
17.03 (8.82)
2.94 (2.24)
10.80 (5.52)
6.66 (3.21)
4.23 (.042)
15.50 (<.001)
18.77 (<.001)
.054 (.816)
3.83 (.052)
4.04 (.046)
F (p)
172
Note. “Regular” includes placement in regular education class with no accommodations and regular education class with minimum
accommodation (e.g., aide services for part of day/tutoring services). “Restrictive” includes placement in a developmental
handicapped class, multihandicapped class, a combination of classroom environments (partial inclusion), and placement in a
specialized academy designed to meet educational needs of students with ASDs and other DDs.
172
Table 21. Logistic Regression Predicting Comorbid Anxiety Disorder at T2
Predictor
B
SE(B)
Wald χ2
p
Odds ratio
95% C.I. for odds ratio
173
Age
-.006
.014
.166
.684
.994
.967-1.022
Gender
-.792
.474
2.789
.095
.453
.179-1.147
ASD Subtype
Autism
Asperger’s
-.008
.082
.856
.655
.043
.130
.836
.719
.846
1.251
.188-5.400
.300-3.921
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.084
.187
.017
.089
.023
.028
.040
.061
.062
.102
.048
.062
4.283
9.318
.041
.702
.322
.202
.039
.002
.839
.402
.570
.653
.920
1.206
1.012
1.089
1.027
1.028
.850-.996
1.069-1.360
.902-1.148
.895-1.335
.931-1.124
.910-1.162
General Language
.013
.020
.446
.504
1.013
.975-1.054
173
Table 22. Logistic Regression Predicting Comorbid ADHD at T2
Predictor
SE(B)
Wald χ2
p
.028
.020
1.943
.163
1.029
Gender
-.591
.619
.912
.339
.554
.165-1.862
ASD Subtype
Autism
Asperger’s
.299
-.766
1.231
1.200
.059
.408
.808
.523
1.348
.465
.121-5.048
.044-4.885
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.065
.020
.514
.013
-.041
.041
.051
.069
.112
.129
.061
.084
1.654
.084
11.080
.010
.454
.235
.198
.773
<.001
.921
.501
.628
.937
1.020
1.672
1.013
.960
1.042
.849-1.035
.892-1.167
1.343-2.082
.786-1.305
.852-1.081
.883-1.228
.023
.027
.769
.381
1.024
.972-1.079
B
Age
174
General Language
174
Odds ratio
95% C.I. for odds ratio
.989-1.070
Table 23. Logistic Regression Predicting Comorbid Depressive Disorder at T2
Predictor
SE(B)
Wald χ2
-.004
.019
.039
.844
.996
.960-1.034
.728
.741
.966
.326
2.072
.485-8.852
ASD Subtype
Autism
Asperger’s
1.205
1.420
1.354
1.090
.792
1.697
.374
.193
3.337
4.139
.235-7.407
.488-5.073
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.036
.219
-.215
-.160
-.038
.312
.061
.085
.106
.157
.078
.104
.338
5.734
4.121
1.033
.238
7.933
.561
.009
.042
.310
.625
.003
1.036
1.145
.807
.852
.962
1.166
.919-1.168
1.055-1.470
.656-.993
.627-1.160
.825-1.122
1.113-1.676
General Language
-.035
.030
1.303
.254
.966
.910-1.025
B
Age
Gender
175
175
p
Odds ratio
95% C.I. for odds ratio
Table 24. Logistic Regression Predicting Restrictive Educational Placement at T2
Predictor
B
SE(B)
Wald χ2
p
Odds ratio
95% C.I. for odds ratio
176
Age
-.015
.017
.864
.353
.985
.953-1.017
Gender
.195
.501
.152
.696
1.216
.456-3.243
ASD Subtype
Autism
Asperger’s
-1.256
-.327
.932
.722
1.817
.206
.178
.650
.285
.721
.046-1.769
.175-2.968
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.138
.043
.026
-.115
-.018
-.070
.043
.057
.064
.107
.050
.068
6.460
.580
.165
1.158
.133
1.081
.001
.446
.684
.282
.715
.298
1.148
1.044
1.026
.892
.982
.932
1.056-1.249
.934-1.167
.906-1.163
.723-1.099
.891-1.082
.817-1.064
General Language
-.061
.021
4.559
.003
.941
176
.903-.980
Table 25. Summary of Significant Predictors of T2 NCBRF Outcomes
Predictor s
Conduct
Insecure/
Anxious
Hyperactive
Self-injury/
Stereotypic
Self-isolated/
Ritualistic
Overly
Sensitive
177
NCBRF subscales
Conduct
Insecure/Anxious
Hyperactive
Self-injury/ Stereotypic
Self-isolated/ Ritualistic
Overly Sensitive
Y↑
—
—
—
—
—
—
Y↑
—
Y↑
—
—
—
—
Y↑
—
—
—
—
Y↑
—
Y↑
—
—
—
—
—
—
Y↑
—
—
—
—
—
—
Y↑
ASD subtype
Autism
Asperger’s disorder
PDD–NOS
—
Y↓
Y↓
—
Y↑
Y↑
—
—
—
—
—
—
—
—
—
—
—
—
T1 age
T1 language ability
Y↑
—
—
—
Y↓
Y↓
—
—
—
—
—
—
ADI–R domains
—
—
—
—
—
—
Intervening time between
T1 and T2
—
—
—
—
—
Y↑
Type of interventions
received a
—
—
—
—
—
—
Note. Y indicates the NCBRF outcome; arrows (↑↓) indicate the nature of relationship (direct/inverse)
A “–” indicates that the variable was not a significant predictor of the outcome
a
Interventions received includes Speech and Language, Occupational Therapy/Physical Therapy, and Applied Behavioral Analytic
Therapy
177
Affective Disorders (n = 64)
40
5
17
2
14
5
21
ADHD (n = 45)
Disruptive Behavior Disorder
(n = 26)
Figure 1. Overlap Between Psychiatric Comorbid Conditions at T2
Affective Disorders denote anxiety disorders (n = 54) and depressive disorder
(n = 17). Seven participants had both anxiety and depressive disorders
178
Figure 2. Dot plot of bivariate relation between T1 NCBRF Hyperactivity scores and T2
ADHD diagnosis
On X-axis “1” indicates presence of ADHD diagnosis. The horizontal reference line indicates
a median score of 16 on Hyperactivity at T1.
179
Appendix B: Initial Letter to Potential Participants
180
Initial letter to potential participating families
LETTERHEAD
Date:
Dear Mr. and Mrs. (name of family),
This is Sherry Feinstein and Monali Chowdhury from O.S.U’s Nisonger Center. We had
worked with you when your child, (child’s name), was seen at the Nisonger Autism Spectrum
Disorder Clinic or Family-Directed Developmental Clinic in (year of visit). We hope you and
(child’s name) are doing well.
Monali is conducting a study to follow up on the behavioral status of all children, like
(child’s name), seen at the Nisonger clinics. The results of this study –“The OSU Autism
Spectrum Follow-up”–will provide important information for professionals, service
providers, and families, like you.
Monali would like to call you in the next ten days to see whether you are interested in
participating in this study. You will not need to come to the Nisonger Center to participate;
all study materials will be mailed to you. Your participation will require about 20 minutes,
and compensation will be available for your participation. Monali will give you further
details, and will answer any questions that you may have. Your participation will be very
valuable because it will allow us to learn more about behaviors associated with autism
spectrum disorders. This information, in turn, will help us to better plan for any needed
services for children like (child’s name). So we thank you for your help in advance.
Your participation is entirely voluntary. If you do not wish to participate in this study, please
call us at (614) 685 6702 (Sherry) or (614) 316 4287 (Monali) within the next ten days, and
we will not contact you. Please be assured that your decision to participate (or not) in this
study does not affect your access to services at the Nisonger Center or the Ohio State
University.
Yours sincerely,
Sherry Feinstein, M.S.
Clinical Program Manager
The Nisonger Center
The Ohio State University
Monali Chowdhury, M.A.
Psychology Graduate Associate
The Nisonger Center
The Ohio State University
181
Appendix C: Demographics Form
182
DEMOGRAPHIC FORM
(To be completed by researcher about the individual with ASD as a semi-structured phone
interview with parents/legal guardian)
1) Name of individual with ASD:
2) Does this individual have intellectual disability (“mental retardation”)?
□ Yes
□ No
□ Dont know
3) Does this individual currently have epilepsy (seizures/convulsions/fits)?
□ Yes
□ No
4) Psychiatric diagnosis:
“Yes” to questions specified under each category resulted in endorsement of that
diagnosis
□ Attention Deficit Hyperactivity Disorder (ADHD)
a) Does this individual have problems with attention, overactivity, and/or acting without
thinking? Y / N
b) Does the doctor say that he/she has Attention Deficit Hyperactivity Disorder or
ADHD? Y / N
c) Does this individual take medicine for attention problems or hyperactivity? Y / N
Diagnosis endorsed if “yes” to (b) and/or (c), in any combination with (a)
□ Disruptive Behavior Disorder (Oppositional Defiant Disorder/Conduct Disorder)
a) Is this individual defiant or hostile, and/or have problems losing temper? Y / N
b) Does this individual have problems such as starting fights, bullying, lying, or stealing?
Y/N
c) Does the doctor say that he/she has Oppositional Defiant Disorder or Conduct
Disorder? Y / N
d) Does this individual take medicine for the behaviors mentioned above? Y / N
Diagnosis endorsed if “yes” to (c) and/or (d), in any combination with (a) and/or (b)
□ Obsessive Compulsive Disorder (OCD)
a) Does this individual have repetitive behaviors, such as hand washing, hoarding things
of no value, placing things in a fixed order, and/or checking? Y / N
b) Does the doctor say that he/she has Obsessive Compulsive Disorder Y / N
c) Does this individual take medicine for obsessive and compulsive behavior? Y / N
Diagnosis endorsed if “yes” to (b) and/or (c), in any combination with (a)
□ Depressive Disorder
a) Does this individual have sad mood most of the day, nearly every day? Y / N
b) Does he/she have markedly decreased interest in activities he/she previously enjoyed?
Y/N
c) Does the doctor say the he/she has depression? Y / N
d) Does this individual take medicine for depression? Y / N
Diagnosis endorsed if “yes” to (c) and/or (d), in any combination with (a) and/or (b)
183
□ Anxiety Disorder
a) Is this individual excessively worried about certain situations? Y / N
b) Does the doctor say that he/she has Anxiety Disorder? Y / N
c) Is this individual taking medicine for anxiety? Y / N
Diagnosis endorsed if “yes” to (b) and/or (c), in any combination with (a)
□ Other
Has the doctor ever mentioned that this individual has any other mental health problem? If
yes, what condition? ________________________________________________________
5) Is this individual going to school or is s/he home schooled?
□ Going to school
□ Home schooled
6) If going to school, what is his/her current education placement?
□ Regular class without any accommodations
□ Regular class with tutoring/aide
□ Multi-handicapped/Multiple disabilities class
□ DH class (Developmentally Handicapped)
services
□ ED class (Emotionally Disturbed)
□ ASD specific academy
Please list school: ________________
□ Other (please explain) ________________
7) Does this individual use speech to express what he/she wants? □ Yes
□ No
8) If YES, please describe his/her level of speech compared to other individuals of
the same age:
□ Very much behind; uses simple phrases only, and/or relies mostly on gestures/sign
language
□ Moderately behind; limited vocabulary, uses basic sentences, at times struggles for
understanding
□ Mildly behind; Use of language OK, occasionally struggles with some language concepts
□ Not behind; fluent with language, has acquired speech skills comparable to or better than
other people the same age
9) Has this individual received any of the following while growing up? Please
check all that apply.
□ Speech & Language Therapy
Ages received: ________ Amount: ____________
□ Occupational Therapy
Ages received: ________ Amount: ____________
□ Behavior Therapy (ABA, IBI, EIBI) Ages received: ________ Amount: ____________
184
□ Other (please explain) _____________________________
10) Does this individual use any prescribed psychotropic medication:
□ Yes
□ No
11) If YES, please complete the following:
Medication name
Given for (Type of problem or diagnosis)
12) Informant:
a) Relationship to individual with ASD:
b) Age of informant:
□ 20-30 years
□ 61 years and above
□ 30-40 years
□ Not reported
c) Highest level of education:
□ Less than high school
□High school degree
□ Graduate or professional degree
185
□ 40-50 years
□ 50-60 years
□ College degree
□ Not reported
Appendix D: Cover Letter in Study Packets to Potential Participants
186
Cover letter to families
LETTERHEAD
Date:
Dear Mr. and Mrs. (name of family),
My name is Monali Chowdhury and I am a graduate student specializing in autism spectrum
disorders at the Nisonger Center at The Ohio State University. You have received this study
packet because you agreed to participate in the “OSU Autism Spectrum Follow-up” study in
our recent phone conversation. Thank you very much for taking the time to participate in this
research! You will receive $15 as a small token of our appreciation of your participation.
Please note that your continued participation is entirely voluntary. You may leave the study
at any time. If you decide to stop participating in the study, there will be no penalty to you,
and you will not lose any benefits to which you are otherwise entitled. Your decision will not
affect your future relationship with The Ohio State University.
Enclosed please find the behavior questionnaire that I mentioned in our phone conversation.
If you would like to participate in this study, please fill in this questionnaire. Please be
assured that your responses will be kept strictly confidential and your child’s name will not
appear on any study reports. However, there may be circumstances where this information
must be released. For example, personal information regarding your participation in this
study may be disclosed if required by state law. Also, your records may be reviewed by the
following groups (as applicable to the research): Office for Human Research Protections or
other federal, state, or international regulatory agencies; The Ohio State University
Institutional Review Board or Office of Responsible Research Practices.
Also, please sign and date the enclosed HIPAA Research Authorization Form, and have your
child sign and date the assent form. If you agree to give us permission (optional) to contact
your child’s school, please fill in, sign and date the enclosed Release of Information form.
Please note that providing permission to contact schools is completely optional and you can
still participate in this study without providing us permission to contact your child’s school.
Agreement or not to contact school does not affect the payment of $15.
Please return the completed behavior rating form, and all signed documents in the self
addressed stamped envelope provided. As soon as we receive these documents from you in
the mail, we will mail out the payment of $15 along with a copy of the signed HIPAA
Research Authorization Form and assent form for your records.
Risks and Benefits:
Results of this study will enable professionals to determine which children with autism
spectrum disorders may benefit from early psychosocial intervention so as to minimize
problems later in life. Additionally, the results will help determine whether particular
187
behavior patterns of children with ASDs can predict psychiatric diagnoses and school
placement in future. Such findings have practical implications for families like you and also
for ASD service providers in planning optimal services.
There is a small likelihood that you may experience some psychological stress when
completing the behavior rating form. You completed this same questionnaire when (child’s
name) was first seen at the Autism Spectrum Disorder Clinic. The possibility of experiencing
this stress is very small and unlikely to be severe. You may choose not to answer some or all
of the questions.
Participant Rights:
You may refuse to participate in this study without penalty or loss of benefits to which you
are otherwise entitled. If you are a student or employee at Ohio State, your decision will not
affect your grades or employment status. If you choose to participate in the study, you may
discontinue participation at any time without penalty or loss of benefits. You do not give up
any personal legal rights you may have as a participant in this study.
An Institutional Review Board responsible for human subjects research at The Ohio State
University reviewed this research project and found it to be acceptable, according to
applicable state and federal regulations and University policies designed to protect the rights
and welfare of participants in research.
If you have any additional questions concerning this study or your participation in it, please
contact me by e-mail at [email protected] or by phone at (614) 316 4287, or Dr.
Michael Aman at [email protected] or by phone at (614) 688 4196.
For questions about your rights as a participant in this study or to discuss other study-related
concerns or complaints with someone who is not part of the research team, you may contact
Ms. Sandra Meadows in the Office of Responsible Research Practices at 1-800-678-6251.
If you are injured as a result of participating in this study or for questions about a studyrelated injury, you may contact Dr. Aman at (614) 688 4196.
We really appreciate your time and energy. This study would not be possible without your
cooperation!
Sincerely,
Monali Chowdhury, M.A.
Graduate Associate
Nisonger Center
The Ohio State University
Michael Aman, Ph.D.
Professor of Psychology and Psychiatry
Nisonger Center
The Ohio State University
188
Appendix E: Combination of NCBRF Raters at T1 and T2
189
Table 26. Combination of NCBRF Raters at T1 and T2
T1 Rater/ T2 Rater
n (%)
Available n = 134
Same raters at T1 and T2
Mother/Mother
Father/Father
Grandmother/Grandmother
Aunt/Aunt
Combined subtotal
87 (65.25)
8 (6.00)
3 (2.25)
1(0.75)
99 (74.25)
Different raters at T1 and T2
Mother/Father
Father/Mother
Mother/Stepmother
Mother/Grandmother
Grandmother/Stepmother
Combined subtotal
4 (3.00)
27 (20.25)
2 (1.50)
1 (0.75)
1 (0.75)
35 (26.25)
Note. T1 rater information not available for 9 participants
190
Appendix F: Ancillary HMLR Analyses with IQ as Predictor
191
Table 27. Summary of Ancillary HMLR Analyses Predicting T2 NCBRF Outcomes with IQ as Additional Predictor in Block 2
Predictors
Block 1
R2
∆ R2
F (p)
Conduct
.026
–
.714 (.547)
Insecure/
Anxious
.073
–
2.114(.105)
192
Block 2
R2
∆ R2
F (p)
.371
.349
2.668 (.004)
.434
.361
3.050 (.001)
Block 3
R2
∆ R2
F (p)
.394
.020
.507 (.731)
.461
.027
.803 (.528)
Hyperactive
.036
–
Self-injury/
Stereotypic
Self-isolated/
Ritualistic
Overly
Sensitive
.019
.028
.069
.520 (.669)
.781 (.508)
2.014 (.119)
–
1.001(.397)
.355
.319
2.404 (.006)
.469
.450
4.055 (<.001)
.364
.009
.200 (.938)
.486
.017
.913 (.122)
192
–
.449
.421
3.661 (< .001)
.482
.032
.982 (.424)
–
.375
.306
2.688 (<.001)
.398
.023
.925 (.455)
Table 28. Ancillary HMLR Analysis Predicting T2 NCBRF Conduct with IQ as Additional Predictor in Block 2
B

SE(B)
p
     
193
Block 1
Age (months)
Gender (Male)
Ethnicity
.011
1.752
- 2.648
.040
2.083
2.136
.793
.403
.219
.029
.093
-.138
Block 2
Age (months)
Gender (Male)
Ethnicity
-.030
1.010
-3.664
.052
1.926
2.012
.559
.605
.073
-.084
.053
-.190
ASD subtype
Asperger’s
PDD-NOS
4.030
-2.069
4.359
4.247
.359
.628
.222
-.082
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.654
-.330
-.167
.553
.085
.105
.162
.189
.286
.437
.206
.289
< .001
.086
.561
.210
.680
.716
.494
-.180
-.086
.142
.045
.040
ADI-R
Reciprocal Social Interaction
Communication
RRB
.189
-.542
-.574
.343
.387
.529
.582
.166
.282
.087
-.240
-.118
continued
RRB = Restricted, Repetitive Behaviors and Interests
193
Table 28, Continued
B

Language composite
General Language
Pragmatic Language
IQ score
SE(B)
p
    
-.089
-.079
.101
.104
.376
.452
-.185
-.104
.005
.090
.960
.010
194
194
Table 29. Ancillary HMLR Analysis Predicting T2 NCBRF Insecure/Anxious with IQ as Additional Predictor in Block 2
B

SE(B)
p
    
Age (months)
Gender (Male)
Ethnicity
.049
1.521
.260
.024
1.222
1.253
.042
.217
.836
.223
.134
.022
Block 2
Age (months)
Gender (Male)
Ethnicity
.040
1.022
.770
.029
1.102
1.151
.184
.357
.506
.182
.090
.067
ASD subtype
Asperger’s
PDD-NOS
3.345
5.235
2.494
2.430
.890
.035
-.032
.344
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
066
.366
-.055
.552
-.160
.078
.093
.108
.164
.250
.118
.165
.477
.001
.738
.031
.179
.640
.083
.332
-.047
.235
-.140
.050
ADI-R
Reciprocal Social Interaction
Communication
RRB
-.136
.054
.334
.196
.221
.303
.489
.808
.274
-.104
.040
.114
continued
195
Block 1
RRB = Restricted, Repetitive Behaviors and Interests
195
Table 29, Continued
B

Language composite
General Language
Pragmatic Language
IQ score
SE(B)
p
    
-.039
-.137
.057
.060
.494
.025
-.136
-.298
.062
.052
.231
.237
196
196
Table 30. Ancillary HMLR Analyses Predicting T2 NCBRF Hyperactivity with IQ as Additional Predictor in Block 2
B

SE(B)
p
    
Age (months)
Gender (Male)
Ethnicity
-.034
2.851
-1.639
.041
2.130
2.184
.412
.185
.455
-.091
.146
-.083
Block 2
Age (months)
Gender (Male)
Ethnicity
-.026
2.279
-2.466
.058
2.162
2.259
.658
.296
.279
-.069
.117
-.125
ASD subtype
Asperger’s
PDD-NOS
1.190
-5.791
4.893
4.767
.809
.229
.064
-.222
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.117
-.029
.543
-.002
.056
-.423
.182
.212
.321
.491
.231
.324
.522
.891
.006
.997
.809
.197
.086
-.015
.272
.001
.029
-.158
ADI-R
Reciprocal Social Interaction
Communication
RRB
-.004
-.331
-.914
.385
.434
.594
.992
.448
.129
-.002
-.143
-.182
continued
197
Block 1
RRB = Restricted, Repetitive Behaviors and Interests
197
Table 30, Continued
B

SE(B)
p
    
Language composite
General Language
Pragmatic Language
-.055
.085
.112
.117
.624
.473
-.112
.108
IQ score
-.006
.101
.953
-.013
198
198
Table 31. Ancillary HMLR Analyses Predicting T2 NCBRF Self-injury/Stereotypic with IQ as Additional Predictor in Block 2
B

SE(B)
p
    
Block 1
Age (months)
Gender (Male)
Ethnicity
-.012
.152
.287
.011
.549
.563
.246
.782
.611
-.130
.031
.057
Block 2
Age (months)
Gender (Male)
Ethnicity
-.001
-.088
.201
.012
.466
.487
.910
.851
.680
-.015
-.018
.040
.327
-1.009
1.055
1.027
.758
.329
.069
-.152
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.050
.057
-.057
.495
.055
.008
.039
.046
.069
.106
.050
.070
.206
.218
.411
<.001
.273
.905
.144
.118
-.113
.482
.110
.012
ADI-R
Reciprocal Social Interaction
Communication
RRB
.030
.142
-.016
.083
.094
.128
.717
.135
.898
.053
.239
-.013
Continued
ASD subtype
Asperger’s
PDD-NOS
199
RRB = Restricted, Repetitive Behaviors and Interests
199
Table 31, Continued
B

SE(B)
p
    
Language composite
General Language
Pragmatic Language
.043
-.024
.024
.025
.078
.338
.342
-.122
IQ score
-.023
.022
.288
-.203
200
200
Table 32. Ancillary HMLR Analyses Predicting T2 NCBRF Self-isolated/Ritualistic with IQ as Additional Predictor in Block 2
B

SE(B)
p
    
Age (months)
Gender (Male)
Ethnicity
.038
-.269
-.990
.026
1.365
1.400
.157
.844
.482
Block 2
Age (months)
Gender (Male)
Ethnicity
-.040
-1.455
-1.439
.032
1.186
1.239
.213
.224
.250
-.168
-.117
-.114
ASD subtype
Asperger’s
PDD-NOS
3.042
5.916
2.684
2.616
.261
.127
.256
.356
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.027
-.157
-.106
.699
.498
.306
.102
.116
.176
.269
.127
.178
.785
.181
.550
.062
<.001
.089
.031
-.130
-.083
.273
.399
.179
ADI-R
Reciprocal Social Interaction
Communication
RRB
-.164
-.223
.152
.211
.238
.326
.440
.353
.643
201
Block 1
RRB = Restricted, Repetitive Behaviors and Interests
201
.158
-.022
-.078
-.115
-.151
.047
continued
Table 32, Continued
B

SE(B)
p
    
Language composite
General Language
Pragmatic Language
-.046
.057
.061
.064
.453
.380
-.147
.114
IQ score
-.029
.066
.601
-.102
202
202
Table 33. Ancillary HMLR Analyses Predicting T2 NCBRF Overly Sensitive with IQ as Additional Predictor in Block 2
B

SE(B)
p
     
Age (months)
Gender (Male)
Ethnicity
.035
-.054
.369
.015
.776
.796
.062
.945
.644
.252
-.007
.050
Block 2
Age (months)
Gender (Male)
Ethnicity
.013
-.231
-.386
.016
.606
.633
.411
.704
.544
.097
-.032
-.053
ASD subtype
Asperger’s
PDD-NOS
2.363
1.248
1.371
1.336
.089
.354
.341
.129
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.044
-.021
-.119
.095
-.133
.725
.051
.059
.090
.137
.065
.091
.389
.728
.191
.492
.044
<.001
.087
-.030
-.161
.064
-.183
.730
ADI-R
Reciprocal Social Interaction
Communication
RRB
.029
.085
.035
.108
.122
.166
.788
.486
.836
203
Block 1
RRB = Restricted, Repetitive Behaviors and Interests
203
.035
.099
.019
continued
Table 33, Continued
B

Language composite
General Language
Pragmatic Language
IQ score
SE(B)
p
    
.010
.069
.031
.033
.750
.354
.055
.092
-.025
.028
.384
-.149
204
204
Appendix G: Supplementary Logistic Regression Analyses
205
Table 34. Supplementary Logistic Regression Predicting T2 Comorbid Anxiety Disorder with ADI–R Domain Scores as Additional
Predictors
Predictor
206
B
SE(B)
Wald χ2
p
Age
-.012
.016
.545
.460
.988
.958-1.019
Gender
-.815
.497
2.685
.101
.443
.167-1.173
ASD Subtype
Autism
Asperger’s
-.198
.380
1.087
.767
.033
.245
.856
.620
.820
1.462
.098-6.905
.325-3.582
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.092
.167
.027
.128
.019
.059
.042
.064
.063
.107
.050
.065
4.730
6.712
.183
1.427
.142
.820
.030
.010
.669
.232
.706
.365
.912
1.181
1.027
1.136
1.019
1.061
.840-.991
1.041-1.340
.909-1.161
.921-1.401
.924-1.124
.933-1.206
.015
.021
.529
.467
1.015
.974-1.058
-.044
.078
-.128
.079
.090
.138
.312
.762
.868
.576
.383
.352
.957
1.082
.880
.819-1.117
.907-1.290
.672-1.152
General Language
ADI-R
Reciprocal Social Interaction
Communication
RRB
RRB = Restricted, Repetitive Behaviors and Interests
206
Odds ratio
95% C.I. for odds ratio
Table 35. Supplementary Logistic Regression Predicting T2 Comorbid ADHD with ADI–R Domain Scores as Additional Predictors
Predictor
B
Age
SE(B)
Wald χ2
p
Odds ratio
95% C.I. for odds ratio
207
.029
.023
1.641
.200
1.030
.985-1.077
Gender
-.678
.642
1.115
.291
.508
.144-1.787
ASD Subtype
Autism
Asperger’s
.119
-.365
1.505
1.399
.169
.068
.681
794
1.856
.694
.097-5.493
.045-4.771
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.064
.013
.509
.033
-.050
.047
.055
.072
.113
.140
.065
.090
1.346
.035
10.213
.056
.580
.269
.246
.853
<.001
.814
.446
.604
.938
1.014
1.663
1.034
.951
1.048
.842-1.045
.879-1.168
1.332-2.077
.786-1.360
.837-1.081
.878-1.250
General Language
.022
.029
.595
.441
1.023
.966-1.082
ADI-R
Reciprocal Social Interaction .032
Communication
-.048
RRB
-.253
.101
.101
.173
.101
.224
2.132
.752
.636
.144
1.032
.954
.777
.848-1.257
.783-1.161
.553-1.090
RRB = Restricted, Repetitive Behaviors and Interests
207
Table 36. Supplementary Logistic Regression Predicting T2 Comorbid DBD with ADI–R Domain Scores as Additional Predictors
Predictor
B
SE(B)
Wald χ2
p
Odds ratio
.036
.018
1.945
.067
1.037
1.000-1.074
Gender
-1.535
.621
1.109
.073
.215
.065-.728
ASD Subtype
Autism
Asperger’s
-1.191
.005
1.457
.951
.668
.002
.414
.995
.304
1.005
.017-5.282
.156-6.481
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.049
-.154
.124
.100
.020
-.160
.049
.076
.081
.124
.056
.083
1.007
4.126
2.348
.652
.134
3.702
.316
.042
.125
.419
.715
.067
.952
.857
1.132
1.105
1.021
.852
.866-1.048
.739-.995
.966-1.327
.867-1.410
.915-1.139
.724-1.003
.021
.026
.692
.406
1.021
.972-1.074
.115
.001
-.015
.106
.119
.160
1.176
.001
.009
.278
.998
.924
1.122
1.001
.985
.911-1.382
.793-1.262
.719-1.348
Age
208
General Language
ADI-R
Reciprocal Social Interaction
Communication
RRB
95% C.I. for odds ratio
DBD = Disruptive Behavior Disorder; RRB = Restricted, Repetitive Behaviors and Interests
Note. Test of the model was statistically non-significant (χ2 = 22.94, p = .061). Consequently, significance values and odds ratios
associated with individual predictors were not interpreted.
208
Table 37. Supplementary Logistic Regression Predicting T2 Comorbid Depressive Disorder with ADI–R Domain Scores as Additional
Predictors
Predictor
B
SE(B)
Wald χ2
p
Odds ratio
95% C.I. for odds ratio
209
Age
.023
.016
.545
.460
.988
.958-1.019
Gender
.161
.931
1.556
.212
2.193
.515-9.793
ASD Subtype
Autism
Asperger’s
1.005
.159
1.793
1.382
.314
.013
.575
.909
2.733
1.172
.081-6.868
.078-5.589
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.110
.243
-.251
-.139
-.031
.335
.075
.097
.127
.172
.082
.120
2.190
6.290
3.911
.648
.146
7.829
.139
.012
.048
.421
.702
.005
1.117
1.276
.778
.871
.969
1.398
.965-1.293
1.055-1.543
.606-.998
.621-1.220
.825-1.139
1.106-1.768
General Language
-.054
.036
2.316
.128
.947
.883-1.016
ADI-R
Reciprocal Social Interaction
Communication
RRB
.128
-.213
.339
.145
.170
.232
.783
1.573
2.128
.376
.210
.145
1.137
.808
1.403
.856-1.511
.580-1.127
.890-2.213
RRB = Restricted, Repetitive Behaviors and Interests
209
Table 38. Supplementary Logistic Regression Predicting T2 Educational Placement with ADI–R Domain Scores as Additional
Predictors
Predictor
B
Age
SE(B)
Wald χ2
p
Odds ratio
95% C.I. for odds ratio
-.011
.019
.356
.551
.989
.954-1.026
.171
.542
.143
.543
1.004
.346-2.894
ASD Subtype
Autism
Asperger’s
-2.278
-1.045
1.196
.855
3.627
1.492
.067
.222
.102
.352
.010-1.069
.066-1.881
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
.131
.033
.016
-.105
-.042
-.056
.045
.061
.065
.115
.053
.070
8.651
.299
.063
.832
.636
.627
.003
.584
.801
.362
.425
.429
1.140
1.034
1.016
.901
.959
.946
1.045-1.244
.918-1.164
.895-1.154
.720-1.128
.864-1.063
.824-1.085
General Language
-.059
.022
7.137
.008
.943
ADI-R
Reciprocal Social Interaction
Communication
RRB
.088
.063
.151
.092
.098
.144
.907
.408
1.089
.341
.523
.297
1.092
1.065
1.163
Gender
210
RRB = Restricted, Repetitive Behaviors and Interests
210
.903-.984
.911-1.308
.879-1.290
.876-1.543
Appendix H: Binary Logistic Regression Predicting Disruptive Behavior Disorder at T2
211
Table 39. Binary Logistic Regression Predicting T2 Comorbid DBD
Predictor
SE(B)
Wald χ2
.029
.016
1.238
.072
1.030
-1.302
.581
1.028
.065
.272
ASD Subtype
Autism
Asperger’s
-.296
.422
1.123
.800
.069
.278
.792
.598
.744
1.526
.082 -6.274
.318-7.323
T1 NCBRF
Conduct
Insecure/Anxious
Hyperactive
Self-injury/Stereotypic
Self-isolated/Ritualistic
Overly Sensitive
-.047
-.140
.106
.095
.031
-.152
.047
.070
.080
.120
.055
.080
1.001
4.041
1.742
.625
.327
3.594
.317
.044
.187
.429
.568
.068
.954
.869
1.111
1.099
1.032
.859
.869-1.046
.759-.997
.950-1.300
.869-1.390
.927-1.149
.735-1.005
.012
.025
.251
.616
1.012
.965-1.062
B
Age
Gender
212
General Language
p
Odds ratio
95% C.I. for odds ratio
.997-1.063
.087-.849
DBD = Disruptive Behavior Disorder; RRB = Restricted, Repetitive Behaviors and Interests
Note. Test of the model was statistically non-significant (χ2 = 19.16, p = .074). Consequently, significance values and odds ratios
associated with individual predictors were not interpreted.
212