PREDICTORS OF TREATMENT OUTCOME

1
PREDICTORS OF TREATMENT OUTCOME
IN YOUNG CHILDREN WITH AUTISM
R. L. Gabriels, PsyD, D. E. Hill, Ph.D., R. A.
Pierce, MA, Rogers, S. J., Wehner, B.
2
INTRODUCTION  Autism is a chronic
neurodevelopmental disorder with core deficits
in social interactions, social communication, and
stereotyped/repetitive behaviors and/or
restricted interests1. The heterogeneous symptom
presentation in this population, ranging from
severe global impairments to minor impairments,
complicates the scientific and clinical pursuit
of identifying predictors of early intervention
outcomes. Variability in treatment outcomes for
different subgroups of children who have autism
(i.e., some children making great progress and
others making little, as measured by
developmental cognitive ability), regardless of
the type and intensity of intervention,
complicates the autism outcome research. Although
autism intervention research indicates that early
intervention can have a positive impact on short-
and long-term developmental outcomes, no one type
of program has proved superior over another for
all children with autism and it is still unknown
how to combine and tailor approaches to best
impact treatment outcomes2.
3
These inconsistent results lead to questions
about the actual effectiveness of specific
treatment models, as well as factors that may
also influence treatment outcomes, including age
of treatment onset, intensity, and duration3. In
addition, other factors have been identified that
may directly or indirectly impact treatment
outcome, including pretreatment intelligence and
language levels4, autism symptom severity,
visual-spatial abilities5, cognitive flexibility,
and co-morbid conditions (e.g., mental
retardation, medical syndromes, seizures) 6.
Parent-family factors, such as SES, parent
motivation/involvement, stress levels,
depression, and education level, may also
influence early intervention outcomes7.
Therefore, crucial questions remain regarding an
individuals unique response to treatment and the
effectiveness of certain treatment components.
Such information may refine our understanding of
the factors contributing to early intervention
success and decrease blind preference for
interventions that have not been
well-researched2. Studies such as this project
contribute to the cumulative body of treatment
intervention knowledge to improve understanding
of how to best tailor treatments to the
individual needs of children with autism.  The
current study examined the developmental outcomes
and predictors of those outcomes of 17 children
diagnosed with an Autistic Disorder or a
Pervasive Developmental Disorder/Not Otherwise
Specified (PDD/NOS).
4
METHOD  Participants Seventeen children
diagnosed with Autistic Disorder or PDD/NOS from
an autism specialty clinic participated in this
study. Children were diagnosed at a mean age of
32 months, with all participants receiving
standardized developmental and/or language
assessments early in the preschool period (mean
age 31 months). Children received a combination
of early intervention treatments, with no child
receiving a brand name therapy. See Table 1 for
demographic variables at initial (Time I) and
follow-up (Time II) evaluations.  Materials All
parents of children gave informed consent prior
to the evaluation. Assessment data from the Time
I evaluation were gathered from available
records. Time II evaluation included measures of
cognitive functioning, language levels, academic
achievement, adaptive functioning, autism
symptoms, and family functioning surveys.
Additionally, treatment intervention data,
including type of treatment and number of hours,
were gathered from a detailed parent report. See
Table 2 for assessment tools.  Procedure
Demographic data, an adaptive functioning
interview, and family functioning surveys were
completed via a telephone interview. Assessment
records were copied and placed in files for
review upon testing completion. Children were
tested in up to three sessions with their parents
present.
5
RESULTS  Relationship between Time I, Time II,
and Treatment Measures Relationships between
Time I, Time II, and treatment measures were
examined via bivariate correlation analyses for
the participant group as a whole. The
correlations between treatment variables and Time
II measures were all non-significant (p value
gt.05). Correlations between Time I developmental
IQ scores and treatment variables were also
non-significant, with the exception of a
significant correlation between Time I
developmental IQ scores and total treatment hours
(r .65, p lt .01). See Table 3 for correlations
between Time I and Time II measures.  Distribution
of IQ at Time II The distribution of Time II
developmental IQ scores had two apparent breaks
IQ scores less than or equal to 48 and IQ scores
greater than or equal to 64. Given this break,
participants were divided into two groups (High
and Low outcome) based on their Time II
developmental IQ score. See Figure I for Time II
developmental IQ scores distribution.
6
Comparison between Outcome Groups Potential
differences between the two outcome groups on
demographic variables were examined. There were
no between group differences in terms of gender
(Fischer's Exact test p 1.0) or socioeconomic
status (F(1,15) .29). The two groups did not
differ in terms of the age at which they were
first diagnosed (F(1,15) .655, ns), age they
began treatment (F(1,15) .173, ns), or current
chronological age at Time II assessment (F(1,15)
.006, ns).  Treatment effects that might
account for outcome differences were then
examined. The two groups did not differ
significantly in either the intensity of
treatment received (F(1,15) 1.91, ns) or in the
total number of treatment hours (F(1,15) .75,
ns). Specifically, the High outcome group
received an average of only 26.97 more treatment
hours per month than the Low outcome group. See
Table 4 for participant demographic
characteristics by outcome group.  The groups did
not differ significantly on Time I developmental
IQ scores (F (1,14) 3.11, ns) or on language
levels (F(1,13) 1.27, ns). However, at Time I,
the IQ of the High outcome group was an average
of 21.4 points higher than the Low outcome group.
The observed power was only .38 with a medium
effect size (.18). See Table 5 for group outcome
characteristics from Time I to Time II.
7
Pattern of Change between Groups over Time A
mixed two-factor within subjects design was used
to investigate the pattern of change in
developmental IQ over time for the two outcome
groups. The overall F (1,15) for the interaction
was 17.26 (p lt .01), suggesting that the two
groups differed significantly in the pattern of
change. Several of the tests of simple effects
were also significant. Specifically, the High
outcome group showed a significant increase in
developmental IQ from Time I to Time II (F (1,
15) 10.46, p .01), whereas the Low outcome
group showed a significant decrease in
developmental IQ scores from Time I to Time II (F
(1, 15) 7.02, p .03). Additionally, the
developmental IQ scores were significantly
different from each other at Time II (F (1,15)
48.51, p lt .01). See Figure 2.  Family
Variables Family stress, coping, and resource
differences between outcome groups were assessed
using an ANOVA (n 16). On the Family Inventory
of Life Events and Changes (FILE8), the two
outcome groups differed significantly on the
factor of Finance and Business Strains (F (1,14)
6.27, p .025). The Low outcome group
endorsed more strain on the family's money supply
(M 121.87, SD 46.15) than the High outcome
group (M 66.50, SD 42.24). The outcome groups
did not differ significantly on any other factors
of the FILE, or on any of the factors of the
Family Inventory of Resources for Management
(FIRM9)
8
DISCUSSION  This study raises several important
issues that should be considered in future autism
treatment outcome research        Fifty-two
percent of the total participants demonstrated
plasticity in developmental rates, even without
intensive 40 hours per week of treatment or a
specific treatment model.        Early language
development was not predictive of later outcome.
This may well be due to the age of initial
assessment for participants in this study, as
language had not yet fully emerged. This
highlights the need to develop various methods of
assessing predictive factors (e.g., quantity and
quality of non-verbal communication skills)
depending on the age of the child.
 Discrepancies between the two outcome groups in
family stress levels regarding financial strain
and amount of social support suggest that these
factors may impact treatment outcomes.
9
The results of this study raise questions
regarding the effects of a combination of IQ
scores, treatment intensity, and levels of family
stress variables on treatment outcome. Future
treatment outcome research needs to specify
potential risk factors (e.g., biological,
cognitive, language, adaptive, symptom severity,
treatment intensity and type, family/parent
factors) within the autism population to help
clinicians better identify and implement services
to meet individual child and family needs.
 Plans are underway to re-examine this same
group of participants, now five years following
the initial assessment of their treatment
outcomes. Participants current levels of
functioning will be evaluated in areas of
nonverbal IQ, communication, social interaction,
adaptive living, and academic, autism severity,
and aberrant behavior. Treatment and family
variables will also be measured. Thus far, 15 out
of the 17 participants parents have expressed an
overwhelmingly positive interest in a follow-up
study. This follow-up study would expand the
breath of the previous study10 by assessing both
the quantitative and qualitative aspects of
treatment outcome variables to enhance the
understanding of individual and family factors
related to outcomes. The specific aims and
hypotheses of this follow-up study are as
follows
10
Identifying the long-term developmental
trajectories of the previously defined high and
low outcome groups by comparing participants
previous and current levels of functioning in the
areas of cognition, language, adaptive behavior,
and achievement. The participants in the original
pilot study differed significantly in language
levels and IQ at a mean age of 5 yrs, 9 months.
Hypothesis Participants in the high outcome
group will continue to show significantly better
performance than participants in the low outcome
group across measures of cognition, language,
adaptive behavior, and achievement. Providing
additional comparisons between the high and low
outcome groups of their current levels of
autism-symptom severity and presence of aberrant
behaviors. Hypothesis Measures of autism symptom
severity and aberrant behaviors will not be
correlated with group membership.
11
Comparing outcome groups on the quantity and
type of interventions since their last
evaluation, including medication interventions.
Significant differences of treatment intensity
were not found between groups in the original
pilot study. Hypothesis High and low outcome
groups will continue to show no significant
differences in number of treatment
hours. Investigating the role of family factors
on treatment outcome, including the number of
individuals in the family and family perceptions.
Previous studies of family stress levels
associated with outcomes in children with autism
have been limited by the use of standardized
instruments to assess parent stress and coping.
Family stress measures in the original pilot
study were inconclusive. Qualitative methods
might better capture the complexities of
caregiver and sibling perceptions of living with
a child who has autism and increase the ability
to relate research findings to clinical
intervention. Hypothesis Families of
participants in high and low outcome groups will
identify comparatively different categories of
stress and coping.
12
Table 1 Overall Participant Characteristics at
Time I and Time II
N 17, unless otherwise specified. SES Based
on Hollingshead (1975) four-factor index of
social status. Possible SES scores range from 8
to 88 (lowest SES to highest SES classification).
Ethnicity 1 African American/Native American,
1 Hispanic, 1 Asian
13
(No Transcript)
14
Table 3 Overall Group Correlations between Time I
and Time II Measures  Time I Time I IQ Time I
Language Time II  Time II IQ .704 (.002) .315
(.253)  Time II Language .686 (.003) .203
(.469)  Time II Adaptive .562 (.023) .331
(.228)  Time II Reading .608 (.013) .333
(.226)  Time II Math .760 (.001) .254 (.361)
15
Table 4 Participant Characteristics by Outcome
Group
p lt .05 p lt .01
16
Table 5
Group Outcome Characteristics from Time I to
Time II   High Outcome (N9) Low
Outcome (N8)
Three participants had a change in diagnosis
from a Time I diagnosis of Autism to a Time II
diagnosis of PDDNOS and one participant went from
an initial undefined PDDNOS/autism diagnosis at
Time I to an Autism diagnosis at Time II.
17
(No Transcript)
18
(No Transcript)