Academic Research
Below are studies researched for program evaluation at CH.I.L.D.
Review of Home Based Intervention Studies
The criteria for inclusion in this literature search required peer reviewed published studies from 1985 to present. The participants in the studies must have been up to 10 years of age and diagnosed with autistic spectrum disorder. The intervention must have focussed on promoting comprehensive development of children. In terms of outcome documentation, the studies must have applied some form of objective measures and used experimental or methodological means to analyze data.
The goals for this literature search were to examine previous assessments and existing data and to identify evaluative methods; also, to identify research design and methodologies, variables, procedure and to examine outcome results. Three studies were found that met the above criteria. Since there were insufficient studies in this search, three center-based studies have also been included.
STUDY #1
The first home-based study was the Intensive Early Intervention Project, published by Lovaas in 1987. This popular study, having the strongest scientific design involved 19 children in the experimental condition that employed intensive 1:1behavioral intervention. The intervention took place in the child’s home, and later incorporated community and preschool settings. Of the 19 children in this study, 90% had 2 or more clinical diagnoses of autism from independent psychologists.
This study incorporated a control comparison group that involved less intensive intervention of the same therapy as in the experimental group. This pretest/ posttest design was followed for two years and various tests were administered to measure change. Pretreatment maturational age- (MA) was calculated using either the Bayley Scales of Infant Development (Bayley, 1955) the Cattell Infant Intelligence Scale (Cattell, 1960), or the Stanford-Binet Intelligence Scale (Thorndike, 1972). Behavioral observations were recorded from videotaped data of the child’s free-play behavior.
Scoring was based on the degree of self-stimulatory behaviors; appropriate play behaviors and recognizable words. A 1-hour interview with the parents about the child’s background provided more diagnostic and descriptive information (Lovaas, 1987). Language use, play skills, sensory issues self-stimulatory behaviors, adaptive skills and behavioral challenges were areas addressed during this interview. Mean chronological age of the control group and experimental group were 35 and 41 months respectively at onset of administration of pretreatment measures. Posttreatment measures were administered when child was 6-7 years of age.
The treatment procedure involved assignment of the subjects to either the experimental condition or to the control group based on variables that included availability of therapists, and distance traveled by therapists from UCLA to the family’s residence. The therapist consisted of highly trained undergraduate students who worked with the children for approximately 40 hours per week. There was parental involvement in the therapy procedures and implementation.
The conceptual basis for the therapy was behavioral theory and principles of operant learning (Lovaas, 1989: Sheinkopf & Seigel, 1998) that employed discrete trial training–(DTT) procedure. This was used to teach social, language, cognitive, and adaptive skills and also to reduce undesirable behaviors (Rogers, 1998).
Pretreatment results indicated that 2 of the 19 children scored in the normal range of intellectual functioning.; 7 children scored in the moderately retarded range, and 10 children scored in the severely retarded range. One child showed minimal appropriate speech, 7 children echolalic and 11 children were mute (Lovaas, 1989).
Results were obtained after first grade and were based on both results from IQ tests and educational placement (Lovaas, 1987: Rogers, 1998). Success was later measured on whether or not the children passed first grade. In the experimental group, 9 children (47%) obtained average to above average scores on IQ tests. Eight children (42%) passed in a special educational class and had obtained IQ scores within the mildly delayed range. Two children were placed in a contained class for other autistic/mentally retarded children and scored in the profoundly retarded range on IQ tests (Lovaas, 1987).
The Lovaas study is one of the most intensive and comprehensive studies with scientific evidence foe effectiveness. Currently there are no other methodological replications of his work (Rogers, 1998). There is high quality control training and education for therapists and educators working under the supervision of Lovaas and his team to effectively employ treatment methods (Lovaas, 1996).
The main methodological weaknesses involve experimental design that includes lack of specific documentation of treatment hours for the groups (Rogers, 1998). There is also lack of documentation regarding additional treatments (if any) implemented for the experimental group in addition to the intensity of the EIP, which could have positively influenced the results of this group.
STUDY #2
Sheinkopf and Seigel (1998) conducted the second study, a home-based intervention that also employed behavioral methods. This intervention differed from the Lovaas study in that it was for a shorter period, less intensive and did not have academic support from a university.
This approach was implemented in the community and examined the effects of treatment on IQ and symptom presentation (Sheinkopf and Seigel, 1998). This study explored treatment effects for approximately 19 months. The 11 children in the study were previously involved in behavioral therapy using the Lovaas method. The parents selected current therapies and the authors had no role in treatment selection. A control group (n=11), was matched on pretreatment chronological age (CA), mental age (MA), diagnosis, and posttreatment assessments. All children attended school and were placed in special education classes that received equal intensity of individual speech therapy and occupational therapy.
Pre and post treatment measures consisted of the Merrill-Palmer Scale of Mental Tests, the Bayley Scales of Infant development (Bayley, 1955), the Wechsler Preschool and Primary Scale of Intelligence (Wechsler, 1974), or the Cattell Infant Intelligence Scale (Cattell, 1960). Results indicated significant group differences in IQ. The authors examined the effects of treatment hours on outcome by calculating IQ scores from pre treatment to post-treatment.
Results indicated no significance in number of hours of therapy in the experimental condition per week, versus the number of hours of total intervention per week. Symptom presentation and severity was rated according to the number of observable positive DSM-III-R symptoms as a percentage of the number of age-appropriate “scorable” symptoms and severity. The authors chose to calculate “scorable’ symptoms by giving a percentage score rather than scoring each positive symptom. This method would best represent the sample since many of the children were very young and had low IQ estimates, thus many DSM-III-R symptoms were unable to be assessed. (Sheinkopf and Seigel, 1998). Pretreatment scores revealed no group differences. Although statistical significance was not reached, posttreatment scores did reveal group differences with a positive reduction of symptom presentation and severity for the experimental group.
The authors here presented a study that has important implication for contrasting levels of intensity of intervention as prescribed by the Lovaas study. Here the implementation of a less intensive model without academic support of student therapists, questions beliefs of delivering only highly intensive therapies to children with autism.
Although this study is important to challenge the claims made by the Lovaas study regarding the need for intensive therapies, this study failed to present a therapeutic paradigm that contrasts traditional behavioral methods. Therefore, it is important for studies to compare and contrast methods and philosophies of different therapies in addition to variables that include treatment hours and intensity.
STUDY #3
The third study conducted by Ozonoff and Cathcart (1998), evaluated the effects of a TEACCH- (Treatment and Education of Autistic and related Communication Handicapped CHildren) based home program intervention. The TEACCH model was developed by Schopler & Reichler in 1987, and emphasizes parental-professional collaboration. Parents serve as “co-therapists”, implementing the intervention. The study consisted of 22 children (2-6 years of age), with clinical diagnoses of autism, which were recruited from the Salt Lake City area (Ozonoff and Cathcart, 1998). Assignment to the experimental condition was on first come basis.
The first 11 children comprised the experimental condition and the latter 11 children comprised the no treatment control group. The groups were matched on several variables including age, severity of autism, pretreatment scores and time interval between pre and posttesting. The Psychoeducational Profile Revised (PEP-R; Schopler, Reichler, Bashford, Lansing, & Marcus, 1990) was used to measure functioning in seven developmental domains: imitation, perception, fine and gross motor, eye hand coordination, and verbal and non-verbal conceptual ability (Ozonoff and Cathcart, 1998). The Childhood Autism Rating Scale (CARS) was also used to rate severity of autism. The experimental condition lasted 10 sessions for a 4-month period. The PEP-R was administered pre and posttreatment to both groups. Once the assessment was completed the parents and the therapists developed individual program goals that reflected emerging skills from the assessment and included specific parental concerns. The families participated in the therapy by being trained to work with their child. Parents were expected to continue this treatment regime with their child at home thereafter.
The intervention consisted of common elements that included, structured teaching, there was emphasis on visual strengths of the child to teach language and imitation, visual schedules, a multi-modal communication system, and pre-academic-prevocational activities (Ozonoff and Cathcart, 1998). Staff monitored, and supervised the parents at the child’s home to ensure consistent teaching, and provided direct feedback through modeling. Supervision was eventually reduced from weekly meetings to once in every 2-3 week period.
The results indicated significant group differences of total change scores on the PEP-R scale. The CARS scores were significantly negatively correlated with change scores indicating that subjects with mild autism and better language skills were more likely to have a positive outcome (Ozonoff and Cathcart, 1998).
This study is important as it presents a multidimensional approach involving behavioral, ecological and developmental perspectives. Family involvement and visual support systems present an important dimension to treatment methods involved with children with autism. This study also explored the benefits of simultaneously using different types of intervention with the same child.
It would have been clearer to attribute treatment effects to the model presented here if other interventions were controlled. Another limitation involves the length of treatment (8-12 weeks) and limited follow-up to measure effectiveness.
Review of Center Based Interventions
STUDY #1
The first study was conducted by Rogers & Lewis (1988), and evaluates a day treatment program at the University of Colorado Health Sciences Center, that consisted of 31 children (ages 2-6), with a clinical diagnosis of autism, based on the DSM-III and CARS.
The treatment model emphasized play, interpersonal relationships, pragmatic language development strategies, and activities to foster symbolic thought. Data was collected over a five-year period on the 31 children. The children attended the program for at least 6 months prior to the study.
Pretreatement scores were evaluated using the Bayley Scales of Infant Development, the Leiter International Performance Scale, or the Merrill-Palmer Test of Mental Abilities (Rogers & Lewis, 1988). The treatment program was over a 12-month period and the children attended for 4.5 hours per day. The groups consisted of 6 children, 1 teacher and 2 assistants. Each class had a maximum enrollment of 12 children and consulting to each class were, a speech-language pathologist, an occupational therapist, a child psychiatrist, and a child clinical psychologist. Each child participated in individual speech therapy for two half-hour sessions per week. The child psychiatrist or child clinical psychologist consulted with the child’s parents for approximately one hour, once per week. In addition monthly parent support groups were arranged and parents were encouraged to participate in the classroom (Rogers & Lewis, 1988).
Pre and posttreatment measures consisted of the Early Intervention Profile and Preschool Profile (Schafer & Moersch, 1981), that assess development in 5 domains including cognitive, language, motor, social and adaptive. The second measure consisted of the Play Observation Scale (Rogers et al., 1986), that was administered at approximately 8-month intervals through videotaped observation of structured play between the teacher and the child. The third measure, the CARS was administered also via video observation described above and scored by different raters. The authors controlled for maturation using the prediction index method (Fewell & Sandall, 1986) to account for treatment effects over and above maturation effects. The results indicated significant differences for each of the 5 domains; cognitive, language, social, gross motor, fine motor. For the sixth domain, self-help, the difference in scores approached statistical significance (Rogers & Lewis, 1988).
The authors also administered a third assessment of the Developmental Profiles to 15 children, 12 months after their enrollment into the program. Data was collected on these 15 participants to measure effects of either onset of treatment or gains made attributed to longer intervention period.
The authors performed a two-tailed t-test to compare means at Time 2, with the mean rate at Time 3. The results indicated statistical significance in 2 of the 5 domains; social, and self-help, (Rogers & Lewis, 1988). The results from the Play Observation Scale were calculated and Time 1 & 2 was compared on 3 dimensions; symbolic agent level, symbolic substitute level, and symbolic complexity level. The results from this scale were also calculated and Time 1 & 2 was compared for social and communicative play skills. The CARS was administered via videotaped observation of the 28 subjects at Time 1 & 2. The results indicated a significant reduction in means from Time 1 to Time 2 (Rogers & Lewis, 1988).
The authors also compared the children in the study, to examine the effects of differential response to treatment. This was accomplished by grouping the children according to diagnosis that comprised two groups. Nineteen children comprised the first group diagnosed with Infantile Autism-IA. These children were younger and had more severe autistic symptomology according to the CARS results. The second group consisted of 12 children diagnosed as Pervasive Developmental Disorder-PDD. These children were older than the first group and therefore had higher developmental quotients across most domains, and had higher symbolic play skills at onset. Both groups did not differ significantly on IQ. Significant results were reported using a one-tail t-test to compare the projected scores with the actual posttreatment scores for the IA group and PDD group.
Consistent with the authors findings for overall group scores (IA and PDD combined), statistical significance also resulted in the same 5 developmental domains; fine motor; cognition, language, social, gross-motor. The sixth domain, self-help, also did not result in statistical significance. Consistent with the findings of the IA and PDD group combined, differences in scores for the symbolic play scores also resulted with significant increases in: Symbolic Substitute Level, Symbolic Complexity Level and, Social Communicative Level. There was also a reduction in symptomology on the CARS from Time one to Time two. The PDD group comparison between projected scores with actual posttest scores using a one-tailed t-test resulted in statistical significance in any 2 of the six domains cognitive, language. The comparison between Time one and time two symbolic play skills resulted in Symbolic Agent Level, Symbolic Complexity Level. There was also a reduction in symptomology on the CARS from Time one to Time two.
STUDY #2
The second study conducted by Harris and colleagues (1991), compared children with autism and typically developing peers after one school year in terms of language and cognitive functioning. The Douglass Developmental Disabilities Center, a Rutgers University based program, consisted of two classes. One class was segregated and served only children with autism. The second class was integrated and included typically developing peers. Children involved in this program were first enrolled into the segregated class for approximately 1-2 years prior to enrollment in the integrated class.
This study presented the results of this intervention approach after one year of data collection by exploring changes in IQ and language functioning for the children with autism and their normally developing peers. The classes were structured according to a preschool model that also included periods of small group, large-group and one on one teaching. The segregated class consisted of 6 children with autism, 1 teacher and 3 assistants. The integrated class consisted of 6 children with autism and 7 normally developing peers, 1 teacher and 3 assistants. Incidental instruction was the dominant teaching approach. However, Discrete Trial Teaching was used to introduce new materials. Parents were also supported to maintain consistency with home programming. Equal emphasis on daily outdoor activities, weekly classroom themes, monthly field trips and computer instruction (Harris et al., 1991). Two samples consisting of the Stanford-Binet IV group (n=9), and the PLS group (n=16) were compared. The Stanford-Binet IV group had a control group of 9 typically developing peers (8 boys, 1 female) in each group. The PLS group had a control group of 12 typically developing peers (9 boys, 3 female) (Harris et al., 1991) in each group. The mean time from pre to posttesting for the Stanford-Binet IV was 10.89 months. The mean difference in time for the PLS was 9.82 months (Harris et al., 1991). Independent t-tests were used to measure differences from Time 1(autistic M=67.56, normal M=114.11), to Time 2 (autistic M=86.33, normal M=113.67). The IQ measures administered to the children with autism indicated a significant increase of 18.78 points. There was no significant difference for the normal group. The PLS measure administered to the children with autism indicated a significant increase of 8.03 points. Normal peers also exhibited a significant increase of 7.73 points. The results here suggest that a language-intensive style of teaching can benefit young children with autism and their normally developing peers.
STUDY #3
Fenske and colleagues (1985) conducted the final study presented here. It examined age of intervention and treatment outcomes for 18 children with autism in a center based program at Princeton Child Development Institute-PCDI in New Jersey. The first cohort comprised 9 children that entered into the treatment program prior to 60 months of age. The second cohort comprised children that entered into the treatment program after 60 months of age. The children involved in this study included those already enrolled in the PCDI day school and treatment program during the period 1975-1983. Cohort 1& 2 consisted of 8 boys and 1 girl (n=9 each group). Children in cohort 1 were enrolled in the program for 45.9 months. Children in cohort 2 were enrolled in the program for 72.4 months.
The pretreatment measure included the Peabody Picture Vocabulary Test (PPVT) (Fenske and colleagues, 1985). The intervention model employed an applied behavior analysis approach. This comprehensive program at PCDI included behavioral intervention, Teaching-Family Model group homes, parent-training services, individualized transition program and post discharge follow-up services (Fenske and colleagues, 1985). The 18 children in this study participated in the teaching program that involved school attendance of 5.5 hours per day, per 5-day week for a total of 11 months. The school day was divided into eleven 30-minuite classes. To emphasize generalizability, each class was held in a different room with a different teacher and involved different learning activities. One child from cohort 1 and 4 children from cohort 2 was involved in the Teaching-Family Model group homes. Here, the child resided with professional teaching parents (a married couple) in a family-style environment that is systematically coordinated with the intervention program. For the monthly parent training service, a home programmer instructs parents by serving as home therapists for their children incorporating the same model as in the school program. The individualized transition programs are used to gradually fade in community-based activities and to mainstream the children into the school system (Fenske and colleagues, 1985).
Criteria for positive discharge include generalizability of skills to the new public school environment and manageable behavior. Follow up maintenance is first scheduled monthly and then faded to yearly visits. The authors chose a two-sample, static-group design to measure relation between age of entry and treatment outcome. The authors also used chi-square and correctional tests to determine the effects of age at program entry.
The results indicated positive outcome for 6 children in cohort 1, and 1 child in cohort 2. The authors indicated using a chi-square analysis, a greater probability of positive outcome for children entering a treatment program prior to 60 months (Fenske and colleagues, 1985). Four of the six children that met criteria for positive discharge attended regular public school classrooms and 2 attended part time special education classes in public schools. Three of these children resided at home with their parents and 1 child resided in a Teaching-Family Model group home (Fenske and colleagues, 1985). The child that achieved positive discharge from cohort two attended a mainstreamed class in the public school.
