Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
12 Comprehensive Programs An overview of well-known model approaches to early autism in- tervention reveals a consensus across programs on the factors that result in program effectiveness. Similarities far outweigh differences in ten state- of-the-art programs that were selected for comparison. On the other hand, program differences suggest that there are viable alternatives on many program dimensions. Both differences and similarities among the pro- grams are fundamental. Despite limitations of the outcome research avail- able, it is likely that many children benefit substantially in the different programs reviewed. The national challenge is to close the gap between the quality of model programs and the reality of most publicly funded early educational programs. This chapter begins with a description of the process by which the ten models were selected for review and a brief description of each program. The theoretical backgrounds of the various approaches are then consid- ered, followed by an examination of points of convergence and diver- gence across the program models and consideration of the empirical un- derpinnings of each approach. SELECTION AND OVERVIEW OF MODEL PROGRAMS Representative model programs were selected for the purpose of il- lustrating key features related to program effectiveness; however, this is not an exhaustive review, and not all existing programs are described here. 140
COMPREHENSIVE PROGRAMS 141 Criteria for Selection of Programs In order to select representative programs objectively, the committee established a set of criteria that relied on the availability of recently pub- lished program descriptions (Harris and Handleman, 1994; Handleman and Harris, 2000) and existing reviews of model programs for children with autistic spectrum disorders (Dawson and Osterling, 1997; Rogers, 1998). The committee also reviewed research and program descriptions in recent special issues on autistic spectrum disorders of professional journals, including Infants and Young Children (Neisworth and Bagnato, 1999), School Psychology Review (Harrison, 1999), and The Journal of the Association for Persons with Severe Handicaps (Brown and Bambara, 1999). Programs that had received federal funding for peer-reviewed grants by the National Institutes of Health and by the U.S. Department of Education were also included. Model programs that provided invited representa- tion in the Autistic Spectrum Disorders Forum Workgroup of the Na- tional Early Childhood Technical Assistance Systems were also included. A simple frequency count was conducted of the number of times each program was described in these sources. The programs selected were cited and described as program models between three and nine times in the designated resources. Excluded from the count were publications of isolated procedures rather than overall program descriptions. For ex- ample, references to an incidental teaching or discrete-trial procedure were not counted as a reference to a specific program model. However, references to a model by either title or investigator(s) were counted. These criteria yielded a total of 12 programs, all in the United States. The committee sent an invitation to the director or developer of each, asking for program description materials and peer-reviewed data that they deemed best represented their model. Two of the programs did not respond, leaving ten programs for the committeeâs review. Brief Overview of Programs Most of the ten representative models selected began as research pro- grams in which empirically demonstrated strategies for addressing spe- cific problems were gradually packaged as components of overall clinical models. However, there have been different approaches to the develop- ment of these models. All ten of the models individualize programming around the needs of particular children, and intervention regimens are designed to be imple- mented in a flexible manner. Essential differences in program design pertain to whether the curriculum is aimed at addressing some or all of a childâs needs and whether the program staff provide direct service or serve as consultants to external providers. The following description of
142 EDUCATING CHILDREN WITH AUTISM the ten programs (presented in alphabetical order), and the review that follows it, summarize the similarities and differences across programs. Childrenâs Unit at the State University of New York at Binghamton This program was designed in 1975 as an intensive, short-term program (ap- proximately 3 years) for children with severe behavioral disorders. Con- sistent with the original purpose, the program operates from a deficit- oriented perspective that seeks to identify the factors most crucial in preventing a child from benefiting fully from services in the local commu- nity. The program primarily uses traditional applied behavior analysis techniques, although more naturalistic procedures may be implemented as children progress. An elaborate individualized goal selection curricu- lum has been developed, and there is an extensive computerized assess- ment and monitoring system (Romanczyk et al., 2000). Denver Model at the University of Colorado Health Sciences Center This program originally opened in 1981 as the Playschool Model, which was a demonstration day treatment program. This developmentally oriented instructional approach is based on the premise that play is a primary vehicle for learning social, emotional, communicative, and cognitive skills during early childhood. The role of the adult and the purpose of play activities vary across learning objectives. The overarching curriculum goals are to increase cognitive levels, particularly in the area of symbolic functions; increase communication through gestures, signs, and words; and enhance social and emotional growth through interpersonal relation- ships with adults and peers. In 1998, the treatment unit was closed, and the intervention format was changed to the more natural contexts avail- able in home and preschool environments with typical peers (Rogers et al., 2000). Developmental Intervention Model at The George Washington Univer- sity School of Medicine As in the Denver Model, this relationship-based approach is derived from a developmental orientation. There is a home component of intensive interactive floor-time work, in which an adult follows a childâs lead in play and interaction, and children concurrently participate in individual therapies and early education programs. Intense floor time sessions at home are aimed at âpulling the child into a greater degree of pleasure.â The curriculum is aimed at six developmental ca- pacities: shared attention and regulation; engagement; affective reciproc- ity and communications through gestures; complex, pre-symbolic, shared social communication and problem-solving; symbolic and creative use of ideas; and logical and abstract use of ideas and thinking (Greenspan and Wieder, 1999).
COMPREHENSIVE PROGRAMS 143 Douglass Developmental Center at Rutgers University The center opened in 1972 to serve older children with autism; the preschool pro- grams were added in 1987. Douglass now has a continuum of three programs that serve young children with autistic spectrum disorders, including an intensive home-based intervention, a small-group segregated preschool, and an integrated preschool. The curriculum is developmen- tally sequenced and uses applied behavior analysis techniques, beginning with discrete-trial formats and shifting across the continuum to more naturalistic procedures. Initial instruction is focused on teaching compli- ance, cognitive and communication skills, rudimentary social skills, and toilet training, as well as on the elimination of serious behavior problems. The small-group classroom emphasizes communication, cognitive skills, and self-help skills; social intervention begins in the form of interactive play with teachers. The emphasis in the integrated classroom is on com- munication, socialization, and pre-academic skills (Harris et al., 2000). Individualized Support Program at the University of South Florida at Tampa A parent-training program developed in West Virginia served as the predecessor of this model, which started in its current form in 1987. The Individualized Support Program is implemented in childrenâs homes and community settings during a relatively short period of intensive as- sistance and ongoing follow-up. The program is intended to be adjunc- tive to ongoing, daily, special educational services delivered in preschool and by other clinical providers. Specifically, it is oriented toward helping families gain the knowledge and skills needed to solve problems, as well as the competence and confidence needed to continue effective interven- tion and advocacy over the course of their childrenâs educational history. Essential elements of the model include: development of functional com- munication skills, facilitation of the childâs participation in socially inclu- sive environments, and multifaceted family support (Dunlap and Fox, 1999a, 1999b). Learning Experiences, an Alternative Program for Preschoolers and their Parents (LEAP) Preschool at the University of Colorado School of Education LEAP opened in 1982 as a federally funded demonstration program and soon after incorporated into the Early Childhood Interven- tion Program at Western Psychiatric Institute and Clinic, University of Pittsburgh. In recent years, the original classrooms continue to operate in Pittsburgh, but new LEAP classrooms are now being developed in the Denver Public School System. LEAP includes both a preschool program and a behavioral skill training program for parents, as well as national outreach activities. LEAP was one of the first programs in the country to include children with autism with typical children, and the curriculum is well-known for its peer-mediated social skill interventions. An individu-
144 EDUCATING CHILDREN WITH AUTISM alized curriculum targets goals in social, emotional, language, adaptive behavior, cognitive, and physical developmental areas. The curriculum blends a behavioral approach with developmentally appropriate prac- tices (Strain and Cordisco, 1994; Strain and Hoyson, 2000). Pivotal Response Model at the University of California at Santa Barbara Beginning in 1979, components of the current model were evaluated in applications with children of varied ages. In recent years, the primary focus has been on early intervention. Using a parent education approach, the ultimate goal of the Pivotal Response Model is to provide individuals with autism with the social and educational proficiency to participate in inclusive settings. In early stages, this model used a discrete-trial applied behavior analysis approach, but there has been a shift toward use of more naturalistic behavioral interventions. The overriding strategy is to aim at change in certain pivotal areas (e.g., responsiveness to multiple cues, motivation, self-management, and self-initiations). Intervention consists of in-clinic and one-on-one home teaching, and children concurrently par- ticipate in special education services in the schools. Specific curriculum goals are targeted in areas of communication, self-help, academic, social, and recreational skills (Koegel et al., 1998). Treatment and Education of Autistic and Related Communication Handi- capped Children (TEACCH) at the University of North Carolina School of Medicine at Chapel Hill This program was founded in 1972 as a state- wide autism program that serves people with autistic spectrum disorders of all ages. Regional centers provide regular consultation and training to parents, schools, preschools, daycare centers, and other placements throughout the state. There is one demonstration classroom. TEACCH is based on a structured teaching approach, in which environments are or- ganized with clear, concrete, visual information. Parents are cotherapists and taught strategies for working with their children. Programming is based on individualized assessments of a childâs strengths, learning style, interests, and needs, so that the materials selected, the activities devel- oped, the work system for the child, and the schedule for learning are tailored to this assessment information and to the needs of the family. TEACCH has developed a communication curriculum that makes use of behavioral procedures, with adjustments that incorporate more naturalis- tic procedures along with alternative communication strategies for non- verbal children (Watson et al., 1989; Marcus et al., 2000). The University of California at Los Angeles (UCLA) Young Autism Project The development of this program was based on earlier research with older children and adolescents with autism; its applications to young children with autism began during the 1970s. The behavioral interven-
COMPREHENSIVE PROGRAMS 145 tion curriculum is delivered in a one-to-one discrete-trial format, which is implemented by parents and trained therapists who work in a childâs home. The treatment is focused primarily on developing language and early cognitive skills and decreasing excessive rituals, tantrums, and ag- gressive behaviors. The first year of intervention is aimed at teaching children to respond to basic requests, to imitate, to begin to play with toys, and to interact with their families. During the second year, the focus on teaching language continues; the most recent curriculum descriptions note a shift toward teaching emotion discriminations, pre-academic skills, and observational learning. For children who eventually enter inclusive settings, a paraprofessional assists with participation in regular preschool or kindergarten settings (Smith et al., 2000a). Walden Early Childhood Programs at the Emory University School of Medicine The Walden program was developed in 1985 at the University of Massachusetts at Amherst, where the primary function was as a labora- tory preschool to accommodate research in incidental teaching. Follow- ing relocation to Emory University in Atlanta, toddler and prekinder- garten programs were added to complete an early intervention continuum. The classrooms include children with autism with a majority of typical peers. The incidental teaching approach is based on behavioral research, although there are developmental influences on goal selection. There is a toddler program with both center- and home-based compo- nents, and initial goals include establishment of sustained engagement, functional verbal language, responsiveness to adults, tolerance and par- ticipation with typical peers, and independence in daily living (e.g., toilet training). The preschool is aimed at language expansions and beginning peer interaction training. The prekindergarten emphasizes elaborated peer interactions, academic skills, and conventional school behaviors (McGee et al., 2000). Organizational Structures Irrespective of curriculum content, there are certain organizational similarities in the ten selected programs. For example, all are university- based programs. Four are housed within psychiatry departments (Den- ver, Developmental Intervention Model, TEACCH, Walden; also formerly LEAP), and four are affiliated with psychology departments (Childrenâs Unit, Douglass, Pivotal Response Model, Young Autism Project; also for- merly Walden). The Individualized Support Program is sponsored by a Department of Child and Family Studies, and LEAP is currently in a Department of Special Education. Virtually all of these programs are or formerly were a component of a larger autism center. The Denver Model is operated within one of the
146 EDUCATING CHILDREN WITH AUTISM National Institutes of Health Autism Research Centers. Three of the pro- grams are components of statewide autism centers (i.e., the Individual- ized Support Program in Florida, TEACCH in North Carolina, and Walden in Georgia), and two other preschools (Douglas Disabilities De- velopmental Center and the Childrenâs Unit) are the early childhood com- ponents of programs that serve people with autism through adulthood. Three programs operate out of university-based clinics, although a sig- nificant portion of the interventions take place in homes and community settings (i.e., Developmental Intervention Model, Pivotal Response Train- ing, and the Young Autism Project). The LEAP, TEACCH, and Denver programs are carried out primarily in public schools; all programs pro- vide consultation or technical assistance to schools serving participating children, either concurrently or following early intervention. Many of the selected programs were developed while funded with extramural research support. At least seven of the programsâ directors have or have had funding from the U.S. Department of Education (Dunlap and Fox, the Koegels, Lovaas, McGee, Rogers, Schopler, and Strain). Five of these program directors have had research funding from the National Institutes of Health (Koegel, Lovaas, Rogers, Schopler, and Strain). In addition, virtually all have had state funding, either directly (e.g., Childrenâs Unit, Douglass, TEACCH) or through child or school district tuitions. Trends in the Development of the Programs This review focuses on the most recently published practices of each model; it should be acknowledged that each of these programs has under- gone considerable evolution over the years. Over the past two decades, the development of preschool programs for children with autistic spec- trum disorders has influenced and been influenced by major shifts in intervention approaches (Dunlap and Robbins, 1991). Early behavioral interventions often targeted behavior reduction as a major goal, and some used aversive procedures. However, very few programs for young chil- dren currently report planned use of aversive stimuli as punishments. Another trend includes broadened conceptualizations of family involve- ment, which has expanded from simple participation in parent training to preparation for parental roles as collaborators, advocates, and recipients of family support. There has also been a shift toward instruction in more natural environments, and there has been a growing emphasis on inclu- sion of children with autism with typically developing peers. For ex- ample, virtually all model programs list inclusion among typical peers as a major emphasis of their program, either as a goal or as a strategy for promoting social learning (Handleman and Harris, 2000; Harris and Handleman, 1994). In the past few years, there has been an increased
COMPREHENSIVE PROGRAMS 147 focus on identification and treatment of toddlers with autistic spectrum disorders, in contrast to previous models of early intervention that began when children were in preschool or elementary school. THEORETICAL ORIENTATIONS OF PROGRAM MODELS The ten program models described derive from either developmental or behavioral orientations, which influence goals, intervention proce- dures, and methods of evaluation. Thus, the Denver Model and the De- velopmental Intervention Model were conceptualized from a deductive framework, in which developmental theory was used to organize hypoth- eses regarding the fundamental nature of autistic spectrum disorders. Group design research has been aimed at seeking confirmatory evidence regarding deviations from normal development that need to be addressed in intervention. With the exception of TEACCH, which is eclectic with elements of both developmental and behavioral orientations, the other seven programs derive from the field of applied behavioral analysis. Be- havioral interventions have been developed from a âbottom-upâ approach in which procedures based on principles of learning are subjected to (largely single-subject) empirical tests, and techniques of demonstrated efficacy have then been assembled into program models (Anderson and Romanczyk, 1999). Although these differing conceptual frameworks influence the inter- vention models in substantial ways, in practice, there is also considerable overlap between and across the various models. Within the behavioral approaches, a wide range of applications are used within and across pro- grams, ranging from traditional discrete-trial training procedures to newer naturalistic approaches. Developmental Approaches The Denver Model recognizes the interplay among cognitive, com- municative, and social and emotional development (Rogers and DiLalla, 1991). It was originally based on Piagetâs (1966) experientially based theory of cognitive development, with additional influence from Mahlerâs conceptualization of interpersonal development via the attachment-sepa- ration-individuation process (Mahler et al., 1975). The underlying as- sumption was that, if intervention is directed at establishing strong, affec- tionate interpersonal relationships, then it may be possible to accomplish broad developmentally crucial improvements. From this perspective, it has been argued that the traditional behavioral approach of teaching spe- cific behaviors is too narrow to have an impact on the fundamental nature of autistic spectrum disorders (Rogers et al., 1986). Although the Pivotal Response Model evolved from behavioral research, it arrived at a similar
148 EDUCATING CHILDREN WITH AUTISM conclusion, that it is more efficacious to aim intervention at key autistic spectrum disorders deficits that will yield broad changes in collateral behaviors than to address individual behaviors in an isolated fashion (Koegel et al., 1999a). The approach of the Developmental Intervention Model is based upon the assumption that a childâs symptoms reflect unique biologically based processing difficulties that may involve affect, sensory modulation and processing, motor planing, and symbol formation (Greenspan and Wieder, 1997). Relationships and affective interactions may go awry sec- ondarily, and intervention is aimed at helping a child try to work around the processing difficulties to reestablish affective contact. Behavioral Approaches By far, the bulk of autistic spectrum disorders intervention research has been conducted from the perspective of applied behavior analysis. An exhaustive review of 19,000 published journal articles revealed that there were 500 papers on applied behavior analysis and autistic spectrum disorders, and 90 of these were studies using single-subject designs to evaluate specific interventions for young children with autistic spectrum disorders (Palmieri et al., 1998). Rather than being tied to specific proce- dures, applied behavior analysis includes any method that changes be- havior in systematic and measurable ways (Sulzer-Azaroff and Mayer, 1991). Historically, the behavioral approaches emphasized acquisition of discrete skills, and interventions were evaluated in terms of whether they produced observable and socially significant changes in childrenâs behav- ior (Baer et al., 1968). Traditional behavioral interventions impose structure in the form of distraction-free environments and presentation of opportunities-to-re- spond in discrete trials, and appropriate behavior is rewarded when it occurs. Technically sophisticated discrimination training procedures have been derived from years of research in applied behavior analysis. Lovaasâ Young Autism Project, Harris and Handlemanâs Douglass Center, and Romanczykâs Childrenâs Unit represent classic behavioral interventions, although all now use more naturalistic interventions as childrenâs basic skills improve. In an effort to improve the generalization of skills from teaching set- tings to daily use in the real world, comprehensive behavioral interven- tions have modified traditional applied behavior analysis techniques in a way that permits instruction in natural environments. The LEAP model was the first to recognize the importance of direct instruction in peer- related social behaviors, and that more natural instructional settings were required to accommodate the presence of typically developing classmates (Strain and Hoyson, 2000; Strain et al., 1985). Waldenâs incidental teach-
COMPREHENSIVE PROGRAMS 149 ing approach incorporated the technical procedures generated by behav- ior analysis research into the environmental contexts in which social and communication behaviors typically occur for children without disabilities (McGee et al., 1997, 1999). Both the Individualized Support Program (Dunlap and Fox, 1999b) and the pivotal response model have empha- sized the use of naturalistic procedures as a method to reduce stress on families. Conceptual Differences and Practical Similarities The conceptual differences between developmental and behavioral approaches to intervention are real, yet the gaps in practice appear to be narrowing. Developmental researchers may criticize behavioral ap- proaches for failure to target the specific deficits associated with autistic spectrum disorders (Rogers et al., 1986), and it has been argued that this failure to select target skills within a meaningful developmental frame- work results in isolated skills that are difficult to transfer to other situa- tions and skills (Rogers and Lewis, 1988). Behaviorists counter that the irregularity of skill development in children with autistic spectrum disor- ders decreases the relevance of careful adherence to normal developmen- tal sequencing (Anderson and Romanczyk, 1999). However, develop- mental approaches to autistic spectrum disorders treatment have incorporated methods that recognize the needs of children with autistic spectrum disorders for high levels of structure, adult attention, and con- sistency. At the same time, behavioral interventions are increasingly be- ing used to address complex social and communication goals in normal environmental settings. CONVERGENCE AND VARIABILITY OF PROGRAM DIMENSIONS Common elements among the early intervention models presented here include specific curriculum content, highly supportive teaching en- vironments and generalization strategies, predictable routines, use of a functional approach to problem behaviors, carefully planned transitions across intervention settings, and active family involvement (Dawson and Osterling, 1997). Additional similarities include highly trained staff, ad- equate resources, and supervisory and review mechanisms (Anderson and Romanczyk, 1999). All ten model programs/approaches recognize the importance of individualizing interventions in a manner that meets the needs of each child and family. The similarities and range of variabil- ity of features across the models are summarized in Table 12-1 and dis- cussed below.
150 EDUCATING CHILDREN WITH AUTISM TABLE 12-1 Features of Comprehensive Programs Mean Age at Entry Primary (range), Hours Usual Teaching Program in Months Per Week Settinga Procedure Childrenâs Unit 40 27.5 School (S) Discrete trial (13 to 57) Denver 46 20 School (I), home, Playschool Community (24 to 60) community curriculum Based Approach Developmental 36 10-25 Home, clinic Floor time Intervention (22 to 48) therapy Model Douglass 47 30-40 School (S and I), Discrete trial; (32 to 74) home naturalistic Individualized 34 12 School (I), home, Positive Support (29 to 44) community behavior Program support LEAP 43 25 School (I), home Peer-mediated (30 to 64) intervention; naturalistic Pivotal 36 Varies School (I), home, Pivotal Response (24 to 47) community, clinic response Training training TEACCH 36 25 School (S), clinic Structured (24 and up) teaching UCLA Young 32 20-40 Home Discrete-trial Autism Project (30 to 46) Walden 30 36 School (I), home Incidental (18 to 36) teaching a(S) segregated classroom; (I) inclusive classroom
COMPREHENSIVE PROGRAMS 151 Intervention Begins Early All ten programs emphasize the importance of starting intervention when children are at the earliest possible ages. At least two retrospective studies have found less restrictive placement outcomes for children who began intervention at earlier rather than later ages (Fenske et al., 1985; Handleman and Harris, 2000). Several of the approaches were initially developed for elementary aged students and gradually applied to chil- dren at increasingly early ages (e.g., Douglass, Pivotal Response Training, TEACCH). Others were developed specifically for preschool-aged chil- dren (e.g., Denver, Individualized Support Program, LEAP). Although several programs (e.g., Developmental Intervention Model, Young Au- tism Project) have accepted children at ages younger than preschool, only the Walden toddler program was specifically designed to address the needs of toddlers with autistic spectrum disorders. Questions of how best to modify well-established approaches to fit the needs of very young children and their families are critical in future planning as children are identified at earlier ages. Extension of services to children younger than preschool ages has sometimes been limited by funding mechanisms, which apply when chil- dren turn 3 years of age. In addition, a few approaches have established cutoffs for cognitive functioning that impose some limits on entry to in- tervention at the earliest ages. Despite policy and funding influences, all ten programs show recognition of the importance of early intervention by reporting outcome data on at least some children below the age of 3 years (see below). Intervention Is Intensive in Hours All of the comprehensive program models that are introduced to pro- vide a childâs major educational program report children participating in from 20 to 45 hours of intervention per week (see Table 12-1). The pro- grams usually operate on a full-year basis, across several early childhood years. Lovaas (1987) provided the most direct evidence of the importance of intervention intensity in a comparison of 40 hours per week of tradi- tional behavioral intervention compared with less than 10 hours per week of the same intervention. However, Sheinkopf and Siegel (1998) did not find a dose-response relationship between more than 20 hours of inter- ventions and outcome, in part because childrenâs skills at entry were such strong predictors of improvement. In addition, there is indirect evidence of the importance of intervention hours in a comparison of replications of the UCLA program, which have used fewer hours (18 to 25 hours per week) and obtained positive but more limited results (Smith, 1999; Smith et al., 2000a).
152 EDUCATING CHILDREN WITH AUTISM The Individualized Support Program and Pivotal Response Training Models offer relatively few hours of intervention per week; however, these approaches were designed to be provided in addition to other edu- cational and therapeutic interventions. Similarly, the Developmental In- tervention Model is implemented at widely varying numbers of hours per week (Greenspan and Wieder, 1997). However, the outcome evaluation of this approach included only children who had received at least 2 years of intervention (see below), which serves to illustrate that intensity en- compasses duration as well as hours per week of participation. Hours reported for the TEACCH model were also few, but these are hours of technical assistance provided by TEACCH regarding individual children, and not the total number of hours of educational intervention received by each child. Usually, children identified with autism who receive TEACCH services begin full-day schooling, in a range of different placements and combinations of placements, at age 3 years. Another source of variability in hours is how much emphasis each model places on intervention by families. For example, although the LEAP model has been cited as providing relatively low levels of hours intensity (Dawson and Osterling, 1997), a very active family program component easily expands the intervention to more than 25 hours per week (Strain and Cordisco, 1994). However, there is a range of family capability to provide intervention (Dunlap and Fox, 1999a), and although widely regarded as crucial, family intervention hours are difficult to quan- tify precisely. Intervention intensity cannot be simply measured in terms of hours of enrollment or even attendance in an intervention program (Wolery and Garfinkle, 2000). In other words, hours of participation do not unilater- ally translate to hours of time engaged in intervention. It has been argued that intensity is best thought of in the context of âlarge numbers of func- tional, developmentally relevant, and high-interest opportunities to re- spond activelyâ (Strain and Hoyson, 2000). Evidence from the general education and developmental literature support claims that intensity of education is associated with amount of progress (Ceci, 1991; Frasier and Morrison, 1998). While some level of dose-response relationship might be expected if the âactive componentâ of the âdoseâ could be measured accurately, simplistic dose-response functions of intensity are not charac- teristic of typical child development. Rather, a more productive focus might be how variations in intensity are associated with day-to-day con- texts (e.g., full school day compared with a 2-hour preschool program; full-year programming compared with a 9- or 10-month school year). Families Are Actively Involved in Their Childrenâs Intervention All ten models explicitly acknowledge the importance of having par- ents play a central role in their childrenâs intervention, although how
COMPREHENSIVE PROGRAMS 153 parents are asked to participate differs across programs and approaches. Virtually all of the programs provide parents with at least the opportu- nity to be trained in specialized skills in teaching their children with autistic spectrum disorders. However, there is a wide range of how many hours parents are asked to participate and whether or not that participation is optional or re- quired. At least three of the programs (Developmental Intervention Model, Walden, and the Young Autism Project) require a parental com- mitment to deliver at least 10 hours of intervention per week in their homes or community settings. These programs provide parents with extensive instruction and supervision on the specialized skills needed to effectively teach their child with an autistic spectrum disorder. The Douglass and LEAP programs strongly recommend parent participation for 10 to 15 hours per week, but they do not require completion of the assignments to parents. The format of parent participation varies considerably across pro- grams, but all provide for some individual meetings with professionals at a clinic, center, or home. In some programs (Developmental Intervention Model, Young Autism Project), family intervention requires that parents set aside time to work intensively with their child in a one-on-one format. In others (Individualized Support Program, Pivotal Response Training, Walden), parental instruction is blended into normal daily home and community activities. The Denver Model, which is community-based, aims for a combination of intervention in both one-on-one and natural contexts (Rogers et al., 2000). An example of the range of formats offered is available from the Douglass Center, which offers a workshop on ap- plied behavior analysis, formal clinics with therapist modeling and par- ent-demonstration of skills, preschool observational clinics, home visits twice a month, voluntary support groups, sibling groups, and four educa- tional meetings per year (Harris et al., 2000). Other programs place em- phasis on training advocacy skills (McGee et al., 1999), and a few offer psychological counseling or social work support services (Rogers and Lewis, 1988; Romanczyk et al., 2000). Parentsâ observation of childrenâs school participation is another venue for parent education (Rogers and Lewis, 1988). There is an increasing trend toward providing families with support to deal with the considerable emotional and logistical stresses of raising a child with an autistic spectrum disorder, so that intervention goes beyond parent training. For example, the Individualized Support Program was explicitly designed to accommodate the individualized needs of families (Dunlap and Fox, 1999b). The intervention begins with a family needs assessment to determine whether parents require information to increase their understanding of the disability, assistance in gaining access to ser- vices, skills for improving interactions with their child, or other family, social, or financial support. Families receive home visits for several hours
154 EDUCATING CHILDREN WITH AUTISM per week, along with phone consultations as needed. Parents may also be accompanied on visits to physicians, other providers, or Individualized Educational Plan (IEP) meetings at their childrenâs schools. In sum, all of the model programs reviewed placed a high priority on parental involvement in the early education of their children with autistic spectrum disorders. In addition, the trend towards broadened parent supports reflects an appreciation of the challenges faced by these families. Staff Are Highly Trained and Specialized in Autism All ten programs are directed by at least one doctoral-level profes- sional with a long-standing reputation in the treatment of autistic spec- trum disorders. All the program developers have demonstrated academic productivity as evidenced by their status on a university faculty. The program developers have Ph.D.s in various fields (e.g., clinical psychol- ogy, developmental psychology, experimental psychology, special edu- cation, speech and hearing pathology); one has an M.D. In addition, virtually every program developer is assisted by either doctoral or masterâs-level personnel who have worked collaboratively with the pro- gram director for several years. Professional staff members in the selected programs are broadly in- terdisciplinary, and staffing patterns vary according to local licensing and accreditation guidelines. The two certified school programs (Childrenâs Unit and Douglass) have staff with the most traditional credentials, in- cluding certified teachers, speech and language pathologists, and an adap- tive physical educator (Harris and Handleman, 1994; Romanczyk et al., 2000). The Childrenâs Unit has a social worker, school psychologist, art and music therapists, and a consulting occupational therapist. The inclu- sion of typical children at LEAP and Walden requires that at least some teachers have degrees in regular or special early childhood education (McGee et al., 1994; Strain and Cordisco, 1994). Most of the programs also have an array of bachelorâs level staff (who often have degrees in psychol- ogy). The Individualized Support Program and LEAP have employed a parent of a child with a disability to work directly with families, and LEAP also invites and trains parents to work as staff volunteers in their classrooms (Dunlap and Fox, 1999b; Strain and Cordisco, 1994). The programs vary in terms of their use of specialized therapists and whether or not those therapists are part of the regular staff. In the pro- grams with specialty therapists, there is variability in whether one-to-one direct therapy is provided. In most of the programs, there is an emphasis on the therapistâs role as a consultant to the classroom staff, so that thera- peutic suggestions can be blended into the regular daily intervention program. In almost all of the programs, college students play key roles in their
COMPREHENSIVE PROGRAMS 155 service delivery systems. One advantage of these programsâ university affiliations is the relatively low-cost labor pools of students, who range from undergraduates to graduate students to post-doctoral fellows. In a number of programs, the bulk of the direct services are provided by su- pervised college students. In addition to obvious cost advantages, the reliance on student labor provides the opportunity to expand expertise in the autistic spectrum disorders to future professionals. Elaborate training and supervisory systems have been developed to accommodate the training and supervision needs of the student person- nel. For example, Childrenâs Unit provides a 1-week initial orientation with lectures, observation, and a weekend âimmersionâ training session (Romanczyk et al., 2000). Student trainees must pass a written exam on program policies and procedures, and they are videotaped in interactions with children before and after training. Following initial training, there are weekly supervision and feedback sessions, as well as two objective performance reviews each year. In the UCLA Young Autism Project, the primary therapists are un- dergraduates who have worked for a minimum of 6 months under super- vision (McEachin et al., 1993). Supervisory staff must have a masterâs degree in psychology and 2 or more years of experience with the inter- vention program. This project, like many of the others, has packaged both manuals (Lovaas et al., 1981) and tapes (Lovaas and Leaf, 1981) to standardize personnel training. The Young Autism Project is also en- gaged in large-scale program replication activity. There have been pub- lished outcome reports on systematic (or partial) replications at the May Center in New England, (Anderson et al., 1987) and at UCLA with chil- dren with pervasive developmental delaysânot otherwise specified (Smith et al., 2000b). Standardization of the training protocols has permitted most of the programs to be replicated outside the administrative umbrella of the origi- nal site. Having developed replication formats early on in the process of building a statewide system, TEACCH has now been replicated interna- tionally (in Denmark, France, Norway, Sweden, and Switzerland). TEACCH offers well-known teacher training workshops in North Caro- lina and at other locations around the country. A number of the programs were developed as model demonstration programs (Denver, Individual- ized Support Program, LEAP, and Walden) with support from the U.S. Department of Education, and these grants came with the requirement that the models be packaged and tested in replication sites. In an evaluation of one of these model replications (Rogers et al., 1987), the Denver Model was disseminated to four public schools by us- ing a standardized teacher training approach. Preservice training in- cluded a 6-hour introductory workshop, a 1-day visit to the new site to determine needs and resources, and a 40-hour training institute (which
156 EDUCATING CHILDREN WITH AUTISM included 20 hours of didactic presentations, 8 hours of guided observa- tion, and 12 hours of individual consultation on plans for implementing the model in the new site). Program implementation was monitored with videotaped samples, and formal feedback on teacher performance was provided to replication site staff at 6-week intervals across a period of 4 months. The trainers also conducted at least two 2-day follow-up visits to each replication site. Multidimensional program evaluation data (includ- ing surveys of trainee satisfaction, pre- and post-knowledge tests, a model implementation checklist that was completed with objective scoring of videotaped teacher performances, and measures of child change) docu- mented the effectiveness of this comprehensive training model. The model programs are being directed and implemented by teams of professionals who have had extensive training and experience in early autistic spectrum disorders intervention. It is unlikely that similar child outcomes can be achieved if expertise in autistic spectrum disorders is not readily available. However, the use of student personnel and replication demands have driven the preparation of training formats that could be effective in expanding the number of personnel qualified in education of young children with autistic spectrum disorders. Ongoing Objective Assessment of a Childâs Progress Although the assessment measures varied, all ten programs reviewed have a mechanism for tracking the progress of individual children, and the systems for ongoing assessment permit timely adjustments in the childâs intervention plan. As would be predicted by both the goals and associated methodological trends, the programs with a developmental orientation tend to rely on standardized assessment instruments, while the applied behavior analysis programs include a component for direct observation and measurement of specific target behaviors. However, the behavioral programs also collect standardized assessment data for pur- poses of program evaluation, and the developmental programs have means for ongoing tracking of child progress. Specific issues pertaining to assessment are discussed in Chapter 2 of this report; this section emphasizes the unique methods of assessment that are used by the selected programs. However, as noted in Chapter 2, nearly all of the programs have collected data using the Childhood Au- tism Rating Scale (Schopler et al., 1988), the Vineland Scales of Adaptive Functioning (Sparrow et al., 1984), and one of several available. The Developmental Intervention Model uses an instrument called the Functional Emotional Assessment Scale (FEAS), which is used to assess developmental levels of emotional, social, cognitive, and language func- tioning at the time of the initial evaluation and at each follow-up visit to the clinic (Greenspan and Wieder, 1997). Detailed therapist notes written
COMPREHENSIVE PROGRAMS 157 at the time of each appointment are also viewed as a source of data that is used to monitor child progress. The Denver Model uses an instrument called the Early Intervention Profile and Preschool Profile (Schafer and Moersch, 1981), which is com- pleted by teachers. More objective measures of child progress were also obtained from systematic scoring of videotaped vignettes of a child in play interactions with teachers and parents (Rogers et al., 1986). The Play Observation Schedule was used to rate the developmental level of a childâs play. The traditional behavioral programs (Childrenâs Unit, Douglass, Young Autism Project) tend to rely on trial-by-trial teacher-collected data, which is graphed daily and reviewed weekly or quarterly. Behavior analyses are conducted to provide information regarding the frequency, intensity, and duration of each target behavior, and more detailed func- tional analyses may be accomplished to determine the controlling ante- cedent or consequent events. The Childrenâs Unit has one of the most elaborate data collection systems, in which a rotating videotaping sched- ule is scored for multiple behaviors and subsequently analyzed in a com- puterized database for the rate and pattern of specific behaviors (Romanczyk et al., 2000). A similar system is in place in the Walden programs, with the major exception being a relatively stronger emphasis on tracking ongoing language and social behavior in free-play activities, in contrast to tracking specific problem behaviors or skills during direct instruction (McGee et al., 1997). For programs in which children are learning in the course of naturally occurring early childhood activities, it is difficult to obtain trial-by-trial data. The solution selected by most of these programs has been to obtain videotaped samples and score them according to operational definitions of various behaviors of relevance to the instructional curriculum. For example, the LEAP program obtains 20-minute videotaped probes of par- ent-child interactions, and tapes are scored and reduced in terms of the percentage of intervals the child is engaged in appropriate behavior. In addition, the LEAP program developed a detailed system for analyzing various components of peer interactions (Kohler et al., 1996). The emphasis of the Pivotal Response Model on communication is reflected in the collection of unstructured videotaped language samples (Koegel et al., 1999a), which are analyzed according to Brownâs pragmat- ics criteria (Miller, 1981). Videotapes of parent-child interactions are also obtained under standardized probe conditions and scored for levels of child initiations. In addition, community functioning data is collected by this and other models, including information from report cards and school files regarding school placement, academic achievement, social circles, living situation, and extracurricular activities. The Individualized Support Program obtains systematic videotaped
158 EDUCATING CHILDREN WITH AUTISM probes an average of once or twice per week. The tapes allow tracking of child functioning under conditions in which staff maintains consistent demands and reinforcers. Unique to this model is the family-guided developmental and ecological assessment format, along with systematic tracking of the person-centered planning accomplishments related to the person-centered planning process. As in the more traditional behavioral programs, functional assessments are conducted to develop a plan for reducing significant problem behaviors. In keeping with its community- based emphasis, this approach uses interview and direct observation forms that are more likely to be used in non-research settings than the strict analogue assessment conditions that are described in the research literature (Carr et al., 1994; OâNeill et al., 1997). To summarize, ongoing assessment of childrenâs progress is viewed as a hallmark of each of the model programs, although the methods of measurement logically vary with the curriculum emphasis. Virtually all of the model programs assess cognitive functioning, while relatively few directly assess the effects of intervention on a childâs everyday social functioning. Curricula Provide Systematic, Planful Teaching Each of the program models has a custom-designed curriculum, a term used broadly here to refer to the environment, staffing, materials, and teaching interactions. Several of the programs have commercially packaged portions of their curriculum, including the Childrenâs Unit (Romanczyk et al., 1998), the Developmental, Individual Difference, Rela- tionship-Based Model (DIR; Greenspan and Wieder, 1998), TEACCH (Schopler, 1995; Schopler et al., 1980, 1983; Watson et al., 1989), and the Young Autism Project (Lovaas et al., 1981). The other models have un- published program manuals for use in staff training and program replica- tion activities. Some of the programs make use of other commercially packaged cur- riculum materials. For example, LEAP uses the Creative Curriculum (Dodge and Colker, 1988) to organize activities of interest to typical chil- dren as well as children with autistic spectrum disorders, although these materials are only one component of the overall LEAP curricula (Hoyson et al., 1984). There are many shared features in these varied model curricula. These points of convergence, as well as some interesting points of divergence, are discussed in the rest of this section. Highly Supportive Physical, Temporal, and Staffing Environments As described above, the model programs are implemented in a wide
COMPREHENSIVE PROGRAMS 159 range of environments, including classrooms, homes, clinics, and com- munity settings. The programs also vary on dimensions of environmen- tal stimulation, with traditional behavioral programs generally conducted in distraction-free settings and more naturalistic procedures being imple- mented in more âeverydayâ environments. However, even in the most natural environments, it is common that the curriculum specifies certain environmental arrangements. For example, the early Denver classroom was described as being âchoreographedâ in a manner that reflected pre- cise planning and coordination of physical space, equipment, materials, activities, staff roles, and timing (Rogers and Lewis, 1988). Consistent across programs is the existence of predictable daily rou- tines, which are organized according to written schedules of activities. The center-based programs tend to vary activities from one-to-one to small group to large group, with goals addressed in the most compatible format (e.g., new language can be difficult to teach in a large group situation). For center-based programs, the class size varies from 6 (Denver) to 18 (Walden prekindergarten), although there is variability across childrenâs ages. Class size also varies, depending on the ratio of children with special needs to total number of children (e.g., the Walden preschool program has 18 children, but only 6 have autism). Perhaps more relevant than the number of children with autistic spec- trum disorders is the adult:child ratio, which all of the programs keep high in order to ensure that each childâs individualized needs can be met. Across the ten programs, the adult-child staffing ratios range from 1:1 to 1:8, depending on the program format, class size, and childrenâs develop- mental and chronological age. The Developmental Intervention Model and the Young Autism Project remain nearly exclusively 1:1 throughout the intervention period, including the use of a one-on-one âshadowâ if a child is eventually included in a regular early childhood center. Other programs offer staffing of approximately 1:3, although each of these pro- vides for some 1:1 sessions in the course of each childâs day. A number of programs (e.g., Childrenâs Unit, Douglass, Walden) systematically and intentionally fade the adult:child ratios across time in intervention, in order to prepare children to function independently in future sites. Focus on Communication Goals and Other Developmental Areas All ten programs explicitly address the communication irregularities associated with autistic spectrum disorders, although there is some vari- ability in the specific target objectives and in the strategies for promoting communication. The programs also target other developmental domains, including engagement, social, play, cognitive and academic skills, self- help, behavioral challenges, and motor skills. The distribution of treat- ment time devoted to teaching skills in different developmental areas
160 EDUCATING CHILDREN WITH AUTISM varies across programs, and the programs also vary on the sequencing in which the developmental domains are addressed across treatment years. The developmental areas addressed are discussed below. Communication It is not possible to directly compare verbal abilities of the children across programs because of differing ages and other poten- tial differences in child populations, but reported data suggest that for many of the model programs, the predictions that only 50 percent of children with autism will develop functional speech (Lord and Paul, 1997) are far exceeded. For example, the Denver Model reported that 73 per- cent of their preschool graduates were verbal at exit (Rogers and DiLalla, 1991), and Walden reported that 82 percent of children who began inter- vention as toddlers were functionally verbal by the time they entered preschool (McGee et al., 1999). Most of the programs reported teaching speech as well as alternative means of communication. Children in several programs (i.e., Denver, Douglass, and TEACCH) were taught speech, sign language, and use of the Picture Exchange Communication Systems (PECS; Bondy and Frost, 1994). All the programs that teach alternative forms of communication maintain a vigorous effort (either simultaneous or sequential) at teaching language development as well. Only Walden formally avows a verbal- only approach to language instruction (McGee et al., 1994, 1999). The Developmental Intervention Model stands alone in focusing on nonver- bal communication and interactions rather than teaching verbal language (Greenspan and Wieder, 1997). The Individualized Support Program (and most of the behavioral programs) places a heavy emphasis on develop- ment of communication skills that are functional equivalents of problem behaviors (Dunlap and Fox, 1996). As a rule, the programs that emphasize a naturalistic approach to language intervention focus on conversational language. Thus, both the Douglass (Taylor and Harris, 1995) and the Pivotal Response Model (Koegel et al., 1999a) programs have reported procedures for teaching how to ask questions (e.g., âWhatâs that?â âWhere is it?â âWhose is it?â âWhatâs happening?â). LEAP and Walden also emphasize the impor- tance of directly teaching verbal interactions with typical peers (McGee et al., 1992; Odom and Strain, 1984). Engagement Although the terminology in which it is discussed and achieved varies, from the outset of intervention, all of the ten programs either explicitly or implicitly teach engagement. Engagement is defined as sustained attention to an activity or person. The traditional behavioral programs emphasize compliance with one-step directions (e.g., âSit down,â âStand upâ) as a first step of intervention, with a goal of prepar- ing the child to follow teaching instructions (Lovaas et al., 1981). In the
COMPREHENSIVE PROGRAMS 161 Developmental Intervention Model, a child is encouraged to initiate pur- poseful behavior, and the therapist follows the childâs lead to extend engagement (Greenspan and Wieder, 1997). At Walden, an engineered environment provides high-preference toys, which are dispensed by teachers in a systematic manner to ensure that childrenâs engagement levels are maintained at least 80 percent of the time (McGee and Daly, 1999). The Pivotal Response Training approach uses a variety of proce- dures (e.g., interspersing previously learned tasks among newer and more challenging tasks, reinforcing attempts to respond) to keep childrenâs motivational levels high (Koegel and Koegel, 1986; Koegel et al., 1988). Social Interactions Virtually every program lists social interactions as an intervention priority, although the programs differ as to whether the con- centration is on interactions with adults (i.e., Developmental Intervention Model, Young Autism Project) or on interactions with peers (i.e., LEAP, Walden). Because the Individualized Support Program is a family sup- port model, this approach emphasizes social interactions with parents and siblings (Fox et al., 1997). Because all of the programs have an overriding goal of promoting childrenâs long-term independent functioning in the community, all rec- ognize inclusion of children with autistic spectrum disorders in classes with typical children as a desired long-term outcome. The major differ- ences center around whether the program takes a âreadinessâ position, which assumes that certain prerequisite skills are needed for a children with autistic spectrum disorders to benefit from inclusion (e.g., the Childrenâs Unit and TEACCH [Marcus et al., 2000; Romanczyk et al., 2000]), or a position that early social skills are most feasibly developed when children are included with typical children from the outset in inter- vention (Strain et al., 2001). The Developmental Intervention Model and the Young Autism Project tend to include successful children with autism with typical peers near the end of their early intervention period (Greenspan and Wieder, 1997; Smith et al., 2000a). Douglass now pro- vides a continuum of settings, across which children move from one-to- one, to small-group segregated classes, to an integrated class that includes a majority of typical peers (Harris et al., 2000). One reason the original Denver center-based treatment program closed was in recognition of the importance of including children with autism with their typical peers, which is now done through technical assistance in preschool settings (Rogers, 1998). A number of models maintain an a priori position that favors inclu- sion from the outset, based on various arguments for inclusion (Strain et al., in press). Both LEAP and Walden have developed their curricula with a major emphasis on promoting normal social learning opportunities from the earliest possible moment, when all young children are learning how
162 EDUCATING CHILDREN WITH AUTISM to interact socially. The Pivotal Response Model takes the position that inclusion is most easily accomplished when children are in preschool because this is the period when academic demands are lowest (Koegel et al., 1999a). Although the Individualized Support Program is philosophi- cally committed to inclusive education, the reality of very limited inclu- sion options for children with autistic spectrum disorders leads to a prag- matic approach of providing intervention in the most natural settings available. Play Play skills are closely related to both social and communication domains, and the ten models vary considerably in how play is addressed. Thus, play is a major emphasis of the Denver approach (Rogers and Lewis, 1988). Teaching in the course of play activities is also intrinsic to the models that primarily use incidental teaching or other naturalistic in- structional procedures (i.e., Individualized Support Program, LEAP, and Walden), and inclusive programs are most likely to target creative or interactive play with peers (McGee et al., 1992; Odom and Strain, 1984). In fact, most programs target goals related to recreation (e.g., Pivotal Response Training [Koegel et al., 1999a]) and leisure skills (e.g., Childrenâs Unit [Romanczyk et al., 2000]), which, for young children, involve toy play. A review of published curriculum materials and program descrip- tions suggests that basic functional play skills (such as stacking rings and putting pegs in a pegboard) are routine goals at the Childrenâs Unit, Douglass, TEACCH, and the Young Autism Project. Cognitive and Academic Skills Virtually all of the programs teach cogni- tive skills, although the distribution of treatment time to this area varies considerably. Cognitive growth is a major emphasis of the Denver, Douglass, TEACCH, and Young Autism Project models. Although cogni- tive abilities tend not to be a major curriculum priority in programs that focus on peer interaction skills (i.e., LEAP, Pivotal Response Model, and Walden), skills such as mathematics, reading, and writing are taught be- cause academic preparation may help secure a childâs placement in a regular kindergarten classroom (Koegel et al., 1999a). Self-Help The behavioral programs use an array of procedures of dem- onstrated efficacy in teaching self-help skills. The developmental pro- grams tend to place less emphasis on self-help skills, probably because self-help skills are not viewed as core autism deficits. Although there are relatively few published studies on self-help skills that are specific to young children with autism, virtually all of the selected model programs were found to track the development of independent daily living skills.
COMPREHENSIVE PROGRAMS 163 Behavioral Challenges To a growing extent, strategies for promoting en- gagement have become nearly synonymous with methods of preventing challenging behaviors (McGee and Daly, 1999), because the behavioral challenges presented by young children with autistic spectrum disorders are usually not of a severity to warrant more intrusive intervention proce- dures (see Chapter 10). However, the Young Autism Project acknowl- edged use of aversive procedures with children participating in a 1987 outcome study. In a recent replication, aversives were discontinued after the first few subjects (Smith, 2000b). In another replication of the Lovaas (1987) outcome study, there was speculation on the possibility that the absence of aversives could have accounted for less positive child out- comes (Anderson et al., 1987). At least five approaches (i.e., Denver, Individualized Support Pro- gram, LEAP, Pivotal Response Model, and Walden) rely exclusively on positive procedures for preventing challenging behaviors or for building incompatible appropriate behaviors. Because the Individualized Support Program model is a more short-term, problem resolution approach (Dunlap and Fox, 1999a), a comprehensive positive behavior support strategy has been developed to accomplish demonstrable improvements in relatively short time-frames (see Chapter 10). Motor Skills The Developmental Intervention Model places a major em- phasis on motor skills, including motor planning and sequencing. Most of the programs teach age-appropriate gross and fine motor skills. The UCLA program encourages gestural and vocal imitation. The Denver Model emphasizes motor imitation and motor planning. Carefully Planned, Research-Based, Teaching Procedures Include Plans for Generalization and Maintenance of Skills The ten representative programs use a range of research-based teach- ing procedures. The behavioral programs use procedures based on prin- ciples of learning, but the format of instruction falls along a continuum of discrete-trial procedures to incidental teaching. At the ends of the con- tinuum, the Young Autism Project has historically used discrete-trial pro- cedures nearly exclusively (Lovaas et al., 1981), while Walden provides all instruction using an incidental teaching approach (McGee et al., 2000). The other five behavioral programs use a mixture of discrete-trial and naturalistic teaching procedures, although the Individualized Support Program (Dunlap and Fox, 1999a), LEAP (Strain and Cordisco, 1994), and the Pivotal Response Model (Koegel et al., 1999a) models use predomi- nately natural context procedures, and the Childrenâs Unit most com- monly uses a highly structured discrete-trial approach (Romanczyk et al.,
164 EDUCATING CHILDREN WITH AUTISM 2000). The Douglass Centerâs treatment continuum moves children from discrete-trial instruction to eventual placement in a classroom that uses mostly natural contexts teaching formats (Harris et al., 2000). The trend toward use of naturalistic teaching procedures began as an attempt to improve generalization of skills to use in everyday life. Proce- dural comparisons of discrete-trial instruction and incidental teaching have indicated that, given comparable reinforcement procedures, acquisi- tion occurs at approximately the same pace for both of the procedures (McGee et al., 1985). However, generalization or transfer of skills from the teaching setting to unprompted use in new settings or with new people is enhanced when skills have been learned through incidental teaching. Incidental teaching is a systematic protocol of instruction derived from principles of behavior analysis, and haphazard or unplanned instruction of any type is unlikely to produce acquisition in children with autism (McGee et al., 1999). A method called structured teaching is used at TEACCH (Marcus et al., 2000). Structured teaching shares features common to discrete-trial instructional procedures but also emphasizes instructional formats de- rived from the developmental literature and psycholinguistics, as well as some incidental teaching (Watson et al., 1989). The focus is on environ- mental structure, visual schedules, routines, organizational strategies (e.g., working from left to right), and visual work systems that help a child achieve independence in various skills. With respect to reinforcement, the TEACCH model works from the idea that task performance and task completion will be motivating for children if they understand a task that is at an appropriate developmental level (e.g., supporting the develop- ment of emerging skills) and that builds on individual interests. The TEACCH structured teaching approach focuses on helping parents and teachers adapt the environment while helping children to develop skills. The two developmental programs use somewhat different ap- proaches, although both are delivered during play interactions between adults and children. The technical foundation for the Denver Model and the Developmental Intervention Model differ significantly from the be- havioral approaches, yet each involves teaching in natural contexts. Meaningful differences, however, tend to center on the role of reinforce- ment in the instructional process. The use of discrimination training tech- niques is most common in both discrete-trial and incidental teaching pro- cedures. Individualized Intervention Plans Are Needed to Adjust for the Wide Range of Childrenâs Strengths and Needs All ten programs give explicit attention to the importance of indi- vidualizing treatment; their methods vary. In general, the procedural
COMPREHENSIVE PROGRAMS 165 approaches tend to be entirely custom-designed for each child, while the âprogramsâ provide for individual adjustments within an overall pack- aged curriculum. Transitions from Preschool to School Are Planned and Supported Most of the selected programs report specific preparation for childrenâs transition from intensive intervention into school programs. For example, Douglass reports a process that occurs across the childâs last 9 months prior to program exit. Transition preparation begins with staff visits to future schools to assess the match of child with placement and to determine specific skills the child will need to function successfully in the next environment (Handleman and Harris, 2000). Receiving teachers are invited to visit Douglass to get an understanding of the childâs interven- tion history, and follow-up consultation is offered to receiving classrooms. In some cases, children make transitions gradually, with either partial- day placements or accompaniment by familiar staff. Nearly all of the programs report placement outcomes, although it is recognized that a childâs progress is not the sole determining factor in placement decisions. The range of children going to typical classrooms following intervention differs widely across the programs, with program evaluation data reporting a range from 15 percent of children treated at the Childrenâs Unit (Romanczyk et al., 2000) to 79 percent of the children from Walden (McGee et al., 2000). It should be noted that children at the Childrenâs Unit were selected on the basis of severity of problem behav- iors. Programs that exclude or do not encourage children with autism and other severe difficulties have tended to have more uniform positive outcomes. The political climate and local policies are also factors that influence placement outcomes. For example, 35 percent of the first 20 children treated in the Denver Model went to nonspecialized schools or daycare centers with normally functioning peers (Rogers et al., 1986); however, today, those numbers would be higher, because Colorado now has a state- wide policy of including the vast majority of children with disabilities in regular settings. The Walden program was able to replicate placement outcomes achieved in Massachusetts, an area in which inclusion was well accepted and promoted, when the program relocated to Georgia, where inclusion of children with autistic spectrum disorders was rare. How- ever, the policies of the program itself also play a role. Walden, for ex- ample, recommends inclusion for nearly all children with autistic spec- trum disorders, irrespective of level of functioning, due to a program policy emphasizing that all children with autistic spectrum disorders have social needs that require exposure to normal social behavior.
166 EDUCATING CHILDREN WITH AUTISM INTERVENTION STUDIES There is a need for well-controlled clinical outcome research on these and other models of service delivery. The available research strongly suggests that a substantial subset of children with autistic spectrum disor- ders are able to make marked progress during the period that they receive intensive early intervention, and nearly all children with autistic spec- trum disorders appear to show some benefit. However, the research to date is not at a level of experimental sophistication that permits unequivo- cal statements on the efficacy of a given approach, nor do the data sup- port claims of ârecoveryâ from autistic spectrum disorders as a function of early intervention. There is no outcome study published in a peer- reviewed journal that supports comparative statements of the superiority of one model or approach over another. Rather, with a few exceptions, much of the current outcome information is in the form of program evalu- ation data or measures of childrenâs progress when comparisons are made before and after intervention without control groups or blinded assess- ments of outcome. Although many children have participated in the ten model programs, outcome data is generally based on small samples, and the small sample size has also prohibited analysis of the role of individual differences within children in the effectiveness of different models. The components of the ten program models discussed above are em- pirically grounded. Researchers working with each of them have pub- lished numerous peer-reviewed findings specific to the procedures devel- oped in their programs, although the level of standards for intervention studies varies considerably across journals. In some cases, the programs originated as applied laboratories in which to develop and test interven- tion procedures, so research about the effects of specific procedures was the natural output. As reviewed in other sections of this report, this cumulative body of procedural research serves as evidence that early educational interventions do enable young children with autistic spec- trum disorders to acquire a variety of skills. However, the quality and quantity of research that evaluates the over- all efficacy of these models has lagged behind the procedural research. The paucity of outcome data may be due to the fact that early education programs for children with autistic spectrum disorders are relatively new. The ethical and logistical complexity of conducting clinical outcome re- search with young children is also a major contributing factor. Examples of the outcome data generated by the ten selected models to date are presented in this section; the models are covered in alphabeti- cal order. The studies discussed were published in peer-reviewed jour- nals; these journals vary widely in the experimental rigor of their review process. In several cases, published data were provided to augment pro- gram description information rather than as results of experimental tests.
COMPREHENSIVE PROGRAMS 167 Childrenâs Unit Although this program regularly collects a comprehensive set of both observational and standardized measures of child progress, outcomes have been reported primarily in non-peer-reviewed book chapters (Romanczyk et al., 1994, 2000). There have also been a number of con- trolled evaluations of the computer data systems, staff training efforts, and clinical procedures (Romanczyk, 1984; Taylor et al., 1994; Taylor and Romanczyk, 1994), but these are beyond the scope of the model outcome data considered here. Denver Model There are at least four peer-reviewed outcome reports on the Denver Model, including the evaluation of a comprehensive training model de- scribed above (Rogers and DiLalla, 1991; Rogers et al., 1986; Rogers and Lewis, 1988; Rogers et al., 1987). An evaluation of the progress of 49 children treated in the Denver Playschool Model reported better than predicted gains in all developmental areas assessed by the Early Interven- tion Developmental Profile and Preschool Profile (Schafer and Moersch, 1981), with the exception of self-help skills. The developmental assess- ment was based on ratings by classroom teachers obtained early and late in treatment (Rogers and DiLalla, 1991). In addition, impressive language gains were demonstrated on standardized language assessments (one of five commonly used instruments) conducted by the childrenâs speech and language pathologists. An earlier assessment of the progress of the first 31 children treated in this model revealed small but statistically significant improvements in symbolic and social and communicative play skills, as rated on an objec- tive observational system by blind observers (Rogers and Lewis, 1988). Moreover, there were indications that the intervention had impacted the severity of autism, as measured in the Childhood Autism Rating Scale (CARS). Douglass Developmental Center There have been four peer-reviewed publications of data on the Douglass Center (Handleman and Harris, 2000; Harris et al., 1990, 1991, 1995). These studies include documentation of progress as measured on the Stanford-Binet (Thorndike et al., 1986), the Preschool Language Scale (Zimmerman et al., 1979), and the Vineland (Sparrow et al., 1984). The most recent report is on 27 children who entered intervention between the ages of 31 and 65 months (Handleman and Harris, 2000). After 4-6 years following termination of intervention, the childrenâs place-
168 EDUCATING CHILDREN WITH AUTISM ments were analyzed in relation to their entry data to determine whether reliable predictors of treatment outcome could be identified. Both IQ scores and age of entry into treatment were found to be predictors of long-term placement. Of 11 children who entered intervention before the age of 48 months, pre- to posttreatment IQ score gains averaged 26 points, and all but one child were in regular placements (seven with support and three without support). For those who entered intervention at or older than 48 months, the average IQ score gain was only 13 points, and only one child was in a regular class placement at follow-up. Developmental Intervention Model Greenspan and Wieder (1997) provide a detailed review of the case records of 200 children who had participated in the Developmental Inter- vention Model for 2 or more years. Based on clinical notes and scores on the Functional Emotional Assessment Scale, 58 percent had âgood to out- standingâ outcomes, 25 percent had âmediumâ outcomes, and 17 percent had âlowâ outcomes. Overall, this pattern of outcomes was better than that of a comparison group of children who entered treatment with the Developmental Intervention Model following treatment with traditional behavioral services. However, there was a major confounding element in use of a comparison group: their parents had been dissatisfied with their previous intervention. Ratings were also not blind to intervention status. A more in-depth examination of 20 of the highest functioning children detailed marked gains on the Vineland (Sparrow et al., 1984) and CARS (Schopler et al., 1988). Somewhat inconsistent with the outcomes reported by others, expressive language scores were reported to be above those for receptive language, and self-care skills were lower than communication and socialization abilities. Individualized Support Program The Individualized Support Program model has reported single-sub- ject data on the first six participating children (Dunlap and Fox, 1999a). Although this report was in a peer-reviewed journal, only one of the childrenâs interventions was evaluated with an experimental design. Posi- tive pre-post changes were reported on the Autism Behavior Checklist (Krug et al., 1980), and proportional change index scores (Wolery, 1983) were computed for pre-post scores on the Battelle Developmental Inven- tory (Newborg et al., 1984). LEAP The LEAP programâs effect on childrenâs cognitive growth (Hoyson
COMPREHENSIVE PROGRAMS 169 et al., 1984) and social interaction development (Strain, 1987) was com- pared with that of a comparison group treated at another autism treat- ment program, with results documenting more positive developmental progress by children in the experimental program. More recently, a sum- mary of case reports of the long-term progress of the first six children in the LEAP program is now available (Strain and Hoyson, 2000), but with- out comparisons or controls. The children ranged in age from 30 to 53 months at the onset of treatment, and they scored in the moderate to severe range of autism on the CARS (Schopler et al., 1988). By the time of program exit, and continuing until the children were 10 years old, the CARS scores for these children fell beneath the cutoff for autism. Large decreases in noncompliance were demonstrated in videotaped samples of parent-child interactions, both at program exit and when the children were 10 years old. There were also clinically significant increases in the amount of time the children spent engaged in positive peer interactions, both at program exit and at age 10. Five of the six children spent their school careers in regular education placements. Pivotal Response Model The Pivotal Response Model has reported long-term follow-up on a total of ten children (Koegel et al., 1999b). The first six children had similar language ages at entry, but they differed in their levels of initiat- ing interactions. At the time of follow-up, it was found that higher initia- tion levels at entry predicted less restrictive school placements, higher adaptive and language test scores, and more appropriate parent-child interactions. The next four children, who displayed low levels of initia- tion at the time of entry into intervention, were provided with specific training on how to independently initiate interactions. At follow-up, three of the four children trained in self-initiations had placements in regular education settings, as well as impressive outcomes on measures of language pragmatics, adaptive behavior measured by the Vineland (Sparrow et al., 1984), and lower levels of autistic behaviors reflected on the CARS (Schopler et al., 1988). As a group, the 10 children treated with Pivotal Response Training had very good outcomes, but the absence of experimental design leaves it unclear whether these improvements can be attributed directly to the programâs intervention; this is a difficulty that holds true for almost all of the data reported for the ten model programs. TEACCH Program evaluation information on the TEACCH model has included consumer satisfaction data from parents, trainees, and replication sites (Mesibov, 1997), as well as objective assessment of parent teaching skills
170 EDUCATING CHILDREN WITH AUTISM (Marcus et al., 1978) and child progress (Schopler et al., 1982). There have been a number of studies describing progress in follow-up samples of young children who received services at TEACCH (Venter et al., 1992), and substantial IQ score gains have been commonly reported for nonver- bal children who were diagnosed at early ages (Lord and Schopler, 1989). However, these studies are not direct evaluations of treatment outcomes. Most recently, a 10-hour home-based TEACCH program training teachers to serve young children with autism was compared with a dis- crete-trial classroom without the home-based program (Ozonoff and Cathcart, 1998). The focus of intervention in both programs was cogni- tive, academic, and prevocational skills. Following 4 months of interven- tion, the group served in the TEACCH home-based program showed more improvement than the comparison group on imitation, on fine and gross motor skills, and on tests of nonverbal conceptual skills. UCLA Young Autism Project Although the UCLA program has generated the most rigorously con- trolled early intervention research published to date, there has been con- siderable controversy due to various methodological and interpretational limitations (Gresham and MacMillan, 1997). In the original report (Lovaas, 1987), 38 children with autism were divided into two treatment groups: half of the children received intervention for at least 40 hours per week for 2 or more years, and the other half received the same interven- tion for less than 10 hours per week. There was a second comparison group who received treatment outside of the UCLA program. Nine of the 19 children who received intensive intervention showed IQ gains of at least 20 points. Gains were far less for children in both of the comparison groups. The Young Autism Project has also reported the longest follow-up tracking of children with autism who have received intensive early inter- vention (McEachin et al., 1993). By age 13, eight of the nine high-outcome children from the Lovaas (1987) study continued to have high IQ scores, and they were functioning unsupported in regular education classrooms. In contrast, only one child who received less intensive intervention had a âbest outcome.â Several peer-reviewed evaluations have been conducted of replica- tions of the Young Autism Project (Anderson et al., 1987; Birnbrauer and Leach, 1993; Sheinkopf and Siegel, 1998; Smith et al., 2000b). The replica- tion results have been generally positive but mixed. With fewer hours of intervention, some of the replication programs were able to achieve simi- larly high IQ sore gains; results were more variable on other measures. For example, the most recent replication (Smith et al., 2000b), which served both children with autism and children with pervasive developmental
COMPREHENSIVE PROGRAMS 171 delayânot otherwise specified, yielded improvements in IQ scores, but, there were no significant changes in the childrenâs diagnoses or their adaptive or problem behaviors (Smith et al., 2000b). Most critiques of the outcome research generated by the Young Au- tism Project do not deny impressive child outcomes (Mesibov, 1993; Mundy, 1993); the debate centers on methodological issues related to subject selection and assessment measures (see Chapter 15). One of the most controversial issues surrounding the program pertains to descrip- tions of the best-outcome children in the 1987 study as ârecoveredâ or ânormal functioning,â especially in light of the paucity of measures of social or communicative functioning (McEachin et al., 1993). Walden Toddler Program Pre-post data on the preschool and overall Walden programs have been reported in non-peer-reviewed book chapters (McGee et al., 1994, 2000), and an evaluation of the family program is described by McGee and colleagues (McGee et al., 1993). Therefore, only the outcome data published on the toddler program is considered here. A total of 28 chil- dren with autism began intervention at an average age of 30 months, and those who participated in the program for at least 6 months were in- cluded (McGee et al., 1999). Pre-post comparisons without other experi- mental controls provide the majority of data. Videotaped observations of each childâs ongoing behavior were obtained daily across the first 10 days and last 10 days of enrollment in the toddler center. Results showed that although only 36 percent of the children were verbal at program entry, 82 percent of the children were verbalizing meaningful words by the time that they exited the toddler program to enter preschool. In addition, by the time of program exit, 71 percent of the children showed increases in the amount of time that they spent in close proximity to other children, with only one child showing levels of peer proximity that were outside the ranges displayed by typical children. Summary of Intervention Studies As a group, these studies show that intensive early intervention for children with autistic spectrum disorders makes a clinically significant difference for many children. The most systematic evaluation data are associated with intensive intervention approaches. However, each of the studies has methodological weaknesses, and most of the reports were descriptive rather than evaluations with controlled experimental research designs. There are virtually no data on the relative merit of one model over another, either overall or as related to individual differences in chil- dren; there is very limited information about interventions for children
172 EDUCATING CHILDREN WITH AUTISM under 30-36 months of age. There is overlap in the levels of intensity with which the models are implemented, and the measures of outcome differ widely across interventions. In addition, as summarized in Figures 1-1,1-2, and 1-3 (in Chapter 1), studies that addressed general aspects of interventions consistently had methodological limitations that were often even more common than in studies about interventions for narrower target areas (see Kasari, 2000). These limitations in part reflect the tremendous scope required in carry- ing out research concerning comprehensive intervention programs. On the whole, issues related to internal and external validity were addressed only minimally in about 80 percent of the published studies, with mea- surement of generalization outside the original setting occurring only minimally in 70 percent of the research reports. Given the difficulty and the cost in time and money of such studies, it seems most useful to con- sider alternative methods to addresses these concerns. The models presented positive and remarkably similar findings, which included better-than-expected gains in IQ scores, language, autistic symptoms, future school placements, and several measures of social be- havior. Although possible changes in diagnosis are implied, these have not been systematically documented or supported with independent ob- servations or reports. Considered as a group, these peer-reviewed out- come studies suggested positive change in the language, social, or cogni- tive outcomes of children with autistic spectrum disorders who received intensive early intervention beginning at young ages. However, only three of the studies (plus one follow-up) had comparison group data, and only one of the studies (Smith et al., 2000b) practiced random assignment of children to conditions, and this procedure was complex. Pre-post as- sessment measures reflected positive outcomes for the majority of chil- dren receiving intervention, and most children showed some progress. However, there was almost no information on the contribution of the other interventions and therapies in which the children participated. In sum, it appears that a majority of children participating in compre- hensive behavioral interventions made significant progress in at least some developmental domains, although methodological limitations pre- clude definitive attributions of that progress to specific intervention pro- cedures.
III Policy, Legal, and Research Context