A perspective on the research literature related to early intensive behavioral intervention (Lovaas) for young children with autism

VICTORIA SHEA
Chapel Hill, North Carolina, USA

autism 2004 SAGE Publications and The National Autistic Society Vol 8(4) 349367; 047223 1362-3613(200412)8:4

A B S T R AC T

Various aspects of the research literature on early intensive behavioral intervention (EIBI) have been poorly understood within the psychological, educational, and advocacy communities. Examination of the studies that are frequently cited by proponents of EIBI suggests that the expectation that 47 percent of youngsters who receive EIBI will reach normal developmental status is questionable.

K E Y WO R D S

ADDRESS

Correspondence should be addressed to: V I C T O R I A S H E A , 1506 E. Franklin Street, Suite 202, Chapel Hill, North Carolina 27514, USA. e-mail: victoria.shea@mindspring.com

autism; early intervention; EIBI; Lovaas; young children

Introduction
Many well-intentioned service providers, parents, and advocates have overlooked, misinterpreted, or inadvertently misstated the results of the early research studies (e.g. Lovaas, 1987; McEachin et al., 1993) that are generally cited as the empirical foundation of early intensive behavioral intervention or EIBI. (The term EIBI will be used for techniques variously referred to in the popular literature as the Lovaas method, discrete trial training,operant learning, and applied behavior analysis, although these terms are not actually interchangeable. For discussions of the relationships among these terms, see Lovaas, 2002, p. 394; McClannahan and Krantz, 2001; Smith, 2001.) This article will consider selected elements of the core research about EIBI and suggest that it is time for certain erroneous impressions to be changed, in particular the perception that 47 percent of youngsters who receive EIBI can be expected to attain normal developmental status.

www.sagepublications.com DOI: 10.1177/1362361304047223

349

AU T I S M

8(4)

History and details of the original research

Chronology The behavioral treatment of young children with autism known as the University of California, Los Angeles (UCLA) Young Autism Project (YAP) and documented in Lovaass (1987) well-known article took place between 1970 and 1984 (Lovaas, 2002, p. 390). Seven of the 19 experimental subjects were referred between 1970 and 1974 (McEachin et al., 1993, p. 362). A follow-up study (McEachin, 1987) of the experimental subjects was the focus of McEachins doctoral dissertation at UCLA and took place between 1984 and 1985. However, the description of the original study was not published until 1987, with the follow-up study published in 1993 (McEachin et al., 1993), which has led some providers to think that the research is quite recent. Initial results and follow-up results for the experimental group At the time of referral, the children were younger than 46 months. Treatment was received for at least 40 hours per week and lasted at least 2 years (mean = 2.5 years: McEachin et al., 1993, p. 52). Standardized testing at age 67 years consisted of an IQ test or a related test of development or receptive vocabulary. The mean IQ score at follow-up was 83.3 (range 30120: Lovaas, 1987). These youngsters were also assessed in 19845 (mean age 12 years 9 months, range 8 years 10 months to 19 years 2 months: McEachin, 1987). At that time, the groups mean IQ was 84.5 (range not given) and their mean Vineland Adaptive Behavior Composite (ABC) was 71.6 (McEachin et al., 1993). Best-outcome subjects In both the 1987 Lovaas report and the 1993 McEachin et al. follow-up, results were reported not only for the whole experimental group (N = 19) but also for the subgroup of nine youngsters (that is, 47 percent of the 19) who had the best outcomes (McEachin et al., 1993, p. 359) from the treatment. These nine youngsters were reported to have achieved normal intellectual and educational functioning, with normal-range IQ scores and successful rst grade performance in public schools (Lovaas, 1987, p. 3). (In the later article it was claried that the educational criterion meant that no special assistance was provided in rst grade: McEachin et al., 1993, p. 362.) The term recovered was used in the text and in a table presenting the results of this group (Lovaas, 1987), and Lovaas recently has written that it is not incorrect to use the term recovered in describing the bestoutcome subjects (2002, p. 405). However, Smith and Lovaas have also
350

S H E A : A P E R S P E C T I V E O N L O VA A S

written that we did not claim recovery (1997, p. 204) and indicated that they regretted unnecessary confusion (1997, p. 204) that may have been caused by using the term recovered in the table of the 1987 article. The documentation of the development of these nine individuals has been the foundation of the recent enormous optimism about the effectiveness of EIBI and the resulting widespread demand on the part of parents for this treatment. Some advocates may have overlooked the fact that McEachin et al. reported that one of these nine best-outcome students eventually went into special education and was no longer considered to be normal-functioning (1993, p. 368). In other words, the often-cited gure of 47 percent normal functioning is not completely accurate (8/19 = 42 percent). Another student in the experimental group eventually moved into regular education, so it could be argued that the gure of 47 percent of subjects in the experimental group in regular classes remained intact (1993, p. 364). However, this student was placed in regular education after sixth grade (Lovaas, 1987, p. 7), so attributing his educational status at that time to his preschool intervention may be overreaching.

Individual test results: intelligence, adaptive behavior, and personality On the intelligence, adaptive behavior, and personality tests, McEachin et al. (1993) presented, among other ndings, the mean scores of the group of nine best-outcome youngsters. That is, the mean group IQ, the mean group Vineland scores etc. were presented. These are not unusual statistics to report, but they can be misinterpreted. For example, if in a different circumstance a group consisted of two scientists both with IQs of 150 and two people with Downs syndrome both with IQs of 50, it could be said that the mean IQ of the group was 100. However, that statistic would not accurately convey the extreme individual differences in the group. Similarly, to understand the developmental status of the nine best-outcome youngsters in the McEachin (1987) dissertation and the McEachin et al. (1993) journal article, it is important to look at the individual proles provided by the authors. On the intelligence tests (McEachin, 1987; McEachin et al., 1993) most of the nine best-outcome youngsters scored in the average range on most subtests, resulting in IQs in the low average to very superior range (mean verbal IQ 108, range 80125; mean performance IQ 111, range 88138; mean full-scale IQ 111, range 99136). This nding of low average to very superior intelligence in a subset of individuals with autism is not surprising, since this is a well-documented occurrence (Asperger, 1991; BallabanGil et al., 1996; Chakrabarti and Fombonne, 2001; DeMyer, 1979; Gabriels et al., 2001; Grandin, 1995; Howlin and Goode, 1998; Jolliffe et al., 1992;
351

8(4) Kanner, 1973; Kobayashi et al., 1992; Rumsey et al., 1985; Rutter, 1970; Sigman and Ruskin, 1999; Szatmari et al., 1989; Venter et al., 1992; Wing, 1981). The individual data for the nine best-outcome youngsters on the other standardized assessment instruments (Vineland Adaptive Behavior Scales; Personality Inventory for Children, PIC) in the McEachin et al. (1993) report seem often to be overlooked. Scores on both of these measures were missing for one youngster and PIC scores were missing for an additional youngster. Further, of the nine best-outcome youngsters, eight had at least one score in the clinically signicant range, and most had more than one such elevated score. Put another way, there was only one best-outcome youngster who had all test scores in the normal range (the youngster identied as L.B.). Thus, the popular notion that all subjects had completely normal test results is not what McEachin et al. (1993) reported (Mundy, 1993). Specically, on the Vineland Adaptive Behavior Scale McEachin et al. indicated that some of the best-outcome subjects had marginal scores (1993, p. 366), including a score of 79 on daily living skills (subject identied as M.M.), a score of 77 on communication (subject J.L.), a score of 74 on daily living skills (subject B.W.), and a composite score of 80 (subject D.E.). On the Vineland Maladaptive Behavior Scale, in the group of nine best-outcome subjects there were three youngsters (M.C., J.L., and D.E.) with signicant elevations; at least one youngster (B.W.) in the intermediate range between signicant and non-signicant scores (it is not possible to determine the exact number of individuals in this range without knowing each subjects age), and one youngster with missing data. On the Personality Inventory for Children, in the group of nine best-outcome subjects three youngsters (M.C., J.L., and B.W.) had at least one clinically elevated scale score; there were two individuals (R.S. and A.G.) with one or more marginal scores only one point below the clinical cutoff, one subject with only one elevated score that might be a reection of early history rather than current functioning (L.B., the only youngster who otherwise had scores in the normal range), and two subjects with missing data.
AU T I S M

Individual test results: Clinical Rating Scale Another misunderstood aspect of the McEachin et al. (1993) report involves the Clinical Rating Scale, which consisted of ratings of atypical behavior in four areas: social, motor, emotional, and language and cognition. Although some community professionals and parents may have assumed that, as part of the research, several independent professionals used this instrument with each subject as a sensitive measure of the subtle decits of autism, nding the best-outcome youngsters indistinguishable from normal controls, the facts are somewhat different. McEachins (1987)
352

S H E A : A P E R S P E C T I V E O N L O VA A S

dissertation indicates that a single Clinical Rating Scale was completed for each subject by a psychology graduate student after a 20 minute interview with the subject. (The student also tested the subject and apparently interviewed the parents, although the order in which these measures were completed is not clear, and the scale was apparently intended to be based solely on the 20 minute interview.) The scale consisted of 22 items, each rated from 0 (normal) to 3 (denitely/markedly deviant) for a possible total score of 66. In describing the Clinical Rating Scale, McEachin et al. wrote that we do not regard the interview as an instrument that by itself yields conclusive results (1993, p. 370) and reported that there are no norms or data on reliability and validity (1993, p. 370) of the scale, which was developed specically for that research project and does not appear to have been used since. It was reported that none of the normal controls scored above 5, while the mean score for the best-outcome group was 8.8. (The scores were presented in McEachins 1987 dissertation, p. 88.) Scores ranged from 0 to 42 in the best-outcome group. In the dissertation the term used was good outcome (1987, p. 48) and this group included both the student who later went into special education (who had the score of 42 on the scale) and the student who eventually moved out of special education (who had the next highest score, 17). Even if the scores of these outliers were disregarded, the mean of the remaining members of the best-outcome group would be 4.63 (range 010), which still seems markedly higher than the mean of 1.8 (range 05) in the normal controls (Mundy, 1993).

Individual test results: teachers and peers observations No measures of the opinions of teachers or peers were reported in either the original research (Lovaas, 1987) or the follow-up study (McEachin, 1987). Thus, there is no apparent foundation for indicating that teachers found the children in the experimental group indistinguishable from typically developing children (Lovaas, 1987, p. 8). Similarly, the suggestion that teachers and peers did not see the best-outcome children as having unusual problems or being different (McEachin, 1987, p. 72) is not supported by test data (Gresham and MacMillan, 1998). Current functioning The nine youngsters described as best-outcome were born between approximately 1965 and 1975 (based on the report in McEachin, 1987, that in 19845 they ranged in age from approximately 9 to 19). That would make them between 28 and 38 years old in 2003 when this article was prepared. Lovaas (2000b) indicated in a book chapter that a second followup study of these nine individuals had begun in 1998. He reported that
353

8(4) preliminary outcome data for six of the nine showed normal intelligence, normal proles on measures of personality (Rorschach and Minnesota Multiphasic Personality Inventory, MMPI) and executive functioning (Wisconsin Card Sort Test), evidence of recognition of mental states of others (2000b, p. 144) and independent functioning and relationships (2000b, p. 144) within the normal range. No other details were given and no further reports of adult outcome of these individuals are available in peer-reviewed journals as of the date this article was prepared.
AU T I S M

Perspectives on replication of initial ndings

As popular interest in EIBI has grown, many training and service delivery models have appeared, including using staff trained in workshops followed by periodic supervision, as contrasted with the UCLA clinic-based training and supervision described in the original (1987) study. The relationship between staff training and the replicability of treatment results was discussed in a chapter by Lovaas titled Clarifying Comments on the UCLA Young Autism Project, published in 2002. (A slightly different version of portions of this material (Lovaas, 2000a) can still be found on various websites including www.ctfeat.org/LovaasRebut.htm.) Lovaas was concerned about EIBI services from providers who used the workshop-based training model while citing his treatment results of 47 percent normal functioning. He wrote that We estimate that valid outcome data documenting normal functioning from such services, if completed and made available, would be less than 10 percent (Lovaas, 2002, p. 397). In the website version of this commentary, concerning workshop-based EIBI services through the UCLA Young Autism Project itself, it is stated that it appears that the rate at which children achieve normal functioning (average levels of intelligence and satisfactory, unassisted performance in a class for typically developing children) is probably closer to 20 percent than that reported for clinic-based treatment at UCLA (Lovaas, 2000a, Appendix, section Two Kinds of Treatment Services; similar comments are found in Mudford et al., 2001 and Smith and Wynn, 2002). These comments again raise questions about the likely replicability of the 47 percent gure particularly when EIBI is modied in this way. In terms of training requirements for staff, Lovaas indicated the following:
Reading the teaching manual (Lovaas et al., 1981), attending a workshop led by UCLA-certied consultants, practicing behavioral treatment on several families or spending a short time at a UCLA-afliated site, does not make a person qualied to provide UCLA-based treatment. Even professional behavioral analysts who have failed to complete an internship at a UCLA-afliated

354

S H E A : A P E R S P E C T I V E O N L O VA A S

replication site would not, in our judgment, have appropriate experience to replicate the UCLA program. (2002, p. 399)

Related research Within the past few years, several studies have been published (some based on data from a number of years ago) on the outcome of EIBI, either with different clinical populations or with treatment variations. Summarized below are the results of standardized tests of students who received EIBI. As a whole, they suggest that most children who received EIBI remained signicantly impaired after treatment.
Children with severeprofound mental retardation and pervasive developmental disorder (PDD) Smith et al. (1997) described the outcome of 11 youngsters under the age of 46 months at referral with both PDD and severeprofound mental retardation. (This was an archival study of children who had been referred to the UCLA YAP or afliated replication sites but excluded from the original 1987 study because their intake IQs were below 35.) The children received 30 hours per week (intensive treatment: 1997, p. 240) of EIBI for at least 24 months. At follow-up (which varied from age 5 to approximately 10 years), the groups mean IQ was 36 (range 1159). Scores on the Vineland were available at follow-up for eight of the youngsters. The mean Vineland ABC for this group, computed for the current article based on data provided in the Smith et al. (1997) article, was 50 (range 3080). The authors reported that of the six youngsters for whom Vineland ABC scores before and after treatment were available, three childrens scores had essentially stayed the same, two childrens scores had declined, and the scores of one child had increased (this was the child who obtained the composite score of 80). Children with Retts disorder Smith et al. (1995) reported on the results of EIBI with three girls (3137 months of age at intake) referred with diagnoses of autism who were later found to have Retts disorder, a degenerative condition with some behavioral characteristics similar to autism. The girls were described as performing in the moderate to severe range of mental retardation at intake, and their functioning declined to the point that they were no longer testable at posttreatment (1995, p. 319). The authors concluded that behavioral treatment does not seem promising as an intervention for Retts disorder (1995, p. 321). Children with pervasive developmental disorder not otherwise specied (PDD-NOS) Smith et al. (2000b; 2001) reported archival outcome data for 15 children with either autism (N = 7) or PDD-NOS 355

8(4) (N = 8) under the age of 42 months when they were referred to the UCLA YAP (between 1989 and 1992). Fourteen of the 15 children received an average of 24.52 hours per week of EIBI for a year,with gradual reductions thereafter (2000b, p. 274). The other child was kept at low levels of service (1015 hours per week) because therapy procedures appeared ineffective for him (2000b, p. 274). Length of treatment averaged 33.44 months. Follow-up assessment took place when the children were 78 years old. The mean IQ at follow-up for the seven children with autism was 55.29 (range not provided). Their mean Vineland ABC declined between intake and follow-up from 68.86 to 62.14 (range not provided). Two of the seven children with autism were in regular classes without special services; only one of the seven (14 percent) students with autism met the best-outcome criteria of McEachin et al. (1993). Follow-up results for the children with PDD-NOS were as follows: average IQ 76.25 (range not provided), average Vineland ABC 60.44 (range not provided); two of the eight children with PDD-NOS were in regular classes without special services; only one of the eight (12 percent) students with PDD-NOS met the best-outcome criteria.
AU T I S M

Older children/school-based EIBI Eikeseth et al. reported on the outcome of 13 Norwegian children with autism who began intensive behavioral treatment based on the University of California, Los Angeles (UCLA) treatment model (2002, p. 51) between the ages of 4 and 7 years at intake (19958). The mean amount of treatment was 28.52 hours per week; this intervention was provided in separate rooms in the childrens kindergartens or elementary schools rather than at home. Outcome assessment after 12 months of treatment yielded the following results: mean IQ 79.08 (range not given); mean Vineland ABC 67 (range not given). Parent-directed/parent-managed EIBI Smith et al. reported results of parent-directed (2000a, p. 297) EIBI, meaning that parents recruited and hired therapists, attended workshops with them to learn EIBI principles and techniques (from the second author, Buch, a UCLA doctoral candidate in psychology), and also arranged for ongoing consultation (the source of consultation was not specied for all children). This study involved six children with a diagnosis of either autism (four children) or PDD-NOS (two children); all children were under 45 months of age at intake. The children averaged 26.2 hours of EIBI for the rst 5 months of treatment, after which one student dropped out of treatment, while the remaining 5 continued to receive approximately the same amount of treatment for the next 23 years (2000a, p. 306). At the post-treatment follow-up assessment, the mean IQ (computed for this article) of the four children with autism was 62 (range 3079; note that it is not clear when the follow-up 356

S H E A : A P E R S P E C T I V E O N L O VA A S

testing for the student who dropped out of treatment was obtained). Their mean Vineland ABC was not reported and could not be computed for the current article, since it ranged from below 30 to 80; the median ABC score was 62. Three of the four children with autism were in regular classes but apparently had individual aides. The fourth student with autism had dropped out of treatment after 1 year, had declined in all areas assessed, and was placed in a residential home (2000a, p. 305). The two students with PDD-NOS had a mean IQ of 51 (range 4458) and a mean Vineland ABC of 62 (range 6064) and were each in regular classes with an aide. In summary, none of the students met the normal functioning criteria. Bibby et al. reported on the outcome of a group of 66 children in the United Kingdom who had received parent-managed (2002, p. 81) EIBI (dened in the same way as parent-directed in Smith et al., 2000a). The impetus for this study, according to the authors, was that:
It was not unusual for research ndings from clinic-based services to be emphasized by advocates for the treatment with the implicit (or explicit) assertion that 47 percent best outcome (normal intellectual and educational functioning) could be expected from workshop model professional services. This study was designed to assess the validity of this extrapolation. (2002, p. 84)

In the Bibby et al. study, record review indicated that 64 of the 66 children had diagnoses of autism or autistic spectrum disorder (2002, p. 85); the other two children had diagnoses of PDD. Treatment began at a mean age of 45 months (standard deviation 11.2 months). Length of treatment varied at the time of the rst assessment (time 1), with 18 percent of children in the rst year of treatment, 52 percent in their second year, and 30 percent beyond their second year. Treatment then continued (for 60 of the 66 children) and a second assessment took place 12 months later (time 2). At the time 2 assessment, the mean IQ for the 60 children was 58.5 (range not given, although 10 children had IQs > 85); mean Vineland ABC was 61 (range not given). At time 2, there were 42 children who met the age and time in treatment criteria (that is, 72 months and 24 months, respectively) used by Lovaas (1987) in looking at outcome. The mean IQ of these 42 children was 53.1; their mean Vineland ABC was 55.2. The authors reported that None of the 42 children aged > 72 months were in mainstream school without individual adult support. That is, there were no educationally normal functioning children in the outcome subgroup, therefore no best outcomes (2002, p. 94). They also reported that according to follow-up information 69 months later, after the study was terminated, three other children aged > 72 months with time 2 IQs > 85 were succeeding in regular schools without any extra adult support 357

8(4) (2002, p. 95). Although the authors did not report the rate of best outcome represented by these three subjects, for the purposes of this article it was estimated at not more than 6 percent (3/48, if 48 was the total number of students aged > 72 months at that time).
AU T I S M

Young Autism Project replication sites Smith et al. wrote that we began collaborating on replication efforts in 1988 (1993, p. 385). In the mid 1990s (Lovaas, 1994; 1995) the Multisite Young Autism Project received federal funding to set up replication sites in various cities in the United States, Europe, and Scandinavia to study whether the results of the original (Lovaas, 1987) study could be replicated (without the use of aversive consequences for unacceptable behavior; see section Treatment procedures in this article). Lovaas indicated that some 13 replication sites across the United States and Europe have employed staff who have passed the academic background [sic] and required a [sic] nine-month full-time internship using the UCLA treatment model (2000b, p. 148). Bibby et al. indicated that more than 10 replications of the UCLA model of treatment are underway (2002, p. 82). According to Lovaas, the rst reports on replicability were submitted to peer-review journals in 1999 (2000b, p. 148). As of the time that this article was prepared, written descriptions of results of replication studies could be located only on the Internet. Preliminary results dated spring 1999 from the rst year of a three-year study by Sallows and Graupner (n.d.) at the replication site in Wisconsin were presented at a professional meeting and are available on several websites (including http://www.londonearlyautism.com/). Results were reported for 11 children with autism (mean age at the beginning of treatment 33 months). After 1 year of treatment the replication groups mean IQ was 69 (4 points lower than that of the control group who received parent-directed behavioral treatment). The mean Vineland ABC of the replication group was 70. On the Reynell Developmental Language Scales, the mean comprehension and expressive language levels of the replication group were 25 months and 24 months, respectively (compared with their mean chronological age of 48 months). A news release (available at http://researchnews.osu.edu/archive/ autism3.htm) described a professional presentation in August 2003 of a study by Mulick, Metz and Butter of eight children who had received EIBI (it was not clear whether this treatment was received at a YAP replication site). According to the news release, after a year of EIBI all students had IQs above 70 (although it was also reported that two students were mentally retarded), seven students made gains in adaptive behavior (mean and range not reported), but at least four of the students apparently continued to have 358

S H E A : A P E R S P E C T I V E O N L O VA A S

signicant language impairments (no details reported). It was not stated whether any of the students met the best-outcome criteria, although the rst author was quoted as being skeptical about the best-outcome rate of approximately 50 percent. As this article went to press, it was learned that results from the Central Valley Autism Project (Modesto, California) were presented at a professional conference in February 2004 (authors Amerine-Dickens and Cohen). The slide presentation, provided on request by the co-author, suggested that of the 21 children who had received EIBI for 3 years, the mean group IQ was in the 80s (range not given) and the mean Vineland ABC was approximately 80. Group mean expressive language standard scores were also below average. Attainment of best-outcome status could not be determined from the slide presentation. Recent independent studies Boyd and Corley (2001), who apparently are not associated with the UCLA group, reported results of follow-up case reviews for 22 children (19 with autism, three with PDD-NOS) whose behavioral treatment was funded by a non-prot agency in California (Golden Gate Regional Center). The authors reported that direct observation of sessions, viewing of videotapes, discussion with parents and inspection of written documents indicated that the treatments provided were clearly representative of typical community-based EIBI programs patterned after the model described and materials disseminated by the UCLA Young Autism Project (2001, p. 434). Nineteen of these students began EIBI before the age of 4 years (mean age 41 months; range 2948 months); the others began between 51 months and 72 months (Boyd, personal communication, November 2002). The average length of treatment was 23 months (range 9 to over 36 months: Boyd and Corley, 2001, p. 434). Services were funded beginning in 1995, and the outcome survey took place at the end of 1998. The outcome report was based only on school records, parent information, and other agency records, not on evaluation by the studys authors. Further, according to the authors, local public schools are prohibited from administering individual IQ tests under the Larry P. v. Wilson Riles decision, and hence IQ scores per se are not a feature of school psychological reports although other appropriate indices of cognitive abilities are reported (2001, p. 436). At the time of the outcome survey, the children ranged in age from 40 to 85 months; thus, some children had not yet completed rst grade (as had the children in Lovaas, 1987). However, at the time of the Boyd and Corley (2001) outcome survey, none of the children met the criteria of normal intelligence and placement in regular education without special assistance. Seven of the 22 students did not appear to be mentally retarded; of these seven children, 359

360
Smith et al., 1997 Yes 11 < 46 m Range 3137 m 2 yd 8 me Mean 33.44 m 12 m 51f 50.13g 61.92 79.08 67 55.29f 76.25g 52f,h 43g,i 62f 51g 63.5f,j 62g Cannot be determined 0 Range 23 y 37 No longer testable Not reported 62.14f 60.44g 0 14f 12g < 42 m Range 47 y < 45 m 3 15 13 6 60 Mean 45 m Yes Yes Yes Yes No? No 22 < 4 yk 46 yl 1+ y to 3+ y Mean 23 m 57.6 58.5 61 Smith et al., 1995 Smith et al., 2000b; 2001 Eikeseth et al., 2002 Smith et al., 2000a Bibby et al., 2002 Boyd and Corley, 2001 No 11 24 m < 28 36 50c Not reported 62.8m Not reported 89.7m 0 0? 6? 0 27?

AU T I S M

Table 1

Characteristics and outcomes of young children receiving variations of EIBI

Scheinkopf and Siegel, 1998

Lovaas, 1987

McEachin et al., 1993

8(4)

Afliated with Yes UCLA YAP?

Yes

No. of children

19

Same as Lovaas, 1987

Age at beginning treatment

< 46 m

Same as Lovaas, 1987

Mean 33.8 m

Length of treatment

Mean 2.5 y

Same as Lovaas, 1987

Mean 15.73 m

Mean initial IQ

53

Same as Lovaas, 1987

Mean 83.3 outcome IQ 107b

84.5a 111b

Mean outcome Not reported 71.6a Vineland 94b ABC

Not reported Not reported

% achieving Lovaas normal functioning

47

47? 42?

y = years, m = months. a Total experimental group (N = 19 children). b Best-outcome group (N = 9 children). c N = 8 children. d N = 2 children. e N = 1 child. f Children diagnosed with autism. g Children diagnosed with PDD-NOS. h N = 3 children; 1 child was untestable. i N = 1 child; 1 child was untestable. j Median Vineland ABC (see text). k N = 19 children. l N = 3 children. m N = 10 children.

S H E A : A P E R S P E C T I V E O N L O VA A S

three were in special classes and four were in regular classes with oneto-one aides. Scores on a computer-generated measure of behavioral difculty used in California (Frankland factor, FF) were reported as follows: The mean post Frankland factor for the participants was 22.45 with a range of 0 to 56. Within agency guidelines an FF of 28 is considered a threshold for designation of very serious behavior problems; eight children met this threshold while two others were quite close (2001, p. 436). Scheinkopf and Siegel reported the results of a study of 11 children (10 with autism, one with PDD) who at mean age 33.8 months (range 2347 months) began receiving home-based behavioral treatment described by their parents as based on methods developed by Lovaas et al. (1981) (1998, p. 18). Reportedly the behavior therapists overseeing the treatment had prior training specic to Lovaass treatment protocol, although no direct support from Lovaass group was reported during the course of treatment (1998, p. 18). Behavioral treatment averaged 19.45 hours per week for an average of 15.73 months; the children also attended school and received other services during this time. At follow-up assessment at mean age 54.4 months, this groups mean IQ was 89.7 (range 65117 with one childs data missing). Five of the 11 children were later reported to have been placed in regular education classes; two of these ve had full-time assistants. It thus appears that three of 11 children (27 percent) may have met Lovaass (1987) best-outcome criteria. The group of children in this study actually had a higher mean IQ at follow-up than that of any of the other experimental groups reviewed in this article (except for the subgroup of best-outcome individuals in the McEachin et al., 1993 study). However, Lovaas has been critical of the behavioral treatment the children received, describing it as unknown and nonoptimal (2000a, section O(3)) and writing of the behavior therapists who guided the treatment that we think it is quite unlikely that they would be able to serve competently as a consultant to families in home-based programs (2000a, section O(2)). Elements of the characteristics and results of the published studies reviewed above are presented in Table 1.

Critiques and controversies

Experimental methodology The Lovaas (1987) article and the McEachin et al. (1993) follow-up have been the subject of extensive analysis and some criticism (e.g. Gresham and MacMillan, 1997a; 1997b; 1998; Howlin, 1997; Jordan et al., 1998; Rogers, 1998; Schopler et al., 1989). Issues raised include:
361

AU T I S M

8(4) lack of random assignment of subjects to experimental and control conditions experimental and control groups that were not equivalent experimental group that was not representative of the population of children with autism lack of data about the degree to which the treatment as delivered actually matched the treatment intended (i.e. the delity of actual treatment to the planned treatment procedures) lack of information about additional interventions received concurrently with behavioral treatment lack of clarity about the amount of treatment received by control groups outcome assessments that took place long after treatment ended assessments that were not independent of the research staff an outcome measure (class placement) that may have reected philosophical or political factors rather than childrens skill levels measurement error due to the use of different IQ or development measures for different children use of an unusual statistic (prorated mental age: Lovaas, 1987, p. 4) and questionable calculation of deviation IQs unspecied human factors (involving parents and/or therapists) that may have signicantly affected results reliance on parental report measures (Vineland, PIC) that may have been inuenced by parental investment in the treatment.

Lovaas and colleagues have published responses to these critiques (e.g. Eikeseth, 2001; Lovaas, 2000b; 2002; Lovaas et al., 1989; Smith and Lovaas, 1997; Smith et al., 1993).

Treatment procedures Some proponents of EIBI insist, understandably, that in order for the normal functioning rate of 47 percent to be achieved, the original treatment methods must be followed precisely. Specically, parents and other advocates for children have often sought the treatment intensity of 40 hours per week outlined in the Lovaas (1987) article, even though Smith and Lovaas indicated that we have never recommended a 40-hour week for everyone (1997, p. 212). However, separate from the issue of intensity, another critical treatment method described in the 1987 article and the treatment manual referenced therein (Lovaas et al., 1981) was the use of aversive consequences (shouting no and/or slapping the youngster on the thigh in response to self-stimulatory, aggressive, and noncompliant behavior: 1987, p. 7). In the 1987 article Lovaas wrote that contingent aversives were isolated as one
362

S H E A : A P E R S P E C T I V E O N L O VA A S

signicant variable. It is therefore unlikely that treatment effects could be replicated without this component (p. 8). Lovaas and colleagues have subsequently made it clear that aversive consequences are no longer used in the YAP (e.g. Smith and Lovaas, 1997; Lovaas, 2002, p. 390). However, while the element of aversive consequences in the original intervention methodology has been signicantly modied by Lovaas, some professionals and families remain adamant about not modifying the element of 40 hours per week of EIBI (Gresham and MacMillan, 1997b). This logic is not compelling. Further, the 1981 Lovaas et al. treatment manual has recently been revised (Lovaas, 2002) which would presumably put additional distance between previously published results such as 47 percent normal functioning (Lovaas, 1987; McEachin et al., 1993) and future results based on newer methodologies (Kasari, 2002).

Conclusion
It is time for advocates and professionals to stop citing the gure of 47 percent and the concepts of normal functioning, being indistinguishable from average children, and having recovered from autism. The reports of the initial research are not consistent with these interpretations; further, over three decades since the research began, other studies have consistently fallen short of the 47 percent gure. Although there is little professional disagreement that early intervention is important and benecial for youngsters with autism (National Research Council, 2001; Rogers, 2001), it remains the case that most children diagnosed with autism spectrum disorders continue to have signicant functional difculties throughout the lifespan (Howlin and Goode, 1998; Nordin and Gillberg, 1998). It is also the consensus of the professional literature that a variety of educational and therapeutic techniques help children with autism at all levels of functioning to develop skills, interests, and relationships (Dawson and Osterling, 1997; Handleman and Harris, 2001; Hurth et al., 1999; National Research Council, 2001; Rogers, 1998; Smith, 2001; Volkmar et al., 1999). It is understandable that parents and professionals have hoped for children to be recovered and normal functioning. It is also understandable that the initial reports (Lovaas, 1987; McEachin et al., 1993) that eight or nine youngsters were doing well were met with guarded optimism (Mesibov, 1993). However, it is time for the professional community to acknowledge to families that although EIBI may be benecial there is no evidence that it results in recovery or normal functioning in 47 percent of its recipients. 363

8(4) References
AU T I S M
ASPERGER, H.

(1991) Autistic Psychopathy in Childhood (1944), in U. F R I T H (ed. and trans.) Autism and Asperger Syndrome. New York: Cambridge University Press. B A L L A B A N - G I L , K . , R A P I N , I . , T U C H M A N , R . & S H I N N A R , S . (1996) Longitudinal Examination of the Behavioral, Language, and Social Changes in a Population of Adolescents and Young Adults with Autistic Disorder, Pediatric Neurology 15: 21723. B I B B Y, P. , E I K E S E T H , S . , M A RT I N , N . T. , M U D F O R D , O . C . & R E E V E S , D . (2002) Progress and Outcomes for Children with Autism Receiving Parent-Managed Intensive Interventions, Research in Developmental Disabilities 23: 81104. B OY D , R . D . & C O R L E Y, M . J . (2001) Outcome Survey of Early Intensive Behavioral Intervention for Young Children with Autism in a Community Setting, Autism 5: 43041. C H A K R A B A RT I , S . & F O M B O N N E , E . (2001) Pervasive Developmental Disorders in Preschool Children, Journal of the American Medical Association 285: 30939. DAW S O N , G . & O S T E R L I N G , J . (1997) Early Intervention in Autism, in M . J . G U R A L N I C K (ed.) The Effectiveness of Early Intervention, pp. 30726. Baltimore, MD: Brookes. D E M Y E R , M . K . (1979) Parents and Children in Autism. Washington, DC: Winston. E I K E S E T H , S . (2001) Recent Critiques of the UCLA Young Autism Project, Behavioral Interventions 16: 24964. E I K E S E T H , S . , S M I T H , T. , J A H R , E . & E L D E V I K , S . (2002) Intensive Behavioral Treatment at School for 4- to 7-Year-Old Children with Autism, Behavior Modication 26: 4968. G A B R I E L S , R . L . , H I L L , D . E . , P I E R C E , R . A . , RO G E R S , S . J . & W E H N E R , B . (2001) Predictors of Treatment Outcome in Young Children with Autism, Autism 5: 40729. G R A N D I N , T. (1995) How People with Autism Think, in E . S C H O P L E R & G . B . M E S I B OV (eds) Learning and Cognition in Autism, pp. 13756. New York: Plenum. G R E S H A M , F . M . & M AC M I L L A N , D . L . (1997a) Autistic Recovery? An Analysis and Critique of the Empirical Evidence on the Early Intervention Project, Behavioral Disorders 22: 185201. G R E S H A M , F . M . & M AC M I L L A N , D . L . (1997b) Denial and Defensiveness in the Place of Fact and Reason: Rejoinder to Smith and Lovaas, Behavioral Disorders 22: 21930. G R E S H A M , F . M . & M AC M I L L A N , D . L . (1998) Early Intervention Project: Can Its Claims Be Substantiated and Its Effects Replicated?, Journal of Autism and Developmental Disorders 28: 513. H A N D L E M A N , J . S . & H A R R I S , S . L . (eds) (2001) Preschool Education Programs for Children with Autism. Austin, TX: Pro-Ed. H OW L I N , P. (1997) Prognosis in Autism: Do Specialist Treatments Affect Long-Term Outcome?, European Child and Adolescent Psychiatry 6: 5572. H OW L I N , P. & G O O D E , S . (1998) Outcome in Adult Life for People with Autism and Aspergers Syndrome, in F. R . VO L K M A R (ed.) Autism and Pervasive Developmental Disorders, pp. 20941. New York: Cambridge University Press. H U RT H , J . , S H AW, E . , I Z E M A N , S . G . , W H A L E Y, K . & RO G E R S , S . J . (1999) Areas of Agreement about Effective Practices among Programs Serving Young Children with Autism Spectrum Disorders, Infants and Young Children 12: 1726 (available from NECTAS at http://www.nectac.org). J O L L I F F E , T. , L A N S D OW N , R . & RO B I N S O N , T. (1992) Autism: A Personal

364

S H E A : A P E R S P E C T I V E O N L O VA A S

Account. Available from The National Autistic Society, 393 City Road, London EC1V 1NG. J O R DA N , R . , J O N E S , G . & M U R R AY, D . (1998) Educational Interventions for Children with Autism: A Literature Review of Recent and Current Research. Report no. 77. London: Department for Education and Employment (available from DfEE at http://www.dfes.gov.uk). K A N N E R , L . (1973) Childhood Psychosis: Initial Studies and New Insights. Washington, DC: Winston. K A S A R I , C . (2002) Assessing Change in Early Intervention Programs for Children with Autism, Journal of Autism and Developmental Disorders 32: 44761. KO B AYA S H I , R . , M U R ATA , T. & YO S H I N AG A , K . (1992) A Follow-Up Study of 201 Children with Autism in Kyushu and Yamaguchi Areas, Japan, Journal of Autism & Developmental Disorders 22: 395411. L OVA A S , O . I . (1987) Behavioral Treatment and Normal Educational and Intellectual Functioning in Young Autistic Children, Journal of Consulting and Clinical Psychology 55: 39. L OVA A S , O . I . (1994) Long Term Outcome of Early Intervention for Autism (Abstract). Grant 1R01MH051156-01A1. Available at http://crisp.cit/nih.gov/, accessed 22 November 2002. L OVA A S , O . I . (1995) Multisite Young Autism Project (Abstract). Grant 1R01MH048863-01A4. Available at http://crisp.cit.nih.gov/, accessed 22 November 2002. L OVA A S , O . I . (2000a) Clarifying Comments on the UCLA Young Autism Project. Available at www.ctfeat.org/LovaasRebut.htm, accessed 1 August 2003. L OVA A S , O . I . (2000b) Experimental Design and Cumulative Research in Early Behavioral Intervention, in P. J . AC C A R D O , C . M AG N U S E N & A . J . C A P U T E (eds) Autism: Clinical and Research Issues, pp. 13361. Baltimore, MD: York. L OVA A S , O . I . (2002) Teaching Individuals with Developmental Delays: Basic Intervention Techniques. Austin, TX: Pro-Ed. L OVA A S , O . I . , AC K E R M A N , A . , A L E X A N D E R , D . , F I R E S T O N E , P. , P E R K I N S , M . & YO U N G , D . B . (1981) Teaching Developmentally Disabled Children: The ME Book. Austin, TX: Pro-Ed. L OVA A S , O . I . , S M I T H , T. & M C E AC H I N , J . J . (1989) Clarifying Comments on the Young Autism Study: Reply to Schopler, Short, and Mesibov, Journal of Consulting and Clinical Psychology 57: 1657. M C C L A N N A H A N , L . E . & K R A N T Z , P. J . (2001) Behavior Analysis and Intervention for Preschoolers at the Princeton Child Development Institute, in J . S . H A N D L E M A N & S . L . H A R R I S (eds) Preschool Education Programs for Children with Autism, pp. 191213. Austin, TX: Pro-Ed. M C E AC H I N , J . J . (1987) Outcome of Autistic Children Receiving Intensive Behavioral Treatment: Psychological Status 3 to 12 Years Later, Dissertation Abstracts International, 48(8-B) 2461. UMI no. AAT 8725043. M C E AC H I N , J . J . , S M I T H , T. & L OVA A S , O . I . (1993) Long-Term Outcome for Children with Autism Who Received Early Intensive Behavioral Treatment, American Journal on Mental Retardation 97: 35972. M E S I B OV, G . B . (1993) Treatment is Encouraging, American Journal on Mental Retardation 97: 37980. M U D F O R D , O . C . , M A RT I N , N . T. , E I K E S E T H , S . & B I B B Y, P. (2001) Parent-Managed

365

AU T I S M

8(4)

Behavioral Treatment for Preschool Children with Autism: Some Characteristics of UK Programs, Research in Developmental Disabilities 22: 17382. M U N DY, P. (1993) Normal versus High-Functioning Status in Children with Autism, American Journal on Mental Retardation 97: 3814. N AT I O N A L R E S E A R C H C O U N C I L (2001) Educating Children with Autism. Committee on Educational Interventions for Children with Autism. Division of Behavioral and Social Sciences and Education. Washington, DC: National Academy Press. N O R D I N , V. & G I L L B E R G , C . (1998) The Long-Term Course of Autistic Disorders: Update on Follow-Up Studies, Acta Psychiatrica Scandinavica 97: 99108. RO G E R S , S . J . (1998) Empirically Supported Comprehensive Treatments for Young Children with Autism, Journal of Clinical Child Psychology 27: 16879. RO G E R S , S . J . (2001) Diagnosis of Autism Before the Age of 3, in L . M . G L I D D E N (ed.) International Review of Research in Mental Retardation. Vol. 23: Autism, pp. 131. San Diego, CA: Academic. RU M S E Y, J . M . , R A P O P O RT, J . L . & S C E E RY, W. R . (1985) Autistic Children as Adults: Psychiatric, Social, and Behavioral Outcomes, Journal of the American Academy of Child Psychiatry 24: 46573. RU T T E R , M . (1970) Autistic Children: Infancy to Adulthood, Seminars in Psychiatry 2: 43550. S A L L OW S , G . O . & G R AU P N E R , T. D . (n.d.) Replicating Lovaas Treatment and Findings: Preliminary Results. Available at www.londonearlyautism.com, accessed 4 August 2003. S C H E I N KO P F , S . J . & S I E G E L , B . (1998) Home-Based Behavioral Treatment of Young Children with Autism, Journal of Autism and Developmental Disorders 28: 1523. S C H O P L E R , E . , S H O RT, A . B . & M E S I B OV, G . B . (1989) Relation of Behavioral Treatment to Normal Functioning: Comment on Lovaas, Journal of Consulting and Clinical Psychology 57: 1624. S I G M A N , M . & RU S K I N , E . (1999) Continuity and Change in the Social Competence of Children with Autism, Down Syndrome, and Developmental Delays. Monographs of the Society for Research in Child Development no. 56, vol. 64. S M I T H , T. (2001) Discrete Trial Training in the Treatment of Autism, Focus on Autism and Other Developmental Disabilities 16: 8692. S M I T H , T. & L OVA A S , O . I . (1997) The UCLA Young Autism Project: A Reply to Gresham and MacMillan, Behavioral Disorders 22: 20218. S M I T H , T. & W Y N N , J . (2002) Considerations for Selecting Consultants for Home-Based Programs, in O . I . L OVA A S (ed.) Teaching Individuals with Developmental Delays: Basic Intervention Techniques, pp. 32731. Austin, TX: Pro-Ed. S M I T H , T. , M C E AC H I N , J . J . & L OVA A S , O . I . (1993) Comments on Replication and Evaluation of Outcome, American Journal on Mental Retardation 97: 38591. S M I T H , T. , K L E V S T R A N D , M . & L OVA A S , O . I . (1995) Behavioral Treatment of Retts Disorder: Ineffectiveness in Three Cases, American Journal on Mental Retardation 100: 31722. S M I T H , T. , B U C H , G . A . & G A M B Y, T. E . (2000a) Parent-Directed, Intensive Early Intervention for Children with Pervasive Developmental Disorder, Research in Developmental Disabilities 21: 297309. S M I T H , T. , G RO E N , A . D . & W Y N N , J . W. (2000b) Randomized Trial of Intensive Early Intervention for Children with Pervasive Developmental Disorder, American Journal on Mental Retardation 105: 26985.

366

S H E A : A P E R S P E C T I V E O N L O VA A S
S M I T H , T. , G RO E N , A . D .

& W Y N N , J . W. (2001) Randomized Trial of Intensive Early Intervention for Children with Pervasive Developmental Disorder: Errata, American Journal on Mental Retardation 106: 208. Abstract retrieved 12 November 2002 from PsychINFO database. S M I T H , T. , E I K E S E T H , S . , K L E V S T R A N D , M . & L OVA A S , O . I . (1997) Intensive Behavioral Treatment for Preschoolers with Severe Mental Retardation and Pervasive Developmental Disorder, American Journal on Mental Retardation 102: 23849. S Z AT M A R I , P. , B A RT O L U C C I , G . , B R E M N E R , R . , B O N D , S . & R I C H , S . (1989) A Follow-Up Study of High-Functioning Autistic Children, Journal of Autism & Developmental Disorders 19: 21325. V E N T E R , A . , L O R D , C . & S C H O P L E R , E . (1992) A Follow-up Study of HighFunctioning Autistic Children, Journal of Child Psychology & Psychiatry & Allied Disciplines 33: 489507. VO L K M A R , F . R . , C O O K , E . H . , P O M E ROY, J . , R E A L M U T O , G . , TA N G U AY, P. E T A L . (1999) Practice Parameters for the Assessment and Treatment of Children, Adolescents, and Adults with Autism and Other Pervasive Developmental Disorders, Journal of the American Academy of Child and Adolescent Psychiatry 38 (12) Supplement: 32S54S. W I N G , L . (1981) Aspergers Syndrome: A Clinical Account, Psychological Medicine 11: 11529.

367