Critically review the evidence base for applied behaviour analysis in the effective treatment of autism

Autism is a chronic developmental disorder which affects essential human behaviours such as the ability to interact socially, to communicate effectively, and to imagine (Mulick, & Butter, 2002). At present education is the primary form of treatment in autism (Roberts, 2003). Education is broadly defined as the promotion of skill and knowledge acquisition in order to support personal independence and social responsibility (Kavale & Forness, 1999). In autism, this education incorporates socialisation, communication, adaptive skills, and the reduction of challenging behaviours.

This essay will focus mainly on critically evaluating the two most widely endorsed and cited educational approaches to the treatment of autism (Roberts, 2003), the behavioural therapies based on applied behaviour analysis (ABA) with a focus on the UCLA Young Autism Project (Lovaas, 1987), and the combined therapies with a focus on the TEACCH project (TEACCH, 1998). Sensory-motor therapies and other less widely endorsed therapies will also be discussed in brief. Applied Behaviour Analysis Based Treatment of Children with Autism

The focus of ABA on promoting socially relevant skills of practical importance to the individual (Baer, Wolf, & Risley, 1968) makes it an ideal candidate for the treatment of autism. The ABA approach conceptualises autism as characterised by behavioural deficits and excesses which are open to change through the utilization of controlled environments (Lovaas & Smith, 1989). Specifically, in ABA the analysis of behaviour is applied to the understanding of the functional relationship between behaviour and environment (Jensen & Sinclair, 2002).

There are numerous empirically supported treatment methods which are used within ABA-based programs such as shaping, chaining, and prompting, as well as techniques which have been specifically designed for use in the treatment of autism (e. g. discrete trial training, incidental teaching) (MADSEC, 2000). Common to all methods is their ability to change behaviour in a systematic, observable and measurable way (Sulzer-Azaroff & Mayer, 1991). Behaviourally-based interventions have proven effective in improving an extensive range of skills in children and adults with autism (Thorp, Stahmer, & Schreibman, 1995).

For instance, a review of 251 studies of ABA based interventions with people with autism from 1980 to 1995 found that ABA was effective in promoting language and communication development, daily and community living skills, social skills and academic skills (Baglio, Benavidiz, Compton, Matson & Paclawskyj, 1996). The effectiveness of ABA in the modification of specific behaviours prompted the authors of a review of 1,100 studies to conclude that a systematic behavioural education program should be the treatment of choice for improving and expanding the behavioural repertoire of a child with autism (DeMyer, Hingten, & Jackson, 1981).

However, if ABA is to be deemed effective in the treatment of autism it must be shown to be effective in treating in concert the core deficits of autism rather than in the modification of specific skills in isolation (Roberts, 2003). There is substantial empirical evidence which supports the proposition that early comprehensive and intensive behaviour analytic intervention can produce substantial, wide-ranging, enduring and meaningful improvements in relevant domains for a large proportion of children with autism (Green, 1996).

A landmark study by Lovaas (1987) showed that 47% of children with autism in the program, achieved normal intellectual functioning as indicated by normal-range IQ scores after 2 to 3 years of intensive 1:1 (instructor: child) behavioural intervention involving 40 hours per week discrete trial training based instruction (Lovaas, 1987). In fact, the intervention group gained and maintained on average 22-31 IQ points more than the control groups over the intervention period (McEachin, Smith, & Lovaas, 1993). The UCLA project demonstrated the benefits of behavioural treatment and the importance of early intervention (before age 5) (Lovaas, 1987).

However, the project has also been criticised on numerous methodological grounds (Gresham, & MacMillan, 1997). Most importantly, the sample selected for the intervention was not representative of individuals with autism and contained systematic bias as 15% of those available for selection were selectively excluded (based on prorated mental age) so as that the resultant sample was higher functioning than a random sample (Gresham et al. , 1997). The original Lovaas (1987) study has also never been replicated using the same treatment intensity which creates another gap in the support for the effectiveness of this treatment.

A number of studies which have been published in the time since the Lovaas (1987) study have supported the efficacy of comprehensive and intensive behaviour analytic treatment for young children with autism. For instance, a number of home-based partial replications of the UCLA project have reported gains in important domains of functioning, though the nature of these improvements varied significantly across studies.

The May program found gains in all areas examined (e. g. anguage, adaptive functioning) (Anderson, Avery, Dipietro, Edwards, & Christian, 1987), the UCSF (University of California, San Francisco) program found large gains in IQ (on average 28 points) (Skeinkopf et al. , 1998), and the Murdoch program found significant IQ gains amongst particular sub-groups but not across the group as a whole (Birnbrauer & Leach, 1993). These findings whilst supportive of the effectiveness of comprehensive ABA based programs, also demonstrate that these programs have yet to yield consistently positive results in support of the effectiveness of ABA in the treatment of autism (Smith, 1999).

However, the lack of sufficient funding and the difficulties in recruiting adequately trained staff and of providing regular supervision of those staff may explain why these partial-replications failed to consistently achieve similar outcomes to those of McEachin et al. (1993). School-based behaviour analytic programs include the Douglas Center program, the LEAP program (Lifeskills and Education for Students with Autism and other Pervasive Developmental Disorders), and the PCDI program (Princeton Child Development Institute) program (Smith, 1999).

The Douglas program has yielded mixed outcome results, indicating that the behavioural intervention impacted more significantly on certain areas of functioning (e. g. cognitive functioning) than on others (e. g. language) (Harris, Handleman, Gordon, Kristoff, & Fuentes, 1991). The LEAP and PCDI programs suggest respectively that the use of ABA with children with autism can improve cognitive functioning and can help children achieve mainstream school placement (Roberts, 2003).

However, the reliance of the later two programs on a single-outcome measure limits the conclusions which can be drawn from these studies regarding the effectiveness of ABA (Gresham, Beebe-Frankenberger, & MacMillan, 1999). The UCLA and home-based programmes suffer from the least number of methodological weaknesses in terms of the reliability of assessments, use of experimental or quasi-experimental design, operational definitions of the treatments, and use of multiple outcome measures, and it is perhaps significant that the UCLA study had both the strongest methodology and the most positive results (McEachin et al. 1993; Smith, 1996). However, the need for replication of this study remains a significant concern.

Non-behavioural Treatment of Children with Autism Whilst behavioural treatment and evaluation methods are constantly being reformed and improved it is of prime importance to identify alternative treatments which may be used in collaboration with these treatments to enhance their effectiveness, to act as an alternative, or to supplant behavioural treatment altogether (Smith, 1996). Several studies have been conducted to evaluate non-behavioural therapies for children with autism.

A number of approaches to the treatment of children with autism such as psychodynamic management or facilitated communication have been shown to be both ineffective and harmful to such children (Maurice, Green, & Luce, 1996; Mesibov, Adams, & Clinger, 1997). Sensory-motor therapies such as sensory integration training, auditory integration training, and music therapy have been associated with both anecdotal and research support for their effectiveness in treating symptoms of autism such as ritualistic and self-injurious behaviour (Roberts, 2003).

Such therapies operate from the empirically supported and theoretically intuitive stance that for some children with autism environmental overstimulation or understimulation may contribute significantly to both social and behavioural problems (O’Neill ; Jones, 1997). However, a review of the research on sensory-motor therapies concluded that the research studies which exist in the area are characterised by serious methodological flaws (e. g. non-random selection of participants, no control group, participant characteristics not explained) (Dempsey ; Foreman, 2001). At present the potential benefits from sensory-motor training have not been demonstrated scientifically (Dempsey et al. , 2001) and a need exists for carefully controlled studies with procedural transparency and reliable and operationally defined measures of intervention based change (Roberts, 2003). A number of combined therapies have incorporated a combination of techniques to support children with autism (Gresham et al. , 1999).

For instance, the TEACCH (Treatment and Education of Autistic and Related Communication Handicapped Children) program focuses on the strengths in functioning of children with autism (TEACCH, 1998). However, at present there are no comprehensive treatment outcome studies which have evaluated the effectiveness of the TEACCH model in a systematic and controlled manner (Smith, 1999). Parents have indicated their satisfaction with the program and reported that only 7% of those who enrolled in the program were institutionalised in later life (Schopler, Mesibov, ; Baker, 1982).

More recently, a study has reported developmental gains of 9. months in a four month intervention period, suggesting that TEACCH interventions in the home increase developmental functioning (Ozonoff ; Cathcart, 1998). However, the methodology of this study was compromised by the lack of supervision to ensure that the skills implemented in the home cohered with those as outlined by the TEACCH program (TEACCH, 1998). A further three TEACCH follow-up outcome data studies have documented substantial IQ increases (15-24 points) over a 4 to 5 year period amongst children with lower functioning at program commencement (aged 3-4, IQ;50) (Smith, 1999).

However, the overall effect of the TEACCH program on cognitive functioning across all participants was only 3-7 IQ points suggesting that the program is ineffective or only minimally effective for higher functioning children with autism (Smith, 1999). These three studies did not use a control group and it therefore remains unclear at present whether the significant gains observed in the lower functioning sub-group were due to inclusion in this program or due to maturational factors. Further combined therapy programs include the Higashi School Program, the Giant Steps Program and the Colorado Health Sciences Program (Dempsey et al. 2001).

The former two programs have not published in-house or independent research supporting the effectiveness of their interventions, and claims of integration of children into mainstream schools, and of parental satisfaction with the program are the sole evidence in regard to the effectiveness of these treatments (Dempsey et al. , 2001). The Colorado Health Sciences Program emphasises learning through play and is based on object relations and Piagetian theory (Smith, 1999). Two outcome studies have identified IQ increases of 3-9 points amongst children with autism (Smith, 1996).

The comparatively small gains made in this program, the failure of these studies to present data in support of their findings, or to use a control group, combined with the brief follow-up period of 5-6 months to assess gain maintenance makes it impossible at present to infer that this therapy is effective. Discussion of Treatment Effectiveness The above review has evaluated a number of therapies for the treatment of young children with autism. ABA based therapy has been shown to be effective in producing large gains in various aspects of functioning (e. . cognitive functioning, langauge) (Smith, 1999) and such gains have been shown to be maintained after the termination of treatment (McEachin et al. , 1993). Non-behavioural therapies have at present not conducted scientifically sound research documenting either the effectiveness or long-term benefits of such treatment, and thus remain unvalidated at this time (Smith, 1999).

The research evaluating the effectiveness of non-beahvioural treatments, whilst methodologically problematic, does suggest that some such therapies (e. . TEACCH) may be somewhat effective in the treatment of autism. From a service delivery perspective it is important that carefully controlled research identify which treatments are more efficacious, and for which populations (Roberts, 2003). Eikeseth, Smith, Jahr and Eldevik (2002) have compared the effectiveness of intensive 1:1 ABA based treatment with “eclectic” treatment which combined TEACCH-based procedures with discrete trial training, sensory integration therapy and a number of other methods.

After one year the children who received behaviour analytic treatment had made significantly greater gains in cognitive, language, and adaptive functioning than those who recieved “eclectic” treatment. These findings are particularly strong as children were randomly assigned to either group and both interventions were provided for approximately 30 hours per week by staff with similar levels of training. This study also demonstrated that substantial gains in functioning (e. . 17 point gain in IQ) could be made in only one year with less intensive intervention than that advocated by Lovaas (1987) (30 vs. 40 hours per week).

Howard, Sparkman, Cohen, Green and Stanislaw (2005) also found significantly greater gains in cognitive, language, and adaptive functioning amongst children with autism who received 14 months of behaviour analytic treatment as compared to a group who received an “eclectic” treatment similar to that in the Eikseth et al. 2002) study, or to a third group who received nonintensive public early intervention. These findings contradict inferences made by reviewers that non-treatment specific factors such as supportive and structured teaching, predictability and routine, and family involvement are central to the determination of treatment effectiveness (Gresham et al. , 1999; Roberts, 2003). They instead suggest that it is the type of intervention, not common elements identified across treatments or even the intensity of treatment that determines effectiveness.

Overall, the behaviour analytic studies have demonstrated that ABA based treatment with young children with autism can result in significant gains in cognitive functioning, language, and adaptive functioning (Green, 1996; Smith, 1999). Parents have reported high levels of satisfaction with these treatments and have experienced lower levels of stress than parents of children with autism not receiving behaviour analytic treatment (Anderson et al. , 1987; Birnbrauer et al. 1993). However, there are a number of issues of concern regarding the existing evidence for the effectiveness of behaviour analytic interventions in the treatment of autism (Smith, 1999). Of particular concern are the differences in treatment responsivity which characterise the behaviour analytic programs discussed. The number of children that made substantial gains in regard to various aspects of functioning varied both between and within studies.

For instance, Lovaas (1987) noted that the children in the UCLA study could be split into two groups one of which was labelled the “best outcome” group who were higher functioning before treatment and made major gains on standardised tests, and another group who were lower functioning initially and whose scores on standardised tests remained quite stable. The evidence suggests that certain children are more likely to benefit from ABA treatment than others, and that some domains of functioning are likely to be more amenable to change due to ABA intervention.

Specifying child characteristics which will predict response to such treatment is thus an important concern for service providers who seek to match specific interventions to the needs of the child (Lord ; Schopler et al. , 1982; Mundy, 1993). One reviewer (Smith, 1999) has proposed that the body of evidence suggests that as in the Lovaas (1987) study those with the highest functioning at pre-treatment will make the largest gains through ABA instruction. However, the nature of the behavioural intervention may explain the failure of a sub-group of children to make significant gains in the UCLA project (Roberts, 2003).

In particular discrete-trial training has been criticised for being poor at promoting the generalisation of learning from one environment to another, particularly in those with a lower level of functioning at program inception (Schriebman ; Pierce, 1993). Incidental teaching, the natural language paradigm and pivotal response training have been developed to promote such generalisation through the provision of a more naturalistic approach to the behavioural treatment of children with autism (Hart, 1995).

At present, though these techniques hold promise, no outcome studies documenting the effect of the intensive and comprehensive use of these interventions have been carried out (Smith, 1999). There is also a need for more comprehensive outcome measures in order to link both child characteristics and ABA instruction characteristics to gains in specific areas of functioning. The studies discussed focused principally on global improvements in cognitive functioning and ignored important aspects of skill acquisition and change across domains such as social development, aggression, rituals and daily living skills (Dempsey et al. 2001). It would also perhaps be useful to evaluate the effects of ABA interventions on measures which assess change in regard to the core deficits of autism (e. g. pragmatics, theory of mind, emotional regulation) (Baron-Cohen, 1995). Such assessment is more consistent with the proposition that ABA focuses on the promotion of socially relevant behaviours (Baer et al. , 1968) rather than on changes in the child’s intelligence quotient.

Similarly, a wider focus on assessing the change in both the child’s and parents’ quality of life would help evaluate the practical importance of the changes brought about by ABA instruction (Schalock, 1996). In conclusion, the use of applied behaviour analysis in the treatment of autism has been shown to be effective in generating substantial gains in diverse areas of functioning amongst children with autism. ABA based intervention is currently the sole approach to the treatment of autism which is significantly supported by empirical data (New York State Department of Health Early Intervention Program, 1999).

However, the medical comorbidities, the range of child abilities and diverse family characteristics make the tailoring of treatments to the needs of the child a complicated and challenging task for the future. Carefully designed outcome studies with comprehensive evaluation measures coupled with research investigating the development and improvement of behavioural interventions will help achieve the important goal of identifying methods which are likely to help specific populations improve significantly in specific areas of functioning.