Behaviour Therapies

The behaviour therapies have relied on sophisticated individual experimental case designs applied to therapies for intractable personal problems. The literature is replete with outstanding examples although, as an extensive and successful series of treatment approaches, they have not been without criticism. The main criticisms have been that many behavioural treatments tend to focus on reducing single-target problems whereas, in reality, clients are likely to have multiple difficulties related to lack of communicative skill, a range of problem behaviours and so on. The second trenchant criticism has been that successful case examples are often derived from highly staffed, specialist units whereas most individuals with intellectual disabilities are cared for, managed and treated within generic community facilities. More recently, there have been a number of reports that are larger in scale both for individuals and groups. Graff, Green & Libby (1998) reported on the intensive behavioural treatment of a severely disabled, destructive and autistic four-year-old boy who had no functional or communicative skill. Treatment was comprehensive employing the full range of behavioural methods to reduce problem behaviours and increase pro-social and communicative skills. Following treatment, he was able to take instruction reliably, showed no aberrant behaviour, generalised his skills and developed some non-verbal communication. As independent validation, his medication was discontinued after treatment. Given the level of disability and destructiveness described by Graff et al., these changes and improvements are impressive indeed.

One of the most influential behavioural therapies has been based on a seminal study by Carr & Durand (1985) in which they developed the methods of functional communications training. In this study they hypothesised that problem/challenging behaviour, rather than being viewed as an aberrant condition, intrinsic to the individual, should be construed as a means of communicating the needs of the individual to others. Carr and Durand developed functional communication training, whereby the individual was taught alternative ways of communicating his or her needs. This was a very attractive hypothesis and generated a considerable research and treatment literature. However, studies have tended to find that functional communications training on its own may be of limited effectiveness without concurrent extinction of the target behaviour. Therefore, in addition to developing new ways of communicating, researchers have found it necessary to implement procedures to reduce the problem behaviour (Drasgow, Halle & Ostrosky, 1998; Fisher et al., 1998). Thompson et al. (1998) employed functional communications training with extinction for a range of problem behaviours in a six-year-old child. They found that the combined procedures reduced all forms of aggression except the one that their functional analysis indicated was maintained independent of contingencies. Following their functional analysis, they hypothesised that the remaining behaviour (chin grinding) was automatically reinforced. Following the provision of an alternative source of reinforcement, this problem was also eliminated. Hagopian et al. (1998) reported a series of 21 case studies of participants who had a range of intellectual disabilities and behavioural problems. Using carefully balanced research designs they found that functional communications training on its own was of limited value. When combined with extinction (not delivering a consequence for a problem behaviour) or punishment (for example, time out in a room or a chair) directed at the target behaviour, there were massive (90 %) reductions in target behaviour for 50 % of clients in the extinction condition and 100 % of clients in the punishment condition. It should be noted that extinction on its own as a treatment has been shown to produce increases in aggression and destruction and the importance of matching extinction procedures to target behaviours has been demonstrated frequently (Horner et al., 2002).

Positive Behavioural Programming

Positive behavioural programming is a behaviour therapy based on the influential writings and teachings of a number of behaviour analysts including LaVigna & Donellan (1986)

and Carr & Durand (1985). These procedures were developed to concentrate solely on interventions designed to increase desired behaviours with the theoretical argument that these would then replace problem behaviours. Put simply, if someone persistently bangs their head with their right hand, they should be taught to engage in an adaptive, constructive act with their right hand that would then replace the maladaptive behaviour. Functional communications training is clearly an example of positive behavioural programming. While there have been many encouraging reports and case studies concerning positive behavioural programming, research has suggested that methods for promoting prosocial, constructive abilities are more successful when they are employed in combination with methods directed at reducing antisocial, maladaptive behaviours. Time and again it has been shown that functional assessment increases the likelihood that treatment intervention will be successful (Horner & Carr, 1997).

Iwata et al. (1994) instigated the most influential system of functional analysis for the assessment of problem behaviours and subsequent research has developed and refined these methods. These authors set up analogue systems to identify the reinforcement contingencies that maintain certain challenging behaviours. In a series of experimental trials, the assessor introduces a variety of reinforcing conditions such as social reinforcement, tangible reinforcement, automatic (intrinsic) reinforcement, escape from demands (negative reinforcement) and so on, in order to assess their effect on the target behaviour in question. When the assessor ascertains which kind of reinforcement is controlling the behaviour, treatment is based on these findings. Clearly this work is of a highly individual nature and, as a result, most of the studies in the field are small N experimental designs). Recent research has focused on variables, setting conditions and events that can predict or maintain the occurrence of problem behaviour. Functional assessment has developed beyond analogue assessments into considering the same variables in natural settings, thus addressing one of the criticisms mentioned above.

Meta-analyses of Behavioural Interventions

Several meta-analyses have been conducted exclusively on outcome studies of behaviour therapy. For example, Carr et al. (1999) published a meta-analysis on interventions using specific types of behavioural interventions - those using positive behavioural support (PBS). This is a specific development of ABA. Like typical ABA it emphasises skills training and intervention based on functional assessment. However, PBS eschews punishment procedures and other procedures such as differential reinforcement of other behaviours because of their potentially aversive nature. Positive behavioural support also places at least equal weight on antecedent control and environmental redesign procedures as do contingency-based procedures. It is also concerned not merely with changing specific behaviours but to improve the quality of a person's life. Hence, it is interested in the broader ecology, social validity and systems issues related to positive client outcomes.

Carr et al. (1999) identified 216 articles from 36 journals using multiple methods including online and hand searches and contacting leading researchers in the field. Positive behavioural support articles included at least one intervention using stimulus intervention methods such as interspersed training, expansion of choice or curriculum modification, or reinforcement based interventions including functional communication training, self-management and differential reinforcement of alternative behaviours. Articles using non-PBS methods included differential reinforcement of other behaviour, extinction and time out were excluded. All of these methods were included in the Scotti et al. (1991) and Didden et al. (1997) meta-analyses. Other exclusion criteria were articles that had no data, used AB designs, group studies with no individual data, group design with no control group and n designs with fewer than three baseline data points. The dependent variable used in this meta-analysis was the percentage reduction over baseline calculated using the last three intervention data points. Of the 216 articles identified by Carr et al., 107 were excluded using the above criteria. The remaining 109 yielded 366 outcomes, because some studies used more than one dependent variable.

Changes in positive behaviours were reported in only 45 % of papers. They found that positive behaviours often increased substantially during PBS interventions. However, they noted that there were wide variations in this outcome measure. They found that 68 % of PBS interventions were associated with an 80 % or greater reduction in the target behaviour. This was true for interventions that used both stimulus-based and reinforcement-based interventions. Fifty-two per cent of studies achieved 90 % or greater reduction in the target behaviour. The outcome literature was limited in that response and stimulus generalisation was rarely addressed in the PBS literature and maintenance data were generally absent and few studies reported maintenance data beyond 12 months. Data on lifestyle change was taken in only 3.5 % of the papers reviewed and social validity data were taken in only approximately 10 % of papers. Input from clients, family members and staff on evaluations was rarely included in papers on PBS.

The effectiveness of PBS was mediated by several salient variables. Interventions that included significant others as change agents, which used environmental re-organisation and used typical agents of change were associated with larger effect sizes. Paradoxically, interventions in atypical settings were associated with larger effect sizes when compared with interventions in typical settings. This is particularly problematic because it suggests that interventions in experimental situations were more effective than interventions in naturalistic situations. However, it is the latter that are crucially important in any treatment intervention.

Carr et al. (2000) were modestly optimistic that interventions often have a large impact on a wide range of target behaviours. However, they were more cautious about the broader social validity, generalisation and maintenance of PBS interventions.

Although there are relatively few randomised controlled studies within the field of behaviour modification and the behaviour therapies, as we have argued, the scientific integrity of the field is beyond reproach. Where there are problems, they are in the generalisability of the treatment methods from experimental situations where those implementing the programmes are highly trained, to realistic situations where it is likely that individuals who are supporting clients are relatively untrained and not sophisticated in relation to behavioural methods. The Carr et al. (1999) meta-analysis has identified this weakness. However it is being addressed increasingly by those involved with applied behaviour analysis and behaviour modification. Corrigan (1991) conducted a meta-analysis of 73 social skills training studies conducted with people with intellectual disabilities, psychotic disorders, other psychiatric disorders and offenders. They found that studies of social skills training with people with intellectual disabilities were associated with the largest effect sizes. Hence, social skills training seems to be especially effective with people with intellectual disabilities.

Campbell (2003) conducted a meta-analysis of behavioural interventions of behavioural interventions for persons with autism using single subject designs. He identified 117 articles involving 181 individuals from 15 journals. Campbell again found strong evidence that behavioural interventions could reduce maladaptive behaviours in people with autism. Horner et al. (2002) review the existing literature on problem behaviour interventions for young children with autism. Reviewing peer reviewed research from 1996 to 2000, they used the following criteria for studies: subjects with autism less than 97 months old; problem behaviour as a dependent variable; an experimental design that allowed identification of a causal relationship between reduction and problem behaviour and intervention; data for individual subjects and at least three data points for pre-intervention and three data points for post-intervention phases. Nine articles were identified fulfilling these criteria with a total of 24 participants and 37 comparisons evaluated. The most frequent problem behaviour identified was tantrums (76 %) followed by aggression (59 %), stereotyped behaviour (14 %), and self-injury (11 %). Some functional assessment was conducted in 68 % of comparisons with a full functional analysis in 14 %. Interventions were stimulus based (43 %), instruction based (81 %), extinction (51 %), reinforcement of appropriate behaviour (30 %), punishment (32 %) and systems change (27 %). Multiple intervention components were included in 62 % of the comparisons and none included pharmacological procedures. In 62 %, intervention agent and context were typical and most likely to occur in the home or at school with the agent a parent or teacher. Of the 37 comparisons, the mean reduction in problem behaviour was 85 % with a median of 93.2 % and a mode of 100 %. Fifty-nine per cent of the comparisons recorded behaviour reductions of greater than 90 % and 68 % recorded reductions of 80 % or greater. Fifty-seven per cent of comparisons recorded maintenance data at an average of 12 weeks (longest one year) and in all cases, the level of improvement remained within 15 % of initial improvements during the intervention phase. There was no pattern for generalisation data. In this very careful review, the authors noted several limitations to their very positive conclusions. The two most important of these were that only nine studies were identified and that 'both single-subject and group design studies are more likely to be published if important effects are demonstrated than if no effects are found. This logical standard for publication means that the pattern of effects (or effect size) experienced in clinical contexts may not match that observed in published research' (Horner etal., 2002, p. 436).

Early Intervention Studies on Autism

Autism is a pervasive developmental disorder that has many incapacitating behavioural and cognitive features including an association with low IQ, chronic disruption of social function, severe behavioural problems and several features of developmental delay. Intervention research on individuals with autism is both extensive and far reaching and clearly illustrates the challenges for EBP at every level - scientific, political, legal, emotional and personal. By far the most comprehensive and detailed intervention is the early intervention project (EIP) developed by Lovaas (1987). EIP itself is designed to run for 40 hours or more per week, 365 days a year over a period of three years. Treatment during the first year is directed at reducing behavioural problems such as aggressive behaviours and self-stimulatory behaviours and also begins to promote prosocial behaviours such as appropriate play with toys and imitation responses. During the first year the family is also involved in treatment. Social deficits are addressed during the second year with programmes to encourage peer interaction and individual approaches to teaching both abstract and expressive language. In the third year, while continuing to maintain gains made during the previous part of the programme, pre-academic skills such as attending, concentration and basic reading, writing and arithmetic skills are taught. In addition, emotional recognition and expression are taught through the final year of the programme. Outcome claims by Lovaas (1987) and McEachin, Smith & Lovaas (1993) indicated that there were massive benefits in an EIP experimental group when compared to a control group. By using two outcome measures, IQ and educational placement, they found that 47 % of the experimental group achieved normal IQ and these improvements were maintained at 13 years of age (mean IQ = 84.5) compared to a mean IQ for the control subjects of 54.9.

Gresham & MacMillan (1998) review the Lovaas studies and the political and legal sequelae. They note a large number of internal and external validity threats to the experimental designs. Subjects were not randomly assigned to treatments but rather were matched in pairs. The measures should have been more targeted and discreet, giving information on individual adaptive function rather than the very global measures that were used. According to Smith & Lovaas (1997), those who deliver the programme require extensive, detailed training. Therefore the fidelity of replications or the fidelity of service developments far removed from the Lovaas Centre is very difficult to achieve. More recent developments in the field differentiate clinical subtypes. Although these developments were not available to the original authors, they have not modified their claims in light of these advances. Consideration of these subtypes might allow a better understanding of those individuals who might and might not achieve benefit from an EIP. As Gresham & MacMillan (1998, p. 11) note the stakes are high as these exchanges are not mere academic exercises; rather, families with the children with autism deserve critical analysis as do taxpayers who will be asked to finance the estimated $60 000 per child per year and... limited financial resources are frequently diverted into attorney's fees and high cost programmes such as EIP.

They note ruefully 'if anything positive has accrued from these disputes, it only may be the gainful employment of attorneys who appear more than willing to take these cases to fair hearings and/or the courts' (Gresham & MacMillan, 1998, p. 12). This clearly illustrates the challenges facing EBP on several fronts.

Subsequent research has supported the positive outcomes for EIP but, because of the length of treatment and the obvious nature of the programme, blind comparisons with a control group are not possible. Sheinkopf & Siegel (1998) addressed some of the criticisms by extending the work to home-based programmes implemented by parents and community based clinicians. The programme was less intensive with 27 hours being delivered over 11 weeks and post-treatment assessments were conducted between 18 and 20 months' follow-up. Therefore the researchers addressed the limitation that treatment had only been tested in specialist centres delivered by highly trained staff. Eleven children in the experimental treatment group were matched to 11 children in a control group who received conventional school-based and individual interventions. Strong cognitive gains were made by children receiving treatment despite the fact that it was far less intensive than in the Lovaas studies and much smaller, but statistically significant, effects were recorded on symptom severity. After treatment all experimental subjects still met the diagnostic criteria for autism. On this occasion, the authors note several study limitations including the low intensity school-based nature of the control group, the confounding of treatment intensity and treatment type in the experimental group and difficulties in establishing treatment fidelity.

Other programmes based on behavioural principles have also been assessed with promising results, but similar limitations. Ozomoff & Cathcart (1998) evaluated the effectiveness of a well-established home-based intervention. In this case, parents were taught the principles and methods in the programme typically lasting 10 weeks with around 10 hours intervention per week. Eleven children were assigned to the treatment group and 11 matched children received their normal programme (nine attended specific programmes for children with autism and two attended non-categorical programmes). Progress in the treatment group was three to four times greater than in the control group on a range of outcome tests including abilities on imitation, perception, fine motor skills, gross motor skills, eye hand integration and cognitive performance. Follow up was only four months. Jocelyn et al. (1998) reported a randomised treatment study comparing an experimental treatment group of 16 children and a control group of 19 children. Treatment consisted of a 12-week programme of educational seminars and support with consultations to parents and childcare workers who also worked with the children. Results indicated that the experimental group improved over the control group on language skills and that their parents felt more in control of their children. Follow-up was only 12 weeks.

McConnell (2002) has similarly reviewed interventions designed to facilitate social interaction for young children with autism. As with other reviewers in these areas, he notes that studies are predominantly single-case designs but also records that the studies typically maintain the highest standards of internal validity and scientific integrity. He reviews five general categories of methodology: ecological variations, collateral skills interventions, child-specific interventions, peer behavioural interventions and comprehensive interventions. The 11 studies identified under the category of ecological variations suggest that simply locating children with autism with more competent peers does not increase social interaction. However some weak-to-moderate effects were noted in initiations received. The nine studies identified in collateral skills interventions again suggested some effects on increasing social interaction by activating natural processes for social development. However, again the effects were modest. Fifteen studies were identified under the category child-specific interventions, most frequently social skills training, and the author concluded that child-specific interventions increased social interaction and promoted generalisation in maintenance of improvements. However, since they focus more on social initiations, these interventions may be constrained regarding long-term effectiveness. Thirty studies were identified in the category of peer mediated interventions and treatment effects were consistently substantial and robust. Treatment effects generalised to untrained peers and novel situations and maintenance of improvements was extensive. Seven studies employed comprehensive interventions (two or more of the previous categories) and the evidence, though small, suggested significant effects with some evidence of generalisation to other settings. One study (Odom et al., 1999) randomly assigned 98 pre-school children with a variety of disabilities including autism to five intervention conditions: ecological variation, child specific intervention, peer mediated intervention, comprehensive intervention and a control condition (pre-existing arrangements). At one-year follow-up only the peermediated condition demonstrated an effect in comparison to the control condition in terms of increased interaction. The authors considered that the comprehensive intervention may have been overly complex for a field trial with subsequent challenges on fidelity.

The work on treatment for children with autism is extensive and robust. There have been several reviews grading evidence, evaluating the best studies, making recommendations for future experimental and field trials and setting the work in a cultural, political and legal context. As has been pointed out, work has then been conducted addressing these criticisms. The available evidence suggests that interventions are effective although the criticisms of Gresham & MacMillan (1998) remain relevant. Indeed, because of the transparent and intensive nature of interventions, blind trials are impossible. However, efforts have been made to add randomised, controlled trials to the research literature. All of this work continues to attest to the effectiveness of interventions but the caution of Horner et al. (2002) regarding the publication of results, should be borne in mind.

Funny Wiring Autism

Funny Wiring Autism

Autism is a developmental disorder that manifests itself in early childhood and affects the functioning of the brain, primarily in the areas of social interaction and communication. Children with autism look like other children but do not play or behave like other children. They must struggle daily to cope and connect with the world around them.

Get My Free Ebook


Post a comment