The purpose of the present review is to consider recent evidence on intervention for behaviour disorders in children and adults with intellectual disabilities. We have included studies published during 2004 and the first few months of 2005, and used a broad definition of behaviour disorder. Thus, we have included studies of classic ‘challenging behaviour’ (such as self-injury, aggression towards others or the environment, stereotyped/repetitive behaviours), studies of general problem behaviours in people with intellectual disabilities, and behaviour problems labelled with common psychiatric terminology (especially conduct disorder, ADHD/Hyperactivity). The main sections of the review focus on evidence for interventions using methods derived from Applied Behaviour Analysis (ABA), pharmacological interventions, and also a number of service delivery models. However, the literature review also identified a number of general issues that have implications for our understanding and treatment of behaviour disorders in intellectual disability and it is to these that we turn first.
General issues in behaviour disorders and intellectual disability
Behaviour disorders in individuals with intellectual disabilities clearly constitute a significant clinical concern. A three-year longitudinal study of 13 preschool children by Green et al. (2005) suggest that behaviour disorders can emerge early and are highly persistent. When behaviour disorders are not treated, they can develop into serious, life-threatening actions once children develop physically and enter adolescence and young adulthood. Many psychological intervention studies of adults with intellectual disabilities are focused on cases where behaviour disorders are likely to have been present for many years, posing significant challenges for intervention. Early intervention for behaviour disorders is thus clearly warranted. However, there were no studies in the review period on early psychological intervention for behaviour disorders.
A further significant issue is illustrated by research from Rojahn et al. (2004), who studied the relationships between psychiatric conditions and behaviour problems in adults with intellectual disabilities. As indicated earlier, a broad range of terms are used to describe behaviour disorders in individuals with intellectual disabilities some, but not all, of which refer to psychiatric conditions. Rojahn’s research data suggest that behaviour problems may well be symptoms of underlying mental health problems but that the relationships between behaviour problems and psychiatric conditions are select and differential. Thus, aggressive/destructive and self-injurious behaviours (SIB) were related to impulse control and conduct problems, and stereotyped behaviours were related to pervasive developmental disorders and less strongly to psychosis. Of particular interest was the finding that depression, mania and anxiety were unrelated to SIB and aggression/destructive disorders, suggesting that these behaviours are not atypical manifestations of mood and anxiety disorders. However, there is as yet no consensus as to whether behaviour disorders are separate clinical conditions in intellectual disability or merely a part of the symptom pattern for co-existing psychiatric problems. More conceptual and empirical work is needed to explore this question, but the main implication for now is that it is important to clearly define the behaviour of interest in any intervention study if we are to evaluate the strength of an evidence base. Studies using ABA methods are characteristically strong on definitions of behaviour.
Within the review period, there were two useful studies of large numbers of cases of people with behaviour disorders exploring their access to intervention. In a Canadian sample of 625 adults and children with intellectual disabilities and behaviour disorders, Feldman et al. (2004) found that 97% were receiving some form of intervention but that the majority of these were informal (55%) (i.e., poorly recorded, and typically not evaluated). Formal recorded interventions were more likely to be present where the individuals engaged in more dangerous behaviours, and where formal interventions were in place 60% included the use of behaviour control drugs. Data from Australia also shows that many individuals with intellectual disabilities are given a range of drugs, often involving polypharmacology, to control their behaviour (McGillivray and McCabe 2004).
These studies illustrate that many individuals with intellectual disabilities are not in receipt of therapeutic interventions for their behaviour disorders, let alone interventions with an established evidence-base. Thus, a review of evidence for interventions is needed, including how treatment approaches might be developed into effective service models. The frequent use of pharmacological interventions supports the need for a review of the evidence base in this domain. Each of these issues is reviewed in the following sections.
Interventions based on the principles of Applied Behaviour Analysis
Central to contemporary applied behaviour analytic interventions is the prior identification of the cause or function(s) of behaviour. Function in this context refers broadly to the change brought about by a given behaviour in the individual’s external or even internal environment. Studies in the review period continue to support the efficacy of function-based interventions for challenging behaviours such as aggression, self-injurious behaviour and property destruction. Interventions in the review period included Functional Communication Training (FCT), choice-making, non-contingent reinforcement, and extinction. Overall, a theme of several of the studies identified is that low rates of reinforcer quality or density in the environment operate as setting events for the occurrence of challenging behaviour. While there remains an emphasis on demonstrating the effectiveness of single behavioural interventions on the frequency of a given target behaviour, a small number of studies reflected the use of combined behavioural intervention. This begins to approach the complexity of the multi-element behaviour support plans more common in service settings. Unfortunately, quality-of-life outcomes as a result of behavioural interventions were rarely reported, suggesting that researchers are not yet routinely using this important dimension as an outcome variable.
FCT is one of the most frequently recommended behaviour analytic interventions for the treatment of challenging behaviour, and was the most common intervention reported in the review period. Accumulating evidence from prevalence studies suggests that escape, from demands or social situations, is one of the most common functions of challenging behaviour (Hanley et al 2003). This is reflected in studies in the review period where escape or task avoidance was identified as the primary function of challenging behaviour. Peck Peterson and colleagues (2005) demonstrated that while the implementation of FCT to request breaks from educational tasks (i.e., escape) can, alone, reduce the frequency of challenging behaviour, it also increases task avoidance—which is itself is often a problem in education and work settings. However, when FCT was combined with increased reinforcement associated with the avoided task relative to avoidance itself, task engagement increased without a concurrent increase in problem behaviour and work tasks were more frequently requested. This study demonstrates that FCT for escape is ineffective in increasing task engagement without concurrent changes to the motivative dimensions of avoided tasks, and that in such cases combined interventions are warranted to address both environmental deficits and skills deficits in the individual. In another study investigating FCT but using a modified analogue assessment procedure to identify the function of low rate occurrence of challenging behaviours, Tarbox et al. (2004) demonstrated that non-contingent attention and FCT to request task termination was associated with reductions in attention and escape-maintained aggression in two adults with severe and profound intellectual disability in a sheltered workshop setting. Collateral recording of task engagement did not occur but the absence of a higher rate, or increased quality, of reinforcement for task engagement suggests task engagement is unlikely to have been affected.
Two other studies contributed to an increased understanding of the conditions under which augmentative communication strategies such as FCT are most effective. Schindler and Horner (2005) demonstrated that FCT does not automatically reduce problem behaviours in settings other than those where FCT is taught. When three 4-5 year old children with autism underwent FCT in one setting to reduce problem behaviour in school, problem behaviour reduced in that setting, but not in three other targeted settings in school and in the family home. However, with prompting and with reinforcement contingent on the augmentative communication response in these other settings, a reduction in problem behaviours in those settings was observed. Mildon and colleagues (2004) combined FCT with non-contingent escape (NCE) in the treatment of escape maintained problem behaviour and to increase compliance with task demands. Their results suggest that the initial implementation of NCE creates an opportunity to subsequently teach a functionally equivalent response to the problem behaviour. NCE combined with FCT was subsequently changed to NCE and FCT with a changing criterion (i.e., a changing number of tasks had to completed before the functionally equivalent response was reinforced). In this condition, no increase in problem behaviour was observed and compliance increased.
One of the more interesting findings in the review period concerns the use of extinction. Fisher et al.(2004) reported that for four individuals with intellectual disability, destructive behaviour was maintained by access to adult interaction. Both non-contingent attention and non-contingent access to items of high interest combined with extinction were equally effective in bringing about rapid reductions in destructive behaviour, and both were differentially more effective than extinction alone. That there was no difference between the two interventions suggests that setting or motivating events, such as environmental deprivation or unpredictability of reinforcer delivery, may have increased the salience of adult interaction as a reinforcer at the outset. A second study by Long and colleagues (2005) also demonstrated that the introduction of non-contingent access to items of high interest to the individual can bring about reductions in the frequency of problem behaviour maintained by automatic reinforcement and problem behaviour maintained by escape during hygiene routines. Non-contingent access to items of high interest probably reduces escape maintained problem behaviour by making the demand context less aversive. However, both studies in the review period on competing stimuli suggest that environmental deprivation operates as a setting or motivating event for problem behaviour. A concern, in terms of social validity of these studies, is that they do not contain an explicit contingency to teach an alternate response to the challenging behaviour.
Meta analysis and reviews
Shogren and colleagues (2004) conducted an excellent review on choice making as an intervention for the treatment of problem behaviour. Specifically, they identified 13 studies that examined the effects of one of two choice interventions: interventions that allowed individuals to choose the order in which they completed assigned tasks, and interventions in which individuals chose between two activities. Meta-analysis results indicate that facilitating individuals with intellectual disability to make choices significantly reduced problem behaviour to below baseline rates of behaviour. Interestingly, choice-making as an intervention was associated with greater reductions in aggressive problem behaviours than non-aggressive problem behaviours. Overall, these results suggest that when individuals with intellectual disability increase their self-determination, behaviour problems can reduce.
Adam and colleagues (2004) conducted a review of 26 studies evaluating behavioural interventions in the treatment of pica from 1975 onwards. Earlier studies were associated with an absence of functional assessment and the use of procedures that could be classed as aversive (contingent aversive presentation, over-correction, physical restraint, time-out, facial screening, and negative practice). Only four of the studies reported the use of functional assessment, and functions identified for pica were primarily physiological and social. All interventions reported reductions in pica. However, there were a large number of interventions that could be classed as aversive. Non-aversive approaches, such as non-contingent presentation of food/attention, discrimination training and differential reinforcement, appear equally effective in the treatment of pica. Thus, the need for and validity of aversive procedures is in question.
The prevalence rate for the prescription of psychotropic medication for the treatment of challenging behaviours has been a cause for concern, given the lack of empirical data on their effectiveness (McGillivray and McCabe 2004). Over the review period, the vast majority of articles focused on the effects of atypical anti-psychotic medication and, in particular, the effects of risperidone on behaviour disorders. In a review of empirical research on the effectiveness of risperidone from 1992 to 2004, Singh et al. (2005) identified only six studies that met the methodological criteria for sound psychopharmacological investigation. They conclude, with some caution, that risperidone may be an effective treatment for some individuals with certain behavioural presentations.
Over the review period, four studies were identified which investigated the effects of risperidone on behaviour disorders occurring in children with intellectual disability and/or pervasive developmental disorders (PDD). A strength of this cohort of studies is the use of common outcome assessment instruments, notably the Aberrant Behavior Checklist (ABC) and the Nisonger Child Behavior Rating Form (N-CBRF), and very similar age ranges of children. In an eight-week, randomised, double-blind investigation of children aged between 5 and 12 years, Shea and colleagues (2004) reported significantly greater improvements in the risperidone group on subscales of the N-CBRF and ABC than the placebo group. The most prevalent side effect was somnolence reported in 72.5% of the medication sample. However, approximately 90% of the risperidone group received at least one concurrent medication during the trial such as analgesics or sedatives. A study by Aman, Binder and Turgay (2004) combined data from two previously published six week placebo-controlled trials and investigated risperidone in the treatment of children in the same age range with disruptive behaviour disorders and comorbid ADHD. Compared to the placebo groups, risperidone treatment was associated with greater improvement on measures of conduct problems and hyperactivity. Concurrent stimulant medication was not associated with greater improvements on these measures, thereby questioning the efficacy of such medication in the treatment of hyperactivity.
In two open-label studies with children with intellectual disabilities, a similar pattern of results was observed but these studies also allowed for an investigation of long-term effects of risperidone. Findling and colleagues (2004) reported improvements in disruptive behaviours as measured by the conduct disorder subscale of the N-CBRF over a 48 week period. Croonenberghs and colleagues (2005) reported improvements on the conduct disorder, insecure/anxious, hyperactive, self-injury/stereotypic, self-isolated/ritualistic and overly sensitive subscales of the N-CBRF in 363 children aged between seven years and 14 years prescribed risperidone over a 12-month period. In this latter study, there was no significant difference between gains observed at four weeks and at 12 months suggesting that improvements are unlikely to be observed after an initial four weeks. Gains were observed on basic measures of memory and attention at 12 months providing some tentative evidence that learning may not be not be affected by risperidone. However, these measures were administered prior to commencement of medication and at the completion of the study and consequently it is unclear how attention and learning were affected after initial medication and throughout the trial. A high rate of side effects was noted in both studies (97% and 91%), with the most common side effect being somnolence reported in 30% and 33% respectively (Findling et al. 2004; Croonenbergs et al 2005). Though somnolence was reported as mild and generally transient, the relationship between somnolence and behaviour reduction is unclear. Elevated levels of serum prolactin were noted in both studies after approximately four weeks. Though levels returned to the non-clinical range at approximately nine months, they remained elevated compared to baseline. The incidence of extraprimidal effects was low, but weight gain above typical developmental gain was observed. The number of children receiving behavioural interventions concomitantly with the use of risperidone or placebo was not specified in either study.
Guidelines in respect of the use of risperidone were also published in the review period (Aman and Gharabawi 2004). These are to be welcomed as they standardise dosage and titration schedules for risperidone with respect to aggression, irritability, and problems of impulse control in individuals with intellectual disability. These guidelines also specify that functional assessment should be used in the evaluation of behaviour disorders. However, the guidelines do not identify the criteria under which risperidone should be discontinued. This is unfortunate, as data in the review period identify optimal clinical response within four weeks of treatment onset. Furthermore, the use of medication must be balanced against relapse on discontinuation and failure to specify the necessity of function based behavioural interventions, which have been shown to bring about reductions in challenging behaviour and improve adaptive skills.
The role of a second atypical anti-psychotic, Olanzapine, in the treatment of behaviour disorders exhibited by children and adults with PDD was reported by Stavrakaki and colleagues (2004). Though improvements were noted using the Clinical Global Impressions scale (CGI) and the Global Assessment of Functioning across seven individuals aged between eight and 52 years over a 26-week period, all had co-morbid Axis I diagnoses and were receiving unspecified behavioural interventions, making firm conclusions about the efficacy of olanzapine impossible. The use of use basic measures of behaviour change such as the CGI has also been noted to over-estimate improvements (Adam et al. 2004). A case study by Symons and colleagues (2004) reported a decrease in self-injurious behaviour in a 13-year old girl as a result of treatment with clonidine. Substantial clinical effects were observed within four weeks, which, like earlier studies, suggests that if atypical anti-psychotics are to be effective, it will be within a short time frame.
A final study by Zarcone and colleagues (2004) is particularly notable for the inclusion of functional analysis methodology (i.e., analogue assessment) in determining the efficacy of risperidone on destructive behaviour. In contrast to the majority of other studies, which relied on rating scales to measure outcomes, this study tracked the actual rate of clearly defined behaviours on a per minute basis across five analogue conditions and across medication phases for 13 individuals. Risperidone was effective in reducing destructive behaviour (compared to placebo) for ten individuals. For seven of these ten, the occurrence of destructive behaviour did not co-vary with specific environmental/analogue conditions.
Behaviour disorders have also received some interest at a broader level of intervention—that of service design and service models, whether these are directed explicitly at behaviour disorders or where behaviour disorder may be a measured outcome. There were four studies of this kind published in the review period. One key service management issue is whether to co-locate people with behaviour disorders or to include them in community settings with other service users. Robertson and colleagues found that co-locating people with behaviour disorders in community-based supported accommodation cost more (Robertson et al. 2004) and was associated with more physical restraint and interventions (Robertson et al. 2005). In both types of settings, behaviour disorders were stable over a 10-month period and there were few ABA technologies in place to reduce behaviour (<15% of service users) but high levels of antipsychotic medication in use (Robertson et al. 2005).
A general service model that is becoming popular in western societies is the use of person-centred planning (PCP) with service users. Holburn and colleagues (2004) studied 20 adults with intellectual disabilities and behaviour disorders receiving PCP and a matched group of service users who received standard interdisciplinary service planning. The data showed that quality of life improved more for the PCP group when compared with the matched controls, and many more of the PCP group were able to relocate to a community service setting. However, no data are presented on changes in behaviour disorders as a result of the two different service models.
Most informative are data from two large-scale studies of the application of ABA technologies in the treatment of behaviour disorders. Asmus and colleagues (2004) summarised data on 138 cases referred to a short-term inpatient unit that used functional analysis and behavioural intervention methods. These cases were those that had not been successfully treated in an outpatient clinic using similar techniques. Most cases were for the treatment of aggression, disruptive behaviours, and self-injury. For 66% of cases, behaviour problems were reduced by 90% when the average rates of behaviour in baseline were compared to the final three treatment session observations. These effects were achieved in an average of 10 days of assessment and treatment. This was a highly specialised ABA assessment and treatment service with input from other professionals and thus is not easy to replicate. However, the data are supportive of a high degree of effectiveness with a sample of difficult-to-treat cases. Unfortunately, follow-up data were not available on a consistent basis so the extent to which these gains are maintained when individuals leave the inpatient service is unknown.
The final study of interest adopted a different approach to the use of ABA technologies. A staff training model was adopted by McClean and colleagues (2005) whereby ABA assessment and intervention skills were taught and staff members developed behaviour support plans, as a part of their training, for 138 service users with behaviour disorders. As this was a clinical service model, there were no control group data available but 77% of cases were deemed a success (70% or more reduction in rates of behaviour disorder from baseline levels) and these changes were shown to maintain over a two-year follow-up period. This is one of the few studies to provide data with respect to maintenance of reductions in challenging behaviours over a significant period of time.
The present review identifies the dominance of two approaches for the treatment of behaviour disorder in people with intellectual disabilities. First, there is gathering evidence that risperidone can have relatively fast suppressing effects on behaviour disorders in children with intellectual disabilities. However, there remain questions about the effects of somnolence on reductions in behaviour disorder and, despite emphasising the issue in treatment guidelines (Aman and Gharabawi 2004), there are few examples of studies combining the analysis and treatment methods of ABA with risperidone intervention (Zarcone et al. 2004). Second, assessment and treatment approaches from ABA perspectives continue to be the focus of clinical research. The evidence base here is well-established (Carr et al. 1999), and the necessity of pre-intervention functional assessment has been confirmed. Themes in the review period were attention to broader environmental variables, and multi-element treatments. These themes fit very well with the translation of controlled intervention studies into service models for people with intellectual disabilities and behaviour disorders.
In terms of implications for services, there was no clear evidence of benefits to the co-location of people with behaviour disorders, and no data on the effects of PCP on behaviour disorder (although this model seems to help achieve quality of life improvements). However, there were encouraging data in the review period about using ABA principles as a service model. In particular, a short-term in-patient service relying on comprehensive behavioural assessment and intervention may contribute to reductions in behaviour disorder for hard-to-treat cases, and a competency-based staff skills training model may be an effective method for disseminating ABA skills and achieving service wide impact with positive changes on behaviour problems.