The effectiveness of applied behaviour analysis interventions for people with Autism Spectrum Disorder

Publication Details

This systematic review considers the evidence for the effectiveness of interventions grounded in the principles of applied behaviour analysis for people with autism spectrum disorder.

Released on Education Counts: April 2010

Author(s): Marita Broadstock and Anne Lethaby, New Zealand Guidelines Group.

Date Published: 19 December 2008

Please consider the environment before printing the contents of this report.

This report is available as a download (please refer to the 'Downloads' inset box).  For links to related publications/ information that may be of interest please refer to the 'Where to Find Out More' inset box.


The majority of primary studies identified by the current review considered early intensive behavioural interventions clearly based on ABA principles contrasted with standard care programmes.

Comparison of early intensive behavioural intervention with standard care

A larger number of studies compared behavioural treatment with eclectic treatment or standard care that was available in the community for autistic children. It is possible that the eclectic care arms of the included studies varied in the content, delivery and intensity of their programmes but it is useful to compare the findings of the studies to see if patterns can be determined. Some of the studies attempted to replicate the number of hours per week recommended by Lovaas (1987) (35 to 40 hours per week) in the Young Autism Project but most averaged 25 to 30 hours per week. A variety of outcomes were measured from one to three years follow up. Results were not always consistent and the extent of improvement varied across the studies. 

Only four studies (all non randomised experimental or cohort studies) attempted to broadly match intensity between behavioural and eclectic/usual care arms. Of the three considering treatments of high intensity averaging between 25-40 hours per week, two (of good and fair quality respectively) reported benefits for children having behavioural treatment in cognitive development/IQ, adaptive functioning and language development at 12-14 month, and 2 ½ year follow-up (Eikeseth et al. 2002; Eikeseth et al. 2007; Howard et al. 2005). The other study (Magiati et al. 2007) rated as good quality found no group differences after two years in cognitive ability, language, play or severity of autism, although a trend was found for the EIBI group (approaching significance) for daily living skills. Finally, a fair quality study comparing less intensive behavioural treatment with eclectic treatment (average of 12 hours a week) found benefits for the behavioural group in IQ, language, communication but not in nonverbal intelligence, adaptive functioning, daily living or socialisation (Eldevik et al. 2006). These results are consistent with the conclusions of two recent high quality systematic reviews. BMJ Clinical Evidence concluded that ABA may improve IQ and language skills compared with eclectic treatment (a recommendation based on low quality evidence) (Parr 2007). The review conducted by SIGN (Scottish Intercollegiate Guidelines Network 2007) gave a qualified recommendation that behavioural interventions should be considered to address a wide range of specific behaviours in children and young people with ASD. 

This potential benefit of behavioural treatment when compared with usual care or mixed approaches must be treated with caution as results in similar studies were not consistent and many of the studies noted that there was a large variation in individual response within the treated groups. Several studies attempted to ascertain the factors that predicted response but results from these analyses were not always consistent and/or conclusive. 

Comparing variations of EIBI programmes 

There is clear evidence that behavioural therapy produces positive effects in young children with ASD. What remains to be identified are the particular features, intensity and duration of behavioural therapy that give optimal results.

One RCT (Kasari et al. 2006) of good quality reported that the addition of joint attention and play skills interventions to a standard early intervention programme resulted in greater joint engagement, joint attention and functional play skills than the control group which had similar intensity treatment, thus suggesting that the content of the intervention was responsible for the benefit. This finding needs to be confirmed in further research studies. 

The three primary studies that attempted to compare intensity and mode of delivery of treatment did not reach clear conclusions about optimal treatment. Although one study (Reed et al. 2007) found that children treated for an average of 30 hours per week achieved stronger gains in educational functioning than those treated for an average of 13 hours per week, there were no reported differences for other types of functioning such as adaptive functioning and changes in autism severity. Moreover, the authors found that there was an inverse relationship as hours increased in the intensively treated group. This study was non randomised and of fair quality and the possibility of bias cannot be excluded. Another good quality RCT (Smith et al. 2000), which reported more benefits for children treated intensively for an average of 25 hours per week compared to those having less intensive behavioural treatment (hours not reported), had a variable source of delivery of the intervention between the groups (clinic staff versus parents who had received some training). This means that either mode of delivery or intensity of treatment could be associated with the significant differences in intellectual functioning, visual spatial skills, language and academic achievement. The third study (Sallows and Graupner 2005), a good quality RCT, found no differences between a clinic directed group with intensive EIBI delivery (39 hours/week) and a slightly less intensive (32 hours/week) parent directed group (receiving relatively less supervision from clinic staff). Overall, there is insufficient evidence to determine the optimal intensity of behavioural therapy and whether mode of delivery (clinic directed vs parent directed) influences achievements. 

Future research should concentrate on both defining the specific features of behavioural treatment (intensity, duration, settings) that lead to optimal gains and determining the predictors of response so that the effectiveness of treatment can be enhanced through the targeting of individuals who will most benefit.

Behavioural communication interventions

The results from the studies focusing on communication outcomes varied. Three studies investigated the effects of PECS (Carr and Felce 2007; Howlin et al. 2007; Yoder and Stone 2006a; Yoder and Stone 2006b) on functional communication skills and one RCT of fair quality investigated the effects of a different type of communication intervention (Drew et al. 2002). Two of the PECS studies included older non verbal children with autism, mostly of primary (elementary) school age. These studies found a benefit for two sessions of PECS in these children but outcomes and results varied. A good quality RCT (Howlin et al. 2007) found a benefit for rates of initiation and usage of PECS (but not for frequency of speech, ADOS scores and language) and the other found a benefit for a range of communicative interactions between children with ASD and the teaching staff. Although some benefits were found for PECS in these studies, there was no evidence that these benefits were maintained after treatment ceased. The Howlin et al (2007) RCT was designed to measure maintenance of effects and found that an average of 10 months after treatment ceased, there were no differences between children having PECS and those in the control group. The other study (Carr and Felce 2007), of fair quality, was not designed to measure maintenance of effects and so it is unknown whether the benefits found were maintained after treatment. 

The third PECS study, a very good quality RCT, compared PECS with Responsive Education and Prelinguistic Milieu Teaching (RPMT) in preschool children (Yoder and Stone 2006a; Yoder and Stone 2006b). There was no significant difference between groups six months after treatment finished and results from both groups indicated strong growth on both measures of spoken communication. However, communication outcomes significantly favoured PECS immediately after treatment was finished suggesting that PECS had a more rapid effect on spoken communication than did RPMT. This finding can be further qualified: relative treatment efficacy varied by initial object exploration. The initial advantage of PECS for children who began treatment with high object exploration and the initial advantage of RPMT for children who began treatment with relatively low object exploration levels emphasises the importance of targeting and considering object play skills. For both treatments, effects were maintained after treatment ceased, in contrast to the other studies. The finding of relative effects has implications for the individualising of treatment for children to maximise communication outcomes. In summary, the included studies all found benefits for PECS in preschool and primary (elementary) school children with ASD but these effects were not maintained in all studies after treatment stopped. More research is needed to determine whether the benefits of PECS can be maintained, particularly when taught to primary (elementary) school children with ASD.

The other study used a different type of communication intervention, a psycho-linguistic and social pragmatic approach (Drew et al. 2002). This RCT found no benefits for the experimental intervention. 

Interventions for challenging behaviour

Behavioural interventions have also been used successfully to reduce challenging behaviour, with clear evidence of a potential overall benefit (ranging from 76% to 90%) in the secondary studies specifically measuring this outcome. However, there was little evidence that effects were generalised and maintained. Studies have generally not determined what participant, treatment or experimental characteristics identify the more successful behavioural approaches in reducing challenging behaviour in people with autism. High quality reviews of SCED studies have identified one feature as predictive of intervention success: the use of functional behaviour assessment (FBA) prior to the intervention, and particularly the more rigorous form of FBA known as experimental behaviour analysis. The findings of one recent review of SCED studies (Campbell 2003) suggested that functional communication training preceded by FBA reduces challenging behaviour and increases communication, though the generalisability of these studies to more natural environments is limited by their being conducted predominantly in clinic settings. The social validity of behavioural interventions for challenging behaviour needs to be evaluated.

Other behavioural approaches

There was insufficient evidence to recommend other behavioural approaches that could be distinguished from EIBI. One good quality RCT (Jocelyn et al. 1998) found that there was a benefit for children receiving the Autism Preschool Program in their language development but blind assessment of autistic symptoms did not find evidence of a difference in these other outcomes. The addition of a computer to a behavioural programme in an RCT (Moore and Calvert 2000) reported a benefit for the number of nouns learned and attention to task but the methodology was poor and the sample very small. 

The effects of interventions on communication, play and social skills were investigated by two other primary studies of fair quality (Kroeger et al. 2007; Wetherby and Woods 2006). Using a quasi-randomised, controlled design, direct teaching using a videomodelling format (compared to unstructured play) found benefits for social skills (Kroeger et al. 2007). The benefits of video modelling have been confirmed by three reviews of SCED studies in this domain (Ayres and Langone 2005; Bellini and Akullian 2007; Delano 2007) but concern was expressed that the critical features of the interventions were not clearly identified and thus specific recommendations could not be made. The three reviews (Bellini et al. 2007; Matson et al. 2007; McConnell 2002) that assessed the benefits of social skills interventions of various types found qualified support for these interventions and attempted to categorise the different types but there was little quantitative analysis. In the only review that provided quantitative analysis (Bellini et al. 2007), there was a low to questionable treatment effect, low to questionable generalisation effect and a moderate maintenance effect. These findings need confirmation in further research and the specific features responsible for the effectiveness of programmes need to be defined and standardised. The other primary study addressing social skills development (Wetherby and Woods 2006) was non-randomised and had significant flaws in design. It assessed the effects of ESI, a programme based on NRC recommendations in the UK compared to no treatment. The benefits found for ESI, social signals, rate of communicating, communicative functions and understanding, also need to be confirmed in trials with a more rigorous design. 

Summary & Conclusions

This systematic review considered the evidence for the effectiveness of interventions grounded in applied behavioural analysis for people with autism spectrum disorder. 

From 1517 articles identified by the search strategy, 43 publications were identified as eligible for appraisal and inclusion in the review. These were: 21 systematic reviews or evidence-based guidelines (10 reviews of either exclusively SRs and RCTs or broader inclusion criteria and 11 reviews of lower order evidence), and 20 primary studies (8 RCTs, one quasi randomised study, 3 cohort studies and 8 non­randomised experimental studies).

There was consistent evidence across a range of studies of reasonable quality and in different settings that behavioural approaches (predominantly grounded in ABA principles) can produce positive treatment outcomes in young (particularly pre­school) children with ASD. Further, of the few studies identified that compared ABA with eclectic/usual care approaches of similar intensity, the majority found greater benefit for the ABA group in terms of language skills, IQ and adaptive behaviour, although there was wide variation in individual responses. Whilst the evidence base is small, there is emerging evidence that behavioural approaches appear to hold the most promise for providing positive outcomes for children with ASD compared to eclectic programmes of similar intensity. This conclusion is consistent with several recent high quality systematic reviews which cautiously recommended the use of behavioural approaches for this population.

Identifying patterns in the study results was limited by the heterogeneity of the evidence base, reflected in varying intervention approaches, intensity of treatment, comparators, study settings, sample characteristics (particularly age of participants) and outcomes. As more than one variable often varied between comparators (such as treatment type, intensity and delivery), it was difficult to determine the precise cause of any group differences identified. 

The majority of primary studies evaluated variations on early intensive behavioural intervention programmes. Evidence from two of three studies which compared interventions of similarly high intensity suggests that EIBI can improve language skills, IQ and adaptive behaviour, although individual responses were highly variable. From two primary studies comparing programmes of varying intensity (keeping source of delivery constant), there was insufficient evidence to recommend the optimal number of intervention contact hours, although one study found that effectiveness diminished beyond 20 hours per week. Comparisons between interventions delivered by clinic staff and those of similar intensity directed by parents (with some specialist training or supervision provided) did not reveal differences in outcomes for participants with ASD.

Three studies investigating Picture Exchange Communication System found positive effects for preschoolers and primary school children with ASD, although these were not necessarily maintained beyond treatment. One study found differential results for two communication approaches, suggesting that treatment choice should take into account the abilities of participants.

There was generally consistent evidence from a primary group study and three reviews of SCED studies that video modelling can lead to positive impacts for children with ASD. Three secondary studies found qualified support for social skills interventions generally. There was insufficient evidence to determine the effects of a range of other behavioural intervention approaches evaluated in single studies. 

Four secondary studies provided evidence that behavioural interventions can reduce challenging or problem behaviour from between 76% to 90%, although it is not known whether these effects are maintained or generalised.

Whilst there was some variation in findings relating to functional behaviour analysis, two of the better quality systematic reviews found that FBA conducted prior to an intervention increases its effectiveness, with one review providing evidence that experimental functional analysis was the most effective FBA approach of those considered.

The current research base was limited by variability of results between studies and between individuals. Further controlled experimental research is needed to investigate the characteristics, intensity and duration of behavioural interventions that lead to the most positive impact on outcomes for people with ASD (and their families). Whilst the current review identified particular intervention programmes and strategies that can be effective, further work is needed to determine the critical features that are both necessary and sufficient components of an effective intervention. A notable limitation of the evidence base is the lack of studies comparing ABA-based interventions with high quality non ABA-based interventions also delivered by experts.

Additional research into the characteristics of the person with ASD which best predict response would also assist in directing behavioural interventions toward individuals who are most likely to benefit or to benefit more significantly. Research in this field considered by the current review has tended to focus on young children, and investigating the use of behavioural approaches for assisting older children and adults with ASD is a research gap deserving of further attention. 

Evidence identified in this review is broadly consistent with the following recommendations of the New Zealand ASD Guideline (Ministries of Health and Education 2008) that:

  • “decisions about the type of intervention and the degree of intensity should be informed by a skilled team and reflect the child’s developmental stage, characteristics, teaching goals and family preferences” (Recommendation 3.1.3);
  • “educational interventions should incorporate principles of positive behaviour support, particularly a focus on understanding the function of the child’s behaviour” (Recommendation;
  • “behaviour management techniques should be used to intervene with problem behaviours” (Recommendation 4.3.4);
  • “all behavioural interventions should be of good quality and incorporate . . . functional assessment” (Recommendation 4.3.5).

The current review’s findings should be considered in conjunction with those of an independent systematic review conducted in parallel for the Ministries that included SCED primary studies. It is hoped that the conclusions from these overlapping streams of evidence complement and enrich each other to provide a comprehensive account of the evidence base.

The findings from the current review extends and strengthens the ASD Guideline’s (Ministries of Health and Education 2008) recommendations relating to ABA. It identifies emerging evidence for superior benefits of behavioural approaches over eclectic/standard care approaches in education, treatment and managing problem behaviour for people with ASD in their early childhood. However this finding should be treated with caution as results were not always consistent, the number of studies where potential confounding factors were controlled were few, and responses between individuals were highly variable. Future research is needed to maximise treatment success and the direction of available resources by identifying the specific characteristics of behavioural treatment and the individuals receiving it that lead to best results.

Development and refinement of interventions for people on the autism spectrum as well as experimental investigations of treatment effectiveness are active areas of research interest. It is recommended that the findings of this report are reviewed and updated as relevant high quality evidence emerges.