THE EFFECTIVENESS OF APPLIED BEHAVIOUR ANALYSIS INTERVENTIONS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER - AN UPDATE OF SECONDARY LITERATURE Report to the New Zealand Ministry of Health 30 September 2009 UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 2 PRINCIPAL AUTHOR Marita Broadstock ACKNOWLEDGEMENTS Acknowledgements This systematic review was undertaken by staff of the New Zealand Guidelines Group, and completed in September 2009. The scope of the review was agreed with the Living Guidelines Group project’s sponsor, Joanna Curzon (Ministry of Health) and was funded by the Ministry of Health to contribute to the work of the Living Guideline Group. Marita Broadstock (Senior Researcher) conducted the review and prepared the report and Evidence Tables. Margaret Paterson (NZGG Information Specialist) conducted the search strategy and managed document retrieval and referencing. A list of excluded publications is available upon request. Anne Lethaby (Acting Manager, Research Services) provided methodological input and peer review. Please cite this report as: New Zealand Guidelines Group. The effectiveness of applied behaviour analysis interventions for people with autism spectrum disorder – an update of secondary literature. Wellington; 2009. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 3 SYNTHESIS OF RECENTLY PUBLISHED SECONDARY LITERATURE ON APPLIED BEHAVIOUR ANALYSIS BACKGROUND AND SCOPE In 2008, the New Zealand Ministries of Health, and Education, commissioned two technical reviews on interventions and strategies for people with Autism Spectrum Disorder (ASD) grounded in the principles of applied behaviour analysis (ABA) (Ministries of Health and Education, 2007). These two reviews were completed in parallel by Uniservices and New Zealand Guidelines Group (NZGG) respectively and based on research published to 31 December 2007. These reports represent the prime evidence for the Living Guideline Group (LGG) to consider in determining whether any revision of Recommendations relevant to ABA is needed in the New Zealand Autism Spectrum Disorder Guideline (2008). However, as this field has been the subject of significant research interest, it was considered valuable to supplement the New Zealand commissioned research with a synthesis of other high level evidence that has emerged since the reviews’ search cut-off period. In a hierarchy of evidence for studies of intervention effectiveness, the highest level evidence (Level I) consists of systematic reviews and meta analyses which include (but are not limited to) randomised controlled trials (RCTs) (level II evidence) (National Health and Medical Research Council, 2008). This current update is limited to level I evidence. Such secondary evidence offers the opportunity to consider a range of approaches to reviewing and synthesising the evidence relating to ABA and compare conclusions with those of the two comprehensive parallel technical reviews. OBJECTIVES AND RESEARCH QUESTION The objective of this update is to critically appraise secondary research evidence published since December 2007 relating to the effectiveness of interventions and strategies grounded in the principles of applied behaviour analysis for people with autism spectrum disorder. ABA-based interventions can be defined as ‘those in which the principles of learning theory are applied in a systematic and measurable manner to increase, reduce, maintain and/or generalise target behaviours’ (Ministries of Health and Education, 2007). Well-established principles and techniques of ABA include (a) reinforcement, (b) shaping, (c) chaining, (d) fading, (e) response and stimulus prompting, (f) discrimination training, (g) programming, and (h) functional assessment (Broadstock & Lethaby, 2008). The clinical question employed for this update of secondary literature was that used for the parallel technical reviews (Ministries of Health and Education, 2007): To what extent are interventions and strategies based on the principles of applied behaviour analysis effective in leading to the following outcomes for people with autism spectrum disorders: UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 4 social development and relating to others development of cognitive (thinking) skills development of functional and spontaneous communication which is used in natural environments engagement and flexibility in developmentally appropriate tasks and play and later engagement in vocational activities development of fine and gross motor skills prevention of challenging behaviours and substitution with more appropriate and conventional behaviours development of independent organisational skills and other behaviours generalisation of abilities across multiple natural environments outside the treatment setting maintenance of effects after conclusion of intervention improvement in behaviours considered non-core ASD behaviours, such as sleep disturbance, self mutilation, aggression, attention and concentration problems. REVIEW METHODS A systematic method of literature searching, selection and appraisal was employed in the preparation of this report, consistent with New Zealand Guidelines Group (NZGG) review processes (Broadstock & Lethaby, 2008). The search was limited to articles published in the English language between January 1 2008 and August 6 2009 (when the search was conducted). Sources included general bibliographic databases (Medline, PsychINFO, EMBASE, CINAHL, ERIC, Cochrane Library) as well as various health technology assessment/guideline databases, and cross-checking of references from retrieved references. Selection criteria were based on those used for the NZGG technical review (Broadstock & Lethaby, 2008), modified to solely identify on secondary studies. Studies were included if they: were secondary research (systematic reviews and meta-analyses) were published on or after January 1st 2008 had a clear review question used at least two searching sources and reported on studies: of eligible interventions (studies which evaluated interventions which were predominantly based on the principles of applied behaviour analysis) considering comparators including usual care, another intervention or application of interventions considering individuals with a diagnosis of autism spectrum disorder or where results are reported separately for this group. Research papers were excluded if they: UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 5 were non-systematic reviews, correspondence, editorials, expert opinion articles, comments, case reports, book chapters, articles published only in abstract form, conference proceedings, news items, unpublished work were primary studies (except where reported in included systematic reviews did not provide separate analyses/syntheses of results for eligible interventions and eligible population were not deemed appropriate to the research question or nature of review, including those reporting on outcomes solely relating to safety; the acceptability of, or ethical, economic or legal considerations relating to ABA; or the impact on persons other than those diagnosed with ASD. STUDY SELECTION AND APPRAISAL Selection criteria were applied by a single reviewer to abstracts/titles identified by the search strategy, and again to those retrieved as full text, to identify the final set of included papers for critical appraisal. All included studies represented level I evidence (National Health and Medical Research Council, 2008) as specified in the selection criteria, being systematic reviews or meta analyses of studies that included randomised controlled trials. Included studies were formally critically appraised using NZGG’s quality checklists which are based on the GATE Frame tools designed by Effective Practice, Informatics and Quality improvement (EPIQ), within the School of Population Health at the University of Auckland. Twenty questions addressed methodological quality, including whether the review’s methods were internally valid, precision of results, and the applicability/external validity of the review. An overall quality code was applied based on the appraisal to summarise the quality of each included study. Overall study quality was categorised qualitatively based on summary questions on the GATE checklist and using one of the following descriptors: “very good”, “good”, or “fair”. Details of each study were entered into Evidence Tables, including aspects of methodology, results, authors’ conclusions, reviewer’s additional comments, and the summary study quality descriptor. The included studies are presented alphabetically by first author within two intervention type categories, (1) applied behaviour analysis interventions (defined broadly), and (2) Picture Exchange Communication System (PECS) interventions (see Evidence Tables). DATA SYNTHESIS Studies were narratively synthesised to determine the strength of evidence. Strength of evidence is determined by three domains (West, King, & Carey, 2002): - quality (the extent to which bias was minimised); - quantity (magnitude of effect, numbers of studies, sample size or power); - consistency (the extent to which similar findings are reported). UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 6 KEY FINDINGS AND CONCLUSIONS Overview Of 461 articles identified by the search strategy, 35 publications were retrieved as full text. Twelve of these met selection criteria and were deemed eligible for critical appraisal and inclusion in this synthesis. Of the 12 included secondary studies, 10 reported on evaluations of interventions broadly based on applied behaviour analysis, and two reported on evaluations of the Picture Exchange Communication System (PECS) intervention specifically (see Table). The 10 reviews reporting on ABA-based interventions (in general) used varying descriptive terms for included interventions, reflecting that there is no universally accepted definition of ABA-based interventions. For example, some studies refered specifically to Early Intensive Behavioural Interventions (EIBI) as being restricted to interventions based on the work of Lovaas (1987), whereas others included as EIBI any intensive behavioural intervention directed at children in pre-school years. In this report the broader definition is used. Of the 10 reviews, four included meta analyses (Sigmund Eldevik et al., 2009; Ospina et al., 2008; Reichow & Wolery, 2009; Spreckley & Boyd, 2009b); all four were rated as being of good (Reichow & Wolery, 2009) or very good quality (Sigmund Eldevik et al., 2009; Ospina et al., 2008; Spreckley & Boyd, 2009b). Four of the 12 reviews (33%) were classified as being of “very good” quality (Blue Cross and Blue Shield Association, 2009; Sigmund Eldevik et al., 2009; Ospina et al., 2008; Spreckley & Boyd, 2009b), another third were of “good” quality (Svein Eikeseth, 2009; Howlin, Magiati, & Charman, 2009; Reichow & Wolery, 2009; Rogers & Vismara, 2008), and the remaining final four systematic reviews were of “fair” quality, including two considering ABA in general (Case-Smith & Arbesman, 2008; Seida et al., 2009), and two reporting on PECS (Ostryn, Wolfe, & Rusch, 2008; Sulzer-Azaroff, Hoffman, Horton, Bondy, & Frost, 2009). It should be noted that as these reviews were all published across an 18 month period on the same broad topic area, they tend to report on the same studies, and also overlap with the studies included in the parallel technical reviews of NZGG and Auckland Uniservices. For this reason it is important to be aware that the results and the studies appraised should not be summated as independent sources of evidence as this would misrepresent the “quantity” of studies and give the individual studies undue weight. No additional primary studies published post December 2007 and eligible for inclusion in either of the two technical reviews were identified by the 12 reviews, and so the evidence “catchment area” is broadly the same as that considered by the parallel technical reviews, although some included studies prior to 1998, the earliest date for publications considered in the parallel technical reviews. In considering secondary reviews which consider overlapping but different selection criteria and methodological approaches, attention should be given to the consistency of UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 7 conclusions. More weight/emphasis should be placed on the reviews that are of higher quality. For this reason, the following synthesis will consider the findings and conclusions of the “very good” quality reviews first. Reviews of “very good” quality Four reviews were rated as being of “very good” quality using the systematic critical appraisal process outlined above. All considered ABA interventions in general, though the focus was on studies of early intensive behavioural analysis. The systematic review published by Blue Cross and Blue Shield Association’s (2009) Technical Evaluation Centre considered evidence published 1987 - July 2008. The review considered the use of EIBI based on applied behaviour analysis. It excluded single-case experimental studies (SCED), arguing that such designs lack generalisability across individuals, and are of limited value in evaluating EIBI due to methodological constraints. The reviewers included 16 studies, including 2 RCTs. They concluded that, “weaknesses in research design, differences in the treatments and outcomes compared, and inconsistent results, mean that the impact of EIBI versus other treatments on outcomes for children with autism cannot be determined.” They argued that the weakness of the evidence base and variability in findings meant that they could also not reach conclusions about the impact of EIBI on any specific domains/outcomes, or on whether greater intensity of EIBI can lead to better outcomes. The authors called for research to identify what aspects of an intervention and its delivery lead to greatest positive effect, including content, technique, intensity, starting and ending age. A greater emphasis was said to be needed on randomized, controlled trials with substantially larger sample sizes, uniformity of outcomes and instruments, and consistent treatments. The systematic review by Eldevik and Hastings (2009) considered nine controlled group studies of EIBI in their meta analyses. They reported moderate to large effect sizes for adaptive behaviour composite (ABC) and full-scale IQ scores, respectively, compared with controls or eclectic comparators. It is concluded that, “in the absence of other interventions with established efficacy, EIBI should be an “intervention of choice for children with autism.” However the authors cautioned that the improved outcomes for people receiving EIBI may be due to the EIBI group having received, in general, more frequent and more total hours of supervision and training than comparison groups. They noted that this remains a threat to the validity of their conclusions and called for RCTs where the comparison intervention is of similar intensity and where staff receive similar training and supervision. Another SR with meta analysis rated as being very good quality was authored by Ospina et al (2008). It considered literature published to May 2007 and included 101 controlled studies across a range of interventions ranging from behavioural to developmental. Those interventions relating to ABA specifically included those based on Direct Trial Training (DTT) and Lovaas (1987) therapy. With respect to ABA, the UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 8 reviewers concluded from descriptive synthesis that there was “some support” for discrete trial training in terms of motor and functional skills but not for communication skills. From meta analyses of 3 controlled clinical trials the authors reported that Lovaas’ therapy appeared to show benefits when compared to special education on several domains (adaptive behaviour, communication and interaction, comprehensive language, daily living skills, expressive language, overall intellectual functioning, socialisation) with the exception of non-verbal intellectual functioning. Based on a meta-analysis of two RCTs, they found no difference in communication skill outcomes for children receiving EIBI compared with a Developmental Individual Difference Relationship based intervention. Lower level evidence from meta-analysis of two retrospective cohort studies showed greater effects for “high intensity” versus “low intensity” Lovaas. These findings were tempered by the limitations of the evidence base, with the authors noting that, “these findings are based on pooling of a few, methodologically weak studies with few participants and relatively short-term follow-up”. The heterogeneity of the study populations, interventions and outcomes was also noted, raising questions as to the appropriate use of meta analysis to synthesise these results. The overall conclusion from across interventions was that there is no clear answer regarding the most effective therapy to improve symptoms associated with ASD. The authors called for replication in RCTs to substantiate the use of Lovaas (1987) and to assess the effect of treatment intensity on outcomes of children with ASD. They further recommended that clinical management be guided by individual needs and availability of resources. The fourth review of very good quality also included a meta analysis (Spreckley & Boyd, 2009b), and reported on applied behaviour interventions (ABI) for preschool children with ASD. Randomised or quasi-randomised trials were included, and of the 13 studies meeting initial selection criteria, only six met minimum quality criteria and four had adequate data to be included in the formal meta analysis. Results suggested no significant improvement on outcomes for those in ABI programs compared with standard care on cognitive outcomes, expressive language, receptive language, or adaptive behaviour. Low to moderate heterogeneity existed across outcomes which prompts uncertainty about the suitability of meta analytic techniques here, a limitation the authors acknowledged given high variability in participant characteristics and programme content. Other limitations of the evidence base disussed by the authors included difficulty in establishing genuine control groups, lack of strict selection criteria, and limited information on participant retention. The authors concluded that there is inadequate evidence that applied behaviour intervention has better outcomes than standard care for children with autism. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Summary These high quality reviews offer some agreement and some divergence in their interpretations of the evidence base. Two reviews, which included meta analyses, gave tentative evidence of positive benefits of ABA-based interventions over comparison treatments from a limited evidence base. EIBI was reported as having beneficial effects on adaptive behaviour and IQ compared with control or eclectic comparators (Sigmund Eldevik et al., 2009). Some support was reported for discrete trial training in terms of motor and functional skills but not for communication skills. Lovaas’ (1987) therapy was presented as having benefit over no treatment, and greater effect for higher intensity treatment (Ospina et al., 2008). However these conclusions are qualified by methodological weaknesses of the individual studies and the small number of studies and participants. In particular, the lack of control for the intensity of treatment delivered between intervention and comparator is problematic. In contrast, the two other reviews suggested there was inadequate evidence of effectiveness of ABA-based early interventions. These conclusions were based on a lack of significant results in the meta analysis of four studies by Spreckley and Boyd (2009b). The Blue Cross and Blue Shield Association (2009) report argued that the limitations and inconsistent results of the evidence base precluded making any conclusions about the relative effectiveness of EIBI compared with alternative treatments, about its effectiveness for particular outcome domains, or about the impact of intensity on outcomes. The report also argued that study heterogeneity ruled out the use of meta analysis. All four reviews emphasised the need for additional research addressing methodological weaknesses, and suggested the need for appropriately powered RCTs to systematically investigate what characteristics of treatment lead to the best outcomes. Particularly called for was the use of comparison interventions of similar intensity and quality to those of EIBI. Reviews of “good” quality Four reviews were rated as being of “good” quality, having more flaws and limitations than the “very good” reviews discussed above, often in terms of a limited search strategy, lack of systematic quality assessment, and/or lack of detail and precision in reporting study characteristics and synthesising results. However these studies are still of reasonably good quality and of value in contributing to the evidence base. A systematic review of comprehensive psycho-educational research (Eikeseth, 2009) included 20 studies on ABA-based interventions. The reviewers concluded that intensive ABA treatment carried out by trained therapists is effective in enhancing “global functioning” in pre-school children with autism, and those with PDD-NOS. In Howlin et al’s (2009) review of 11 controlled (any comparator) group EIBI studies, in 9 of 10 studies, IQ scores were significantly higher for children at follow-up in the EIBI group compared with those in the control/comparator group, with moderate to large effect sizes in the majority, but not all studies. However the authors also noted that children receiving EIBI received significantly more (mean) hours of intervention UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 10 per week than those in the comparison group (30 c.f. 19 hours, respectively). It was concluded that whilst EIBI is highly effective for some children, there is wide individual variation in response, and some failed to make progress at all. Howlin et al (2009) also noted that chronological age and length of intervention did not appear to be related to outcome, with varying results for initial language level as a predictor. Initial IQ strongly predicted improvement after EIBI treatment in four studies, but was unassociated with outcome in a fifth. A systematic review and meta analysis by Reichow and Wolery (2009) evaluated 13 controlled group studies on EIBI interventions based on the Lovaas (1987) model (notably, Eldevik and Hastings, 2009, criticised this as too restrictive a definition for a review of EIBI’s effectiveness). Meta analyses of 12 studies considering change in IQ within treatment group suggested a moderately large effect size for improvement at followup, however the authors cautioned that this finding was limited by excluding reference to a comparison group. As such, improvements cannot be attributed to EIBI exclusively; for example, they could relate to maturation effects. As there was significant variability between study characteristics, the appropriateness of using meta analytic techniques for synthesising results is questionable. Descriptive and summative syntheses suggested that EIBI can be an effective treatment, on average, for some children with autism compared with control/various comparators. However individual data typically were not presented and the authors noted that the intervention has not worked for all children, and gains in one domain may not be accompanied by gains in another domain within individuals. Moreover, comparison groups were of limited quality and poorly described. The authors concluded that it is not possible to determine whether EIBI is more or less effective than other treatment options without comparisons between EIBI and empirically validated treatment programs. The systematic review by Rogers and Vismara (2008) considered early interventions including psychopharmacological ones for young children with autism, reporting on seven ABA based interventions. From four controlled studies evaluating Lovaas’s (1987) treatment, the authors argued that this intervention meets criteria for a ‘‘well-established’’ psychosocial intervention for improving the intellectual performance (specifically IQ scores) of young children with ASD. “Well established” interventions were defined as those requiring treatment manuals, clearly specified participant groups, and either (a) two independent well-designed group studies showing the treatment to be better than placebo or alternative treatment, or equivalent to an established effective treatment, or (b) nine or more single-subject design studies using strong designs and comparison to an alternative treatment (Chambless et al., 1998). There was less consistency in the data for improvements in behavior, adaptive skills, and language skills. Overall, the review concluded that focused daily early intervention programs of several different kinds are beneficial for young children with autism, however the lack of comparative studies precludes determining which comprehensive treatment approach is best. Further, the authors suggested that other well-known interventions may be as or more efficacious as Lovaas’s (1987) model but they have not been rigorously evaluated. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Summary The review by Eikeseth (2009) concluded that intensive ABA treatment carried out by trained therapists is effective in enhancing global functioning in pre-school children with autism, and those with PDD-NOS. Rogers and Vismara (2008) suggested that Lovaas-based (1987) EIBI met criteria for a ‘‘well-established’’ intervention for improving the IQ of young children with ASD, but that there was less consistency for improvements in behavior, adaptive skills, and language. Howlin et al (2009) and Reichow and Wolery (2009) both concluded that whilst EIBI is highly effective in increasing IQ for some children, some appear to make no progress, suggesting wide individual variation in response. The former reviewers also found that age and length of intervention do not appear to predict response, but that initial language level did in some studies, and initial IQ appears to be related to impact on outcome in most studies evaluated. Despite these broadly positive conclusions about the potential for benefit among (at least some) children with ASD, the reviewers tempered their findings with reference to the methodological limitations of the evidence base. Major concerns centred around the need to disentangle the effects of programme content from those of programme intensity, which has not been well controlled for. To address this, Howlin et al (2009) called for equivalence trials, where high quality interventions are compared witho other high quality interventions of a similar intensity. Other calls for further research echoed those of the “very good” studies reported earlier, including the need to systematically assess characteristics of children for whom EIBI is most effective, optimal duration and intensity of treatment, and age of commencement of treatment, so that better targetted and cost-effective interventions could be developed (Howlin et al., 2009). Another repeated theme concerned the limitations of the comparator or control groups with which ABA based interventions are compared. These included their lack of standardization, being poorly defined, lack of measures of procedural fidelity, and lack of data on any supplemental treatments participants may also receive. The apparent superiority of EIBI may be an artifact of the lack of rigorous empirical evaluation of alternative interventions of good quality. As fully randomised controlled studies in this area were extremely rare, the nonrandom assignment of participants to treatment groups was also said to severely limit conclusions about treatment effectiveness. This is because factors affecting group allocation (such as parental chioce) may have contributed to improved outcomes (such as increased parental motivation and support), thus confounding any effects attributed to the treatment. Only one review (Rogers & Vismara, 2008) made specific reference to cultural factors, observing that treatment programmes for children with autism have been developed for and evaluated primarily with children from European American backgrounds, and generalization of effects across ethnic groups from this research is premature. It was suggested that cultural issues may moderate the effects of autism intervention programs. Reviews of “fair” quality Four reviews were rated as being of fair quality, two considering ABA interventions broadly defined, and the only two reviews considering a specific ABA based intervention, both relating the the Picture Exchange Communication System. These intervention categories will be considered separately, as they are in the Evidence Tables. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Applied Behaviour Analysis interventions in general Two reviews on ABA interventions were classified as being of “fair” quality due to flaws in methodology and reporting such that results are likely to be susceptible to bias. It should be noted that some of these studies had supplementary or alternative goals to providing comprehensive systematic reviews of the literature. Whilst the narrative critiques of the literature in these reviews may be of interest, conclusions about the evidence base from these studies may not be reliable. The review by Case-Smith and Arbesman (2008) reviewed ASD interventions relevant to occupational therapy. The authors concluded that behavioural interventions are successful in training children in basic academic and life skills, but noted that generalisation of skills to natural environments was unproven. They also commented on the lack of studies into the promotion of effective transition to work and independent living. Discussing barriers to widespread application of behavioural interventions, the reviewers pointed to the time and resources required to implement a 30- to 40-hour/week intervention as well as the lack of definition of the most appropriate candidates. Seida et al (2009) conducted an “umbrella review” of systematic reviews of psychosocial interventions for people with ASD. Broadly favourable outcomes from behavioural interventions versus no treatment were reported. Critiquing the literature, the authors noted the absence of data on the relative effectiveness of treatment options, and the lack of research attention to factors in choice of treatment including cost, convenience and family burdens. Picture Exchange Communication System Two systematic reviews of fair quality specifically considered interventions based on Picture Exchange Communication System (PECS). The review by Ostryn et al (2008) considered 15 PECS studies concluded that whilst PECS is widely implemented it lacked a strong empirical base. Criticisms of the literature included lack of reporting of statistical significance and precision of results, lack of scope and definition of key outcomes, and lack of data on maintenance. The authors recommended that PECS was best used as an initial intervention to teach manding as part of a multimodal system but is not recommended as a long term intervention. The other PECS review by Sulzer-Azaroff et al (2009) concluded from 34 studies that professionals and parents can teach individuals to successfully initiate exchanges of pictures for tangible and non tangible reinforcers. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER REFERENCES The references for this report and the appended Evidence Tables are presented below. Publications included for critical appraisal in this review are indicated by ** prefacing the citation. Bassett, K., Green, C. J., & Kazanjian, A. (2000). Autism and Lovaas treatment: A systematic review of effectiveness evidence. Retrieved 17 June 2008, from http://www.chspr.ubc.ca/node/351 ** Blue Cross and Blue Shield Association. (2009). Special report: early intensive behavioral intervention based on applied behavior analysis among children with autism spectrum disorders. Technology Evaluation Centre Assessment Program, 23(9), 1-61. Broadstock, M., & Lethaby, A. (2008). The Effectiveness of Applied Behavioural Analysis interventions for people with autism spectrum disorder, Systematic Review. Wellington, New Zealand: New Zealand Guidelines Group. Carr, D., & Felce, J. (2007). Brief report: Increase in production of spoken words in some children with autism after PECS teaching to phase III. Journal of Autism and Developmental Disorders, 37(4), 780-787. ** Case-Smith, J., & Arbesman, M. (2008). Evidence-based review of interventions for autism used in or of relevance to occupational therapy. American Journal of Occupational Therapy, 62(4), 416-429. Chambless, D. L., Baker, M. J., Baucom, D. H., Beutler, L. E., Calhoun, K. S., Crits-Christoph, P., et al. (1998). Update on empirically validated therapies, II. The Clinical Psychologist, 51(1), 3-16. Cohen, H., Amerine-Dickens, M., & Smith, T. (2006). Early intensive behavioral treatment: Replication of the UCLA model in a community setting. Journal of Developmental and Behavioral Pediatrics, 27(2 SUPPL. 2), S145-S155. Drew, A., Baird, G., Baron-Cohen, S., Cox, A., Slonims, V., Wheelwright, S., et al. (2002). A pilot randomised control trial of a parent training intervention for pre-school children with autism: Preliminary findings and methodological challenges. European Child and Adolescent Psychiatry, 11(6), 266-272. Eikeseth, S. (2009). Outcome of comprehensive psycho-educational interventions for young children with autism. Research in Developmental Disabilities, 30(1), 158-178. Eikeseth, S., Smith, T., Jahr, E., & Eldevik, S. (2002). Intensive behavioral treatment at school for 4- to 7-year-old children with autism: A 1-year comparison controlled study. Behavior Modification, 26(1), 49-68. Eikeseth, S., Smith, T., Jahr, E., & Eldevik, S. (2007). Outcome for children with autism who began intensive behavioral treatment between ages 4 and 7: A comparison controlled study. Behavior Modification, 31(3), 264-278. Eldevik, S., Eikeseth, S., Jahr, E., & Smith, T. (2006). Effects of low-intensity behavioral treatment for children with autism and mental retardation. Journal of Autism and Developmental Disorders, 36(2), 211-224. ** Eldevik, S., Hastings, R. P., Hughes, J. C., Jahr, E., Eikeseth, S., & Cross, S. (2009). Metaanalysis of early intensive behavioral Intervention for children with autism. Journal of Clinical Child & Adolescent Psychology, 38(3), 439-450. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 14 Horner, R. H., Carr, E. G., Strain, P. S., Todd, A. W., & Reed, H. K. (2002). Problem behavior interventions for young children with autism: A research synthesis. Journal of Autism and Developmental Disorders, 32(5), 423-446. Howard, J. S., Sparkman, C. R., Cohen, H. G., Green, G., & Stanislaw, H. (2005). A comparison of intensive behavior analytic and eclectic treatments for young children with autism. Research in Developmental Disabilities, 26(4), 359-383. Howlin, P., Gordon, R., Pasco, G., Wade, A., & Charman, T. (2007). The effectiveness of Picture Exchange Communication System (PECS) training for teachers of children with autism: A pragmatic, group randomised controlled trial. Journal of Child Psychology and Psychiatry, 48(5), 473-481. ** Howlin, P., Magiati, I., & Charman, T. (2009). Systematic review of early intensive behavioral interventions for children with autism. American Journal on Intellectual & Developmental Disabilities, 114(1), 23-41. Jocelyn, L. J., Casiro, O. G., Beattie, D., Bow, J., & Kneisz, J. (1998). Treatment of children with autism: A randomized controlled trial to evaluate a caregiver-based intervention program in community day-care centers. Journal of Developmental & Behavioral Pediatrics, 19(5), 326-334. Lovaas, O. I. (1987). Behavioral treatment and normal educational and intellectual functioning in young autistic children. Journal of Consulting and Clinical Psychology, 55(1), 3-9. Magiati, I., Charman, T., & Howlin, P. (2007). A two-year prospective follow-up study of community-based early intensive behavioural intervention and specialist nursery provision for children with autism spectrum disorders. Journal of Child Psychology and Psychiatry and Allied Disciplines, 48(8), 803-813. Ministries of Health and Education. (2007). Request for tender (RFT). Review of published research on applied behaviour analysis (ABA) interventions for people with autism spectrum disorder (ASD). Wellington, New Zealand. Ministries of Health and Education. (2008). New Zealand autism spectrum disorder guideline. Wellington. Morris, S. B. (2000). Distribution of the standardized mean change effect size for meta-analysis on repeated measures. Br J Math Stat Psychol, 53 ( Pt 1), 17-29. Nathan, P. E., & Gorman, J. M. (2002). A guide to treatments that work (2nd ed.). New York, NY: Oxford University Press. National Health and Medical Research Council. (2008). NHMRC additional levels of evidence and grades for recommendations for developers of guidelines. Pilot program 2005-2007. NHMRC. ** Ospina, M. B., Krebs Seida, J., Clark, B., Karkhaneh, M., Hartling, L., Tjosvold, L., et al. (2008). Behavioural and developmental interventions for autism spectrum disorder: a clinical systematic review. PLoS ONE [Electronic Resource], 3(11), e3755. ** Ostryn, C., Wolfe, P. S., & Rusch, F. R. (2008). A review and analysis of the Picture Exchange Communication System (PECS) for individuals with autism spectrum disorders using a paradigm of communication competence. Research & Practice for Persons with Severe Disabilities, 33(1/2), 13-24. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 15 ** Reichow, B., & Wolery, M. (2009). Comprehensive synthesis of early intensive behavioral interventions for young children with autism based on the UCLA young autism project model. Journal of Autism & Developmental Disorders, 39(1), 23-41. Remington, B., Hastings, R. P., Kovshoff, H., degli Espinosa, F., Jahr, E., Brown, T., et al. (2007). Early intensive behavioral intervention: Outcomes for children with autism and their parents after two years. American Journal on Mental Retardation, 112(6), 418-438. ** Rogers, S. J., & Vismara, L. A. (2008). Evidence-based comprehensive treatments for early autism. Journal of Clinical Child & Adolescent Psychology, 37(1), 8-38. Sallows, G. O., & Graupner, T. D. (2005). Intensive behavioral treatment for children with autism: Four-year outcome and predictors. American Journal on Mental Retardation, 110(6), 417438+497. ** Seida, J. K., Ospina, M. B., Karkhaneh, M., Hartling, L., Smith, V., & Clark, B. (2009). Systematic reviews of psychosocial interventions for autism: an umbrella review. Developmental Medicine & Child Neurology, 51(2), 95-104. ** Smith, T., Eikeseth, S., Sallows, G. O., & Graupner, T. D. (2009). Efficacy of applied behavior analysis in autism. The Journal of Pediatrics, 155(1), 151-152. Smith, T., Groen, A. D., & Wynn, J. W. (2000). Randomized trial of intensive early intervention for children with pervasive developmental disorder. American Journal of Mental Retardation, 105(4), 269-285. Spreckley, M., & Boyd, R. (2009a). Efficacy of applied behavior analysis in autism. Reply. Journal of Pediatrics, 155(1), 152-153. ** Spreckley, M., & Boyd, R. (2009b). Efficacy of applied behavioral intervention in preschool children with autism for improving cognitive, language, and adaptive behavior: a systematic review and meta-analysis. Journal of Pediatrics, 154(3), 338-344. ** Sulzer-Azaroff, B., Hoffman, A. O., Horton, C. B., Bondy, A., & Frost, L. (2009). The Picture Exchange Communication System (PECS): what do the data say? Focus on Autism & Other Developmental Disabilities, 24(2), 89-103. West, S., King, V., & Carey, T. S. (2002). Systems to rate the strength of scientific evidence. (Vol. Evidence Report/Technology Assessment): Research Triangle Institute-University of North Carolina Evidence-based Practice Center under Contract No. 290-97-0011. AHRQ Publication No. 02-E016. . Yoder, P., & Stone, W. L. (2006). A randomized comparison of the effect of two prelinguistic communication interventions on the acquisition of spoken communication in preschoolers with ASD. Journal of Speech, Language and Hearing Research, 49(4), 698-711. Yoder, P., & Stone, W. L. (2006). Randomized comparison of two communication interventions for preschoolers with autism spectrum disorders. Journal of Consulting and Clinical Psychology, 74(3), 426-435. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 16 EVIDENCE TABLES Evidence tables for included Level I secondary studies assessing the effectiveness of applied behaviour analysis (ABA) for people with autism spectrum disorder. 1. Applied Behaviour Analysis interventions in general Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Blue Cross Blue Shield (2009) Technology Evaluation Centre Evidence level: I Inclusion criteria: reported on the use of EIBI compared to another treatment strategy; attempted to identify features of EIBI that had the most impact on its effectiveness; or sought to identify children most likely to benefit from EIBI. 16 studies met selection criteria, including 2 RCTS, 9 nonrandomized, comparative studies, and 5 single-arm studies. No studies were found that included children with Asperger’s disorder; 4 studies explicitly included children with PDD or PDD-NOS. US SR Aim: To conduct a systematic review of the research literature on the use of early intensive behavioural interventions based on applied behavioural analysis (EIBI) among young children with ASD. Asked: (1) How effective is EIBI in improving the functioning of children with autism spectrum disorders, and how does it compare to other early intervention approaches? (2) Can patient characteristics be identified that predict better outcomes from EIBI? (3) Does the effect of EIBI vary with the intensity of treatment? Search period: From 1966 (varied, see below) – July 2008. A more narrowly defined update was performed in January 2009 using Medline but identified no additional eligible studies. Exclusion criteria: sample size <10, including single-subject studies; interventions were very poorly described; interventions were not comprehensive, where a number of domains affected by ASD were addressed; intervention within a treatment group was heterogeneous, combining a variety of methods; experimental intervention was not intensive (< 20 hrs/wk); study did not directly measure outcomes through a direct assessment of the child’s achievement; The strongest evidence was provided by two randomized, controlled trials (Smith et al. 2000; Sallows and Graupner 2005) and they compared different interventions, had small sample sizes, and came to different conclusions. 3 of 4 studies examining the impact of pretreatment cognitive functioning found that it significantly predicted outcomes, while one (a randomized, controlled trial) did not. Some studies suggested that younger age at the start of therapy is a predictor of better outcomes (e.g., Howard et al. 2005), while others found no difference based on initial age (e.g., Magiati et al. 2007). The findings on whether more intense treatment leads to better outcomes were inconsistent, and no conclusions were drawn. Authors’ conclusions: Unfortunately, only two randomized, controlled trials have been conducted. Weaknesses in research design, differences in the treatments and outcomes compared, and inconsistent results mean that the impact of EIBI versus other treatments on outcomes for children with autism cannot be determined. The body of evidence overall is too weak to reach conclusions regarding any of the domains/outcome areas, and too variable to assess whether greater intensity of EIBI can lead to better outcomes. The heterogeneity of the interventions used in studies on EIBI and the significant methodological weaknesses preclude the use of meta-analysis and of reaching any general conclusions about the effectiveness of EIBI among children with ASD. About half of the studies followed children for approximately 2 years or less, and some for only 1 year. This is not sufficient follow-up time to assess the potential impact of an intervention over a lifetime. Research is needed to identify those characteristics of treatment - content, UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score Clearly defined methodology though only two databases considered Study quality coded and rated, apparently by single reviewer. Justification given for excluding SCED studies including lack of generalisability across individuals, inability to use blinding, variability of the ASD condition over time, and potential for carryover effect between intervention and comparator (given that interventions are intended to have long lasting effects) Comprehensive Tables including details of study characteristics, outcomes, results, and aspects of quality. Thorough critical discussion of included studies and of recent systematic reviews Discussion of the methodological limitations of studies Suggestions made for future research 17 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Note that selection criteria then excluded studies published <1987. Search strategy: Medline (from 1966) PsycINFO (from 1970) (search terms provided). References of retrieved papers were cross checked to identify additional papers. study was published before 1987, when the seminal article on EIBI by Lovaas was released. Study quality was appraised and graded (as good, fair or poor) using detailed criteria and formal grading systems. Results and authors’ conclusions technique, intensity, starting and ending age, etc. - that maximize its effectiveness. (Need) a greater emphasis on randomized, controlled trials; substantially larger sample sizes; uniformity of outcomes and instruments; and consistent treatments that do not vary widely within the experimental or control group. The cost of continuing the current course of assuming that EIBI works may not be obvious. EIBI is costly financially for society and requires a large time commitment from children, their families, and their teachers or therapists. However, these programs may not appear to pose any harm for the children themselves. Nevertheless, the opportunity costs could be high, indeed, of providing suboptimal care to these children, simply because we as a society do not know what works best. The children may be treated with an intervention that is not as effective as the alternatives. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score Overall quality: Very good 18 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Case-Smith & Arbesman (2008) Evidence level: I Inclusion criteria: Provided evidence for an intervention approach used with children or adolescents with ASD; Peer reviewed; Addressed a performance area or intervention approach within the domain of occupational therapy; RCTs, non randomized clinical trials, and beforeand-after, one group designs. 49 studies met selection criteria. Six categories of research topics were identified, including intensive behavioral intervention, which is reported on here. USA SR Aim: To identify, evaluate, and synthesize the research literature on interventions for ASD of relevance to occupational therapy. Search period: 1987 –2007 Search strategy: CINAHL Cochrane SRs database ERIC Medline PsychINFO Social Science Abstracts Sociological Abstracts RehabData Latin American and Caribbean Health Sciences Literature EBSCOHost (search terms not provided). Exclusion criteria: Case series, single subject design, case reports and expert opinion, narrative reviews, consensus statements; Used qualitative methods to the exclusion of quantitative methods; Had serious design limitations. EIBI using one-on-one discrete trial training is widely applied to children with ASD and has evidence of moderate to strong effects (Lovaas, 1987; Sallows & Graupner, 2005; Smith, Groen, & Wynn, 2000). Positive behavioral support has moderate to strong positive effects in reducing problem behaviors in children with ASD. A systematic approach that includes prevention of problem behaviors by applying consistent instruction and consequences to behavior, modifying the environment to promote appropriate behaviors, and collecting data to monitor children’s progress appears to be highly effective (Horner et al., 2002). Authors’ conclusions: Although behavioral interventions are successful in training children in basic academic and life skills concepts, it is not known how well these skills transfer into the natural environment, and the findings are mixed regarding changes in behavior (Smith et al., 2000). Barriers to widespread application are as follows: (1) time and resources required to implement a 30- to 40hr/wk intervention and (2) definition of the most appropriate candidates (Bassett, Green, & Kazanjian, 2000). When children exhibit problem behaviors, functional analysis is essential to determine the basis for the behavior. “Level I” studies of interventions to promote adolescents’ and young adults’ success in work and independent living were virtually absent from the research literature. Single reviewer extracted data. Used a hierarchy of evidence based on study design to rank studies. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Search strategy restricted to databases, no search terms given. No Tables Narrative discussion of results with minimal methodological critique and minimal detail of quantitative results. Framed selection and discussion of studies in terms of relevance to practice of occupational therapists. Overall quality: Fair 19 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Howlin et al. (2009) Evidence level: I Inclusion criteria: Case-control comparison study of EIBI for children with autism; UCLA affiliated (home or clinic based) or UCLAmodeled (homebased) EIBI program; At least 10 people in EIBI group; age < 6 years at start of treatment; intervention at least 12 hrs/wk; intervention for 12 mths or longer adequate data on IQ or standard outcome measures. 11 studies (reported in 13 reports) met selection criteria, including one RCT (Smith et al, 2000). UK SR Aim: To examine the findings from controlled early intensive behavioural intervention (EIBI) studies published in peer reviewed journals. Search period: 1985 – May 2007 Search strategy: CINAHL Cochrane library ERIC Medline PsychINFO Embase (search terms provided). Searched websites of several major health, autism and research organisations. References of review papers were cross checked to identify additional papers. Exclusion criteria: SCED or case series study without comparison group, studies focused on a specific behaviour, use of pharmacological or non ABA interventions, outcomes for/views of therapists or parents. Summaries/analyses based on comparisons of the published group means. Intervention intensity The mean intervention period for the 9 studies which reported intensity clearly was 27.4 months (SD=10.7, range 14-48 mths). Intensity (number of hours of intervention per week) varied greatly by individual within study. Estimated that EIBI children received significantly more hours of intervention per week than controls (EIBI Mean=29.8 hrs/wk cf Control Mean=19.1 hrs/wk, p=.007). Impact of intervention on IQ IQ did not vary at baseline between treatment groups, but at final follow up was significantly higher for EIBI group compared to the Control group in 9 of 11 studies, with estimated effect sizes moderate to large in the majority of studies, but not all studies. Variability in outcome There was considerable variation between individuals, with a minority achieving marked improvement and even educational independence, whilst the majority had less dramatic improvements. Some failed to make progress at all. Predictors of outcome Length of intervention did not appear to be related to outcome, and most impact on IQ occurred in the first year of the intervention after which the effect tended to plateau. Initial (baseline) IQ strongly predicted improvement in 4 studies, but was unassociated in another study that investigated this. Chronological age did not predict outcome (though age range tended to be limited to 3-5.5 yrs) which does not support suggestions that EIBI must start at 3 years. Mixed results were found for initial language level which was associated with outcome in 4 of 7 studies investigating this. Methodological limitations Comparisons were made difficult by varying time lag of follow-up after ending of intervention, and use of different instruments and scoring reported (raw, standard, age equivalents). Few researchers have assessed outcome in terms of behavioural difficulties/severity of autism, or impact on family life. Few studies reported on quality of alternative therapies in the Control arm, which tended not to be autism specific. Authors’ conclusions: There is little question that EIBI is highly effective for some children, however gains are not universal. Crucial need to systematically assess for which children is EIBI most and UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score Clearly defined methodology and comprehensive search strategy No rating of study quality Detailed Tables and discussion of results Good attempt to draw out patterns of results and explain findings. Discussion of the methodological limitations of included studies Suggestions made for future research Overall quality: Good 20 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions least effective. Clearer evidence concerning the optimal duration of therapy and the age at which is should begin could result in the development of better targeted, more cost effective programmes. Need for more RCTs, and need for equivalence trials, where high quality interventions are compared with each other, rather than comparing high quality interventions with low quality/low intensity interventions. Failure to control for time in intervention means that for some studies, EIBI may be more effective due to differences in intensity, not quality. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score 21 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Eikeseth (2009) Evidence level: I Inclusion criteria: study was published in a peerreviewed journal; children had a mean age of 6 years or less at intake; participants received comprehensive psycho-educational interventions; studies contained outcome data. 25 studies met selection criteria, including 20 which evaluated interventions based on Applied Behavior Analysis (ABA) treatment, 3 studies evaluated Project TEACCH, and 2 studies evaluated the Colorado Health Sciences Model. Only those relating to ABA are discussed here. Norway SR Aim: Evaluates comprehensive psycho-educational research on early intervention for children with autism. Search period: Not reported Search strategy: ERIC Medline Psyclit (search terms not provided). References of “recent papers” were cross checked to identify additional papers. Researchers known in the field were contacted by email and asked for relevant articles recently published or “in press”. Comprehensive psychoeducational interventions were defined as interventions addressing social behaviors, communication and ritualistic/stereotyped behaviors. Exclusion criteria: None reported. Outcome studies were graded according to their (i) scientific value and (ii) according to the magnitude of results documented in the studies. Scientific Merit was evaluated based on: (a) diagnosis, (b) study design, (c) dependent variables and (d) treatment fidelity. Four levels were used. Comments and quality score Scientific merit 1 ABA study, an RCT, received Level 1 scientific merit (Smith, Groen, & Wynn, 2000) 4 ABA studies received Level 2 rating (Cohen, Amerine-Dickens, & Smith, 2006; Eikeseth, Smith, Jahr, & Eledevik, 2002, 2007; Howard, Sparkman, Cohen, Green, Stanislaw, 2005; Remington et al., 2007) 9 received Level 3 evidence support 6 ABA outcome studies were classified as having insufficient (level 4) scientific value. Magnitude of treatment effects 4 ABA studies received Level 1 rating demonstrating that children receiving ABA made significantly more gains than control group children on standardized measures of IQ, language and adaptive functioning (Cohen et al., 2006; Eikeseth et al., 2002, 2007; Howard et al., 2005; Sallows & Graupner, 2005). Several studies also included data on maladaptive behavior, personality, school performance and changes in diagnosis. 3 ABA studies received Level 2 rating (Eldevik et al., 2006; Lovaas, 1987; Smith, Groen, & Wynn, 2000), demonstrating that ABA treated children made significantly more gains than the comparison group on one standardized measures of IQ or Adaptive Functioning. 5 ABA studies received Level 3 rating. Authors’ conclusions: Recommended practice parameter ABA treatment is demonstrated effective in enhancing global functioning in pre-school children with autism when treatment is intensive and carried out by trained therapists. ABA treatment is demonstrated effective in enhancing global functioning in children with PDD-NOS (one Level 1 study; Smith, Groen, & Wynn, 2000). Guideline practice parameter ABA can be effective for children who are up to 7 years-of-age at intake (one Level 2 study; Eikeseth et al., 2002, 2007). UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comprehensive search strategy, but no search terms or dates given to allow replication Very detailed Tables, describing studies, interventions, and discussion of results Lacks detail on participant characteristics at baseline. Discussion of the methodological limitations of included studies Suggestions made for future research directions Insufficient quantitative information on the extent of the gains experienced, the clinical meaningfulness of these gains and whether there were subgroups of children who did not improve. Overall quality: Good 22 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Eldevik et al (2009) Evidence level: I Inclusion criteria: intervention generally adhered to defined common EIBI elements; aged between 2-7 years at intake; diagnosed with autism, PDD NOS; full-scale measure of intelligence and/or standardized measure of adaptive behaviour at intake and post intervention; intervention lasted 12-36 mths; included Control (no intervention, considerably less intensive one, or poorly described intervention) or Comparison (non EIBI intervention of similar duration and intensity) group. 9 controlled designs met selection criteria, including one RCT (Smith et al, 2000). Howard et al (2005) included both a Control and Comparison group and so 10 effect sizes were calculated. Four studies had a Comparison group, and six had a Control group. Wales & Norway SR/MA Aim: To conduct a systematic literature search for studies reporting effects of Early Intensive Behavioral Intervention (EIBI). Also aimed to replicate and extend (with methodological improvements) the MA of Reichow and Wolery (2009). Search period: - March 2008 Search strategy: ERIC PsycINFO PubMed (search terms provided). References of retrieved papers were cross checked to identify additional papers. Exclusion criteria: Primarily using nonverbal intelligence measure; Case study or case series; Data from people in another included study. Inter-rater reliability was high for selection and coding of articles. Approached authors for raw individual data which was re-analysed. Completed a meta-analysis yielding a standardized mean difference (SMD) effect size for two available outcome measures: change in full-scale intelligence and/or adaptive behavior composite. Effect sizes were computed using Hedges’s g. The average effect size was 1.10 for change in full-scale intelligence (95% CI .87, 1.34) and .66 (95% CI=.41, .90) for change in adaptive behavior composite. These effect sizes are generally considered to be large and moderate, respectively. As the two sets of comparative data from Howard et al were compared to the same intervention group and not independent, an analysis excluding the control group from Howard et al (2005) was also conducted, with effect sizes not significantly altered. Homegeneity was tested and the Q statistic was not significant, indicating that studies could be combined, and a fixed effect model was used. No statistical or visual evidence of publication bias was observed. Authors’ conclusions: EIBI produces large to moderate effect sizes for changes in IQ and ABC scores for children with ASD when compared with no intervention controls and eclectic provision. The results support the clinical implication that at present, and in the absence of other interventions with established efficacy, Early Intensive Behavioral Intervention should be an intervention of choice for children with autism. The difference in outcome between EIBI and the comparison intervention may be due to differences in the amount and frequency of supervision and training. There was insufficient data to control for this in the present study, however the EIBI group in general received more frequent and more total hours of supervision and training. This remains a threat for the validity of conclusions about the superiority of EIBI in relation to comparison intervention. Randomized controlled trials comparing EIBI to other interventions are still needed. In particular, studies are needed where the comparison intervention is of similar intensity and where staff receive similar training and supervision. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Moderately broad search strategy Excluded Sallows and Graupner (2005) as it compared two EIBI groups of somewhat different intensity varying in programme delivery (clinic- or parent-directed) Used more inclusive definition of EIBI than Reichow and Wolery (2009) Cited papers of those retrieved which were excluded and noted why Detailed account of outcome measures considered Approached authors for individual data Fully detailed Tables, describing design, allocation to group method, interventions, outcome scores Moderator variables were not considered. One author is from Lovaas institute. Overall quality: Very good 23 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Ospina et al. (2009) Evidence level: I Inclusion criteria: RCTs, non randomized controlled trials, and prospective and retrospective cohort studies with a control group; published in English; reporting on the efficacy of any behavioural or developmental intervention; for individuals with ASD (including those with dual diagnoses); reporting on outcomes relating to the core features of ASD (though additional outcomes may also be reported). Canada SR/MA Aim: To identify, appraise and synthesise the evidence on the effectiveness of behavioural and developmental interventions for improving core symptoms of ASD. Search period: – May 2007 Search strategy: 22 electronic databases, including: BioSYS Previews CINAHL Cochrane central register of controlled trials Embase ERIC Medline ProQuest Dissertations and theses PsychINFO Web of Science (search terms provided). Hand searching journals, cross-checking of reference lists, databases of theses and dissertations, and contacting experts in the field were used to identify additional papers. Exclusion criteria: Individuals with Rett’s Disorder, or Childhood Disintegrative Disorder. Two independent reviewers made the final study selection, extracted data using pre-tested checklists, and reached consensus on study quality. Metaanalyses of the study results were conducted when two or more trials assessed the same intervention, used similar comparison groups, and had data for common outcomes of interest. If the same measure was reported, reported weighted mean differences (WMD); otherwise, reported standardised mean differences (SMD). SMD of 0.2 indicated a small effect, 0,5 a medium, and 0.8 a large effect. Results and authors’ conclusions 101 studies met selection criteria, and evaluated 8 broad types of intervention defined on a continuum of behavioural to developmental interventions, the most relevant reported here being the ABA category, which included discrete trial training (DTT) and UCLA/Lovaas therapy as sub-categories. There were 31 ABA studies, including 12 trials and 9 cohort/observational studies. Descriptive analyses The reviewers judged studies to be at predominantly high risk of bias and reported inconsistent results across various interventions. Discrete Trial Training effects were inconsistent, but motor and functional outcomes were often positive compared to speechrelated outcomes which were generally negative. For Lovaas, 7 of 8 non-RCTs reported positive findings, whereas 3 of 4 RCTs reported no significant findings. The reviewers note that as non-RCT evidence can be more prone to bias and overestimate treatment effects, this observation has serious implications for interpreting the results from non-RCTs. Heterogeneity was moderate for some outcomes (for overall intellectual functioning, non verbal intellectual functioning, socialization). Meta Analyses Meta-analyses of 3 controlled clinical trials showed that Lovaas treatment was superior to special education on measures of adaptive behaviour (WMD = 11.8; 95%CI, 6.94 to 16.67), communication and interaction (WMD = 16.63; 95% CI, 11.25 to 22.01), comprehensive language (WMD = 12.84; 95% CI, 6.38 to 19.30), daily living skills (WMD = 5.61; 95% CI, 0.54 to 10.67), expressive language (WMD = 15.05; 95% CI, 6.19 to 23.90), overall intellectual functioning (SMD = 0.95; 95% CI, 0.44 to 1.46), and socialization (WMD = 9.17; 95% CI, 2.16 to 16.19). High-intensity Lovaas was superior to low-intensity Lovaas on measures of intellectual functioning in two retrospective cohort studies (SMD = 0.92; 95% CI, 0.61 to 1.24). No statistically significant differences were found for: Lovaas versus special education for non-verbal intellectual functioning in a meta analysis of three controlled clinical trials (SMD = 7.83; 95% CI, 22.86 to 18.52); or for Lovaas versus Developmental Individual-difference Relationship-based intervention for communication skills based on two RCTs (SMD = 0.73; 95% CI, 20.26 to 1.72). UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score Clearly defined methodology and extensive and comprehensive search strategy Review reports extensive details of individual studies including sample characteristics, as well as assessments for each study across several quality domains. Careful synthesis of results. Discussion of the methodological limitations of review approach and studies. In particular, notes risk of “expectancy bias” for studies with non active controls (eg, wait list) where participants (and their families) in control group would not expect improvement. Moderate heterogeneity evident for some outcomes. Suggestions made for future research Overall quality: Very good 24 Reference, country, design Evidence level, aim and search method Results and authors’ conclusions Inclusion and exclusion criteria Authors' conclusions: There is no clear answer regarding the most effective therapy to improve symptoms associated with ASD. The evidence seems to provide some support for discrete trial training in terms of motor and functional skills but not for communication skills. Lovaas' therapy showed benefits when compared to ''no treatment'' and evidence from meta-analysis of retrospective cohort studies showed greater effects for High versus Low intensity Lovaas. Whilst Lovaas may improve some core symptoms of ASD compared to special education, these findings are based on pooling of a few, methodologically weak studies with few participants and relatively short-term follow-up. Replication in RCTs is needed to substantiate the use of Lovaas and to assess the effect of treatment intensity on outcomes of children with ASD. There is considerable potential for heterogeneity in the population, intervention, comparator and outcomes of interest, as ASD is a spectrum disorder, therapy is not always reported in detail, comparators are difficult to control for, and outcomes are somewhat subjective. As no definitive behavioural or developmental intervention improves all symptoms for all individuals with ASD, it is recommended that clinical management be guided by individual needs and availability of resources. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score 25 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Reichow and Wolery (2009) Evidence level: I Inclusion criteria: evaluated EIBI based on the UCLA/Lovaas (1987) model; participants had diagnoses of autistic disorder, ASD, PDD, or PDD-NOS; participant samples receiving EIBI had a mean chronological age < 84 months at the beginning of treatment; mean duration of EIBI was 12 months; at least one child outcome measure was reported; experimental (e.g., pre/post-test multiple-group design) or quasiexperimental designs (i.e., nonequivalent control group, one-group pre/post-test design); publication in English in a peer-reviewed journal. sample adjustment. 14 studies met selection criteria, although due to patient overlap, only 13 publications were appraised, reporting on 14 distinct samples. The studies included 10 studies allowing between group analyses, including two RCTs (Smith et al, 2000; Sallows and Grupner, 2005). USA SR/MA Aim: To provide a 3-part comprehensive synthesis of the early intensive behavioral intervention (EIBI) for young children with autism based on the Lovaas University of California at Los Angeles Young Autism Project method (Lovaas, 1987). The synthesis included: (a) descriptive analyses, (b) effect size analyses, and (c) a meta-analysis. Search period: Not reported Search strategy: An “electronic database search”, databases not identified. (Search terms not provided). References from review articles and eligible reports were cross checked to identify additional papers. Descriptive analyses EIBI is an effective intervention for children with autism. On average, children present fewer or less severe autism symptoms after intervention. Effect size differences between treatment groups The between group (standardized mean difference) effect sizes suggest children receiving EIBI made more gains than children receiving minimal behavioral intervention, eclectic treatment, or treatment as usual. Children receiving EIBI made large gains on multiple domains of behavior, and made better progress than children with autism who receive less intense behavioral intervention or other treatments. Meta analysis for change in IQ within treatment group A meta-analysis was conducted on 12 samples using standardized mean change effect size for IQ within the intervention group (because comparison groups varied across studies). Using a random effects model, the mean effect size was 0.69. This is statistically significant (p < 0.001) and represents a large effect. This suggests EIBI is, on average, an effective intervention for increasing IQ scores for children with autism, though the effect is likely to be inflated by publication bias. Tests of homogeneity suggest that there was significant variability between studies. Moderating variables identified a priori suggest the greatest results on IQ change might be seen when supervisory staff were trained using the UCLA model (p<0.01). Hand search of selected journals. Contact with experts. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Limited detail provided on search strategy restricting ability for replication. Detailed Tables, figures Precision details (confidence intervals) given in Figure only Discussion of the methodological limitations of included studies and of approach of reviewers. Studies used a range of measures of IQ Included quasi experimental studies without an equivalent control group in the MA. Heterogeneity present, which raises questions about whether combining results through a meta analysis was appropriate. Suggestions made for future research directions Overall quality: Good 26 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Study characteristics and effect size data were coded using a manual. Two independent coders for 4 of the studies had mean interobserver agreement of 92%, range 85-93%. Studies rated on experimental rigour. Meta analysis conducted for studies reporting within-group comparisons using standardized mean change effect size with small sample adjustment. Authors’ conclusions: While these findings were strong, the nonrandom assignment of participants to groups limit conclusions about the superiority of EIBI to other treatments. Further, the comparison groups often lacked standardization within the group, were poorly defined, had no measures of procedural fidelity, and had no data on whether participants received supplemental treatments. The findings suggest EIBI can be an effective treatment, on average, for some children with autism. However the intervention has not worked for all children, and individual data typically were not presented. It is unclear if individuals making change in one domain (e.g., IQ) also made gains in another (e.g., adaptive behavior). These results should be taken with caution because of the small number of studies, and because the standardized mean change effect size is calculated without reference to a comparison or control group. Thus, the threats to internal validity of history, maturation, lack of procedural fidelity, and instrumentation threats cannot be eliminated. Thus, while the effect sizes were often large, they cannot be attributed to EIBI exclusively. Effect sizes tend to be inflated when using standardised mean change effect sizes (Morris, 2000). Without comparisons between EIBI and empirically validated treatment programs, it is not possible to determine if EIBI is more or less effective than other treatment options. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Comments and quality score 27 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Rogers and Vismara (2008) Evidence level: I Inclusion criteria: study involved comprehensive (addressing core deficits in autism) treatment approaches for children with autism; predominantly ages 5 years or younger; controlled group designs or singlesubject multiple baseline designs using 3 or more subjects with published data. 22 studies met selection criteria. Studies of non ABA interventions, including psychopharmacological interventions, are not discussed here. 5 RCTs, included 4 relating to ABA were identified: Jocelyn et al (1998), Drew et al (2002), Smith et al (2000), and Sallows and Graupner (2005). Non-randomized controlled group design studies included 2 of Lovaasbased interventions: Eikeseth et al (2002), Cohen et al (2006); and one of an EIBI approach using a mixture of didactic and naturalistic behavioral teaching approaches by Howard et al (2005). US SR Aim: What is the empirical evidence supporting efficacy of early intervention or young children with autism. Search period: 1998 - 2006 Search strategy: PsycINFO (search terms provided). References of all reviewed articles were cross checked to identify additional papers. Exclusion criteria: studies that did not report as outcomes analyses of child progress using general measures of children’s language or intellectual development; studies targeting only one domain (eg; unwanted behaviours); case reports; studies whose data were published only in book chapters; studies included in review by Rogers (1998). Studies were graded according to methodological criteria based on Chambless et al (1998) as being (i) “well established” or (ii) probably efficacious. Studies also graded using Nathan and Gorman (2002) criteria as Type 1 through to Type 6, ranging from most to least rigorous. Results suggest that young children with autism, as a group, demonstrate accelerated developmental gains in response to focused daily interventions of several different kinds. Significant increases in language and communication abilities in the treated group occurred in most studies and interventions with many targeted hours per week resulted in increases in IQ at the group level as well. The best-designed, controlled studies evaluating Lovaas’s treatment met criteria for a ‘‘well-established’’ psychosocial intervention for improving the intellectual performance of young children with autism spectrum disorders, based on the significant increase in IQ reported in these four studies compared to control groups. Although some of these studies also reported significant improvements in behavior, adaptive skills, and language skills, reviewers argue that there is less consistency in the data in these areas. No treatment met the ‘‘probably efficacious’’ criteria, although 3 treatments met criteria for being ‘‘possibly efficacious’’. Most studies were either Type 2 or 3 in terms of their methodological rigor based on Nathan and Gorman’s (2002) criteria. Limited search strategy, 1 database and cross-checking Detailed Tables, describing studies, interventions, and discussion of results Discussion of methodological limitations of included studies and findings. Detailed suggestions made for future research directions, especially around mediators and moderators of effect Practice recommendations are offered for psychologists. Discussed the value of both RCTs and SCED studies in the pathway of intervention development. Overall quality: Good Authors’ conclusions: The evidence suggests that early intervention programs are beneficial for children with autism, often improving developmental functioning and decreasing maladaptive behaviors and symptom severity at the level of group analysis. Lack of comparative studies prevents determining which comprehensive treatment approach is best for young children with autism. Other wellknown interventions may be as or more efficacious as Lovaas’s model but they have not been rigorously evaluated. Because the majority of studies did not include ethnically diverse participant groups, generalization of effects across groups is premature. Given the few randomized controlled treatment trials that have been carried out, the few models that have been tested, and the large differences in interventions that are being published, it is clear that the field is still very early in the process of determining (a) what kinds of interventions are most efficacious in early autism, (b) what variables moderate and mediate treatment gains and improved outcomes following intervention, and (c) the degree of both short-term and long-term improvements that can reasonably be expected for a child with autism. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER 28 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Seida et al (2008) Evidence level: I Inclusion criteria: Systematic reviews (defined as having defined a search strategy, and attempted to synthesise data of primary studies quantitatively or qualitatively); Data reported for participants with autistic disorder, Asperger syndrome, atypical autism, high functioning autism, PDD-NOS, and/or suspected autism; Have covered a psychosocial intervention aimed at improving the functioning of individuals in any of the impairments characteristic of ASD. 30 studies met selection criteria representing five intervention domains, including one of interventions based on behavioural theory (n=9) reported on here. Canada SR (of SRs) Aim: To present an “umbrella review” of the clinical findings of systematic reviews of psychosocial interventions for ASDs. Search period: – May 2007 Search strategy: 25 electronic databases including Medline (listed with search terms in an online supplement). References of articles and of “personal collections” were cross checked to identify additional papers. Exclusion criteria: editorials, correspondence, abstracts and review summaries All 9 SRs on interventions based on behavioural theory were rated as having low methodological quality. The reviews with meta-analyses showed favourable outcomes for the behavioural intervention. Reductions in problem behaviour were found in 3 reviews and an increase in adaptive, cognitive, and language skills was observed in 1 review. In the reviews without meta-analysis, positive findings were reported for intelligence, developmental gains, functional skills, and communication outcomes. The authors of 3 reviews concluded that there is uncertainty about whether behavioural interventions produce ‘normal functioning’ and improvement on various intelligence and developmental measures. The remaining 2 reviews provided no information on efficacy outcomes; instead trends in the use of behavioural treatments over time were described. Authors’ conclusions: The reviews reported positive outcomes for many of the interventions, suggesting that some form of treatment is favourable over no treatment. Little evidence for the relative effectiveness of these treatment options. Many of the systematic reviews had methodological weaknesses that make them vulnerable to bias. Even if differences in the therapeutic effectiveness of the interventions exist, differences in cost, convenience and family burdens associated with the interventions are likely to be important factors in individual decisionmaking. Future studies and reviews that break down the characteristics of the individuals with autism and the components of the programmes are needed in order to provide more meaningful and stronger conclusions. Abstracts scanned by single reviewer but eligibility criteria applied on full text independently by two reviewers. Studies were graded for methodological quality by two reviewers using the Overview Quality Assessment Questionnaire. Data extracted by single reviewer, and verified by another. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Broad search strategy Use of two reviewers and quality checklist Only secondary studies considered Only sparsely detailed Tables, listing intervention domains and the outcome areas where positive, negative or unclear results were found. No data provided to allow estimation of effect or precision of findings. Suggestions made for future research, including on reporting and conduct of SRs Overall quality: Fair 29 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Spreckley & Boyd (2009) Evidence level: I Inclusion criteria: Systematic reviews, randomized or quasirandomized controlled trials (RCT); Preschool children with diagnosis of ASD or PDD; Interventions focused on ABI approaches to behavioural management; Interventions delivered to parents/carers and/or directly to the child by special educators, teachers, speech pathologists, psychologists, or allied health professional students; Interventions occurred between 18 mths and 6 years; Cognitive, language, and/or adaptive behaviour outcomes measured. 13 studies met selection criteria, 6 were trials with adequate methodologic quality (PEDro6), and 4 of these had adequate data to be included in the meta analysis (Smith et al, 2000; Sallows and Graupner, 2005; Eikeseth et al 2002; and Eikeseth et al, 2007). Australia SR/MA Aim: To review the effectiveness of applied behavior intervention (ABI) programs for preschool children with autism spectrum disorder (ASD) in their cognitive, adaptive behavior, and language development. Search period: From 1982 but varied (see below) - Nov 2007 Search strategy: Cochrane Library Medline (from 1996) PsycINFO (from 1985) CINAHL (from 1982) AMED (from 1985) (search terms provided). Meta-analysis of 4 studies concluded that, compared with standard care, ABI programs did not significantly improve outcomes of children in the experimental group compared with those who received standard care: For cognitive outcomes, a standardized mean difference (SMD) of 0.38 (95%CI �-0.09 to 0.84; P=.1) For expressive language; SMD of 0.37 (95%CI - 0.09 to 0.84; P=.11) For receptive language; SMD of 0.29 (95%CI �- 0.17 to 0.74; P=.22) For adaptive behavior; SMD of 0.30 (95%CI �-0.16 to 0.77; P=.20). Heterogeneity (measured by I2) was low for cognitive and receptive language outcomes, and moderate for expressive language and adaptive behaviour outcomes. Authors’ conclusions: Currently there is inadequate evidence that ABI has better outcomes than standard care for children with autism. Appropriately powered clinical trials with broader outcomes are required. Limitations of the meta-analysis and evidence base include: high variability in the studies included (in ages of children involved, programme content), difficulty in establishing control groups all received some form of intervention), poor homogeneity, limited information on retention in the interventons groups, and lack of strict selection criteria for participants. What is too often forgotten is that the overwhelming majority of children with ASD change over time as part of their development. Exclusion criteria: Studies with PEDro score <6; RCTs with no useful data for meta-analysis; and/or those that did not include discrete trial training (DTT) as part of their intervention. 2 reviewers independently applied the Physiotherapy Evidence Database (PEDro) Scale of quality assessment to critically appraise the studies. Data synthesis used RevMan, reported effect sizes and standardised mean differences (SMD). UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Broad database search strategy, but no other search technique Applied minimum quality criteria for inclusion in the MA Tables described design, interventions, comparators, and intensity. Estimates of precision provided. Too few studies included to consider sub-group or sensitivity analyses or to consider moderator variables. Follow-up correspondence about this article from Smith et al (2009) suggested that the Sallows and Graupner (2005) study be excluded from the MA as its comparison group was a lower intensity form of ABA. Spreckley & Boyd’s (2009b) reply argued that they were evaluating standard (i.e., high intensity) ABA with any standard or eclectic comparator. Limitations and heterogeneity of the evidence base raises questions about whether a meta analysis was appropriate. Overall quality: Good 30 2. PECS Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria Results and authors’ conclusions Comments and quality score Ostryn et al. (2009) Evidence level: I 15 studies met selection criteria. Most studies were single subject within subjects or multiple baseline designs, with one RCT (Yoder and Stone, 2006) and one non-randomised, between-group experimental study (Carr and Felce, 2007). Aim: To examine research involving the use of PECS with individuals with ASD to identify dependent variables and outcomes reported in the research, and to apply the communication competence paradigm to the PECS research. Inclusion criteria: English language; Used experimental or quasiexperimental design; Included individuals aged under 18 years old with ASDs; evaluated PECS; published in a peerreviewed Journal. Search period: 1985 –2007 Exclusion criteria: Not reported USA SR Search strategy: ERIC PsychINFO PubMed (search terms provided). Also searched PECS website. References of retrieved papers were cross checked. Narrative synthesis Noted that whilst “positive outcomes” were reported in all but four studies, though these were not defined in terms of statistical significance. Noted that acquisition rates (through phases of PECS 1-6) were reported as being “fast” by several studies, though “fast” was not defined. Commented that outcomes such as spontaneous communication were only defined in two studies in relation to visual prompts. Outcomes associated with joint attention were measured in only three studies. Notes that few studies examined maintenance of PECS outcomes. Where they are reported, results are encouraging. The authors applied the communication competence paradigm to PECS outcomes and argue that additional strategies to those of manding or tacting could enhance PECS. The authors suggest that PECS strategies may be too limited for some individuals and people with ASD should be encouraged to use any communication attempts, including speech, gestures, and vocal approximations. Authors’ conclusions: Results of this study reveal that the PECS is widely implemented with individuals having ASDs but without a strong empirical base. Suggest that PECS is best used as an initial intervention to teach manding and the basic elements of communicative exchange, but that other communication systems may be needed to permit communication beyond this. PECS is not recommended as a long term intervention and is best implemented as part of a multimodal system for when picture communications are more socially appropriate. UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER Broad search strategy Detailed Tables describing outcome measures and results, but lacking data on participant characteristics and interventions. Narrative discussion of results framed around how outcomes measures in PECS studies relate to the communication competence paradigm. No formal checklists used to assess or grade quality. Minimal synthesis of findings, although note that this was not specifically the aim of the review. Suggestions made for future research. Suggestions for how to use PECS as part of a multimodal approach including other communication systems Overall quality: Fair 31 Reference, country, design Evidence level, aim and search method Inclusion and exclusion criteria SulzerAzaroff et al. (2009) Evidence level: I Inclusion criteria: Articles including data and protocol containing the key features of PECS (as specified by Frost and Bondy). UK SR Aim: To synthesise key features of the published, peer-reviewed, databased research reports of Picture Exchange Communication System (PECS) applications, to answer the question “how solid is the scientific evidence supporting PECS effectiveness?” Search period: Not reported Search strategy: ERIC PsycINFO Science Direct Google scholar. Minimal search terms used including PECS, Bondy, Frost (who developed PECS). References of retrieved papers were cross checked to identify additional papers. Exclusion criteria: Papers written in languages the authors couldn’t read (n=3) Studies describing different pictorial interventions to the original PECS. The review summarised the studies in a narrative way, with little quantification. Results and authors’ conclusions Comments and quality score 34 studies met selection criteria, including two RCTs (Howlin et al. 2007) (Yoder and Stone 2006b) (Yoder and Stone 2006a) and one non randomised experimental study (Carr and Felce 2007). Studies reported on 386 study participants, the “majority” of whom had received diagnoses of ASD. The authors presented a number of summarising statements in the Results section asserting positive research findings, however these were not attended by quantitative supporting data, analysis or synthesis of study findings. For example, “researchers reported improvement in communication among the vast majority of their participants”. Also, “several investigators provided evidence that learning to use PECS was associated with some of their participants increasing their speaking and social approaching” and “a number of investigators cited decreases in disruptive or dangerous behaviours”. Thus whilst these and similar statements stated that there were examples of positive findings, the authors did not demonstrate the extent to which such findings occurred or the precision of this effect, how effects varied between studies and individuals, or why. The authors urged PECS researchers to report more methodological details, and identified questions to be explored in future research. Authors’ conclusions: Analysis of the available research leads to the conclusion that the majority of participants who lacked functional comunication skills did acquire extensive functional vocabularies. Analysis of the studies strongly supports the conclusion that by adhering to the PECS protocol, professionals and parents can teach individuals to successfully initiate exchanges of pictures for tangible and non tangible reinforcers. Limited search strategy, particularly with respect to search terms and reporting. Narrow definition of PECS. Lacks detail of outcome measures and follow-up periods of inidivudal studies. Results and Discussion focussed on narrative, imprecise description of studies. Detailed Tables Emphasised the need for additional reporting in future research Note that two of the authors were the original developers of PECS. Overall quality: Fair Key: ABI CI EIBI IQ PEDro PECS SCED SMD TEACCH WMD Applied behaviour intervention Confidence interval Early intensive behavioural intervention Intelligence quotient Physiotherapy evidence database Picture exchange communication system Single case experimental design Standardised mean difference Treatment and education of autistic and related communication-handicapped children Weighted mean difference UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER ABA DTT Hrs/wk MA PDD NOS RCT SD SR UCLA Applied behaviour analysis Discrete trial training Hours per week Meta analysis Pervasive developmental disorder – not otherwise specified Randomised controlled trial Standard deviation Systematic review University of California, Los Angeles 32 UPDATE OF REVIEWS: APPLIED BEHAVIOUR ANALYSIS FOR PEOPLE WITH AUTISM SPECTRUM DISORDER