Positive behavioural support for children and young people with developmental disabilities in special education settings: A systematic review

Abstract Background Positive behavioural support (PBS) can be effective in supporting children and young people (CYP) with developmental disabilities. This systematic review focused on describing the components and nine characteristics of PBS that have been used with CYP with developmental disabilities in special education settings, and the evidence for PBS effectiveness in these settings. Additionally, facilitators and barriers to PBS implementation, and experiences of stakeholders, were investigated. Method Systematic searches followed a registered protocol, and 30 studies were identified, narratively synthesised, and critically appraised. Results From the 30 studies included, 10 reported the presence of all 9 PBS characteristics, 17 reported on 8 PBS characteristics, and 3 reported on 7 characteristics. Overall, 28 studies demonstrated significant decreases in behaviours that challenge and increases in alternative behaviours, if increasing alternative behaviours was part of the interventions. Conclusions There was a lack of evidence on facilitators and barriers, and a lack of qualitative studies exploring experiences of stakeholders with PBS in special education settings. The available evidence suggested that not all studies reported on all PBS characteristics when describing the approach followed. In addition, available evidence suggested that most studies demonstrated effectiveness of PBS regarding the measured outcomes. Implications and future directions are discussed.

and the Centers for Disease Control and Prevention (CDC) (Centers for Disease Control and Prevention, n.d.) as a group of conditions that can cause impairments in learning, social and occupational functioning, and can also cause physical impairments. The term includes a wide range of conditions such as intellectual disability, autism, specific learning disorder, attention deficit/hyperactivity disorder (ADHD), and many more. The term developmental disabilities will be used in the text to refer to the specific conditions of intellectual disability (or genetic syndromes predominantly manifesting with an intellectual disability) and autism, that belong to the general category of developmental disabilities. Children and young people (CYP) 2 with comorbid conditions of which at least one is either intellectual disability (or genetic syndrome causing an intellectual disability) or autism, will also be referred to as CYP with developmental disabilities. However, it is acknowledged that the term 'developmental disabilities' is commonly used to refer to the abovementioned conditions (i.e., intellectual disability, genetic syndrome causing an intellectual disability, autism) as well as other conditions (e.g., including ADHD). PBS places emphasis on understanding behaviour that challenges at a functional level, using that understanding to develop systems of support that increase the likelihood of preventing or reducing behaviour that challenges, and increasing the quality of life for the person and their family (Carr et al., 2002;Gore et al., 2013). Definitions of PBS reflect the different contexts in which the framework has been applied, including mainstream/inclusive schools in the USA and residential or community settings in the United Kingdom Kincaid et al., 2016;), but also include three components: values, systems change, and the underpinning science and technologies.
The development of PBS was motivated by movements advocating equality for people with developmental disabilities, protection of their human rights , and their valued role in society (Wolfensberger, 2011). There are three dimensions to the values component, and these remain at the core of PBS: a constructional approach, a non-aversive approach, and stakeholder involvement. PBS, as a proactive and constructional approach (Kincaid et al., 2016), focuses on expanding repertories of social and other functional competencies by building new skills, as well as redesigning environments to increase life opportunities for people in need of support Kincaid, 2018). Behaviours that challenge may still occur (Carr et al., 2002), but skills-building and increased life opportunities are a focus in their own right. PBS is guided by the philosophy of non-aversive, respectful practices, and advocates for the use of alternatives to punitive technologies (Carr et al., 2002). Stakeholder participation is also an integral part of the PBS approach (Carr et al., 2002) and is incorporated at every step of the PBS implementation process, to ensure social validity and adherence to the supports provided.
The systems change component of PBS relates to the establishment of a 'continuum of positive behaviour support' (Lewis et al., 2002, p. 181) with interventions applied within a multi-tiered framework at both the individual and a larger systems level (such as family and school contexts; Kincaid et al., 2016). A three-tiered model for prevention and intervention informs supports progressively more individualised according to need, allowing organisations to scale-up the support provided whilst achieving consistency across settings, support providers, and time.
The science and technologies component of PBS is underpinned by behaviour analytic principles, combined with other evidence-based practices Kincaid et al., 2016). Thus, PBS utilises knowledge derived from behavioural science and other scientific fields (such as pedagogical and implementation science) to develop an understanding that all behaviours have a function, what that function is, and how to design and scale-up implementation of supports. These scientific principles define the technologies that are then applied to achieve socially valid outcomes. A strong theoretical and evidencebased understanding of the function of behaviours that challenge is a primary characteristic of PBS Iwata et al., 1994). Functional behaviour assessment (FBA) 3 procedures are designed to identify the function of behaviour and interventions informed by FBA are more effective than those not functioninformed (Ingram et al., 2005). PBS also utilises a data-informed approach (Carr et al., 2002) to guide decision making (Horner & Sugai, 2018;Kincaid et al., 2016). Using data to make critical decisions about established supports minimises bias in intervention decisions and reduces influence of personal perspectives . Therefore, decision making is evidence-informed by quantitative and qualitative data collected, aiming to avoid unsubstantiated personal assumptions.
PBS is not a single intervention or programme. It is rather a framework that incorporates a strong values base, systems change, and supports utilising scientific knowledge and technologies based on primarily, but not limited to, behaviour analytic principles. Behaviour analytic principles, deriving from behavioural psychology, underlie applied intervention programmes such as applied behaviour analysis (ABA) interventions as well. However, from the available evidencebased interventions and supports only those relevant to individual needs and adhering to the core values of PBS are selected as appropriate for implementation within a PBS framework. PBS places emphasis on offering non-aversive, respectful, and socially valid supports, that are proactive and constructional, to increase the quality of life of CYP (Carr et al., 2002;Gore et al., 2013;Kincaid et al., 2016). Therefore, only supports guided by these values are incorporated as part of the PBS framework. Moreover, PBS also includes the use of other evidence-based practices, and incorporates systems change to scale-up and achieve consistency of supports Kincaid et al., 2016), making it distinct from single intervention programmes.
School-wide positive behavioural support (SW-PBS) is a successful preventative model in mainstream schools for all children, including those with special educational needs . A three-tier model of prevention has been adopted, incorporating universal supports across the whole educational setting at the primary level (Tier 1), targeted supports for selected at risk groups at the secondary level (Tier 2), and specialist supports for CYP that require individualised support at the tertiary level (Tier 3; Sugai & Horner, 2006). However, PBS (in its individualised, group level, or school-wide form-following the SW-PBS model) has been implemented in special education settings internationally as well, in support of CYP with developmental disabilities. Moreover, there is also a research base evaluating PBS in alternative and special education settings (e.g., Clarke & Duda, 2019;Jolivette et al., 2012;Paris et al., 2019;Simonsen et al., 2010;Simonsen & Sugai, 2013;Wienen et al., 2019).
To date, systematic reviews on PBS and its effectiveness have either focused on PBS implementation to support CYP with disabilities in various settings, including PBS implementation in noneducational settings (Snell et al., 2005), aggregate reporting of studies conducted in both mainstream and special educational settings (Goh & Bambara, 2012;Noltemeyer et al., 2019), reporting of the implementation of one specific tier of support, such as Tier 1, in alternative settings (including juvenile justice facilities; Grasley-Boy et al., 2020), or have focused on effectiveness of function-based PBS for CYP diagnosed with specific conditions such as emotional and behavioural disorders (Lane et al., 2009). Some of these systematic reviews included data on the effectiveness of PBS in special education settings as part of the aggregate reporting of studies conducted in various settings for CYP with various diagnoses. However, we were unable to find reviews that were conducted in special education settings addressing the implementation of PBS for CYP with developmental disabilities. The current systematic review focused on special education settings because it aims to describe the PBS framework characteristics, including essential adaptations made to achieve contextual fit for this specific setting, when supporting the target population. It also explores the effectiveness of these characteristics. This is particularly important considering the increasing number worldwide of special educational settings that choose to implement the PBS framework to support CYP with developmental disabilities, a group at risk for behaviours that challenge. This systematic review focused on the following questions:

| METHODS
For the systematic review, the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines (Page et al., 2021) were followed. The protocol was registered with the international prospective register of systematic reviews (PROSPERO 2019: CRD42019131954).

| Search strategy
Seven electronic bibliographic databases were searched (PsycINFO, ERIC, MEDLINE, Applied Social Sciences Index and Abstracts (ASSIA), Social Science Citation Index (SSCI), Web of Science, and Scopus). The search strings included terms related to developmental disability (a): Intellectual Disability (e.g., Learning Disab*, Intellectual* Impair*, Down* syndrome) and autism (e.g., Autis* Spectrum Disorder*, Asperger* syndrome); (b): behaviour that challenges (e.g., Challenging behav*, Aggressive behav*); and (c): special education settings (e.g., Special educat* provision*, Alternat* educat* setting*, Resource* room). Terms within the same search group were separated by OR, and then combined using AND. An example search string can be found in the Table S1.
Backward and forward reference searches were also used. The forward reference search was conducted using the Social Sciences Citation Index (SSCI) and Google Scholar databases to identify citations of included studies. Specialists in the field of PBS in special education settings were also contacted and asked for copies of recent relevant research that had not yet been published. No copies of relevant research were provided, and therefore none were included in this systematic review. Given that the International Journal of Positive Behavioural Support (IJPBS) was not included in the major databases, hand searching of the IJPBS was conducted.

| Inclusion criteria for studies
Empirical studies were eligible for inclusion if they met the inclusion criteria relating to Population, Intervention, Comparator, Outcomes, Study design, and Setting (PICOSS): 1. This systematic review focused on the implementation of the PBS framework to support specifically CYP with either intellectual disability, or autism, or both, exploring adaptations of the framework for these CYP groups. CYP with multiple diagnoses were eligible if they had received at least one of the above diagnoses. CYP with intellectual disability (including genetic syndromes such as Down's Syndrome or Angelman Syndrome, that predominately manifest with a comorbid intellectual disability) and/or autism who were aged 3-25 years old (or this age group consisted at least 70% of the sample, or data were reported separately for this age group) were included if: (i) they had a formal diagnosis of an intellectual disability (or genetic syndrome) and/or autism according to DSM-5, ICD-11, or equivalent, or (ii) they were administratively defined as having intellectual disability by being enrolled in a special education setting (e.g., special school), or in a special education provision that exists within a mainstream school. Studies were included if they reported data separately for CYP with intellectual disability and/or autism, or if they did not report data separately for them but this group was at least 70% of the research sample. In the latter case, a cut-off of 70% was selected to ensure that any aggregate data included, reflected the majority of data from CYP with developmental disabilities.
2. The approach being implemented included the three components of PBS (Table 1). This multicomponent framework was developed to provide a defining and synthesis framework for the current review, and it was informed by an analysis of existing PBS definitions and the components and characteristics of the PBS approach (Allen et al., 2005Carr et al., 1999Carr et al., , 2002Dunlap et al., 2009;Gore et al., 2013;Kincaid, 2018;Kincaid et al., 2016;LaVigna & Willis, 2005;Morris & Horner, 2016). Papers describing characteristics of PBS to ensure contextual fit for special education settings (e.g., Simonsen et al., 2010) were also consulted. The use of this evidence-informed a priori defined multicomponent framework, when coding the PBS components and characteristics in studies, allowed for an over-estimation of their presence to be prevented. Approaches used were defined as PBS if: (i) it was clearly stated in the study that the intervention(s) in question had been implemented as part of a PBS framework used within that setting, and/or (ii) if they exhibited sufficient components that aligned with the PBS framework (Table 1). To be defined as PBS, interventions and supports had to include at least the primary characteristic of each PBS component. Therefore, the included studies had to report on the implementation of a proactive constructional approach utilised to establish consistent supportive contexts and environment.
The approach also had to exhibit understanding of the function(s) of behaviour(s) to inform behavioural support procedures (including skills teaching), and thus design functionally informed supports. In addition to the primary characteristic of each PBS component, interventions and supports could also have any number of the remaining characteristics (secondary characteristics).
3. The types of studies that were eligible for inclusion included experimental designs (single case experimental designs or group designs), non-experimental designs with a control group, and single group studies with baseline data without a control group. Any study reporting on quantitative data or qualitative data of relevance to the description and effectiveness of the PBS framework in special education settings delivered to CYP with developmental disabilities, the facilitators and barriers to the PBS framework implementation, and the views of stakeholders about PBS, was also eligible for inclusion. where classes take place during the school day.

Reported outcome was change in behaviours that
g) No restrictions were applied for the demographic characteristics (other than age and diagnosis) of the participants.
h) Only articles published in the English language were included in the systematic review. However, no restrictions were applied considering the country where the studies were conducted. There was also no restriction regarding publication date, thus the date range was not decided a priori.

| Exclusion criteria
Studies were not included if they did not meet the inclusion criteria, or if they were reviews, or opinion articles, or studies not reporting any quantitative or qualitative data, or studies not reporting on sufficient methodological information (such as studies that did not meet the required standards of demonstrating intervention effectiveness across at minimum three different data points: multiple baseline design across only two participants, settings, or behaviours, and AB case studies with only one participant, etc.).

| Review strategy
Electronic searches were conducted (see Figure 1: PRISMA Flow Diagram) and the identified records were exported to the EndNote reference software, which was used for the management of references.
After deduplication, title and abstract records were screened by the first author. The fourth author independently screened 1500 randomly selected records, verifying inclusion/exclusion for full text screening, which resulted in 1490 agreements (inter-rater reliability 99.33%, Cohen's k = 0.811). Following the title and abstract (stage 1) screening, full-text screening was conducted. The fourth author verified inclusion/exclusion eligibility for full texts following a two-step screening process. First, 15 records were screened resulting in 12 agreements (interrater reliability 80%, Cohen's k = 0.526). Following revision of the inclusion criteria guide used for full text screening, 10 additional records were screened by the second reviewer, which resulted in 9 agreements (interrater reliability 90%, Cohen's k = 0.800). Disagreements were discussed and resolved.

| Data synthesis
Both quantitative and qualitative data available were narratively synthesised. For RQ1, a narrative synthesis was conducted focusing on the description of the PBS components implemented in special education settings, and the effectiveness of interventions. A narrative synthesis was also undertaken for RQs 2 and 3, focusing on facilitators and barriers of PBS implementation and the experiences of stakeholders with PBS implementation, respectively.
F I G U R E 1 Preferred reporting items for systematic reviews and meta-analyses flow diagram. Thirty studies were eligible for addressing review question 1 (RQ1), of which 20 were also utilised to address review question 3 (RQ3), and zero studies were eligible for addressing review question 2 (RQ2) For conducting the quality appraisal of the identified studies, types of bias (Reichow et al., 2018). Selection bias refers to the systematic differences between the characteristics of the participants during baseline conditions such as the allocation procedures and the selection criteria of the participants. Performance bias refers to systematic differences between the participants in areas related to the intervention and factors other than the intervention, such as blinding of personnel and participants, and procedural fidelity. Detection bias refers to systematic differences between participants in procedures for determining outcomes such as blinding of outcome assessors, outcome reporting and reliability, and data sampling. 'Other' types of bias refer to additional sources of bias relating to the area of research that cannot be classified under the previous categories of bias. The choice of a suitable methodological tool selected to appraise the quality of the studies was dependent on the research design of each study. The CASP critical appraisal checklist for randomised control trials without the randomisation questions was used for the group study included, and the SCD RoB was used for the single case studies. The third quality indicator is related to the feasibility and costeffectiveness of the intervention. The fourth quality indicator is related to the long-term implementation of the intervention by typical providers of supports using available resources in typical contexts.
Additional measures of social validity included generalisation, maintenance, and explicit measurements of the intervention's social validity.

| Study selection
The electronic searches identified 17,437 records (see Figure 1: PRI-SMA Flow Diagram). After deduplication, 13,354 title and abstract records were screened by the first author. The fourth author independently screened 1500 randomly selected records, verifying inclusion/ exclusion for full text screening. Following the title and abstract (stage 1) screening, a total of 74 articles were selected from the database searches for full-text screening, and from these, 54 articles were excluded (see Figure 1 for exclusion reasons). The fourth author verified inclusion/exclusion eligibility for full texts. A total of 20 articles, identified from the database searches were included in the data extraction phase. Ten additional articles were identified after the backward and forward searches, and the hand search of the IJPBS.
Thirty articles were thus included in the systematic review, of which all of them were utilised to answer RQ1, 20 were utilised to answer RQ3, whereas no papers were eligible for answering RQ2. The characteristics of the included studies are summarised in Tables 2 and 3, and more details are provided in Tables S2 and S3.

| Quality appraisal
The risk of bias for participant selection and selective outcome reporting was low, while for procedural fidelity it was high in 50% (n = 15) of the studies. The blinding of participant and personnel was 'unsure' in all studies, and the blinding of outcome assessors was either unsure or high. When considered individually, no single-case design study had an overall high risk of bias. Even when a study showed high risk for a specific type of bias, it still had either low or unsure risk for most other types of bias. The literature on PBS implementation in special education settings appears, therefore, to include single-case studies of generally acceptable quality.
The one group study included in the review that used a pre and post intervention within-group design (Pitts et al., 2019), addressed a focused issue, the participants were all accounted for in the conclusion, and the application of the results to the local population or in a particular context (e.g., special education setting) was possible.  (Martens & Witt, 1982). TA and parent gave qualitative responses Teacher strongly agreed with 11 items (M = 5.67; range 4-6), TA with 12 items (M = 5.73; range 4-6), and parent with 13 items (M = 5.87; range 5-6). TA reported that student occasionally held blanket but relinquished without selfinjury, and parent was pleased with progress.

T A B L E 3 (Continued)
Authors (year; country) Methods Findings Dunlap et al. (1995) (USA) Six-point Likert scales (Dunlap, 1984) to assess from videotapes interest and happiness of student.
The intervention targeted meaningful outcomes (reduced challenging behaviours, increased on-task behaviour, increased happiness, and interest of student) Flynn and Lo (2016)  . Two teachers agreed or strongly agreed that intervention increased students' replacement behaviours (M = 3.7) and reduced challenging behaviours (M = 3.7), while one teacher neither agreed nor disagreed ('3'). Open-ended questions: Training and feedback were beneficial (Teachers 1 and 3), 'understanding behavioral function was useful' (Teacher 2), implementation of intervention produced positive outcomes.

| RQ1: Implementation of the PBS framework in special education
The studies included reported on interventions to support CYP with an intellectual disability or genetic conditions generally manifesting with intellectual disability (15%), or with autism (34%), or having multiple diagnoses including autism and/or intellectual disability (51%).
Publication dates of the studies were between 1993 and 2019.
Most were conducted in the USA (n = 27, 90%) with three studies conducted in the UK (n = 3, 10%).
Five studies (16.67%) were multiple baseline design studies, 19 studies (63.33%) used reversal designs, and one study was labelled as a quasi-longitudinal (quasi-experimental) descriptive case study that had an ABA design (Moore et al., 2009). Moreover, one study included a case series approach (Foran et al., 2015), one utilised an alternating treatments design (Lang et al., 2010), and one study utilised both reversal and multiple baseline design elements depending on the setting in which the intervention was implemented (Mueller & Nkosi, 2007). One study was described as using single-subject design with multicomponent interventions for three children (Paris et al., 2019), and one study included a single group study design with pre and post intervention measures (Pitts et al., 2019). Details of the PBS characteristics of the reported interventions can be found in Table 2.

| Description of interventions following a PBS framework
The    were reported in two studies (Clarke & Duda, 2019;Dunlap et al., 1995).

Systems
In total, 20 studies (66.67%) reported on appropriate organisational infrastructure, as part of a systems approach, to establish ecologically valid supports that exhibit contextual fit. Appropriate organisational infrastructure within the educational setting refers to the supports in place (e.g., training, coaching, performance assessments, supervision, team meetings, and materials such as adherence checklists) for assisting support providers to offer high quality of services. To achieve the required systems change, appropriate organisational infrastructure should be in place, which will increase capacity for and support of the change. These types of supports are related to suitable protocols and policies of the organisation/school that operationalise its vision, effective and efficient processes, resources, skills and roles of staff members, and the external supports for systems change to scale-up implementation capacity.
These promote buy-in of support providers and scale-up of the approach to a school-wide level.

Science and technologies
The second and third Science and Technologies secondary PBS characteristics, related to function-informed behaviour analytic interventions and data-driven processes, respectively, were reported across all studies (100%). The fourth characteristic, related to complementary evidence-based approaches, was reported in 17 (56.67%) of the studies included.

| PBS implementation processes
Two of the included studies (Clarke & Duda, 2019;Paris et al., 2019) explicitly identified the implemented approach as PBS by naming it as PBS, and described the procedures involved during the implementation of a PBS framework in two special education settings.
Clarke and Duda (2019)  iours that challenge and non-statistically significant decrease in others (Pitts et al., 2019); or decrease in behaviours that challenge and improvement in some alternative behaviours and not on other alternative behaviours for some participants (Bethune & Wood, 2013).
For the studies reporting percentages of reduction, four studies showed up to 20% decrease in behaviours that challenge, seven studies exhibited more than 20% decrease, and seven studies showed about 50% or more decrease. The group study (Pitts et al., 2019) showed statistically significant results for decrease in stereotypy with large effect sizes and in self-injurious behaviours with medium effect sizes. For adaptive behaviours, there was up to a 20% increase in four studies, in four studies there was more than 20% increase, and in four studies there was a 50% or more increase. Two studies ( assessed formally in two studies (Clarke & Duda, 2019;Dunlap et al., 1995), and informally in four more studies (13.33%) (Banda et al., 2009;Flynn & Lo, 2016;Foran et al., 2015; by taking into consideration anecdotal data. Quality of life was formally measured by assessing the happiness and interest of the student as rated by researchers after watching videotaped sessions (Dunlap et al., 1995), and by assessing positive affect of the student after observation of positive affect behaviours and after using quality of life indicators scores provided by peer buddies (Clarke & Duda, 2019). Engagement of students, including on task behaviour, measured in four studies (13.33%; Cihak & Gama, 2008;Clarke & Duda, 2019;Larkin et al., 2016;Pennington et al., 2012) and two studies (Bethune & Wood, 2013;Dunlap et al., 1995), respectively, was the measure most closely related to quality of life outcomes when an explicit measure of quality of life was not included.

| RQ2: Perceived barriers and facilitators to the PBS framework implementation
No studies reported data on facilitators and barriers for the implementation of a PBS framework in special education settings to support CYP with developmental disabilities.

| RQ3: Experience of PBS framework implementation
For answering RQ3, 20 studies (66.67%) out of the 30 studies included in the systematic review were utilised (  (Banda et al., 2009;Banda et al., 2012), one study collecting peer perceptions on quality of life indicators (Clarke & Duda, 2019), two studies (10%) utilising observational data to assess the positive affect of CYP with developmental disabilities (Clarke & Duda, 2019;Dunlap et al., 1995), and only one study (Lane et al., 2006) collecting student provided social validity data.
School staff views were in general favourable towards the interventions. Parental, peer, and student views indicated that they also

| DISCUSSION
The main aim of the current systematic review was to review, synthesise and critically appraise the available evidence on PBS implementation within special education settings to support CYP with developmental disabilities. A meta-analysis was not part of this systematic review. This is because, as outlined in the protocol, the most significant question that the review aimed to address was the description of the components of PBS that have been used in interventions in special education settings. A narrative synthesis was therefore more appropriate. Moreover, the review aimed to address the description of the facilitators and barriers to PBS implementation, and social validity/ experience of the interventions, which also required a narrative synthesis. In addition, the measurement of outcomes exhibited heterogeneity across the studies, preventing a meta-analytic synthesis.
The findings on the outcomes of PBS interventions suggest that they are generally effective in increasing adaptive behaviours and decreasing behaviours that challenge of CYP with developmental disabilities in special education settings. The lack of more robust high quality methodological designs employing a control group, such as randomised controlled trials, and the heterogeneity of outcome data across the included studies, make it difficult to assess precisely the effectiveness of PBS in special education settings. The majority of the included studies that employed a single-case study design were generally of acceptable quality. However, there were elements of singlecase study designs, such as blinding procedures and procedural fidelity, for which the risk of bias was on average unsure or high. These types of bias in studies of PBS implementation in special education settings contribute directly to reduced certainty in the available effectiveness evidence.
It is worth noting that indicators of quality of life, an outcome that PBS aims to promote, were only assessed formally in two studies and informally taken into consideration in four more studies. Considering the emphasis that the PBS framework places on quality-of-life outcomes, the lack of evidence of reporting of these outcomes is disappointing. All the more so given that in one of the first attempts to systematically synthesise the literature base on PBS more than 20 years ago, Carr et al. (1999)

| Limitations
In terms of limitations, these can be identified at the review level and at the level of individual studies included in the systematic review. At the review level, one article (Peterson et al., 2002) considered during the abstract screening stage as eligible for proceeding to full text screening could not be retrieved. Therefore, we cannot be certain if this study would have been included in the systematic review and impact on the current results or not. Moreover, the identification of studies describing the PBS framework was made possible by evaluating them against the components and characteristics of the framework, because not all studies explicitly named as PBS the approach they were following. This may have led to studies not being identified by the search strategy, especially in cases such as when reporting in the articles of the PBS framework primary components was inadequate.
At the study level, there were very limited group studies eligible for inclusion (one group study), and no randomised controlled trials.
The small sample of this group study, the lack of a control group, and the absence of blinding or absence of reporting on blinding procedures decreased the quality of the evidence that the study provided for this systematic review. Moreover, the available data for addressing RQ3 were brief social validity data (provided by quantitative scales and sometimes supplemented with brief comments on open-ended questions) most of which were associated with unsure risk of bias but acceptable level of reporting information on social validity related indicators, and no detailed qualitative studies were identified that could be utilised to address RQ2. Regarding the single-case studies, there were certain types of bias in the risk of bias assessment, such as blinding of outcome assessors (that were consistently rated unsure for all studies), and procedural fidelity (which was high risk of bias in half of the studies included), that contributed to decreased methodological quality and thus certainty in the evidence from single-case studies.
More group and qualitative studies are needed for PBS implementation in special education settings, which will describe additional to Tier 3 supports, that may be implemented in the setting, and will follow more robust designs. Additional consideration should also be given when designing and reporting single-case studies regarding blinding procedures and procedural fidelity. These design and methodology considerations were most commonly vulnerable to risk of bias in the current review.

| Implications and future directions
The findings of this systematic review suggest that PBS holds the potential for decreasing behaviours that challenge and increasing adaptive behaviours of CYP with developmental disabilities in special education settings. These findings extend the research base on PBS and are relevant for practitioners and administrators aiming to promote an increased quality of life for CYP with developmental disabilities, who can benefit from implementing PBS.
Although there is a wider literature on PBS in special and alternative settings, the evidence base on PBS implementation in special education settings to support specifically CYP with developmental disabilities remains insufficient for the target context and population. Future research should explore further the implementation of school-wide PBS in special education settings by using high quality methodological designs, so that studies exhibit low risk of bias (especially for blinding procedures and procedural fidelity), and include more robust group designs, such as randomised controlled trials. Areas that future research should focus  40-year long history of implementation, and yet a lack of clarity on how PBS is defined and reported in the evidence base, and issues related to its social validity remain present. These limitations should be addressed by future research.
Overall, the results of this systematic review suggest that PBS can be successfully implemented in special education settings to support CYP with developmental disabilities. Future research should explore further the implementation of PBS in a school-wide level in special education settings, by employing high quality group designs and more comprehensive qualitative methods.

ACKNOWLEDGEMENTS
The systematic review reported in this article is conducted as part of a doctoral dissertation for the fulfilment of the requirements of a PhD funded by the University of Warwick and the Ambitious about Autism Charity.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article.