A quantitative evaluation of a regional Positive Behavioural Support workforce development approach

Background: Research suggests that providing staff with input in relation to Positive Behavioural Support (PBS) can have beneficial outcomes. Much of this research, however, fails to take account of systemic issues and does not include a control group. Method: We used a non-randomised, controlled group design to evaluate accredited PBS programmes, delivered as part of a systemic, regional and workforce development approach. We compared outcomes of those attending the programmes ( n = 240) with a control group ( n = 54), pre-and post-intervention and at 3-months follow-up. Results: The programme and its wider impact were rated positively. Significant intervention effects were found for staff practice and retention, but not for staff knowledge and attributions, or behaviours that challenge and quality of life of those being supported. Conclusions: The results are discussed in the context of the study limitations and restrictions resulting from the Covid-19 pandemic.


| INTRODUCTION
Positive Behavioural Support (PBS) is a well-established, evidencebased approach that aims to increase the quality of life of people with an intellectual disability, particularly those who present with behaviours that challenge (CB).As a functional approach, PBS views such behaviour as serving a purpose for the person and provides a framework for identifying, understanding and addressing the range of complex factors that can contribute to the expression of CB (Gore et al., 2013).PBS has its basis in applied behaviour analysis and, as the name suggests, is underpinned by a positive, person-centred value base.These principles, such as respect for the individual and the use of non-punitive approaches, have been identified as key to good-quality support by people with an intellectual disability and their families (McKenzie et al., 2017(McKenzie et al., , 2018)).
Despite this, some people with an intellectual disability continue to be subject to abusive and restrictive practices, as exemplified by the Winterbourne View scandal.As a result, the 'Transforming Care' agenda (NHS England, 2014) identified the need to develop high-quality, community-based support.A key requirement for the success of this policy is to ensure that staff have appropriate values, knowledge and competence to provide this support.
There is a body of evidence demonstrating that providing staff with input in relation to PBS can have beneficial outcomes.A review by MacDonald and McGill (2013) of 14 studies found reductions in levels of CB, improvements in staff knowledge and in behaviour support plans, changes in causal attributions about, and emotional reactions to, CB.The duration of the input varied across the reviewed studies from a single day of training, to 10 h per week over a year.
The format also varied, from predominantly didactic teaching, to role play and practice-based assignments.Most of the studies used a repeated measures design, with only one being noted by the authors as having a control group.
Subsequent research, with staff working in community settings, has also shown positive results, again using different types of input.Wills et al. (2013) reported on an evaluation of input comprising five core PBS modules for half a day per week for 5 weeks.
They found that social care staff showed significant improvements in attributions about 'controllability', that is, the extent to which the CB was under the control of the person with an intellectual disability, PBS knowledge, helping behaviours and optimism about reducing CB. Rose et al. (2014) also reported significant improvements in care staff attributions about 'controllability' and attitudes, following 1 day of training, which was 'embedded in a PBS approach' (p39).Significant improvements in knowledge, attributions and in some attitude domains were also found by Wardale et al. (2014).This followed a 4-day programme, delivered over 4-6 weeks, with practice-related activities and ongoing support from the trainers.Stocks and Slater (2016) evaluated a six and a halfday PBS informed course.The authors found increases in staff scores on self-efficacy and positive outcome expectations in relation to their understanding, functional assessment and management of CB, and creating and implementing a PBS plan.None of these studies included a control group.
More recent research has, however, not always found significant changes as a result of PBS input to staff.Hassiotis et al. (2018) reported no significant differences in outcomes, including in CB and use of psychotropic medication, between staff who had received PBS input and those who had not.In this case, the input comprised three 2-day workshops, delivered over a 15-week period, the content of which was supported by a manual.
The staff also received support, on at least a monthly basis, by a mentor for a year.
Similarly, MacDonald et al. (2018) found changes in only some of the areas that they measured.In this case, the authors used a nonrandomised control group design to evaluate a year-long university accredited programme, which consisted of an initial 2-day workshop followed by eight 1-day workshops delivered at 8-week intervals.Participants were service managers and each session had associated tasks to embed the learning in practice.The authors found a significant reduction in CB, but no changes in the quality of life of those being supported.There were also no significant changes on any of the staff measures.These included staff knowledge and attributions, implementation of active support, quality of staff support and engagement with the people they supported.
Many researchers in this area are increasingly highlighting the importance of systemic factors in the success or otherwise of providing PBS input to staff.These include level of managerial support and commitment, the extent of work-based support, levels of staff turnover, workload demands and communication between team members (Bosco et al., 2019;Hassiotis et al., 2018;MacDonald et al., 2018;McKenzie et al., 2005), The need for PBS approaches to take into account the wider systems and contexts within which staff, organisations and those being supported operate, henceforth referred to as 'systemic factors', has been highlighted over many years (Allen et al., 2013;Denne et al., 2015;McGill et al., 2018;Olivier-Pijpers et al., 2019).
PBS applied as a systemic framework for broader service delivery and change appears to be more frequent in countries other than the United Kingdom (UK), such as the United States.Here, it is generally referred to as Positive Behaviour Interventions and Support and is commonly introduced on a school-wide basis to enhance behavioural and academic outcomes for children and young people (e.g., Grasley-Boy et al., 2020).
By contrast, PBS input to staff in the United Kingdom, who support people with an intellectual disability, has largely been delivered based on a staff training model, most commonly taking the form of short training courses that have a focus on the individual learner.
More recently, however, researchers have begun to take account of systemic factors and to explore alternative models of delivering input.
For example, McGill et al. (2018) developed standards, in collaboration with services, that addressed areas that have been found to influence CB.In addition to areas such as activities and skill development, health, communication and relationships, some organisational factors were included under the areas of 'service management', and 'wider organisation'.These included supervision, leadership, management, payment and induction of staff.The researchers then worked with services over an 11-month period to help them achieve the standards as part of their intervention.The extent to which this was successful was included as an outcome measure.The authors found that the average percentage of standards that were achieved, post-intervention, was 80.3%, for service management and 68% for wider organisation targets.
Similarly, MacDonald et al. ( 2018) developed a cascade model of training that took place over 12 months, whereby it was expected that managers would disseminate their learning to their staff teams.
As noted above, the only significant change was a decrease in CB of those being supported.The authors suggest that this may be because levels of service user engagement and of staff assistance were already high prior to the intervention, allowing little scope for change.
The present paper reports on one aspect of a region-wide, wholesystem, PBS approach, which was underpinned by a workforce development (WFD) model.In contrast to staff training approaches, WFD recognises and takes account of the different contexts in which staff work and services are delivered, as well as the organisational, structural, attitudinal and other factors that influence the individual learner (Denne et al., 2015;Jacobs & Hawley, 2009).The model was influenced by the results of a scoping exercise across the North East and Cumbria (McNall et al., 2016) that identified a range of areas that could act as barriers to the successful implementation of PBS.These included limited or no emphasis on PBS in commissioning or monitoring requirements, and limited provision of standardised, transferable, input that was assessed or accredited.This resulting WFD model involved engaging stakeholders, including commissioners, care organisations, NHS staff, families and people with an intellectual disability, to develop a systemic PBS approach (see McKenzie, McNall, et al., 2021 for details).
One strand of this approach was a collaboration between NHS and university staff to develop and deliver three accredited PBS programmes to social care staff in the region who supported people with an intellectual disability and/or autism.An independent evaluation of the programme was also commissioned.
An initial qualitative evaluation of the impact of the wider PBS approach, based on feedback from a range of stakeholders (McKenzie, Martin, et al., 2020) found that it was seen as having resulted in many benefits for people with an intellectual disability, individual staff, teams and organisations.The reported benefits included improvements in staff practice and knowledge; the adoption of better staff recruitment, supervision and performance management approaches; a reduction in staff turn-over and sickness and improved quality of life of the people being supported by the organisations.
The present study reports on a further quantitative evaluation of the PBS programme and aims to evaluate whether the PBS programme, in the context of a regional, system-wide WFD approach to PBS, resulted in benefits that were consistent with those reported in the qualitative evaluation.

| Ethics statement
Ethical approval for the study was obtained from the first author's university ethics committee.All those taking part provided informed consent.

| Design
The study adopted a non-randomised, controlled group design, comparing those who had attended the PBS programme (hereafter referred to as the PBS group), with a control group of those who had not, on a range of outcome measures.Data were collected at three time points: prior to the start of the PBS programme (baseline), at the end of the programme (follow-up 1) and 3 months after the programme ended (follow-up 2).The evaluation was structured based on the updated Learning Evaluation model outlined by Kirkpatrick and Kirkpatrick (2006), as depicted in Table 1.This table also illustrates how each component was measured, who completed the measures, and at which time points.

| Participants
Participants were from organisations across the North East and Cumbria regions in the United Kingdom, which provided community-based support to people with an intellectual disability and/or autism, and who had committed to supporting the PBS WFD approach.Participants were included in the PBS group if they had attended the PBS programme.As the number of places available to each organisation was limited, to allow equity of access, the control group participants were those who were unable to be offered a place on the programmes during the period of the evaluation.
In total, 299 individuals participated, of whom 240 were in the PBS group and 54 were in the control group.The group status of five participants was missing.Both groups comprised staff working in social care/day care organisations and roles ranged from service managers/organisational leads to front-line support staff.Of the total sample, 241 participants reported that they provided direct support on a regular basis to a person with an intellectual disability and/or autism (194 = PBS group, 46 = control group, 1 = missing data).Table 2 provides the demographic information for the two staff groups and the people they supported.

| Intervention
The intervention was the PBS programmes.The programmes used a blended learning approach, with participants receiving 3 days of face-to-face teaching for each module that they undertook, access to online materials and activities relating to each topic, and The first two awards comprised three modules and the third comprised two modules.All modules lasted for 3 months.Table 3 out The provision of support and supervision was based on a cascade model.Three WFD managers, who were experienced PBS practitioners, provided input into the programmes, as well as individual supervision and support to staff in their local areas across the region.The aim of this was to help the participants to apply their learning in practice.The amount and type of support varied according to the needs of the individual, but could include modelling good practice, providing guidance in relation to areas such as the functional analysis of behaviour, and data collection and analysis.In addition, those who had completed the 'Leading PBS' award provided support and supervision to those undertaking the 'Facilitating PBS in teams' award.These staff then provided this, in turn, to the front-line staff who were completing the 'Award of Competence' in PBS.There were both academic and competency-based assessments included as a part of the programmes to address staff learning about the theoretical basis of PBS as well as their implementation of this in practice (see McKenzie, Martin, et al., 2020 for further details about the programmes).The team delivering the PBS programmes was the same throughout the period of the evaluation.

| Measures
The measures used for each level of the Learning Evaluation model are outlined below and summarised in Table 1.Measures that were T A B L E 1 The four levels of learning evaluation outlined by Kirkpatrick and Kirkpatrick (2006)  developed for the project (see Table 1) do not have information available about their psychometric properties.Measures were completed individually by participants online, unless otherwise specified.

| 'Reaction'
The PBS group rated the quality of the teaching, supervision and overall programme on a 6-point scale (very poor = 1 and very good = 6), with a higher score indicating better perceived quality.

| Self-efficacy
This was measured using the 'Challenging Behaviour Self-Efficacy Scale' (Hastings & Brown, 2002).This is rated on a 7-point scale and the response to each question is added to give a total score (range 5-35), with a higher score indicating greater feelings of self-efficacy in relation to managing CB.This scale has been found to have good internal reliability (α = .81:Hutchinson et al., 2014).
T A B L E 3 Overview of the programme structure and content

| Attributions
These were measured using the 'Challenging Behaviour Attributions Scale' (Hastings, 1997).Participants rate the likelihood of different casual explanations for CB (learned behaviour, biomedical, emotional, environmental and stimulation) from very likely to very unlikely.The average score for each category is calculated (range À2 to 2).A score less than 0 indicates that the person thinks the particular category is unlikely to be the cause of the behaviour, while a score above 0 indicates the person thinks it is applicable.

| PBS in practice
The practice of a subgroup of participants who supported a person who displayed CB was assessed in relation to that person's behaviour support plan.This was an individual assessment conducted with the participant.The assessment was structured in line with a periodic service review model (La Vigna et al., 1994) and was adapted from that used by McKenzie et al. (2002).The assessed categories included: a description of the nature and identified function of the target behaviour(s); how this was recorded and reviewed; the proactive and reactive strategies that were used to support the person; any functionally equivalent behaviours that the person was taught; any other factors that were taken into account, for example, the person's physical health; how the plan was reviewed and the involvement of the person being supported in the plan.Possible scores ranged between 0 and 28, with a higher score indicating better practice.
Each participant was also rated on the extent to which their responses evidenced that the behaviour support plan overall: reflected positive approaches; was based on observable and measurable behaviour and had responses related to, and which were consistent with, the identified function of, discrete target behaviours.Participants could score between 5 and 25 points, with a higher score indicating a more robust plan.The assessment was conducted by an experienced clinical psychologist, who was blind to the group status of the participants.
2.9 | 'Results' 2.9.1 | Ratings of impact Participants in the PBS group rated the impact of the PBS programme on a range of factors (see Table 4), from extremely positive = 5 to extremely negative = 1.

| Participants' views on CB
Participants reported on the number of topographies of CB and the frequency with which each type of CB displayed by the main person they supported had occurred in the previous month.No independent observations or interobserver agreement were gathered for these outcome measures.

| Quality of life of those being supported
This was evaluated in two ways.In both cases, the participants reported on the quality of life of the main person that they supported.The 'Guernsey Community Participation and Leisure Assessment-Revised' (GCPLAR: Baker et al., 2016) measures frequency of community use and other activities.It provides domain scores and a total score (possible range 0-230).A higher score indicates greater engagement in activities.In line with previous researchers (see Bowring et al., 2020), the 'social activities' subscale was analysed separately to provide an indication of quality of life, specifically in relation to community participation.The GCPLAR is reported to have acceptable psychometric properties (Baker et al., 2016).
In addition, a measure was designed for the project, which included questions that related to the 'Five Accomplishments' (O'Brien, 1992) and the categories identified in the 'Seven Keys to Citizenship' (Duffy, 2006).The areas covered in relation to the former were as follows: Respect, Relationships, Competence, Choice and Community Presence.The areas addressed in the latter were as follows: Direction/ Purpose, Freedom, Money, Home, Help, Life and Love.The measure has 48 items, which are scored 'yes' or 'no' (range 0-48), with higher scores indicating a greater quality of life.
Participants rate their stress-related thoughts and feelings on a 5-point scale (never to very often).Total scores can range from 0 to 40, with a higher score indicating greater stress.Participants were asked to respond to the questions in relation to their work.

| Psychological well-being
This was measured using the 7-item version of the Warwick-Edinburgh Mental Wellbeing Scale (Stewart-Brown et al., 2009).This has acceptable psychometric properties (e.g., Ng Fat et al., 2017).Participants rate positively worded items on a 5-point scale, with a higher score indicating better mental well-being.

| Staff retention
Participants reported at follow-up whether they remained in the same job, were actively seeking a new job or had left their previous job.
Participants were also asked to provide some basic demographic information, to indicate whether they provided direct support on a regular basis to a person with an intellectual disability and/or autism and, if so, to provide some information about their current systems of data collection in relation to CB.
Different participants completed different aspects of the evaluation.Table 5 illustrates the number of participants in each group, completing each type of measure at each time point.

T A B L E 4
The mean score and SD for the PBS and control groups for the variables explored in each of the levels outlined Kirkpatrick and Kirkpatrick's (2006)  The participating organisations identified potential participants from staff who would be enrolling in the PBS programme during the period of the evaluation and a control group of staff who would not.Both groups were contacted by email and/or phone, provided with information about the evaluation and invited to take part.Those who were interested in participating were emailed a link to the online evaluation.
This provided more detailed information about the study.Participants provided consent by clicking on a link in the survey.They were then asked to provide an identifying code before completing the relevant online measures.Once participants had completed the relevant measures, they were redirected to a separate survey where they were asked for their name and code.This code was used to match responses over the three time points of the evaluation.
A subgroup of those who provided direct support were contacted and invited to take part in the assessment relating to staff practice.
These were conducted by a clinical psychologist/researcher with extensive experience of working in intellectual disability services.
Participants were sent the links to the online survey and arrangements were made to conduct the practice assessments, where applicable, at the two subsequent follow-up points.

| Data analysis
The impact of the PBS intervention on most of the intervention study outcomes was tested using a series of multi-level models (MLMs).For each outcome, an MLM was fit in which time was included as a level-1 predictor and group (intervention versus control) as a level-2 predictor.
Random intercepts and random slopes for time were included in the model.To test whether the intervention led to a relative increase in the outcomes over time, a group-by-time interaction was specified.The statistical significance of this parameter was used to indicate whether the intervention had an effect on a given outcome.Models were estimated using maximum likelihood estimation using the lme4 package (Bates et al., 2014) in R statistical software (R Core Team, 2017).
Multi-level multiple imputation with m = 100 imputations was used to deal with missing data in the outcome variables.The full mixed effects model approach described by Grund et al. (2016) was used.This method can be described as an example of a joint modelling approach (in contrast to fully conditional specification approaches) in which a single model is used for imputing all incomplete variables simultaneously.In our application of the approach, we used a separate imputation model (and associated imputed datasets) for each MLM.In each case, the imputation model used was identical to the analysis model.Between 10,000 and 20,000 burn-in iterations were used depending on convergence.Convergence was checked using the potential scale reduction statistic, which we required to be 1.05.Parameter estimates and standard errors were then pooled using Rubin's rules (Rubin, 2004).This method provided unbiased parameter estimates provided that data are missing at random (Rubin, 1976).All multi-level imputation steps were implemented using the pan and mitml packages in R statistical software (Grund et al., 2016;Grund, Lüdtke, & Robitzsch, 2018;Grund, Robitzsch, & Luedtke, 2018).

| Participant group differences
There were no significant differences between the PBS and control groups at baseline, in respect of demographic and other variables (see Table 6).

| Outcome measures
The results are presented according to the Learning Evaluation model.
Table 4 illustrates the range, mean scores and standard deviation (SD) for the PBS and control groups for the variables explored in each of the levels, with the exception of staff turn-over, the results of which are reported separately.
Table 7 presents the results for the MLMs.In all these cases, the results are for the group-by-time interaction in the MLM (i.e., comparing the PBS and control group scores over time).Ideally, from a theoretical standpoint to avoid potential spurious detection of group-by-time interactions, it would have been ideal to include the random effects for time for all analyses (Barr, 2013).However, it was necessary to remove the random effect for time due to convergence issues in some of the imputed datasets, for all but the following outcome variables: self-efficacy, behaviour support plan response ratings and mean number of CB.For these outcomes, results must be interpreted more cautiously.
T A B L E 5 The number of participants in each group, completing each type of measure at each time point    The mean scores indicate that the participants reported that the PBS programme had a positive impact on all of the factors identified at both follow-up 1 and follow-up 2.

| Work-related stress and psychological wellbeing
There was no statistically significant difference between the two groups over time on stress or well-being.The PBS group rated the PBS programme as having a positive impact on work-related stress.

| Behaviours that challenge
At baseline, only 47.2% and 33.3% of the PBS and control groups, respectively, had recording methods for CB that were updated on at least a daily basis and summarised on at least a weekly basis.This suggests that the baseline data on CB are likely to be somewhat inaccurate for some participants.No significant intervention effects were found in relation to the number or frequency of CB.

T A B L E 7
The results from the MLM for the group-by-time interaction

| Quality of life
No significant intervention effects were found in the reported quality of life of those being supported, as measured by the quality of life measure designed for the study, the GCPLAR social activities scores or GCPLAR total scores.

| Staff retention
Table 8 illustrates the number and percentage of participants who remained in their job, were actively seeking another job or left their job at follow-up.A series of chi-square tests indicated a significant association between group and leaving work/actively seeking another job, with fewer of those participants in the control group remaining in their job, as compared with the PBS group.This result was found at follow-up 1 (χ 2 = 11.4,df = 1, p = .001)and follow-up 2 (χ 2 = 16.7,df = 1, p < .001).

| DISCUSSION
The evaluation of the PBS was structured using the four levels of learning evaluation outlined by Kirkpatrick and Kirkpatrick (2006).The results in respect of each area are discussed below.
In terms of 'reaction' to the PBS programme, the teaching, supervision and overall experience were rated highly by the PBS group participants.The success of the programme, in this respect, may be because it combined important elements that have been found to be effective in WFD, that is, 'classroom'-based teaching, practice leadership and workplace coaching (Bosco et al., 2019;van Oorsouw et al., 2009).
Three areas were explored in relation to 'learning'-self-efficacy, attributions and knowledge about the best ways to manage CB.No significant intervention effect was found for any of these areas in the present study.Some previous research has found positive changes in attributions (see MacDonald et al., 2018) and self-efficacy following PBS training (e.g., Stocks & Slater, 2016), although this latter study did not include a control group.In general, previous studies have shown PBS training to result in increased staff knowledge, although many of these did not include a control group (MacDonald & McGill, 2013).
More recent research by MacDonald et al. (2018), that did include a control group, found that PBS input appears to have a differential impact on staff, depending on their role.They found that managers experienced increases in their knowledge, compared to the control group, while the staff they managed did not.It may be that the overall results of the current evaluation masks any differences in knowledge between groups with different roles.Unfortunately, the small number of staff in the control group, precluded further analysis of this data by role.
It has been highlighted that, while improvements in areas such as knowledge and attributions are positive, they will not necessarily lead to changes in staff behaviour and in the support provided (MacDonald et al., 2018).Our evaluation found significant intervention effects in the outcome measures relating to staff practice, with increases over time in the intervention group compared with the control group.
There have only been a few studies that have explored the impact of PBS input on staff practice.MacDonald et al. ( 2018) used a periodic service review approach to measure staff practice but did not find any significant change.McGill et al. (2018) worked with staff to develop, implement and monitor practice according to a set of standards.The researchers found significant changes in practice as measured by achieving the standards.Neither of these studies, however, included a control group for the aspect of the study that explored changes in practice.
The wider 'results' of the PBS programme were evaluated in a number of different ways.The impact of the programme, as rated by the PBS group, indicated that they perceived it to have had a positive impact on those they supported, on themselves as individuals, and on their organisations.These results were consistent with those highlighted in a qualitative evaluation of the programme (McKenzie, Martin, et al., 2020).
In relation to CB, no significant intervention effects were found.
Previous research has found that PBS input has resulted in improvements in CB (e.g., MacDonald et al., 2018;MacDonald & McGill, 2013); however, Hassiotis et al. (2018) found no significant effect.Our own results may have been influenced by the fact that fewer than half of those in both groups had robust methods for recording CB at baseline.This suggests that the majority of the baseline data was likely to be inaccurate to some extent, meaning that Left job 4 (2.9) 9 (4.9) 9 (20.9) 12 (34.3) Actively seeking another job 9 (6.5) 3 (1.6) 4 (9.3) 3 (8.6) Note: The number of those who left their job at follow-up 1 is also included in the total figure for those who have left their job by follow-up 2. Abbreviation: PBS, Positive Behavioural Support.
ongoing practice-based support and supervision.The online materials were completed individually by participants, while face-toface teaching took place in a group setting.The programmes led to one of three accredited qualifications.Organisational leads and managers completed a Postgraduate Certificate or Advanced Diploma in 'Leading PBS', team leaders and other senior staff completed a Certificate in 'Facilitating PBS in teams' and frontline support staff completed an 'Award of competence in PBS'.
lines the structure and topic areas in relation to each of the awards.
This was measured by asking participants to provide an open-ended written responses to a question, which asked them what they thought were the best ways to successfully manage CB.Participants' written responses were coded in accordance with the PBS Competence Framework (Positive Behavioural Support Coalition UK, 2015).For example, a response that included the phrase 'Put in proactive strategies to ensure an individual has a meaningful and good quality of life' would receive the codes of PR, indicating that proactive strategies had been identified and VL to indicate the response was values-led.Each relevant code received a score of 1.If the code was repeated in the response, only 1 point was given.Fifty percent of responses were coded by two raters to help ensure consistency, and differences in coding were resolved through discussion.The possible range of scores was between 0 and 32.
effects were found in relation to attributions, self-efficacy scores or responses to CB.
support plans Significant intervention effects were found, with the PBS group having a greater increase in scores for both the behaviour support plan practice score and response ratings.

T A B L E 8
The number and percentage of participants, who remained in their job, were actively seeking another job or left their job at followups as applied to the PBS programme evaluation T A B L E 2 Demographic information for the PBS and control groups Abbreviation: PBS, Positive Behavioural Support; SD, standard deviation.
QualificationShared topics across all three qualifications Shared topics across 'Leading PBS' and 'Facilitating PBS in teams'Award of competence in PBS As for the 'Facilitating PBS in teams' award, but with no third module Abbreviations: FBA, Functional-Based Assessment; PBS, Positive Behavioural Support.
learning evaluation model Abbreviations: GCPLAR, Guernsey Community Participation and Leisure Assessment-Revised; PBS, Positive Behavioural Support; SD, standard deviation.