What Do We Know about Bullying in Schoolchildren with Disabilities? A Systematic Review of Recent Work

Bullying is a serious problem that particularly affects schoolchildren with disabilities. However, studies in this group have been carried out on smaller cohorts and the results obtained are, therefore, less representative and sometimes inconclusive. The purpose of this paper is to perform a systematic review of the work carried out in recent years, including the analysis of several variables related to the sample, the methodology applied and the type of bullying. The guidelines set down by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement were followed in three phases. The total sample consisted of 55 children who fulfilled the inclusion criteria. The results reveal that half the studies were conducted in cohorts of less than 250 schoolchildren and drew no distinctions between the different types of disabilities. Furthermore, there is no consensus regarding the methodology used, and no specialized instruments were used. Hardly any specific interventions have been performed into the type of bullying investigated, in which victimization is the predominant mode. We concluded that there is an urgent need to increase the number of studies, including a larger number of individuals and using specialized instruments, in order to obtain more solid results. Such studies will allow us to create specific prevention and intervention programs to address the bullying of schoolchildren with disabilities.


Introduction
Bullying, which is a type of violence among peers, is expressed by physical, psychological or relational abuse, in which the bully's actions are clearly intentional. Bullies take advantage of the unequal power status over their victims, who suffer this abuse over an extended period of time [1]. It is a complex, immoral phenomenon which affects children all over the world and has a prevalence of around 36% (see meta-analysis by [2]). This social interaction in which schoolchildren are involved can have academic, social and personal consequences, regardless of whether they adopt the role of victims, bullies, bully-victims or witnesses [3]. For these reasons, the number of studies into this antisocial behavior is on the rise [4].
The World Health Organization estimates in its World Report on Disability that more than 5% of children aged 0 to 14 have a disability [5]. Autism spectrum disorders, developmental delays, conduct disorders and learning difficulties are among the problems mentioned [5]. Schoolchildren with disabilities are more vulnerable to an imbalance of power, which could be a risk factor for bullying [6]; however, studies into this group of people are few compared to those into students without disabilities [7]. There could be several reasons for this: firstly, providing them with the necessary support to cover their educational needs is given priority, in order to meet academic targets and have a better chance of inclusion [8]; secondly, the fact that there are fewer researchers specialized in bullying in the field of disability [9]; thirdly, the idiosyncratic characteristics of the students' sample, the participants' age and the resources used, among other variables. Based on the revision carried out by [14], our first hypothesis is that there will be an absence of studies and that these will be mainly produced in developed countries. Following the revision on children with ASD by [14,17], the second hypothesis will be that the sample sizes will be small and mainly from primary or secondary grades. The third hypothesis, based on the work of [10], is that there will be major differences between the instruments used for the different studies.

Research Strategy
The current systematic review has followed the criteria pointed by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [19] through an ad hoc Excel control list. The Population Intervention Comparison Outcome (PICO) strategy, specially adapted for this topic, was used to find the question formulation. Consequently, the three main investigators established three phases: phase 1 (initial exploratory research), phase 2 (systematic research) and phase 3 (manual research).
In phase 1, exploratory research was conducted, by browsing "bullying disab *" on Web of Science (WOS) and SCOPUS without defining the year. This asterisk is a truncation that will incorporate all possible combinations of the word "disab". The result was a total of 966 articles; "systematic review" was then included in the search, which produced 30 reviews, some of which were not related to the topic, did not answer the question or were not recent enough to account for new changes, such as the inclusion of Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the new intervention programs. From the first phase, the protocol was designed and the inclusion and exclusion criteria were established to increase the accuracy of the methodology [20].
WOS from Thomson Reuters and SCOPUS from Elsevier were used as scientific databases in the systematic research phase, since they had been used before to carry out studies into similar subjects [16,21]. The terminology used for both databases were "bullying" and "disab *" adding "and" to find research relating both concepts. The asterisk symbol "*" was used to include words containing the same prefix, such as "disability," "disabilities," "disabled" or "disabling." The research period established covered June to December 2019.

Inclusion and Exclusion Criteria
The inclusion criteria used for the articles in the final sample were: (a) studies focused on the educational field, (b) studies which included a school age population (childhood, primary and secondary education), (c) studies in which the disability was the main variable of the bullying, regardless of whether the methodology was qualitative, quantitative or mixed (d) accurate studies of publishing criteria, taking into account only articles published in relevant magazines, subjected to a process of double-blind peer review, (e) studies published after 2011.
The exclusion criteria were: (a) studies published before 2011, (b) studies focused on different conditions such as dyslexia, dyscalculia, attention deficit hyperactivity disorder, etc., or those articles focused on ASD, since a systematic review of this has recently been carried out [17], (c) studies with no previous sample in the educative field or with a sample of over 18-year-old students (d) theoretical papers and systematic reviews, (e) research based on bullying of non-disabled students, (f) articles published as chapters of books, conference abstracts, notes, etc.

Data Compilation
In phase 2, each investigator conducted blind systematic research, following the established protocol. They found 746 articles published after 2011, of which 353 were found in WOS and 393 in SCOPUS, with 260 articles were duplicated in both databases, which, consequently, had to be excluded. After revising the titles and summaries of the remaining Sustainability 2021, 13, 416 4 of 18 486 articles, 282 were excluded, since the topic was not related to coexistence or experiences at school, or the main aim, despite bearing some relation, was not about inclusion. In addition, chapters of books, conference abstracts, notes, etc. were ruled out, which left only 175 articles, which were analyzed under the established inclusion and exclusion criteria. In total, 63 of these 175 articles were rejected, since they were based on students with normal development, and 28 articles more since they were based on students over 18 years old who attended university or were in employment. A further 27 articles were discarded, 18 of which were theoretical articles and 9 systematic reviews. Finally, two more articles were excluded since they were instrumental or intervention proposals which had not been carried out. Thus, the final sample was composed of 55 articles. After a blind research review process, 98.18% coincidence was found among the authors and discrepancies were dealt with using the Delphi Method.
In phase 3, the researchers used Google Scholar to find the authors who had written more than four articles in phase 2. The objective was to learn about their career and check if there were any studies that could be classified as grey literature [20]. However, no more articles were included in the final sample (see Figure 1).
In phase 2, each investigator conducted blind systematic research, following the established protocol. They found 746 articles published after 2011, of which 353 were found in WOS and 393 in SCOPUS, with 260 articles were duplicated in both databases, which, consequently, had to be excluded. After revising the titles and summaries of the remaining 486 articles, 282 were excluded, since the topic was not related to coexistence or experiences at school, or the main aim, despite bearing some relation, was not about inclusion. In addition, chapters of books, conference abstracts, notes, etc. were ruled out, which left only 175 articles, which were analyzed under the established inclusion and exclusion criteria. In total, 63 of these 175 articles were rejected, since they were based on students with normal development, and 28 articles more since they were based on students over 18 years old who attended university or were in employment. A further 27 articles were discarded, 18 of which were theoretical articles and 9 systematic reviews. Finally, two more articles were excluded since they were instrumental or intervention proposals which had not been carried out. Thus, the final sample was composed of 55 articles. After a blind research review process, 98.18% coincidence was found among the authors and discrepancies were dealt with using the Delphi Method.
In phase 3, the researchers used Google Scholar to find the authors who had written more than four articles in phase 2. The objective was to learn about their career and check if there were any studies that could be classified as grey literature [20]. However, no more articles were included in the final sample (see Figure 1).

Data Encoding
The following categories were encoded to extract the results: (a) year of publication, (b) country where the study took place, (c) sample size: here, we used only the number of students with a disability, since some studies with large samples also included them, (d) age, by identifying the average age and/or the students' age range, (e) educational stage, by indicating whether the subject belonged to special education, (f) type of disability, in which category "general" has been used when any type of disability was included, (g) type of school violence, in which intervention studies were included, and (h) instruments used, without specifying the name of the instrument.

Articles Quality Assessment
It was not possible to make use of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) method due to the scope of the study. Instead, the quality of the articles was guaranteed using only WOS and SCOPUS, which include high-impact quality journals and the only selection of scientific articles. Chapters of books were excluded (this was one of the inclusion criteria) for not being subject to peer review. At the discussion stage, possible biases are analyzed as limitations of the study.

Results
Fifty-five articles were included in the final sample for the current systematic review according to the previously established inclusion criteria (see Appendix A).

Year and Country of the Studies
A total of 76.37% (n = 42) of the articles in the final sample were published in the last five years (see Figure 2), while 18.18% (n = 10) of the works were carried out in continental Europe. The studies were from the UK, Ireland, France, Sweden, Greece and Slovenia, and one study included 11 countries, 10 of them European. A total of 14.55% (n = 8) of the papers in the final sample were carried out in the following countries: Taiwan, Hong Kong, Israel and Saudi Arabia; 7.27% (n = 4) of the papers were from Australia. Finally, 3.64% (n = 2) of the papers were from Canada and 1.82% (n = 1) from both Colombia and Nigeria.
age, by identifying the average age and/or the students' age range, (e) educational stage, by indicating whether the subject belonged to special education, (f) type of disability, in which category "general" has been used when any type of disability was included, (g) type of school violence, in which intervention studies were included, and (h) instruments used, without specifying the name of the instrument.

Articles Quality Assessment
It was not possible to make use of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) method due to the scope of the study. Instead, the quality of the articles was guaranteed using only WOS and SCOPUS, which include highimpact quality journals and the only selection of scientific articles. Chapters of books were excluded (this was one of the inclusion criteria) for not being subject to peer review. At the discussion stage, possible biases are analyzed as limitations of the study.

Results
Fifty-five articles were included in the final sample for the current systematic review according to the previously established inclusion criteria (see Appendix A).

Year and Country of the Studies
A total of 76.37% (n = 42) of the articles in the final sample were published in the last five years (see Figure 2), while 18.18% (n = 10) of the works were carried out in continental Europe. The studies were from the UK, Ireland, France, Sweden, Greece and Slovenia, and one study included 11 countries, 10 of them European. A total of 14.55% (n = 8) of the papers in the final sample were carried out in the following countries: Taiwan, Hong Kong, Israel and Saudi Arabia; 7.27% (n = 4) of the papers were from Australia. Finally, 3.64% (n = 2) of the papers were from Canada and 1.82% (n = 1) from both Colombia and Nigeria.

Characteristics of the Sample
According to the sample size, 47.27% (n = 26) of the studies-almost half of the selected papers-contained samples of less than 250 students with a disability. Of this percentage, 21.82% (n = 12) had a sample that was less than 100, while 12.73% (n = 7) had samples of between 250 and 1000 students with a disability, and over 40% (n = 22) of the Articles per year

Characteristics of the Sample
According to the sample size, 47.27% (n = 26) of the studies-almost half of the selected papers-contained samples of less than 250 students with a disability. Of this percentage, 21.82% (n = 12) had a sample that was less than 100, while 12.73% (n = 7) had samples of between 250 and 1000 students with a disability, and over 40% (n = 22) of the papers had samples over 1000. The participants' age was not classified, and this varied widely. In 47.27% (n= 26) of the studies, the samples of boys were larger, while in 20% (n = 11), the sample size of girls was larger, and the sex of the participants with disabilities was not specified in 30.91% (n = 17). Table 1 shows the data corresponding to the educational stage. It should be noted that only 16.36% (n = 9) were in special education. In relation to the type of disability, 60% (n = 33) of the papers used the category "general". 18.18% (n = 10) were carried out in intellectual disability, 3.64% (n = 2) in physical disability and the same percentage in sensory disability. Moreover, 10.91% (n = 6) were conducted in disability together with chronic illness; finally, 3.64% (n = 2) dealt with students with double characteristics (disability and high capacity).

Methodology Used
Scales and questionnaires were the most widely used instruments, with both used in 50.91% (n = 28) of the papers analyzed, and in 21.82% (n = 12), scales were the only instrument used, while 30.91% (n = 17) used questionnaires. Interviews were used in 16.36% (n = 9), tests in 12.73% (n = 7) and self-reports in 7.27% (n = 4). More detailed information on the specific type of instrument according to the study is shown in Appendix A.

Type of Intimidation
A total of 45.45% (n = 25) of the research papers focused exclusively on victimization, while 30.91% (n = 17) included both perpetration and victimization. The papers on bullying and cyberbullying accounted for 7.27% (n = 4), while 3.64% (n = 2) focused solely on online bullying and 1.82% (n = 1) on dating violence. Finally, 7.27% (n = 4) of the papers focused on interventions, one of which was aimed at students with intellectual disabilities and the other three at general disabilities. They all worked with aggressors and victims. In one of the programs, Cognitive-Behavioral Therapy was used, while in two, different competences were studied, such as empathy, communication skills, problem solving, friendship skills, emotional regulation and, in the last paper, resilience and "Stop Bullying Now", which is a program that includes 12 webisodes.

Discussion and Conclusions
The objective of this study was to find out how research into bullying in disabled studies has been carried out over recent years. For this purpose, a systematic review of papers published between 2011 and 2019, both inclusive, was carried out following the PRISMA criteria and using inclusion criteria that helped guarantee the quality of the selected documents. In this way, a total of 55 articles from recent years was found.
The first hypothesis was confirmed, although more papers were found in this review than in the review by [14]: the research found so far is still scarce and has mainly been carried out in recent years, with more than three-quarters in the last five years. There is a greater awareness of the inclusion of students with disabilities in mainstream schools [22] and it is apparent that disability is a risk variable for bullying [6]. However, the studies are not plentiful, despite the fact that papers on bullying in non-disabled students are on the rise [4]. Additionally, many of the papers were produced in developed countries, such as the review by [14]; in fact, three-quarters of the studies were published in the USA or Europe. However, research in underdeveloped countries is scarce, even though it has been shown that bullying is a problem that also affects children in these countries [23], where it is possible to find higher rates of disability [24]. However, the lack of economic resources in Sustainability 2021, 13, 416 7 of 18 these countries may be causing their governments to invest less in research and not permit projects into what bullying is like in their schools.
The second hypothesis was also confirmed. Although the number of samples has increased over the last years (for example, the systematic revision carried out by [14], in which the samples were of less than 250 participants in almost three out of four papers), the current revision includes little over half of the studies performed. However, the samples are still too small to extract convincing conclusions, in fact; one out of five articles have samples with less than 100 students. These results are similar to those found in the systematic review by [17] carried out in schoolchildren with ASD. This could be a reflection of the difficulty in finding large samples of disabled people and, sometimes, having to resort to incidental criteria samples and convenience samples, rather than probabilistic sampling. This hypothesis is supported by the that most of the articles found had a wide range of ages, which would have enabled them to obtain a larger number of participants. Moreover, this idea is backed up by the characteristics of the sample, with three out of five considering all types of disabilities, although it could also refer to the effort made to obtain a larger number of samples. Although one out of five studies into intellectual, physical or sensorial disability were found, this is still insufficient [24,25]. More work, therefore, needs to be done in order to prevent attacks on students that have physical disabilities by their schoolmates [26].
As in studies of traditional bullying, most of the research was conducted in primary and secondary grades [4]. Only nine articles focused on special education schools, while eight of these compared special education with mainstream education, so this remains an under-researched context, despite the high percentages of involvement in such schools [27]. Only two studies were carried out in children's education, so this may be a novel scenario to study unjustified aggression, which can occur at this age, as in other papers published on non-disabled students [28]. On the other hand, in relation to the type of bullying in this age group, most research has been carried out on victimization and few studies have focused on interventions, despite the fact that scientific evidence has shown how this is a particularly vulnerable group [7].
Similarly, the third hypothesis has been confirmed. A wide variety of instruments is available to study this type of violence, although these are not specialized, as was shown in the meta-analysis by [10]. In this study, scales and questionnaires were the most common instruments, although they often inevitably resort to information from third parties such as teachers or relatives, especially with more serious disabilities [29]. Otherwise, interviews were used less, in spite of being useful for some papers in which students' idiosyncratic features made it impossible to use scales and questionnaires [29].
This work presents certain limitations that must be taken into consideration when interpreting the results, such as those produced by the differences between the educational systems in the different countries, which means that schoolchildren of the same age may be attending primary or secondary school, depending on the country. Additionally, some papers included the same sample of students but were published in different articles, so the number of original papers published was, in fact, lower. Likewise, not specifying the type of disability in most of the studies and the search for a larger number of participants leads, in many cases, to a wide heterogeneity of the sample and the use of different instruments, some of which were not validated, which may affect the results. Finally, since it is a systematic review, it is important to consider the bias risk, which is high in this case, where there are specific instruments available for use with schoolchildren with disabilities. In addition, there is considerable publication bias, as a great number of the articles found appeared in the same scientific journals.
However, this up-to-date systematic review provides us with valuable information about how bullying studies have been carried out in students with a disability over the last few years and could inspire specialized researchers to take up future lines of research. One of the lines which would help us advance in our knowledge of this problem is to design specialized instruments for students with different degrees of disability. These may include pictures or pictograms that make the questions easier for students with more severe disabilities to understand. The use of the same instrument can allow comparison between studies, such as between different countries, through cross-cultural studies that would serve to determine the influence of cultural and social variables on schoolchildren with disabilities who have suffered bullying. Another line of study would be on how other forms of violence such as cyberbullying or dating violence affect students with disabilities. The review has highlighted the fact that few studies exist, despite the fact that young people, including schoolchildren with disabilities, increasingly use digital devices to interact with each other. A third line would be to carry out studies into a specific type of disability, since the review has shown that most of these works are general and include all types of disabilities. In this way, with more specific works, it would be possible to know if a certain type of disability leads to a greater risk of bullying. Finally, although this work has implications mainly for research, since the main objective was to learn about the nature of studies into bullying in schoolchildren with disabilities, its findings also stress the importance of continuing to work with specific programs for the prevention and intervention of bullying in students with disabilities and especially due to the victimization that many of these schoolchildren suffer. In short, inclusion and diversity in the classroom is a common goal to achieve a more sustainable society and education.

Conflicts of Interest:
The authors declare no conflict of interest. The funding body had no role in the design of the study, in the collection, analyses or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.