The Association between Deliberate Self-Harm and School Bullying Victimization and the Mediating Effect of Depressive Symptoms and Self-Stigma: A Systematic Review

Background Identifying deliberate self-harm in the young and its relationship with bullying victimization is an important public health issue. Methods A systematic review was performed to explore evidence of the association between deliberate self-harm and school bullying victimization in young people, as well as the mediating effect of depressive symptoms and self-stigma on this association. An advanced search in the following electronic databases was conducted in January 2018: PubMed/Medline; CINAHL; PsycINFO; PsycARTICLES; Science Direct; Scopus, and Cochrane Library. Studies that fulfilled the inclusion criteria were further assessed for their methodological integrity. The Norwegian Knowledge Centre for Health Services tool was applied for cross-sectional studies and the Critical Appraisal Skills Programme instrument for the cohort studies. Only empirical quantitative studies published in the English language in peer reviewed journals during the last decade (2007-2018) aimed at exploring the association between deliberate self-harm and school bullying victimization in community-based schoolchildren with a mean age of under 20 years were included. Results The reviewed cross-sectional and cohort studies (22) revealed a positive association between school bullying victimization and deliberate self-harm, including nonsuicidal self-injury, which remained statistically significant when controlled for the main confounders. The mediating role of depressive symptoms in the association between deliberate self-harm and school bullying victimization was confirmed. A dose-response effect was shown in the association between nonsuicidal self-injury and school bullying victimization, whilst the mediating effect of depressive symptoms needs to be further explored. No studies were found directly exploring the mediating effect of self-stigma in the association between deliberate self-harm and bullying victimization. Conclusion Targeted interventions aimed at eliminating victimization behaviours within the school context are therefore proposed, as well as interventions to promote healthy parenting styles for the parents of schoolchildren. Moreover, school healthcare professionals should screen students involved in bullying for self-injury, and vice versa.


Introduction
The construct of deliberate self-harm predating the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) encompassed a broad spectrum of behaviours that could damage the body in every possible nonfatal way, regardless of intention to die [1]. Moreover, the term "deliberate self-harm (DSH)" is often used interchangeably with "self-harm" and "self-injury" [1]. In contrast, nonsuicidal self-injury (NSSI), recently included in the DSM-5, is defined as an act of intentional self-directed harm to the surface of one's body without conscious intention to die 2 BioMed Research International [2]. Suicidal behaviour, including suicidal thoughts and/or suicide attempts, covers actions aimed at deliberately ending one's own life [2].
Based on the above, deliberate self-harm may be viewed as a wide spectrum of behaviours, characterised by different levels of suicide intent and a variety of motives [2][3][4]. Some of the self-harming behaviours include ingestion of a substance in excess or a nondigestible object; jumping from a height; and self-cutting or self-burning [5]. The main groups of motives behind these actions may be revenge against others and manipulation, intention to die, seeking attention, or escape from intolerable and agonising conditions [3]. Thus, the intention of deliberate self-harm ranges from no intention to end one's own life (NSSI) to strong intention to end one's own life (suicidal attempt/completed suicide), while it is also possible for multiple motives to coexist, for example, wanting to send a message and at the same time to obtain relief from suffering [2].
The presence of mental and behavioural problems has been linked with engagement in deliberate self-harm [2,[17][18][19][20][21][22][23]. Additionally, increased prevalence of deliberate selfharm as a socially deviant behaviour has also been identified in nonclinical populations [24]. Among others, dysfunctional relations with peers, peer rejection, bullying, and victimization have been associated with deliberate self-harm in the young [25]. Bullying victimization is defined as the experience of repetitive, aggressive behaviour towards an individual by her/his peers, such as unprovoked attacks, rejection and social isolation, humiliation and ridicule, malicious rumours, and name-calling, leading to severe distress in the victim, whilst the victim is unable to defend him/herself [26]. Bullying constitutes a public health problem, as approximately 32% of schoolchildren across 38 countries report experiences of peer victimization and subsequent emotional distress, whilst 10% of them may develop self-harming behaviour [6]. Different types of bullying victimization have been identified, such as verbal, relational, physical, or cyber, while the most common environment in which it takes place is school [1].
Bullying victimization during childhood or adolescence is a risk factor of poor physical health, mental health disorders, deliberate self-harm, and suicidal symptoms at any age [16,[27][28][29][30][31]. Bullying victims experience severe emotional distress associated with the psychological and physical violence they are subjected to, as well as social marginalization and decreased status among peers [27,[32][33][34]. A negative selfconcept and decreased self-esteem, combined with perceived weakness and rejection by peers, have been described [35]. This process may be linked with the development of selfstigma. Developmental psychology supports that maltreatment and victimization during childhood and adolescence may be a key factor associated with self-stigma [36]. Selfstigma takes place when individuals internalize the negative public attitudes and stereotypes about their status and further experience a wide range of adverse costs related to these prejudices [36]. Thus, self-stigma evokes intense feelings of embarrassment, worthlessness, and self-blame influencing one's self-esteem. Decreased self-esteem, associated with selfstigma, has been associated with both deliberate self-harm and depressive symptoms [37].
Furthermore, self-stigma among those living with mental problems is also reported [38][39][40][41]. Data show that adolescents who present symptoms of borderline personality disorder, depression, psychosis, and conduct disorder are more likely to experience self-stigma [18,[40][41][42][43]. Additionally, these groups more frequently report both school bullying victimization and self-harming behaviour [44]. Based on this, one may hypothesise that bullying victimization, self-stigma, and deliberate self-harm may be present simultaneously. However, the link among these variables has not been explored adequately, either in clinical or in nonclinical populations [45,46]. For instance, a cross-sectional study in 224 adolescents aged 13-17 years in rural Uganda provided evidence on the link among self-stigma, bullying victimization, severe depressive symptoms, and suicidality [45]. However, this study took place in adolescents with HIV; thus further investigation in the general population is needed. Additionally, other types of self-harming behaviours were not included in this study.
Nevertheless, preexistence of mental health and behavioural problems constitutes a risk factor for peer rejection and bullying victimization, mainly due to the social stigma related to mental illness [42][43][44]. At the same time, bullying victimization may have an additional negative influence on the clinical outcome of the symptoms experienced by those living with mental health problems or who have behavioural problems [2,47]. Thus, the causality between bullying and mental health problems, such as depressive symptoms, self-harming behaviour, low self-esteem, and negative selfconcept, seems to be bidirectional [43,48,49]. Consequently, it would be interesting to explore the mediating effect not only of depressive symptoms in the association between selfharming behaviours and bullying victimization but also of "mental health problem self-stigma" and/or "bullying rejection self-stigma" [34,37,40,41,50]. For instance, the study by Pantelic et al. [46] revealed the complex nature of selfstigma development, suggesting the existence of multilevel mechanisms prominent to it. In particular, Pantelic et al. [46] found that although self-stigma was associated with depressive symptoms, bullying victimization was not directly associated with internalized stigma, implying multiple risk pathways from personal problems to psychological distress; violence within homes, communities, and schools; and selfstigma development.
Although several studies have explored the association between bullying victimization and deliberate self-harm or suicidality [28,[51][52][53][54][55], only a limited number of them have systematically reviewed the articles exploring these associations [28,51,52]. In particular, two focused specifically on suicidal behaviour [28,51] and one solely on NSSI. However, better understanding of the complexity of the association between deliberate self-harm and bullying victimization is 3 needed, since an overlap between all types of self-harming behaviour and motives, especially in the young, has been reported [3]. Moreover, further research is needed regarding possible mediators in the association between deliberate selfharm and bullying victimization [28,51,52]. Comprehensive knowledge of the factors mediating the association of bullying victimization with self-harming behaviour in the young is crucial to further inform current health policies, preventive screening, and educational strategies, as well as developing targeted treatment programmes [56]. Thus, it is important to further explore and better understand the correlates of selfharming behaviours and their outcomes [4,57]. The present review aims to present data regarding potential difference between those who engage in deliberate self-harm with and without clearly reporting intention to die, in relation to school bullying exposure and possible mediators, that is, depressive symptoms and self-stigma. Possible differences may further inform the phenomenology of these two selfharming groups.
The theoretical framework underpinning the present review is derived from the social-ecological theory [58]. Such a framework supports the hypothesis that a link exists between the school environment, risk of involvement in bullying, and the adverse impact of bullying on students, namely, self-harming behaviour, while highlighting the importance of related factors, such as depressive symptoms and self-stigma [58].
For the purpose of the present review, the term "deliberate self-harm" defines every nonfatal act performed intentionally by an individual with the aim of causing physical or psychological harm to her/himself, irrespective of suicidal intent [5]. As a result, the terms self-injury and suicidal behaviour are not used interchangeably in the present study [2]. Additionally, for the purpose of the present study selfstigma is defined as the situation in which a person believes the negative stereotypes related to her/his mental health problems and/or accepts the negative attitudes related to her/his bullying victimization and rejection [41].
School bullying victimization is defined as victimization of any type, that is, verbal, physical, and relational victimization within the school environment or among students [59].

Aim.
The aim of the present systematic review was to assess the evidence of the association between deliberate selfharm, nonsuicidal self-injury, and school bullying victimization in young people, with special focus on the following research questions: Is the effect size of the association between "deliberate self-harm and school bullying victimization" different from that of the association between "nonsuicidal self-injury and school bullying victimization"? Is the association between deliberate self-harm, nonsuicidal self-injury, and school bullying victimization in the young mediated by depressive symptoms? Is the association between deliberate self-harm, nonsuicidal self-injury, and school bullying victimization mediated by self-stigma?

Study Design.
A systematic review of the literature was conducted. The main features of a systematic review comprise precisely reported research questions; implementation and description of a robust and reproducible methodology; a systematic search of scientific data in accordance with precisely stated criteria; assessment of the methodological quality of the reviewed studies; critical and systematic demonstration of the features; and main results of the reviewed studies [60,61]. The above steps were applied in the present study.

Search
Strategy. An advanced search in electronic databases was conducted between the 6th and the 25th of September, 2016, and was repeated in January 18th, 2018. A search in PubMed/Medline, Cumulative Index of Nursing and Allied Health Literature (CINAHL), PsycINFO, Psy-cARTICLES, Science Direct, Scopus, and Cochrane Library (clinical trials) was undertaken using the following keywords alone and in combination, in line with Medical Subject Headings: (self-harm or self-injury or deliberate or self-mutilation or self-abuse or self-injurious or suicidal or self-harming or suicide or "life-threatening behavior") AND (depressive or depression or stress or self-harm or self-stigma) AND (bullying or victimization or "peer aggression" or intimidation) AND (young or youth or juvenile or adolescent or students). The term "self-harm" was used twice aiming to identify additional studies in which "self-harm" was studied as a mediator in the association between suicidality or selfthreatening behaviour and bullying victimization, thus providing indirectly information about the association between self-harm and bullying victimization.
The search was conducted independently by two authors (MK and ES) and validated by a specialist librarian (KM). The latter independently rescreened to assess the rigour of the screening procedure. Studies finally included in (vii) Exploration of school-context bullying victimization, based on the wording of the measurement tool used for assessment of the bullying experience ( i.e., inclusion of the words "school" or "student"). An example is "how often were you left out of things, excluded, or ignored in school?" or "we can say a student is a victim of bullying when another student or a group of peers says malicious or hurtful things to him."

Inclusion and Exclusion
The following studies were excluded after screening for eligibility (n=38) ( Figure 1): (i) Qualitative studies (n=2); theoretical studies (n=0); reviews, meta-analyses, and metasyntheses as systematic reviews focus on primary studies (n=3); protocols; educational studies and programmes (n=0); and monographs, guidelines, or national policy recommendations or guidelines, due to their limited cultural context audience (n=0) (ii) Studies with no direct measurement of the association between deliberate self-harm and school bullying victimization (n=3); studies which did not explain how the variables (self-harm/bullying) were measured or defined (n=4); studies in which the researchers did not provide information about the type of victimization experienced (n=3); studies regarding abuse (i.e., verbal, physical, or sexual abuse) (n= 2); studies conducted in a nonschool context (e.g., bullying among brothers and sisters) (n=1); and studies exploring attitudes on self-harm and bullying (n=1) or retrospective studies in adults (n=2), as well as studies on cyber selfharm (n=1) (iii) Studies exploring indirect experiences of suicidal behaviours or self-harm related behaviours, such as witnessing deliberate self-harming behaviours of others, viewing suicide-related material, underage motorbike riding, or alcohol consumption (n=3); the reason behind this was that these studies lack direct assessment of self-harm. Additionally, studies investigating the association between school bullying victimization and direct suicidal behaviour, whilst not including deliberate self-harming, were also excluded (n=3).
(iv) Studies in vulnerable populations, including prisoners and sex offenders (n=1); people with physical and/or mental illness comorbidities (n=1); homosexuals (n=1); and mental health clinical populations (e.g., bipolar patients or children with Attention BioMed Research International 5 Deficit Hyperactivity Disorder (n=3)), studies including only males or only females (n=2), or studies investigating the same sample as another study that was already included (n=2).

Selection Strategy.
The selection strategy of the included studies was based on the PRISMA procedure [61] ( Figure 1). The combined search identified 474 articles. After removing duplicates (298) and screening titles and abstracts for relevance to the aim and research questions of the review, 59 articles remained. Two researchers (MK and ES) independently screened both titles and abstracts of all retrieved articles for eligibility and resolved disagreements by consensus. One paper was added after a citation search; thus the full texts of 59 articles were studied thoroughly in relation to additional criteria pertaining methods, setting, and target population. Next, the full texts of the selected articles were screened for eligibility by all four researchers and any disagreements were resolved by consensus. When all the inclusion and exclusion criteria had been considered, 22 articles remained and comprised the sample of the present review. The 22 studies were coded for the variables explored herein. Each included study was independently reviewed by two of the four researchers in accordance with the variables presented in Tables 1, 2, and 3, thus corresponding to the measures employed in the aims of the review, that is, context of the bullying (school/ students), target population (schoolchildren), methods (study design, definitions, measures, and tools employed), mediators in the association between school bullying victimization and deliberate selfharm (depressive symptoms and self-stigma), outcomes measured, and important results. A specially designed extraction sheet was used for data collection purposes, whilst the reason for excluding a study was documented.
2.6. Quality Assessment. The 22 studies that fulfilled the inclusion criteria were further assessed for their methodological integrity. The Norwegian Knowledge Centre for Health Services (NOKC) tool was used for assessing the methodological integrity of cross-sectional studies [62][63][64] and the Critical Appraisal Skills Programme (CASP) instrument for cohort studies [65]. Both tools provide an outline for assessment and critical appraisal of risk of bias related to confounding factors, participant selection, measurement, and data analysis. Based on a literature review on the subject of confounders, we identified age, gender, suicidal behaviour/previous suicide attempts, substance abuse, depressive and anxiety symptoms, self-esteem/self-concept, and impulsivity as critical confounders for inclusion to ensure the relevance of the study and reduce the risk of bias.
As the focus was on the association between deliberate self-harm and exposure to school bullying victimization, the NOKC checklist for cross-sectional studies was applied for analytic cross-sectional comparative and noncomparative studies. In cases where two or more groups of students were compared with regard to exposure to school bullying victimization and its association with deliberate self-harm, additional criteria adopted from the NOKC checklist for cohort studies were applied, as previously reported in the literature [62,63]. Moreover, one further criterion regarding the ethical integrity of the included studies has been also added herein [64]. Thus, a modified version of the NOKC assessment tool including 13 questions has been used, addressing both comparative and noncomparative crosssectional analytic study designs. These 13 NOKC checklist criteria are reflected in the following questions. (1): Is the target population of the study clearly defined? (2): Is the sampling method appropriate and is the study sample representative of the target population? (3): Is the nonexposed group selected from the same population as the exposed group? (4): Are the nonexposed and exposed groups comparable regarding the main background variables? (5): Is the degree and way, in which the respondents who consented to participate differ from those who did not, described? (6): Is the response rate satisfactory? (7): Is the method for data collection consistent? (8): Are the measures of the main variable reliable and valid? (9): Are the methods of statistical analysis suitable? (10): Are both exposure and outcome measured consistently in both groups (exposure/ nonexposure)? (11): Is the assessment of the outcome blind to whether participants were exposed or not? (12): Are the main confounders included in the study design? (13): Are the ethical issues properly addressed by the researchers?
The assessment of the rigour of the cohort studies was based on the CASP tool [65] (CASP 2014). This tool includes 12 criteria, organized in three groups of questions. First Section (Validity of Study Results): Is the study addressing a clearly focused topic? Has the cohort been recruited in a valid way? Is the exposure to the risk factors precisely measured to eliminate bias? Is the study outcome precisely measured to eliminate bias? Are all important confounding factors identified in the study design and further included in the data analysis? Is the follow-up procedure comprehensive and long enough in duration? Second Section (Study Results): What are the results of the study? How accurate are the study results? Do you think that they are important? Third Section (Implementation of the Results in the Study Population): Can the results be implemented in the study population? Are the study results in accordance with other available data? What are the implications for clinical practice based on the study results?
Studies that met over 50% of NOKC or CASP criteria were deemed to be of moderate quality, while those that met 70% or more of the criteria were classified as high quality. Studies that met 50% or less of these criteria were categorised as low quality (Tables 3 and 4) [62].

Data
Analysis. Data were analysed in four steps: firstly, all researchers corroborated the definitions of deliberate selfharm, school bullying victimization, and internalized stigma to be used herein, whilst potential theoretical frameworks linked with the aforementioned definitions were then discussed; the next step involved the identification of studies that fulfilled the inclusion criteria and the description of their main methodological characteristics and relevant tables, as well as the assessment of their methodological quality; the  More than 90% of the children exposed to SBV did not engage in SI. Those who did were significantly more likely to (a) have a family member who attempted/completed suicide, ( However, there was no main effect of the SBV on the prediction of NSSI when controlled for depressive symptoms.

DS:
BDI Physical/verbal/relational bullying victimization: Ten yes/no questions about the three different types of victimization (e.g., "others pushed, hit or kicked you"; "others called you names", "others spread rumours about you" & their frequency (occasional/ repetitive)

NSSI:
Deliberate and direct injury of one's own body tissue without suicidal intent, such as scratching, cutting, hitting, and burning oneself.

NSSI:
The NSSI subscale of the

DSH:
A broad definition of self-injurious behaviour including suicidal intent.

DSH:
Two items from the Youth Self-report assessing students 'suicidal ideation and self-injury history were combined into one composite variable.
Each statement referred to the past 6 months and had the response alternatives: True and False: (1) "I deliberately try to hurt or kill myself;" or (2) "I think about killing myself.

Overt/relational bullying victimization:
Four items about the frequency of face-to-face bullying victimization (being picked on, made fun of, called names, and hit or pushed by other students) from the Bullying and Victimization subscales of the University of Illinois Aggression Scales.

DSH:
Cutting and biting arms, pulling out clumps of hair, banging head against walls, and attempted suicide by strangulation.

DSH:
A single question to the mothers of the twins if they had deliberately hurt themselves.  self, or scrapped skin to draw blood without intention to die). Frequency was also assessed.

DS: CES-D scale
Overt/relational bullying victimization: A sociometric peer nomination procedure was used. Each adolescent was presented with a roster of all classmates and asked to identify those who they thought were victims of (i) overt/physical victimization ("Who gets beat up, picked on, or teased by bullies?", "Who gets threatened or hit by others, or has mean things said to them?") & (ii) relational victimization ("Who gets left out of activities or ignored by others because one of their friends is mad at them?", "Who gets gossiped about or has rumours told about them behind their back?")

DSH:
One item asking whether participants had hurt themselves on purpose (e.g., cutting, burning, or bruising)

Bullying victimization:
One single question about the frequency of different types of victimization: "During the last 30 days, how often have you been a victim of fun or teased by another student in a hurtful way or excluded from friends or activities?" Response options were 'never' , 'once or twice' , 'about once a week' , 'several times a week' , 'every day' . Adolescents were asked to identify peers who were targets of the two forms of peer victimization in the school environment. Peer nomination items were used to index overt victimization (e.g., "Who gets threatened or hit by others or has mean things said to them?") and relational victimization (i.e., "Who gets gossiped about or has rumours told about them behind their back?" Janter et al.

NSSI:
Intentional self-inflicted damage to the surface of one's body without conscious suicidal intent, such as cutting or carving the skin, self-biting, or burning skin.

NSSI:
a single question clearly distinguishing NSSI from SB, asking about the intention ("without the intention to kill yourself").

DSH:
An act with a nonfatal outcome in which an individual deliberately did one or more of the following: initiated behaviour (for example, self-cutting or jumping from a height), by which they intended to cause self-harm; ingested a substance in excess of the prescribed or generally recognized therapeutic dose; ingested a recreational or illicit drug that was regarded as self-harm; or ingested a noningestible substance or object.

DSH:
"Have you ever tried to. . .according to the definition used". Additionally, the participants were asked to describe the method(s) used to harm themselves in their own words.

SB:
No direct measurement DS:

Bullying victimization:
A single question about being bullied at school: "Have you ever been bullied at school?" including the timing of the event (more than a year ago or within the past year).

NSSI:
"Have you ever physically hurt yourself on purpose?" If the students responded that they had deliberately hurt themselves in the past, they had to report the way they did it.

SB:
follow-up questions about the self-injurious behaviour, including a question asking if they had hurt themselves with intent to die.

Bullying victimization:
A single yes/no question: "During the past 12 months, has someone bullied you on school property?".           Fisheretal.2012 final stage encompassed the organization of the results of the included studies in relation to the present research questions and the interpretation of data regarding the association between school bullying victimization and deliberate selfharm. Special focus was placed on the mediating effect of depressive symptoms and self-stigma and on the differences between deliberate self-harm and nonsuicidal self-injury behaviour, in accordance with the research questions. Ten out of the twenty-two studies reviewed used a definition that concretely described NSSI [66][67][68][69][70][71][72][73][74][75]. The remaining twelve studies did not focus on intention to die [76][77][78][79][80][81][82][83][84][85][86][87]. The definitions and measurements of the main variables of the reviewed studies are presented in Table 2. Twelve studies assessed lifetime prevalence of deliberate self-harm and nonsuicidal self-injury, while the others explored these experiences during the three to 12 months previous to the study [66,68,71,[73][74][75][76][77][78]87].

Is the Effect Size of the "Deliberate Self-Harm and School
Bullying Victimization" Association Different from That of the "Nonsuicidal Self-Injury and School Bullying Victimization?"  [73]. Moreover, the severity of exposure to school bullying victimization, either verbal or physical bullying, predicted the frequency of NSSI, confirming a prospective, dose-response association. These data were extracted from studies identified as being of high methodological quality [62].

Prospective Association between Deliberate Self-Harm
and School Bullying Victimization. Exposure to frequent school bullying victimization before the age of 12 years (i.e., 7 to 10 years) was associated with increased risk for deliberate self-harm at the age of 12 years and in late adolescence [82][83][84].  [76] reported that deliberate self-harm was statistically significantly higher in school bullying victims compared to students uninvolved in bullying, while depressive symptoms only partially explain that difference. With regard to prospective studies, it was shown that exposure to frequent school bullying victimization before the age of 12 years was a risk factor for deliberate self-harming behaviours in preadolescence for those who also reported depressive symptoms [82]. Data also revealed that adolescents who reported at least one incidence of deliberate self-harming behaviour were more likely to report experiences of all types of bullying victimization, while depressive symptoms only partially mediated this relationship [72]. With regard to NSSI, the cross-sectional study by Claes et al. [69] confirmed that exposure to school bullying victimization predicted NSSI, which was partially mediated by depressive symptoms [69]. Heilbron & Prinstein [68] in their cohort study found no main effect of school bullying victimization on the prediction of NSSI when controlled for depressive symptoms. In contrast, Giletta et al. [73] found a prospective dose-response effect between the frequency of exposure to school bullying victimization and the severity of NSSI, independent of depressive symptoms. These studies were of high quality.

Is
There a Mediating Effect of Self-Stigma on the Association between Self-Injury, Nonsuicidal Self-Injurious Behaviour, and School Bullying Victimization? With regard to this research question, none of the included studies directly explored this association.

Discussion
All reviewed studies confirmed a positive association of deliberate self-harm and NSSI with school bullying victimization, even when controlled for the main confounders (school grade, gender, depressive symptoms, suicidal behaviour, impulsivity, substance use, family history, and abuse). The mediating role of depressive symptoms in the association between deliberate self-harm and school bullying victimization was confirmed [66,69,72,76,82]. In contrast, the role of depressive symptoms in the association between NSSI and school bullying victimization was equivocal. Three studies measured this relationship. While one revealed a prospective, dose-response effect independent of depressive symptoms, the other two, one cross-sectional and one cohort, identified depressive symptoms as a mediator in this association [68,73]. None of the reviewed studies directly explored the mediating effect of self-stigma on the association of selfinjury, NSSI, and school bullying victimization.
The strength of the present results stems from the fact that the reviewed studies were of satisfactory methodological quality (moderate and high quality). Despite this, a lack of consensus regarding the definitions used for deliberate selfharm in the studies reviewed was observed. This is also reflected in the variety of terms applied, which also do not differentiate between self-harm with or without intention to die. The distinction between NSSI and deliberate self-harm is challenging, highlighting the need for future studies to clarify differences in risk profiles [4,57]. A clear definition of NSSI is proposed for future studies aiming to achieve consensus regarding study variables, thus facilitating data comparisons among research groups, and leading to more accurate comprehension of both the context and the outcome of relevant behaviour. However, the way an individual hurts her/himself does not always reflect the intention to die, evidence of which is clearly reported in the CASE study [5]. Thus, it is vital that future studies clearly address the motive(s) behind self-injurious behaviour and not only the means by which it is inflicted [4,57]. Additionally, clinicians need to assess the motives leading to self-injury so that the focus and prioritisation of interventions can be modified accordingly. Yet, one may argue that those who engage in selfharming behaviours, both nonsuicidal and suicidal, report similar clinical features and high rates of mental disorder and the way these two categories of individuals choose to hurt themselves does not seem to distinguish one group from the other. Overall, both groups appear to present similarities in diagnostic and demographic variables, making it extremely challenging to draw conclusions about the differences in the suicide risk among them. Thus, further research on this topic is necessary.
Furthermore, the present findings are in line with the social-ecological theory [58], since the majority of the schoolchildren reviewed herein did not engage in adverse phenomena related to bullying victimization, that is, deliberate self-harm or NSSI [82]. Those who were involved in deliberate self-harm were found to be affected by several personal, interpersonal, and social factors, including depressive symptoms. Other risk factors in that case were (a) having a family member who attempted/completed suicide, (b) experience of being physically abused by an adult, and (c) conduct disorder, borderline personality characteristics, and depressive and psychotic symptoms [82]. A doseresponse effect regarding nonrecurring incidents in the association between school bullying victimization and deliberate self-harm was not confirmed, since only repetitive school bullying victimization was associated with self-harming behaviour.
Furthermore, the results regarding the mediating effect of depressive symptoms on the association between NSSI and exposure to school bullying victimization were equivocal. A meta-analysis on this topic showed that bullying victims reported NSSI more frequently than uninvolved schoolchildren. However, this meta-analysis did not explore the mediating effect of depressive symptoms on this association [52]. In contrast, the mediating effect of age on the relation between NSSI and school bullying victimization was explored, presenting that younger responders reported higher effect sizes than older schoolchildren [52]. The same meta-analysis also found that the association between NSSI and bullying victimization was not moderated by the way the respondents were selected (i.e., randomly or not); the response rate in the reviewed studies; the country in which the studies took place (i.e., the USA or a European country). These issues may be of relative importance for future study designs.
Overall, studies using the definition of deliberate selfharm without discriminating between suicidal and nonsuicidal intention reported a higher effect size of school bullying victimization compared to studies exploring NSSI. One possible explanation might be the fact that the use of a broad definition of deliberate self-harm does not allow distinction between the different types of motive for such behaviour, thus ignoring several types of psychopathology behind this conduct, such as mood disorders. In contrast, a strict definition providing a clear motive behind the selfinjury (i.e., relief from distress) probably excludes cases in which mood disorder symptoms may prevail.
The studies reviewed herein did not directly assess the mediating effect of self-stigma; however, factors indirectly associated with it, such as low self-esteem or sexual and physical abuse, were included in the reviewed studies. The study by Lereya et al. [59] showed that there is an indirect association between exposure to maladaptive parenting and domestic violence before the age of 4 years and self-injuring behavior at the age of 16-17 mediated by school bullying victimization reported at the age of 7,8, and 10. According to these findings [59] one may argue that maladaptive behaviour in adolescents due to a dysfunctional family environment and poor parenting may possibly attract negative peer attention and peer rejection, leading to self-stigma related to social marginalization from peers [42,44,59,88]. At the same time, the study by Heilbron and Prinstein [68] revealed that those adolescents who were less accepted by their peers and had low popularity in terms of both reputation and preference were more likely to be targets of overt and relational bullying victimization and to engage in self-injurious behaviour. This may also imply that social marginalization and consequent self-stigma may be factors involved in the association between school bullying victimization and selfharm. However, the mediating effect of internalized selfstigma due to social marginalization and low peer status was not tested in the group of adolescents who were targets of school bullying victimization and at the same time had reported self-injurious behavior in this study [68].
Further research aiming to explore the association of selfstigma from any cause with both self-injurious behaviour and school bullying victimization may be proposed. Thus, a different perspective is needed in the way possible mediating factors are studied in the association between deliberate selfharm and bullying involvement as well as new etiological links. This is really important in view of the fact that selfharming behaviour is a spectrum rather than a continuum of behaviours for young people, whilst even severe suicide attempts may arise with no warning signs or risk factors as yet identified in the literature. Thus, in order to develop interventions that target specific risk factors it is necessary to identify the etiological pathways from the experience of risk factors during childhood to the development of self-injurious behaviour during adolescence and adulthood. Overall, relevant interventions should focus on educating children and adolescents on how to effectively cope with stressors, including bullying, and how to maintain high selfesteem [59,80,81,86], along with preventive and support policy in community, school environment, and healthcare system [46,88].

Conclusions
School bullying victimization is a risk factor for deliberate self-harm, including both suicidal and nonsuicidal acts. Targeted school-oriented interventions at organizational level aimed at eliminating victimization behaviours within the school context are proposed, as well as community-oriented interventions focused on healthy parenting style and ways of coping with stress. The Saving and Empowering Young Lives in Europe study includes data which show that schoolbased preventive programmes (i.e., mental health educational interventions for pupils) are beneficial in preventing selfharming behaviours [89]. Moreover, as this review revealed a prospective association between exposure to school bullying victimization and deliberate self-harm in the young, school healthcare professionals (i.e., school nurses or educational psychologists) need to screen students involved in bullying for self-injurious behaviour and vice versa.
Psychological and psychosocial interventions may be effective in treating schoolchildren who already engage in self-harming behaviours [56,90]. Cognitive behavioural therapy or dialectical behavioural therapy has been shown to be effective in adults [91]. Thus, further research in children and adolescents may be warranted. Additionally, in cases of limited availability of mainstream psychiatric services interventions from sources other than healthcare professionals (e.g., postcards or online support services) may provide support and relief from distress [90]. However, such interventions need to be investigated in future trials in children and adolescents [92]. Additionally, further longitudinal studies on new possible mediating factors in the association between deliberate self-harm and bullying involvement may provide further data regarding self-harming process in the young.

Limitations
A main limitation of the present study is that different definitions and instruments of deliberate self-harm were used in the studies reviewed. This inconsistency makes it difficult to identify the real extent of the phenomenon across different cultural contexts and compare data. Consequently, an effort was made herein to use a comprehensive, broad definition of deliberate self-harm and further differentiate it from NSSI during the analysis and interpretation of the data [1,93]. Another limitation is that samples of individuals who were both bullies and victims of bullying were not included herein. Despite these limitations, the validity of this review is ensured by a comprehensive literature search, discussions about the search words, and inclusion criteria, as well as assessment of the risk of bias of the reviewed studies by means of the modified CASP and NOKC scores. Furthermore, although these tools provide a thorough assessment of the majority of bias risks, it is worth noting that bias related to missing data and publication bias are not included. Additionally, although we took into consideration the most frequently reported confounders, additional factors exist, such as coexistence of other types of victimization, maltreatment, type of parenting, social support, stressful life events, and interpersonal stress. Although documented, these were not identified among the most important. Moreover, confirmation of the data extracted herein by the primary investigators of the reviewed studies was not performed. However, all sources provided within main texts, additional files, and supplements of the published studies were screened.