Bullying is recognized as a major public health concern impacting the well-being of school-aged youth. Studies estimate that approximately one out of every five students report being the victim of bullying behavior (i.e., ridiculed, threatened, physically injured, or isolated) at school (U.S. Department of Justice 2015). Moreover, approximately 14 to 30% of middle school youth and 15% of high school youth are bullied electronically (via social media platforms or by text; Kann et al. 2018). The potential consequences of experiencing bullying victimization in any form during adolescence can be serious and long lasting. Previous research has identified negative outcomes among victims of bullying, including delinquency, substance use, physical health problems, poor academic performance, and sexual promiscuity (Farrington et al. 2011; Hamilton et al. 2016; Ttofi et al. 2011a, b). Youth who report being bullied have also been found to experience significant mental health problems and are at a heightened risk for suicide (Copeland et al. 2013; Hinduja and Patchin 2010; Klomek et al. 2011), which is the second leading cause of death among school-age youth (Centers for Disease Control and Prevention 2017). To address this growing concern, studies to better understand the processes that connect bullying victimization, mental health, and suicide risk are necessary. Therefore, examining this connection is the aim of the present study.

Empirical Background

Bullying is often defined as behavior that is unwanted, aggressive, repeated over time, or is highly likely to be repeated and involves a power imbalance from youth who are not siblings or intimately involved (Gladden et al. 2014; Olweus 1993). Bullying can be physical (e.g., hitting or pushing) or verbal (e.g., spreading rumors). Bullying behavior can also be categorized into overt and covert forms. Overt bullying is easier to detect in that it includes threatening other students or physically hitting someone; whereas, covert bullying includes acts harder to detect such as gossiping, spreading rumors, or isolating victims from peer groups (Byers et al. 2011).

There is a high prevalence of bullying that occurs within school settings. For example, in a meta-analysis composed of 80 studies, Modecki et al. (2014) found that the mean prevalence of in-school bullying was 36% for youth between ages 12 and 18 (n = 335,519). Other studies have also found the frequency of bullying behaviors to be high: 24% of students were bullied at least once or twice (during the current term) and almost 9% were bullied either sometimes (i.e., moderate bullying) or weekly (i.e., frequent bullying; Nansel et al. 2001). Recent data collected from the 2017 Youth Risk Behavior Survey, a national questionnaire distributed to public, non-public, and Catholic high schools, showed that approximately 19% of youth were bullied on school property during the prior school year (Kann et al. 2018). Moreover, from 2007 to 2016, lifetime rates of cyberbullying nearly doubled (Patchin and Hinduja 2016). Across these studies (Kann et al. 2018; Modecki et al. 2014; Nansel et al. 2001), nearly one quarter to one third of school-aged children reported being bullied at least once during their lifetime, which is of a concern given that being a victim of bullying has been associated with short- and longer-term consequences.

Previous research has also identified a link between bullying victimization and the perpetration of bullying others. Individuals who have experienced bullying victimization have been shown to engage in bully perpetration at a significantly higher rate than individuals who have not been victims of bullying (Álvarez-García et al. 2015; Fanti and Kimonis 2012). Several explanations for the bully/victim overlap have been provided to understand this association. These include similarities in risk factors (e.g., individual, peer, school) for victims of bullying, bullying perpetrators, and bully/victims (Cook et al. 2010; Goldweber et al. 2013) and theoretical perspectives such as self-control and general strain theory (Chui and Chan 2015; Haynie et al. 2001; Jang et al. 2014).

Link Between Bullying Victimization and Mental Health Problems

A large body of research consistently shows that students who are bullied are more likely to experience emotional problems, such as anxiety and depression, compared with students who are not victims of bullying (Arseneault 2017; Brendgen and Poulin 2018; Forbes et al. 2019; Moore et al. 2017; Stapinski et al. 2015; Turner et al. 2013; Wang et al. 2011). For example, in a meta-analysis of 28 longitudinal studies, Ttofi et al. (2011a, b) found that depression was two times higher among victims of bullying. In addition, Copeland et al. (2013) examined the long-term effects of being bullied on depression, anxiety, and other aspects of adolescent well-being and found that bullying victimization was associated with both depression and anxiety. Specifically, victims of bullying were four times more likely to experience an anxiety disorder and three times more likely to experience a depressive disorder. In another study that examined the long-term psychosocial consequences of bullying victimization, Takizawa et al. (2014) found that victimization between ages 7 and 11 led to higher rates of depression and anxiety disorders at ages 23 and 50. Overall, these prior longitudinal studies suggest that internalizing behaviors, such as anxiety and depression, are often experienced by those who report being bullied.

Further, several longitudinal and meta-analytic studies have demonstrated a relationship between bullying victimization and self-harm or engagement in suicide-related behaviors (Barker et al. 2008; Baiden et al. 2018; Fisher et al. 2012; Heerde and Hemphill 2019; Hinduja and Patchin 2019; Holt et al. 2015; Moore et al. 2017). For example, Sigurdson et al. (2018) found that victims of bullying during middle school (mean age = 13) were over two times more likely to report thoughts of self-harm 1 year later and as an adult (mean age = 27). In a meta-analysis of several prospective studies that examined the effects of victimization, bullying victims were found to be at a significantly higher risk for suicidal ideation, attempts, and non-suicidal injury (Moore et al. 2017). Across these studies, the longitudinal relationship between bullying victimization and suicidal risk (i.e., ideation and suicidal attempts), as well as the relationships between victimization and later depression and anxiety (separately), were found to be relatively consistent. Given the robust findings that symptoms of depression and anxiety, independently, are known risk factors for self-harm and suicidal ideation and that being a victim of bullying increases the risk of depression, anxiety, and self-harm, it is possible that victimization leads to thoughts of self-harm indirectly through its impact on depression, anxiety, or other emotional problems. This suggests a process whereby symptoms of internalizing problems (i.e., depression, anxiety, emotional problems) mediate the relationship between being a victim of bullying and suicide risk.

Mental Health Comorbidity, Suicide Risk, and Bullying Victimization

Comorbidity refers to experiencing two or more mental health issues at the same time (e.g., youth presents with anxiety but also has a comorbid, secondary depressive issue or vice versa; Birmaher et al. 1996). As noted by Garber and Weersing (2010), among community samples of adolescents, 25 to 50% of youth with depression meet the criteria for anxiety and 10 to 15% of youth with anxiety meet the criteria for depression (Costello et al. 2003; Scholten et al. 2013; Stockings et al. 2016). Comorbidity often exacerbates symptoms of both mental health disorders and increases the risk for experiencing the other mental health issue over time (Birmaher et al. 1996; Garber and Weersing 2010). For example, in a longitudinal study of 1420 youth (ages 9 to 16) who had co-occurring anxiety and depressive disorders, approximately 23% of adolescents reported suicidal ideation, intentions, and engagement in suicide-related behaviors (i.e., attempts; Foley et al. 2006). Across several psychiatric profiles (e.g., depression and disruptive behavior, anxiety disorder only, depression only, alcohol abuse), depression and anxiety or depression and general anxiety disorder had the highest prevalence among suicidal youth. To date, the literature has largely neglected the co-occurring nature of internalizing symptoms or disorders, particularly depression and anxiety, when studying the consequences of being a victim of bullying. This gap in the literature is an important oversight given that bullying victimization has been shown to be associated with mental health symptomology and self-harm, and that youth with comorbid symptoms have an increased likelihood of engaging in self-harm or suicidal behaviors (Baiden et al. 2018; Heerde and Hemphill 2019; Hinduja and Patchin 2019; Ttofi et al. 2011a, b). While prior studies identify a relationship between bullying victimization and various types of internalizing problems including depression, anxiety, and self-harm, the role of comorbidity in the pathways or processes leading to increased rates of suicidal thoughts among youth who have been bullied has not been fully considered. Therefore, this study seeks to fill this gap in the research by exploring the relationship between having symptoms of depression and anxiety (e.g., emotional problems) and thoughts of self-harm across victims and non-victims of bullying.

Theoretical Background

Several theoretical explanations have been set forth in prior studies to explain the link between victimization and mental health (Bauman et al. 2013; Radliff et al. 2016; Reed et al. 2015). The interpersonal theory of suicide (Joiner 2007; Van Orden et al. 2010), for example, posits that suicidal behavior is the result of thwarted belongingness (i.e., lack of fitting in with peers) and perceived burdensomeness (i.e., hopelessness or loneliness). Victimization by peers can result in the development of these negative feelings and lead to symptoms of anxiety and depression that, in turn, increases suicidal ideation. Beck’s (1967) and Beck et al.’s (1979) cognitive theory of depression and Abramson et al.’s (1989) hopelessness theory of depression have also been used to explain how victimization influences depression through its impact on victims’ perceptions of themselves and the world around them. Based on tenets from these theories, victims of bullying may blame themselves for their victimization and develop self-deprecating feelings about their personality, self-worth, and their future (Baiden et al. 2018). According to Beck (1967), these feelings lead to a “cognitive triad” of negative perceptions of self, the world, and the future (Bauman et al. 2013, p. 342).

Agnew’s (2001) general strain theory (GST) has also been used to explain the relationship between victimization and suicidal behavior (see Hay et al. 2010; Patchin and Hinduja 2011). GST postulates that the manifestations of strain can lead to the expression of negative emotions and symptoms and the development of mental health problems like depression and anxiety. Suicidal thoughts and self-harm behaviors may manifest and serve as an internal coping mechanism to deal with “strain.” According to Agnew (2001), being a victim of bullying is a consequential form of strain because it is often perceived as unjust, high in magnitude, not part of conventional social control, and exposes victims to other deviant role models (i.e., bullies). Although these theories were not directly tested in this study, they provide a theoretical framework for the examining the linkages and temporal ordering among victimization, mental health, and suicidal ideation.

Current Study

The overall goal of the current study is to examine the longitudinal relationship between bullying victimization, emotional problems (reflective of symptoms of anxiety), symptoms of depression, and thoughts of self-harm among a sample of middle and high school students. Thoughts of self-harm may be characterized as a proxy for suicidal ideation and is a risk factor for suicidal behavior (Grandclerc et al. 2016; Herba et al. 2007). Thus, this study contributes to the empirical literature in a number of important ways. First, although several studies have documented the relationship between victimization and thoughts of self-harm, few have included emotional problems and symptoms of depression as mediators. Therefore, this study expands on previous research by assessing potential mechanisms that may influence the relationship between bullying victimization and thoughts of self-harm. Second, since mental health issues tend to co-occur, this study expands existing knowledge regarding the impact of bullying on adolescent mental health by exploring how comorbid emotional problems and symptoms depression influence thoughts of self-harm among bully victims. Third, more research is needed to examine the impact of bullying on community samples of youth in both middle and high school. A large body of the current research is based on samples of middle school students, excluding other age groups (Espelage and Holt 2013; Hinduja and Patchin 2010; Juvonen et al. 2003). There may be important differences in emotional and physical maturity and coping mechanisms as youth transition into higher grade levels. For this reason, it is important to consider mental health functioning across different stages of adolescent development. Finally, this study provides an assessment of the direct and indirect relationships using two time points, whereas the majority of existing research on mental health and bullying victimization is based on cross-sectional analyses (see Hawker and Boulton 2000; Klomek et al. 2010). The use of two time points allows for stronger assertions to be made about the temporal links between being a victim of bullying, emotional problems, symptoms of depression, and thoughts of self-harm. It is important to understand how the co-occurrence of emotional problems and symptoms of depression impacts the relationship between victimization and suicide risk as this may help guide school-based interventions and anti-bullying prevention programs.

Therefore, the following research questions guided the current study:

  1. (1)

    Is being bullied at school associated with thoughts of self-harm?

  2. (2)

    Is the relationship between being a victim of bullying and having thoughts of self-harm mediated by emotional problems or symptoms of depression?

  3. (3)

    Does the co-occurrence of emotional problems and depressive symptoms (i.e., comorbidity) mediate the relationship between being bullied at school and having thoughts of self-harm?

Grounded in prior research (Bauman et al. 2013; Hinduja and Patchin 2010; Klomek et al. 2011), it was hypothesized that: (1) being bullied at school would be significantly and directly related to thoughts of self-harm; (2) the relationship between being a victim of bullying and having thoughts of self-harm would be mediated by emotional problems and symptoms of depression (separately); and (3) comorbidity would mediate the relationship between being bullied at school and thoughts of self-harm.

Methods

Sample and Data

Data for this study were derived from a longitudinal project on school climate and safety that was conducted in one school district in the southeastern USA. The schools involved in the primary study were selected by school administrators due to high rates of discipline issues (e.g., discipline referrals, expulsions) and being located in low-income neighborhoods. The overall goal of the larger study was to improve school climate and safety by placing social workers at each school. Social workers provided behavioral health intervention and community service referrals to students and mental health and crisis intervention training to school personnel. Students selected for behavioral health interventions were mainly referred by guidance counselors or school administrators for delinquency (excessive absences) or disruptive and/or aggressive (fighting, impulsivity, being disrespectful to school personnel) behaviors. Thus, the sample of middle and high school students used in the current study are high-risk students identified as having behavioral difficulties in school.

Data utilized in the current study are from baseline and after-services interviews of project participants (i.e., students working with school social workers). To participate in the project, parental consent and student assent was required. The baseline interview (Time 1) was completed after students were referred to services but before receiving any type of service or intervention from the social worker (i.e., clinical sample). The interview contained several questions about demographic characteristics, victimization experiences, family/peer relationships, and perceptions of school safety. The after-services interview (Time 2) was completed immediately following services and contained the same questions asked in the baseline interview. Interviews were conducted by trained undergraduate or graduate students. The average length of time between pretest and post-test was 6 months. All interview protocols were approved by the university’s institutional review board.

A total of 165 sixth through twelfth graders completed a baseline interview, 112 students also completed an after-services interview (68%). The average age at the baseline interview for the 112 students who had valid data at both time points was 14.14 years (SD = 1.81). Student participants who did not complete Time 2 interviews were excluded from the study. No significant differences (p > .05) were observed on key variables between students who were included in the final sample and those who were omitted due to missing data (n = 53). A description of the sample is in Table 1.

Table 1 Sample characteristics (n = 112)
Full size table

Measures

Independent Variables

The primary independent variable in the current study was bullying victimization measured at the baseline interview. Bullying victimization was comprised of eight items that were adapted from the School Crime Supplement to the National Crime Victimization Survey (Bureau of Justice Statistics 2013). The bullying victimization measure included general and verbal bullying victimization in the past 6 months. Example items include “did another student bully you,” “make fun of you, call you names, or insult you in a hurtful way,” and “bully you electronically.” Responses to each of the eight items were dichotomous. Prior research indicates there is inconsistency in the measurement of bullying victimization and notes that findings are influenced by the way in which bullying is measured (see Esbensen and Carson 2009). Therefore, to account for potential differences in findings due to poor measurement, two measures of bullying victimization were constructed. First, the eight items were summed to create one bullying victimization measure, which ranged from 0 to 7 (M = 1.69; SD = 2.04; α = 0.80). Second, the eight items were then summed and dichotomized to represent whether respondents had ever been a victim of bullying at least once. Thus, the composite measure includes behavior specific incidents and represents diverse bullying experiences, whereas the dichotomous measure captures whether bullying victimization was experienced in the past 6 months (i.e., prevalence).

Five items from Time 2 were used to create an additive scale that measured emotional problems. These questions were based on the Emotional Problems Scale from the Strengths and Difficulties Questionnaire (SDQ; Goodman et al. 1998). Each item is measured using a 3-point Likert scale ranging from not true (0) to certainly true (2). Questions such as “I have many fears, I am easily scared,” “I worry a lot,” and “I am nervous in new situations” are included. The five items were summed to create one measure of internalizing problems representing emotional problems, that ranged from 0 to 10 (M = 3.13; SD = 2.50; α = 0.74). Higher scores represent more emotional problems.

Symptoms of depression was measured using an additive scale that consisted of 12 items adapted from the Children’s Depression Inventory, Short Version (CDI-S; Kovacs 2011) that has shown to have good psychometric properties with diverse youth samples (Allgaier et al. 2012; Thompson et al. 2010). The CDI-S has been validated to differentiate between depressed and non-depressed youth (Kovacs 2011). The depression scale came from the after-services interview. Example items include “I am sad all the time,” “I hate myself,” and “I am tired all the time.” Responses to these questions were measured using a 3-point Likert scale that ranged from 0 to 2, where 2 represented higher levels of depression. The 12 items were summed to create an overall measure of depression which ranged from 0 to 24 (M = 5.40; SD = 3.75; α = 0.78). Higher scores indicate more severe depressive symptoms.

Based on prior research, recommended cutoff scores for the measures of depression and emotional problems were used to create mutually exclusive groups representing participants who reported low/no emotional problems and symptoms of depression, depression symptoms only, emotional problems only, and comorbid symptoms of depression and emotional problems (Balázs et al. 2013; Beck and Beck 1972; Goodman et al. 1998). Emotional problems was dichotomized to distinguish between youth who did not report emotional problems from those who reported emotional problems. Cutoff scores for emotional problems were determined using Goodman et al.’s (1998) recommendations based on national prevalence estimates for anxiety. The most recent national estimates suggest that 32% of youth experience any anxiety disorder, and therefore the cutoff score was set at the 70th percentile (National Institute of Mental Health (NIMH) 2017). Youth who scored below the 70th percentile were categorized as not experiencing emotional problems. Symptoms of depression were converted from a scale into a dichotomous measure that separates youth with little to no symptoms from youth who reported mild to severe symptoms. Cutoff scores for this variable were also based on national prevalence estimates reported by the NIMH. According to the NIMH (2017), approximately 13% of youth experience severe depression in adolescence. Based on this estimate, youth who scored in the 87th percentile or above were categorized as experiencing mild to severe depressive symptoms. Then, comorbidity was constructed to represent youth who reported experiencing both depressive symptoms and emotional problems based on the dichotomous coding described above. These transformations resulted in one, mutually exclusive variable with four categories: low to no symptomology, emotional problems only, symptoms of depression only, and comorbid emotional problems and symptoms of depression.

Dependent Variable

The dependent variable of interest was thoughts of self-harm, a salient factor associated with suicide risk. We used a single dichotomous indicator at Time 2 that asked respondents if they have had thoughts about hurting themselves in the past 6 months (yes/no). Assessing thoughts of self-harm within a 6-month timeframe has been shown to strongly predict adult suicidal ideation (Herba et al. 2007). Of the 112 students in the sample, 25% (n = 28) had thoughts about hurting themselves in the past 6 months.

Control Variables

Several demographic characteristics were included as control variables. Race was coded to differentiate between Whites (48%) and non-Whites (52%). Gender was a dichotomous measure to distinguish between boys (59%) and girls (41%). Grade was also controlled for in the current study and represents the grade the student reported being in at the time of the baseline interview (M = 8.13; SD = 1.60; range = 6–12). Mental health symptomology (emotional problems and symptoms of depression) at Time 1 was controlled to establish time order that might have increased the odds of experiencing bullying victimization. Given that the current study examined the impact of mental health symptomology on thoughts of self-harm, we also included negative life events as a control variable (items were guided by Holmes and Rahe’s 1967 Social Readjustment Scale). This measure included seven items known to trigger stress in youth (e.g., witnessing community violence, being hospitalized for mental health problems, experiencing parental break up/divorce). Items were summed to create one measure to indicate the frequency of negative life events in the past 6 months (M = 1.96; SD = 1.24; range = 0–5). Failing a grade was also controlled for a dichotomous measure that represented poor school performance in the past 6 months.

Analysis Plan

The analyses proceeded in several steps. First, a series of chi-square (χ2) difference tests and t tests were conducted to assess bivariate associations among study variables. Then, stepwise logistic regression was performed to determine the impact of bullying victimization, emotional problems, and symptoms of depression on thoughts of self-harm. Stepwise logistic regression was used to test the components of mediation as outlined by Baron and Kenny (1986). Specifically, we examined the direct effect of bullying victimization on thoughts of self-harm, the direct effects of each mental health measure on thoughts of self-harm, and then examined the indirect effects of bullying victimization on thoughts of self-harm through the effects of depressive symptoms and emotional problems (i.e., mediating variables). Mediation is confirmed if (1) the independent variable is significantly related to the mediating and dependent variables, (2) the mediator is significantly related to the dependent variable, and (3) the strength of the relationship between the independent and dependent variables is reduced when the mediating variable is added to the model (Baron and Kenny 1986; Preacher and Hayes 2008). Since the dependent variable was binary, logistic regression was appropriate for testing each component of mediation (Preacher and Hayes 2008; see also MacKinnon and Dwyer 1993; Samawi et al. 2018). For each of the models, odds ratios (OR) and confidence intervals (CI) are reported in the tables and unstandardized coefficients (b) and standard errors (SE) are in the text, where appropriate.

Results

Bullying Victimization and Thoughts of Self-Harm

To assess whether being a victim of bullying at school was associated with thoughts of self-harm (Research Question 1), a series of bivariate analyses were conducted (see Table 2). Significant differences were observed between youth who endorsed thoughts of self-harm versus those that did not. In general, girls were more likely to report thoughts of self-harm compared with their male counterparts (χ2 [1] = 3.98, p < .05). Youth who had thoughts about hurting themselves were more likely to report experiencing at least one victimization experience in the past year (χ2 [1] = 5.27, p < 0.05); in addition, these youth were also more likely to report diverse bully experiences (F(110) = 7.16, p < .01), such as being punched, kicked, or made fun of. Youth who reported thoughts of self-harm also experienced significantly higher levels of emotional problems (F(110) = 27.41, p < .001), symptoms of depression (F(110) = 33.83, p < .001), and comorbid symptoms (χ2 [1] = 21.58, p < .001). Significant differences were also found among youth experiencing negative life events, such that youth who reported thoughts of self-harm also endorsed having been exposed to multiple life stressors (i.e., family member passed away, saw bad things in their neighborhood). There were no significant differences in having thoughts of self-harm across race, grade, or poor school performance (failing a grade). Therefore, these measures were not included in the multivariate analyses.

Table 2 Bivariate differences across key indicators
Full size table

Bullying Victimization, Mental Health, and Thoughts of Self-Harm

Results of the logistic regression analyses using the dichotomous measure of bullying victimization, which represents prevalence in bully experiences, are presented in Table 3. The results to answer the first research question are provided in the first model. Model 1 shows that the odds of experiencing thoughts of self-harm increased by 28% for every unit increase in bullying victimization (b = 0.24, SE = 0.12). In examining the direct effects of emotional problems on having thoughts of self-harm (Model 2), the odds ratio indicated that a one unit increase in emotional problems increased the odds of experiencing thoughts of self-harm by 54% (b = 0.43, SE = 0.13). The final step in these analyses examined both bullying victimization and emotional problems as predictors of thoughts of self-harm (Model 3). When both measures were included in the model, the effect of bullying victimization on having thoughts of self-harm was no longer significant (b = 0.18, SE = 0.13). The strength of the relationship between emotional problems and thoughts of self-harm was slightly reduced and remained statistically significant (b = 0.40, SE = 0.14). These findings suggest that experiencing emotional problems mediated the relationship between bullying victimization and thoughts of self-harm among study participants (Research Question 2). The logistic regression analyses examining the relationship between bullying victimization, symptoms of depression, and thoughts of self-harm are reported in Table 4 (Research Question 2). As can be seen in Model 2, endorsement of depressive symptoms significantly predicted having thoughts of self-harm (b = 0.34, SE = 0.10). For each unit increase in symptoms of depression, the odds of having thoughts of self-harm increased by 41%. However, when including both symptoms of depression and bullying victimization into the final model (Model 3), the effect of bullying victimization on having thoughts of self-harm was non-significant (b = 0.24, SE = 0.13). The strength of the relationship between symptoms of depression and thoughts of self-harm did not change (b = 0.34, SE = 0.11).

Table 3 Logistic regression results for emotional problems predicting thoughts of self-harm
Full size table
Table 4 Logistic regression results for symptoms of depression predicting thoughts of self-harm
Full size table

Bullying Victimization, Comorbidity, and Thoughts of Self-Harm

To examine whether the co-occurrence of emotional problems and symptoms of depression mediated the relationship between bullying victimization and thoughts of self-harm (Research Question 3), the final objective was to perform the same stepwise logistic regression analyses as reported in Tables 3 and 4, with comorbidity included in the model. However, we were unable to conduct these analyses due to a small sample size, the categorized coding of comorbidity, and low cell counts for youth who did not report victimization but experienced comorbidity and/or thoughts of self-harm. Therefore, we explored the bivariate relationships among comorbidity, victimization, and thoughts of self-harm to obtain a preliminary assessment of these relationships. This led to bivariate comparisons across four groups (n = 112): youth who presented low to no symptoms (n = 73), emotional problems only (n = 19), depression only (n = 5), and comorbid emotional problems and depressive symptoms (n = 15).

Results from the bivariate analyses examining bullying victimization, mental health symptoms, and thoughts of self-harm are reported in Table 5. Cramer’s V was used as a measure of effect size to determine if there were significant differences across the three mental health categories (i.e., low to no symptomology, symptoms of depression only, and comorbidity). Effect sizes for bullying victimization and thoughts of self-harm were both relatively large (Cramer’s V = .31 and .56, respectively). Further, chi-square (χ2) difference tests assessed mental health differences across bullying victimization and suicidal ideation. Youth who reported being a victim of bullying were significantly more likely to report co-occurring symptoms (χ2 [2] = 9.60, p < .01) compared with low/no symptoms and depression only. Furthermore, 17% of youth who were bullied reported comorbid symptoms compared with 9% of youth who did not experience bullying victimization. We also examined bullying victimization utilizing the frequency measure. Results confirmed the findings produced by the dichotomous measure. Moreover, youth who had comorbid symptoms reported an average of 2.87 (SD = 2.42) victimization incidents, youth with emotional problems experienced an average of 2.95 (SD = 2.30) incidents, youth with depressive symptoms reported an average of 0.60 (SD = 0.55) incidents, and youth with low/no symptoms endorsed an average of 1.19 (SD = 1.70) incidents (F [2] = 6.99, p < .001).

Table 5 Bivariate analyses among bullying victimization, thoughts of self-harm, and mental health
Full size table

Next, we examined the bivariate relationship between mental health symptoms and thoughts of self-harm. Youth who reported comorbid symptoms were significantly more likely to report thoughts of self-harm compared with youth who reported symptoms of depression only or low/no symptoms (χ2 [2] = 33.07, p < .001). For example, 10% of respondents in the low/no symptoms group reported thoughts of self-harm compared with 20% of youth with depressive symptoms, 47% with emotional problems, and 73% with comorbid symptoms. To further explain the differences between low to no symptoms, depressive symptoms-only, and comorbidity, we conducted pairwise comparisons using Tukey’s honest significance test and Bonferroni post hoc estimates. Comparisons suggested that youth who reported comorbid symptoms were significantly different from youth who reported low to no symptoms (p < .001) and from youth who experienced depression (p < .001). Additionally, youth who reported emotional problems were significantly different than those who reported low to no symptoms of emotional problems (p < .001). No differences in thoughts of self-harm across the low/no symptoms and depression only groups were revealed. Taken together, results suggest that youth who are victims of bullying are at a higher risk for experiencing emotional problems and symptoms of depression (i.e., comorbidity) and that youth experiencing comorbid symptoms are at a heightened risk for thoughts of self-harm.

Discussion

The aim of the current study was to explore the relationship among being a victim of bullying, symptoms of depression, emotional problems (reflective of symptoms of anxiety), co-occurring emotional problems and depression symptoms, and thoughts of self-harm among a high-risk sample of middle and high school students. In line with prior research (Espelage and Holt 2013; Klomek et al. 2007; Reed et al. 2015), bullying victimization was significantly and directly related to thoughts of self-harm. Specifically, experiencing bullying at school increased the odds of endorsing self-harm thoughts by 30% compared with students who were not bullied. Findings revealed no significant differences in thoughts of self-harm across race or grade level. Prior studies demonstrate inconsistent results regarding race, age, and grade level (Goldston et al. 2016; Hinduja and Patchin 2010, 2019). Youth who had thoughts of self-harm were also more likely to encounter negative life stressors and report emotional problems and depressive symptoms, findings that have been confirmed in other studies (Andover et al. 2012; Avenevoli et al. 2015; Hart et al. 2017).

As expected, the strength of the relationship between bullying victimization at school and having thoughts of self-harm was mediated by emotional problems and symptoms of depression independently. Although it is unclear if other risk factors contributed to these findings, it may be that negative cognitive styles bias the appraisal of stressors (i.e., bullying) and perceptions of the self (Abela and Hankin 2008; Abramson et al. 1989). In studies that have examined the relationship between negative cognitive styles and subsequent mental health problems in late childhood and adolescence, shame proneness (a state of intense pain characterized by feelings of inferiority or worthlessness) and depressogenic thinking (a negative inferential style of responding to adverse events) were found to predict anxiety and depressive disorders (Mills et al. 2015; Muris et al. 2015). Our findings also align with prior theoretical assertions that posit causal pathways whereby experiencing victimization leads to negative emotions such as hopelessness, feelings of not belonging, isolation, and poor self-esteem or negative perceptions of self (see Abramson et al. 1989; Agnew 2001; Joiner 2007). Left unresolved, these feelings can interfere with other domains of functioning and may even lead to suicidal thoughts and self-harm behaviors (e.g., suicide attempts, non-suicidal self-injury (NSSI), and substance abuse). Overall, these findings highlight the significant impact that being victims of bullying can have on adolescent development, particularly socioemotional functioning, a key competency that facilitates learning and increases the likelihood of academic success and overall well-being (Klomek et al. 2007).

Although we were unable to test mediation for co-occurring emotional problems and symptoms of depression using multivariate analyses, bivariate analyses demonstrated that students who reported being a victim of bullying were more likely to report emotional problems and symptoms of depression compared with low/no symptoms or depression only. Students who reported comorbid symptoms, compared with other mental health symptomology, were also more likely to report thoughts of self-harm. Overall, our findings are consistent with other studies that indicate being a victim of bullying leads to adverse mental health consequences among middle and high school students, which in turn, increases the likelihood of thoughts of self-harm or suicidal ideation (Reijntjes et al. 2010).

Implications

Considering that childhood victimization and early-onset disorders, such as anxiety and depression, have been associated with impairments in both adolescence (Baiden et al. 2018; Hager and Leadbeater 2016; Hawker and Boulton 2000) and adulthood (Klomek et al. 2009; Stapinski et al. 2014), schools should play an active role in incorporating programs and policies that focus on student wellness and mental health. Incorporating policies and programs is relevant, particularly as federal mandates have been put in place to improve academic standards and student outcomes (e.g., Every Student Succeeds Act of 2015). For students, the implementation of school-based interventions aimed to develop and enhance socioemotional and mental health skills may be beneficial in promoting resiliency and competencies, which in turn may help to prevent suicide risk and other mental health and behavioral problems (Dray et al. 2017; Goldman et al. 2016). For school personnel, the importance of mental health in protecting against the longer-term effects of bullying should be considered. For example, incorporating mental health and suicide prevention training into professional development standards may help to improve skills and competencies that often interfere with the ability to address emotional and behavioral needs and mental health challenges. Such training can better prepare school officials to identify risk factors and warning signs that may present in the school environment. While all states have laws or policies that protect against bullying, individual schools are given discretion in deciding whether to implement bully prevention programming (Ttofi et al. 2011a, b). Perhaps it is necessary to mandate bully prevention and intervention training in all schools in an effort to decrease the consequences associated with victimization and perpetration, while also protecting against thoughts of self-harm and suicidal ideation. Teaching school personnel how to identify and combat bullying is of critical importance as well (Thornberg et al. 2018; Bradshaw et al. 2018). If victimization (and related symptoms and behaviors) can be quickly and appropriately identified, there is a better chance of alleviating the problem, decreasing mental health problems, and helping youth before they feel helpless and hopeless. Similarly, if suicide risk is detected, clear and standardized protocols are needed to guide school personnel to ensure that youth are properly screened and given adequate community referrals and linkages (Shute 2016).

Limitations

There are some limitations of the current study that should be mentioned. First, our results are based on a small sample of students from one school district in the southeastern USA. The sample was also based on students from five schools located in high-risk neighborhoods who had been referred to services for discipline and attendance issues (i.e., clinical sample). Therefore, our findings are not generalizable to all students in public schools. It is also possible that the selection criteria for this study (i.e., high-risk communities, discipline issues) may have led to higher prevalence rates of the variables of interest including bully victimization, mental health symptoms, or thoughts of self-harm. The small sample size also limited the number of youth who reported being a victim of bullying, experienced comorbidity, and endorsed thoughts of self-harm. As a result, we were unable to perform multivariate analyses when assessing the relationship between comorbid symptoms of depression, emotional problems, and thoughts of self-harm. Though we were able to provide a preliminary analysis of what youth with self-reported emotional problems and symptoms of depression may experience, we caution against using the findings to generalize to students in other public-school districts. In particular, the bivariate analyses presented in Table 5 are based on four small subsamples of high-risk youth and do not account for additional risk factors that may make these relationships spurious. Second, analyses are based on students’ self-reports of victimization, emotional problems, symptoms of depression, and thoughts of self-harm. Students may have underreported occurrences as a way to provide socially desirable outcomes (Brownfield and Sorensen 1993). Additionally, data for this study was derived from a larger study on school climate and safety. The SDQ was incorporated to measure a range of externalizing and internalizing symptoms due to its widespread use among adolescents in research and school-based settings. Future research should seek to include other validated measures commonly used to assess anxiety in community samples. Furthermore, our measure of thoughts of self-harm was based on a single item indicator that asked youth if they had thoughts about hurting themselves but did not ask about the intention or desire to die. Therefore, this item may not accurately reflect thoughts of suicide but instead, may be more reflective of thoughts of self-harm behavior without intentions to die (i.e., NSSI). Since our study was administered in a school setting, the probability of underreporting is increased. Some youth may have believed that answers would be shared with school officials, thereby altering their responses based on fear of punishment or hospitalization. Nonetheless, future studies should also include the use of validated measures to assess suicide risk. Further, due to the small sample size, we were unable to account for additional covariates which might have influenced the relationships assessed in the current study. For example, prior research indicates that peer relationships, negative cognitive appraisals and self-beliefs, and social bonds also impact mental health outcomes and suicidal ideation (Abela and Hankin 2008; Copeland et al. 2013).

Future Research

Additional research is needed to fully understand the causal processes that lead to mental health problems and suicide risk among bully victims. Future research would benefit from examining these public health concerns using larger, more representative samples of youth (including clinical and community samples). Utilizing larger samples will allow for multivariate analyses to examine associations among bullying victimization, comorbidity, and suicide risk. Based on prior research that highlights differences in prevalence of mental health disorders across gender (Balázs et al. 2013), race/ethnicity (Merikangas et al. 2011), sexual orientation (Marshal et al. 2011), and bullying experiences (Arseneault 2017; Turner et al. 2013), future research should explore the mental health consequences of bully experiences across subgroups of youth. The role that negative life events play in the relationship between bullying victimization, mental health, and thoughts of self-harm should also be considered in future research. Our findings suggest that negative life stressors were higher among youth who were victims of bullying. The types of life stressors and the role that these experiences play in the relationship between bullying, mental health, and thoughts of self-harm, in addition to other contextual risk factors, such as perceptions of the school environment and peer relationships, may provide valuable information about the causal pathways that lead to thoughts of self-harm among victims of bullying.

Conclusion

This study expanded on existing literature related to bullying victimization and mental health by investigating how comorbidity impacts thoughts of self-harm in victims of bullying. Overall, our results suggest that victims of bullying endorse thoughts of self-harm, symptoms of depression, and emotional problems at higher rates than youth who do not experience bullying. These finding demonstrate the need for the integration of bully and mental health intervention programs for youth as well as training for all personnel who interact with students. Providing programs and training at the school level that target both bullying and mental health may increase the likelihood of identifying problem behaviors before they escalate. Early detection of mental health problems, victimization, and suicide risk can promote safe and healthy environments that serve to strengthen the overall well-being of all students.