The Association between Deliberate Self-Harm and School Bullying Victimization and the Mediating Effect of Depressive Symptoms and Self-Stigma: A Systematic Review
Table 1
Main results of the included studies pertaining to the research questions in the present systematic review.
Authors & Date (study acronym)
Aim
Study Population
Main Results
Brunner et al. 2014 (SEYLE study)
Exploration of the prevalence and associated psychosocial factors of occasional and repetitive direct self-injurious behaviour.
SBV & SB were both found to be strong predictors of SI in the univariate regression analysis (UA) and also showed an independent effect on the multivariate regression analysis (MA); SBV as a risk factor for lifetime SI: UA/ OR (95% CI): 3.39 (2.96-3.89) & MA/ 1.68 (1.43-1.98), p<0.01
Brunstein Klomek et al. 2016 (SEYLE study)
To examine the association between victimization by bullying and direct self-injurious behaviour.
Community-based adolescents (age range: NR); n= 11.110; RR: 72% of the schools approached.
SBV(physical/verbal/relational) was strong predictors of SI (lifetime, both occasional and repetitive) in the univariate regression analysis and also showed an independent effect on the multivariate regression (MA) model: SBV (physical/verbal/relational) as a risk factor for lifetime SI: UA/ OR (95% CI): 2.10-2.78 (1.90 to2.51 – 2.32 to 3.08) & MA/ 1.33-1.72 (1.18 to 1.49-1.50-2.07), p<0.01 The effect of relational & verbal SBV on SI was partially mediated by depressive symptoms (Sobel test = 7.33; p<0.0001; Sobel test=5.22; p<0.0001).
Claes et al. 2015
Investigation of the association between bullying victimization and NSSI and the mediating effect of depressive symptoms & parental support.
Community-based adolescents - grade; n= 785; RR: NR
The association between SBV and NSSI (β=0.23, p<0.001) was partially mediated by depressive symptoms x2 = 16.44, p=0.003; CFI=0.948; RMSEA=0.063.
Elgar et al. 2014 (2012 Dane County Youth Assessment Study)
Exploration of the association among cyberbullying victimization, traditional bullying, self-injurious behaviour & related mental health problems, as well as the moderating role of family contact.
SBV was a risk factor for SI OR (95% CI): 1.10 (1.08- 1.13).
Espelage & Holt 2013
Exploration of the differences in the frequency of suicidal ideation and suicidal behaviours across a group of verbal bullies, bully victims, victims, physically aggressive bullies, and students not involved in bullying.
SI was statistically significantly higher in victims of school bullying compared to uninvolved students (28.2% vs. 8.7%; x2 = 53.89; p<0.001), while depressive symptoms only partially explain the difference (F= 126.5; p<0.0001; n2 =0.17).
Ford et al. 2017 (K-cohort Longitudinal Study of Australian Children)
Exploration of the association between all types of bullying and adverse mental health outcomes, including self-harming behaviours and suicidality.
The prevalence (%)(95%CI)) of self-harm in adolescents of both genders self-reported as bullying victims (all types of bullying victimization) was higher 18.6 (15.1-22.7)compared to uninvolved adolescents 5.3(4.3-6.5), adjusted score. Similarly, the risk factor for self-harming behaviour was higher in bullying victims 3.4(2.4-4.7), adjusted score (gender, household type & income, home language spoken, parental education, Aboriginal/Torres Strait Islander heritage).The highest risk factors regarded experience of all three types of SBV OR(95%CI): 4.6 (3.2-6.6) and the lowest physical SBVOR(95%CI): 2.9 (1.7-4.8). The mediating effect of DS in the association between SBV and SI was not assessed.
Garish & Wilson 2010
Investigation of potential risk factors for deliberate self-harm among adolescents, specifically focusing on peer victimization and alexithymia, as well as the mediating effect of depressive symptoms.
Adolescents who reported at least one incidence of SI were more likely to report experiences of all types of SBV (df>302, t> 3.30, p<0.005). A positive association was also reported between SBV and SI (0.36, p<0.001). Depressive symptoms only partially mediated the relationship between SBV and SI, as the association between SBV and SI decreased (b weigh from 0.36 to 0.25) after controlling for depressive symptoms.
Giletta et al. 2012
Exploration of risk factors for NSSI including bullying victimization, in adolescents across three different countries (Italy, the Netherlands, and the USA).
Community-based adolescents 14-19 years; n= 1,862; RR:79.7%(n=82) in Italy; 89.3 % (n=675) in the Netherlands; and 50.6% (n=360) in the USA.
In multivariate analysis SBV was a risk factor for NSSI for the entire sample: sociodemographic covariates adjusted score (gender, age, ethnicity, and parental education) OR (95% CI): 1.96 (1.50-2.57), p<0.0001. In the subgroups: Italy: OR95%CI: 1.61 (1.08-2.41), p<0.05 The Netherlands: OR(95%CI) : 2.80(1.74-4.50), p<0.0001 USA: OR95%CI: 1.33(0.67-2.64), p<0.0001
Exploration of the association between the frequency of bullying involvement and both internalizing and externalizing problems, including deliberate self-harm.
Infrequent (1-2 times in ones’ lifetime compared to no SBV was a risk factor for engagement in SI (,, and grade) SI boys OR(95% CI): 1.93 to 2.17 (1.71 to 1.92 – 2.17 to 2.44), p<0.001 SI girls OR(95% CI): 2.07 to 2.23 (1.91 to 2.00 – 2.23 to 2.47), p<0.001 Even occasional SBV is associated with SI during adolescence.
Hay & Meldrum, 2010
Exploration of the hypotheses that bullying victimization is significantly related to NSSI & SI, mediated by DS.
Community- based adolescents 10-21 years; n=424; RR: 93%
NSSI was associated with SBV (B=0.32, p<0.001/ Adjust R2 =0.23), partially mediated by depressive symptoms (B=0.18, p<0.001/ Adjust R2 =0.28) controlled for age, gender, ethnicity/origin, impulsivity, type of parenting, family type, and school performance. reduction of 44% in the association
Jantzer et al. 2015
Exploration of the relationship between school bullying victimization (Type/Frequency) and NSSI/ SB.
Repetitive SBV was a risk factor for NSSI NSSI OR(95% CI): 11.75(5.54-24.94), p<0.001; Occasional SBV was a risk factor for NSSI NSSI OR(95% CI): 4.74 (2.36-9.54), p<0.001; There was a difference in the effect size between repetitive and occasional SBV: Variance explained regarding the prediction of SBV was R2=0.053 and for the prediction of NSSI R2=0.093 (no effect of grade/gender) => Dose-response relationship regarding the frequency of SBV Even occasional SBV had an impact on SI. The mediating effect of depressive symptoms was not assessed.
McMahon et al. 2010 (CASE study)
Exploration of the association between self-injury and psychological, life-style, and stressful life event- related factors in school adolescents.
SBV was a risk factor for a lifetime history of SI in both genders in univariate analysis: girls OR: 2.61 95% CI (1.97-3.46), p<0.0005 / boys OR: 4.07 95% CI (2.57-6.44), p<0.0005. In multivariate analysis to predict a lifetime history of SI, SBV remained a risk factor only for boys OR age-adjusted score (95% CI): 2.83 (1.50-5.36), p=0.001
Although depressive symptoms were reported as a risk factor for SI in univariate analysis for both genders OR 1.25-1.27; 95% (1.18-1.22 to 1.32; p<0.0005), it was not the case in multivariate analysis, implying that when other variables such as SBV exist, the independent effect of depressive symptoms on SI is not significant in adolescents.
Noble et al. 2011/ Kansas City/USA (Kauffman Teen Survey)
Exploration of the association between NSSI and perceived school safety & trust.
SBV in a high school group was a risk factor for NSSI B=0.41, SE:0.22, OR: 1.52, p<0.01. SBV in a middle school group was a risk factor for NSSI B=0.55, SE:0.18, OR: 1.73, p<0.01. The mediating effect of depressive symptoms was not assessed.
O’Connor et al. 2009/ Scotland, UK (CASE Study)
Exploration of the prevalence of DSH and related factors in Scottish adolescents
SBV was found to constitute a risk factor for SI in both genders (The lifetime prevalence of SBV was strongly associated with SI in both boys and girls): girls OR(95%CI): 3.09 (2.06-4.64), p<0.0001 / boys OR(95%CI): 2.18 (1.11-4.28), p<0.005
O’Connor et al. 2014/ N. Ireland (CASE Study)
Exploration of the prevalence of DSH and related factors (exposure to the Troubles & relevant internet/social media pictures) in Northern Irish adolescents
SBV was a risk factor for SI in both genders (The lifetime prevalence of SBV was strongly associated with SI, in both boys and girls): girls OR(95%CI): 2.09 (1.59-2.73), p=0.0001 / boys OR(95%CI): 2.24(1.25-4.01), p=0.007.
Thomas et al. 2017/ Australia (Australian & Adolescent Survey of Mental Health Wellbeing-Young Minds Matter Survey)
Exploration of the association between mental health, including self-harming & suicidal behaviour, and the three classes of bullying.
The risk factor for self-harming behaviour in those self-reported as bullying victims was higher compared to uninvolved participants OR(95%CI):7.32(5.15-10.40), adjusted score (gender, age). The mediating effect of DS on the association between SBV and SI was not assessed.
Fisher et al. 2012/England & Wales, UK (Environmental Risk Study)
Exploration of the association between the frequency of bullying victimization and occurrence of self-harm in early adolescence, as well as identifying which bullied children are at highest risk of self-harm.
Community-based twins 5-12 years; n= 2,232; RR:NR
Exposure to frequent SBV before the age of 12 years was a risk factor for SI at 12 years RRisk(95%CI): 3.53 (2.10-5.93)- reports from mother
RRisk (95% CI): 3.33 (1.91-5.82)-reports from child
Among the 62 children who engaged in self-harm, 35 (56%) had been victimized by bullying, while 42.8% of the self-harm cases could have been prevented 95% CI (23.1%-57.5%) if SBV could be eliminated (other factors remaining constant). 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, (b) have been physically abused by an adult, and (c) exhibit conduct disorder, borderline personality characteristics, depressive, and psychotic symptoms.
Garish &Wilson 2015/ New Zealand
Investigation of the prevalence and correlates of NSSI in adolescents, including school bullying victimization.
Community-based adolescents 15-16 years ( grade); n= 830; RR: 60%
SBV was a significant predictor of NSSI Cross-sectional association between Lifetime NSSI & SBV: r= 0.31; p<0.01; Cross-sectional association between NSSI in the previous 3-8 months & SBV: r= 0.21, p<0.01)
Giletta et al. 2015/ China
Investigation of the predictive effect of school bullying victimization on latent trajectories of suicide ideation and NSSI in adolescents.
Community-based adolescents 15-16 years ( grade); n= 565; RR: 90.5%
Overt & relational SBV significantly predicted NNSI after controlling for gender & depressive symptoms, irrespective of the frequency of these behaviours. SBV differentiated the low (none or very few episodes) from the high frequency OR(95%CI): 2.19 (1.42-3.39), the moderate from the low OR (95%CI): 1.23(0.80-1.89), and the moderate from the high NSSI group 1.02(0.98-1.07) , independent of depressive symptoms. SBV differentiated the low (none or very few episodes) from the high frequency OR(95%CI): 1.71 (1.19-2.47), the moderate from the low OR (95%CI): 0.99(0.68-1.42), and the moderate from the high NSSI group 1.74(01.18-2.56) , independent of depressive symptoms. Adolescents in the high frequency NSSI group had more severe SBV scores than adolescents in the low and moderate frequency group (dose-response effect) (p <0.01).
Heilbron & Prinstein 2010/USA
Exploration of whether overt and relational peer victimization predicts suicidal ideation and NSSI, both concurrently and longitudinally.
A univariate association was reported between overt SBV and NSSI, with boys reporting NSSI being more frequently bullied compared to those who did not report NSSI (MANCONA; p<0.05), while girls who reported NSSI were less frequently victims of overt SBV (MANCONA; p<0.05). However, there was no main effect of the SBV on the prediction of NSSI when controlled for depressive symptoms.
Lereya et al. 2015/ USA & UK (Avon Longitudinal Study of Parents & Children; Great Smoky Mountains Study)
Exploration of the effects of maltreatment and bullying victimization on mental health status (i.e., self-harm, suicidality, and depressive symptoms) in adolescents.
Children who were bullied by peers were significantly more likely to report SI compared to those who were not bullied ALSPAC Cohort: OR (95% CI): 1.7 (1.4-2.2), p<0.0001& GSMS Cohort:; OR (95% CI): 3.0 (1.2-7.7), p=0.002 The experience of SBV has similar, and in some cases even worse, long-term adverse effects on the mental health of young adults compared to abuse by parents.
Lereya et al. 2013/ UK (Avon Longitudinal Study of Parents & Children)
Exploration of the hypothesis that school bullying victimization/SBV between the age of 7 and 10 is directly associated with self-injurious behaviour in late adolescence (16-17 years old).
After controlling for all potential confounders SBV between the age of 7 and 10 was found to be associated with a greater risk of SI in late adolescence, based on reports from the child OR(95% CI): 1.78(1.29-2.46), the mother OR(95%):CI 1.70 (1.27-2.28) and the teacher OR95%CI: 4.57 (1.66-12.54). The association between SBV and SI was partially mediated by depressive symptoms (B=0.21; SE=0.036 p<0.0001), since path analysis showed that SBV indirectly increased the risk of SI through the development of depressive symptoms (B=0.07; SE=0.02; p<0.0001).
SBV: school bullying victimization; NSSI: nonsuicidal self-injury; RR: response rate; n: number of participants; NM: not mentioned; NA: not assessed; SI: self-injury; SB: suicidal behaviour; and RRisk: relative risk.
