The Association between Deliberate Self-Harm and School Bullying Victimization and the Mediating Effect of Depressive Symptoms and Self-Stigma: A Systematic Review
Table 3
Methodological characteristics of the studies in the present systematic review.
Demographic data; income; family type; immigrant status; religiosity; psychopathology; suicidality; anxiety & depressive symptoms; substance abuse; parenting; social relationship problems & loneliness; quality of parenting & communication with parents; impulsivity
Self-reported data. No longitudinal data, thus the study cannot provide information about causality; no triangulation of data with teachers/ parents/peer nomination reports
Moderate quality
Brunstein Klomek et al. 2016/ 10 European countries
Demographic data; income; family type; immigrant status; religiosity; psychopathology; suicidality; anxiety & depressive symptoms; substance abuse; parenting & support; social relationship problems, peer support & loneliness; quality of communication with parents; impulsivity; prosocial behaviour
Self-reported data; the cross-sectional nature of the study does not allow assumptions on causality; no triangulation of data with teachers/ parents/ peer nomination reports
Self-reported data; no triangulation of data with teacher/ parent/ peer nomination reports; only the presence/absence of NSSI was assessed; data were gathered at one point in time; no conclusions on causality; important confounders were not assessed, i.e., impulsivity, drug abuse, self-esteem, mental health problems
Moderate quality
Elgar et al. 2014/USA
Cross-sectional, observational & correlational study
Self-reported data; no triangulation of data with teachers/ parents/ peer nomination reports; cross-sectional design, thus no conclusions on causality; important confounders were not included, e.g. substance use.
Moderate quality
Ford et al. 2017/Australia
Cross-sectional study
Random sample; median age NR (range: 14-15); n=2304
Self-reported data & recall bias; no triangulation of data with teacher/ parent/ peer nomination reports; cross-sectional design, thus no conclusions on causality; important confounders were not assessed, i.e. substance misuse, self-esteem, impulsivity; generalizability limited to adolescents of European origin with a high socioeconomic status.
Age, gender, ethnicity and parents’ educational level; depressive symptoms; family & peer related loneliness; peer preference (interpersonal stressors); substance use
Self-reported data; no triangulation of data with teachers/ parents/ peer nomination reports; cross-sectional design, thus no conclusions on causality; important confounders were not assessed, i.e. self-esteem, impulsivity; low response rate in the subgroups; convenience sample
Moderate quality
Gower & Borowsky 2013/ USA
Cross-sectional study
Convenience sample; mean age: NR; n=128,681
Self-reported questionnaires, psychometric scales
Age; gender; ethnicity; impulsivity; suicidality; personal & parental mental health problems & substance use; emotional distress; family conflict & running away; skipped school; negative self-concept; religious activities; supportive social network; conduct problems; depressive and anxiety symptoms; emotional & physical domestic violence; physical & sexual abuse; witness to domestic violence; parent connectedness; academic performance; family structure & income; residency
Self-reported data; no triangulation of data with teacher/ parent/ peer nomination reports; no causality
Moderate quality
Hay & Meldrum, 2010/ USA
Cross-sectional, noncomparative, correlational study
Age; gender; ethnicity/origin; family type; school performance; impulsivity; authoritative parenting
No triangulation of data with teacher/ parent/ peer nomination reports; no causality; convenience sample; no assessment of mental health variables as confounders
Moderate quality
Jantzer et al. 2015/ Germany
Cross-sectional, noncomparative, correlational study
Entire target population; mean age: 12.8 years; n= 647
No triangulation of data with teacher/ parent/ peer nomination reports; no causality; convenience sample; no assessment of important confounders, e.g. impulsivity; self-esteem, etc.
Low quality
McMahon et al. 2010/Ireland
Cross-sectional, noncomparative, correlational study
Anxiety; impulsivity; self-esteem; DSH of a friend/ family member; drug use; sexual abuse; friendship difficulties; fights with parents; dysfunctional school performance
No triangulation of data with teacher/ parent/ peer nomination reports; no causality assessment; no assessment of important confounders, e.g. substance use; no assessment of social support variables; exclusion of those who did not describe DSH behaviour (risk of underestimation of the prevalence)
Moderate quality
Noble et al. 2011/ USA
Cross-sectional, comparative study
Purposeful (matched groups) sample; mean age: 14.9 years; n= 1,276
Perceived trust in school context (trust in students/teachers/administration/school counsellor) & safety (missed days due to feeling unsafe; carrying a weapon/ threatened/being bullied at school)
No triangulation of data with teacher/ parent/ peer nomination reports; no causality assessment; convenience sample; no assessment of mental health variables as confounders; social support variables were not included
Moderate quality
O’Connor et al. 2009/Scotland, UK
Cross-sectional, noncomparative, correlational study
Anxiety & depression symptoms; impulsivity; self-esteem; DSH of a friend/ family member; drug use; sexual abuse; friendship difficulties; fights with parents; dysfunctional school performance
No triangulation of data with teacher/ parent/peer nomination reports; no causality assessment; no assessment of social support variables
Moderate quality
O’Connor et al. 2014/ Northern Ireland
Cross-sectional, noncomparative, correlational study
Anxiety; depression; impulsivity; self-esteem; DSH of a friend/ family member; drug/alcohol use; sexual/physical abuse; sexual orientation concerns; exercising; living with both parents; exposure to internet/TV DSH images; exposure to difficulties related to “The Troubles”
No triangulation of data with teacher/ parents/ peer nomination reports; no causality assessment; no assessment of social support variables
Moderate quality
Thomas et al. 2017/ Australia
Cross-sectional, correlational study
Random, nationally representative sample; mean age: 14.6 years; n=2967
Cross-sectional design, thus no conclusions on causality; important confounders were not assessed, i.e. substance misuse, self-esteem, impulsivity, etc; over-representation of socially/income advantaged families
High quality
Fisher et al. 2012/ UK
Longitudinal birth cohort, comparative study
Birth cohort sample; mean age: NR; n= 2,232
Clinical interviews of mothers/children /teachers: open-ended questions & psychometric scales
Exposure to physical/ sexual maltreatment; anxiety symptoms; depressive symptoms; withdrawn, aggressive & delinquent behaviour; IQ;
The small number of children who engaged in self-injurious behaviour led to biased estimations about the size of the association between the main variables; no inclusion of important confounders, i.e. substance use; parenting & social support variables were not included
High quality
Garisch & Wilson, 2015/New Zealand
Prospective study with measurement at two time points
Random sample; mean age: 16.34(T1) -16.45(T2) years; n=830
Moderate internal consistency & test-rest reliability of the instruments applied
High quality
Giletta et al. 2015/China
Prospective cohort comparative study
Random sample; grade (mean age: 16 years); n=565
Self-reported questionnaires & peer nominated data
Gender; suicidal ideation; depressive symptoms; stressful peer experiences/ type and quality of friendships; friend support
Peer nominated data regarding overt and relational school bullying victimization, excluding subjective experiences; the low NSSI/ SI trajectory group included those reporting no or very few episodes, thus the degree to which school bullying victimization differentiated those who engaged in SITB from those who did not at all was not clearly reported; no assessment of substance use as a confounder
High quality
Heilbron & Prinstein 2010/ USA
Longitudinal, population-based comparative cohort study
Only partially ethnically diverse sample; no inclusion of important confounders in the analysis/study design, i.e., substance abuse, impulsivity, suicidal behaviour (suicide attempts & plans); parenting/ social support was not assessed; small group sizes in the internal comparisons
High quality
Lereya et al. 2015/ USA & UK
Comparative study of longitudinal birth cohort & population-based data.
Cohort sample; mean age: NR; n= 5,446
Self-reported postal questionnaires about self-harm variables from the adolescents at the age of 16-17 years; face-to-face interviews with the children at the age of 8 & 10 years/mothers/ teachers about predictor variables
Gender; ethnicity; parents’ educational level & marital status; parental mental health problems; parental stress; family conflict; preschool maladaptive parenting (hitting, shouting, hostility); conduct problems; hyperactivity; depressive and anxiety symptoms; emotional & physical domestic violence; borderline personality disorder symptoms; sexual abuse
Self-reported data & recall bias; face-to-face interviews & embarrassment bias
High quality
Lereya et al. 2013/ UK
Longitudinal birth cohort, comparative study
Birth cohort sample; mean age: NR; n= 4,810
Self-reported postal questionnaires about self-harm variables from the adolescents at the age of 16-17 years; face-to-face interviews with the children at the age of 8 & 10 years/mothers/ teachers about predictor variables.