CC BY
0NEW RESEARCH
Being Bullied During Childhood and the Prospective Pathways to Self-Harm in Late
Adolescence
Suzet Tanya Lereya, Ph.D., Catherine Winsper, Ph.D., Jon Heron, Ph.D., Glyn Lewis, Ph.D., David Gunnell, D.Sc., Helen L. Fisher, Ph.D., Dieter Wolke, Ph.D.
Objective: To assess whether being bullied between 7 and 10 years of age is directly associated with self-harm in late adolescence when controlling for previous exposure to an adverse family environment (domestic violence, maladaptive parenting); concurrent internalizing and externalizing behavior; and subsequent psychopathology (borderline personality disorder and depression symptoms). Method: A total of 4,810 children and adolescents in the Avon Longitudinal Study of Parents and Children (ALSPAC) cohort were assessed to ascertain bullying exposure (between 7 and 10 years of age) and self-harm at 16 to 17 years. Results: A total of 16.5% of 16- to 17-year-olds reported self-harm in the previous year. Being bullied was associated with an increased risk of self-harm directly, and indirectly via depression symptoms in early adolescence. The association between an adverse family environment (exposure to maladaptive parenting and domestic violence) and self-harm was partially mediated by being bullied. Conclusions: Being bullied during childhood increases the risk of self-harm in late adolescence via several distinct pathways, for example, by increasing the risk of depression and by exacerbating the effects of exposure to an adverse family environment; as well as in the absence of these risk exposures. Health practitioners evaluating self-harm should be aware that being bullied is an important potential risk factor. J. Am. Acad. Child Adolesc. Psychiatry, 2013;52(6):608-618. Key Words: Avon Longitudinal Study of Parents and Children (ALSPAC), bullying, depression, self-harm, victimization
Self-harm is a widespread problem, with a self-reported prevalence of 14% to 17% among adolescents and young adults in the United States.1,2 It results in a large number of presentations to hospitals, leading to high economic cost.3 Typical self-harm behaviors include cutting, burning, or swallowing pills.4,5 Self-harm may be used to relieve tension or to communicate stress, and, in the most extreme cases, may represent acts with suicidal intent.6 Delineating the developmental antecedents of self-harm and highlighting at-risk groups is important, as single episodes often lead to a repetition of such behavior,7
and self-harm is a key predictor of completed suicide.8
Definitions of self-harm within the extant literature sometimes incorporate suicidal intent9,10 and sometimes exclude this factor.11,12 The extent to which these 2 constructs represent separate behaviors, with different risk and protective factors, rather than extreme variations of the same behavior, remains unclear.13 Recent studies suggest that being bullied from early to mid-childhood is predictive of self-harm (both with and without inclusion of suicidal intent) at 11 to 12 years of age.10,11,14-16 Other factors associated with a risk of self-harm include exposure to domestic violence,17 conflict in parent-adolescent relationships,12 and female sex.18 Furthermore, high rates of psychiatric disorders, including depression9 and borderline personality disorder (BPD),19 have been associated with self-harm. Being bullied has been identified as a consequence, and precursor, of psychopathologies20,21
Clinical guidance is available at the end of this article.
This article can be used to obtain continuing medical education (CME) at www.jaacap.org.
^ Supplemental material cited in this article is available onlir
that are also associated with self-harm, suggesting that being bullied in childhood may represent a marker of present and later psychopathology, rather than a direct cause of self-harm.22 Therefore, further research is required to delineate the etio-logical pathways involving being bullied in childhood to self-harm during late adolescence, while controlling for pre-existing and concurrent risk factors and psychopathology.23
In a previous study,10 we found that being a victim of bullying between 4 and 10 years was associated with self-harm at 11 to 12 years, after controlling for potential confounders. We aim to expand on these findings by investigating the longer-term consequences of being bullied during childhood (between 7 and 10 years), and by delineating multiple pathways to self-harm during late adolescence (16-17 years). Using path analysis, confounding factors occurring before, during, and after being bullied can be controlled for, and the mediating relationships between early risk exposures, being bullied, psychopa-thology and later self-harm quantified. The specific research questions investigated are as follows:
• Is being bullied (child, mother, and teacher report) from 7 to 10 years associated with self-harm during late adolescence?
• Is the effect of being bullied on self-harm direct, or are the pathways mediated by depression or BPD symptoms in early adolescence?
• Does this association vary according to risk exposures occurring before (sex of child, exposure to maladaptive parenting, and domestic violence) and during (internalizing and externalizing behavior) exposure to bullying?
method
Data Source
The Avon Longitudinal Study of Parents and Children (ALSPAC) is a birth cohort study based in the United Kingdom. The cohort comprises children born to residents of the former Avon Health Authority area in South West England who had an expected delivery between April 1, 1991, and December 31, 1992. A total of 13,971 children were alive at 12 months, forming the original cohort. Ethical approval was obtained from the ALSPAC Law and Ethics committee and the local research committees. From the first trimester of pregnancy, parents completed postal questionnaires about themselves and the study child's health and development. Children were invited to attend annual assessment clinics, including face-to-face interviews and psychological and physical tests from age 7 years. Our
study is based on 4,810 children who answered the self-harm questionnaire at age 16 to 17 years.
Outcome Variable
Self-harm, in this study, is defined as an act with nonfatal outcome in which an individual deliberately hurts him- or herself with or without the intention to die.24 The data were collected from participants 16 to 17 years of age (mean = 16.7 years; SD = 0.2 year), using a self-completion postal questionnaire. Participants were asked: "Have you ever hurt yourself on purpose in any way (e.g., by taking an overdose of pills or by cutting yourself)?" Those adolescents who responded positively were asked further questions regarding frequency and how they had hurt them-selves.4 This study focuses on adolescents who harmed themselves in the previous year only (yes = 792 [16.5%]; no = 4,018 [83.5%]) to preserve the time ordering of the analyses, that is, to verify that the risk exposures occurred before self-harming behavior.
Predictor Variables
Being bullied was assessed using child, mother, and teacher reports. Child reports were collected at 8 and 10 years, using a modified version of the Bullying and Friendship Interview Schedule (detailed in Wolke et al.20). There were 5 questions pertaining to experience of overt bullying: personal belongings taken; threatened or blackmailed; hit or beaten up; tricked in a nasty way; called bad/nasty names. There were also 4 questions pertaining to relational bullying: exclusion to upset the child; pressure to do things s/he didn't want to do; lies or nasty things said about others; and games spoiled. Because of the skewed distribution of responses, overt bullying was coded categorically as present if the participant confirmed that at least 1 of the 5 behaviors occurred repeatedly (4 or more times in the past 6 months) or very frequently (at least once per week in the past 6 months). Similarly, relational bullying was coded as present if the child confirmed that at least 1 of the 4 behaviors occurred repeatedly or very frequently.25 The following victimization variables were derived: whether the children experienced any bullying (overt and/or relational versus neither); chronicity of being bullied, defined as unstable (reported only at age 8 years or age 10 years), stable (reported at both age 8 years and age 10 years), or never been bullied (none).25 Mother and teacher reports were derived from a single item of the Strengths and Difficulties Questionnaire26: "child is picked on or bullied by other children." If the response was "somewhat applies" or "certainly applies" at any time point (mother: 7, 8, and 9 years; teacher: 7 and 10 years), the child was considered a mother or teacher reported victim.10 In addition, mother (not bullied; unstable = 1 time point; stable = 2 or 3 time points) and teacher (not bullied; unstable = 1 time point; stable = 2 time points) chronicity variables
were constructed.10 The overall agreement rates between informants were as follows: for mothers and children, k— 0.21, p < .001; for mothers and teachers, k — 0.18, p < .001; and for teachers and children, k — 0.10, p < .001, which are largely consistent with
previous reports. Confounding Factors
A preschool maladaptive parenting variable was constructed using mother reported hitting (daily or weekly at 2 and/or 3.5 years), shouting (daily at 2 and/or 3.5 years) and hostility.27 Hostility (1.8 and/or 4 years) was constructed from 4 items, for example, "mum/ mom often feels irritated by child," "mum/mom has battle of wills with child," previously identified as loading onto 1 distinct factor.27 Hostility was recorded as present if 3 or more items were reported. Malad-aptive parenting was categorized as follows: none, mild (1 or 2 indicators), and severe (3 indicators).28 Domestic violence was considered present if the mother/partner reported that there was emotional and/or domestic physicalviolence (0.7,1.8, 2.8, 4 years) and/or conflictual partnership (2.8 years, e.g., "shouting or calling partner names").28 An internalizing/ externalizing behavior variable was estimated using the sum of negative emotionality, hyperactivity, and conduct problems taken from the Strengths and Difficulties Questionnaire (SDQ),26 reported by the mother across the 3 time-points of 7, 8, and 9 years.
Potential Mediating Factors Between Being Bullied and Self-Harm
Borderline personality disorder symptoms were assessed at 11.7 years using the semi-structured Childhood Interview for DSM-IV Borderline Personality Disorder, UK Version (CI-BPD-UK); based on the borderline module of the Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV).29 The interview comprised 9 sections: intense inappropriate anger; affective instability; emptiness; identity disturbance; paranoid ideation; abandonment; suicidal or self-mutilating behaviors; impulsivity; and intense unstable relationships. A symptom was classified as definitely present if it occurred daily or approximately 25% of the time,30 and as probable if it occurred repeatedly but did not meet criteria for definitely present. The BPD outcome was based on the presence of 5 or more (probable/definite) symptoms.20 The self-harm symptom item was removed to avoid collin-earity between the exposure and outcome. A total of 6.4% participants (n — 224) reported BPD symptoms.
Depression symptoms were assessed using the Short Mood and Feelings Questionnaire (SMFQ),31 administered at 12, 13 (mother report), and 14 (child report) years. Each item is rated on a 3-point scale with respect to events from the previous 2 weeks. Positive items were summed yielding a total score (maximum of 26 points). Scores were collapsed into
a dichotomous variable according to previously identified cut-points (scores of <11 indicated nonclinical symptoms, whereas scores of >11 indicated clinically relevant depressive symptoms).32,33 A total of 9.2% participants (n — 418) had depression symptoms at any time-point.
Statistical Methods
Selective dropout was determined by comparing those participants who completed the self-harm questionnaire to those who dropped out, using logistic regression analyses. Response rates significantly differed according to sex, ethnicity, birth weight, marital status, home ownership, educational level of the mother, and family adversity (see Table S1, available online). Subsequently, we conducted a weighted analysis using inverse probability (of having missing outcome data) weights to account for those lost to follow-up. Using the variables associated with selective drop-out as the independent variables, we fitted a logistic regression model (response vs. nonresponse as outcome) to determine weights for each individual using the inverse probability of response.34 Associations were remarkably similar for the unweighted and weighted data, and thus we used the unweighted data in all subsequent analysis.
Analyses were conducted in 3 stages. First, to assess whether being bullied at school is associated with self-harm, 3 sets of binary logistic regression analyses were conducted (Table 1) using SPSS version 18 software. Model A is based on the full data showing unadjusted analyses. Model B controlled for sex, preschool domestic violence, preschool maladaptive parenting, and internalizing/externalizing behavior. Model C included all of the preceding variables, and also controlled for BPD symptoms and depression symptoms. Analyses were repeated for acts of self-harm with and without the intention to die; but as the results were almost identical, we combined these acts to maximize statistical power. Second, multiple mediation analysis was performed in Stata version 12.1 software to examine the extent to which the association between being chronically bullied and self-harm was mediated by depression symptoms and BPD symptoms, while controlling for sex, preschool domestic violence, preschool maladaptive parenting, and internalizing/ externalizing behavior. The mediational variables were first entered simultaneously to examine their combined effect on the association between victimization and self-harm, and were then entered separately to investigate their individual impact on this relationship. Karlson, Holm, and Breen's khb command was used, which can be used with a combination of dichotomous, continuous and ordinal variables and which provides standardized coefficients.35 Results are presented as odds ratios and 95% confidence intervals. Third, path analysis was conducted using Mplus version 6.12,36 to assess associations between being bullied and self-harm, while
TABLE 1 Crude and Adjusted Associations Between Being Bullied and Self-Harm (Yes vs. No)
Model Aa Model Bb Model Cc
Bullying status OR (95% CI) OR (95% CI) OR (95% CI)
Child report (8 years), n 3,712d 3,037d 2,563d
No [reference]6 [reference] [reference]
Yes 1.37(1.15-1.63) 1.40 (1.14-1.71) 1.30(1.04-1.63)
Child report (10 years), n 3,908 3153 2,713
No [reference] [reference] [reference]
Yes 1.55 (1.28-1.87) 1.47 (1.18-1.84) 1.43 (1.12-1.82)
Chronicity (child report), n 4,181 3,330 2,782
None [reference] [reference] [reference]
Unstable 1.43 (1.20-1.71) 1.44 (1.17-1.77) 1.38 (1.10-1.74)
Stable 1.78 (1.38-2.30) 1.79 (1.34-2.41) 1.66(1.20-2.31)
Mother report (7 years), n 4,114 3,519 2,737
No [reference] [reference] [reference]
Yes 1.46 (1.18-1.79) 1.45 (1.14-1.84) 1.50 (1.14-1.97)
Mother report (8 years), n 4,083 3,609 2,798
No [reference] [reference] [reference]
Yes 1.62 (1.34-1.96) 1.56 (1.26-1.94) 1.60(1.25-2.06)
Mother report (9 years), n 4,093 3,532 2,742
No [reference] [reference] [reference]
Yes 1.51 (1.25-1.82) 1.28 (1.02-1.59) 1.11 (.85-1.44)
Chronicity (mother report), n 4,557 3,623 2,810
None [reference] [reference] [reference]
Unstable 1.18 (0.97-1.43) 1.19 (.95-1.50) 1.13 (.87-1.47)
Stable 1.79 (1.45-2.21) 1.64(1.28-2.11) 1.59 (1.19-2.13)
Teacher report (7 years), n 2,199 1,823 1,423
No [reference] [reference] [reference]
Yes 1.74(1.24-2.44) 1.63 (1.06-2.50) 2.01 (1.22-3.30)
Teacher report (10 years), n 2,721 2,080 1,631
No [reference] [reference] [reference]
Yes 1.59 (1.21 -2.10) 1.37 (.94-1.99) 1.44 (.93-2.22)
Chronicity (teacher report), n 3,513 2,688 2,087
None [reference] [reference] [reference]
Unstable 1.52 (1.19-1.94) 1.24 (.90-1.70) 1.39 (.97-2.00)
Stable 2.68 (1.41-5.11) 3.50 (1.46-8.42) 4.75 (1.72-13.07)
Note: Boldface type indicates significant associations at p < .05. Stable = bullying reported at 2 or more time points; Unstable = bullying reported only at 1 time point. aCrude analysis.
bControlling for sex, preschool domestic violence, preschool maladaptive parenting, and internalizing/externalizing behavior.
cControlling for borderline personality disorder symptoms and depression symptoms in addition to sex, preschool domestic violence, preschool
maladaptive parenting, and internalizing/externalizing behavior. dNumber of participants in analysis.
eReference group in all analyses consists of participants who are not victims.
controlling for all potential confounding associations simultaneously. Domestic violence (emotional and physical domestic violence, and conflicting partnership), preschool maladaptive parenting (maternal shouting, hitting, and hostility), internalizing/externalizing behavior (negative emotionality, hyperactivity, and conduct problems at 7, 8, and 9 years), and being bullied (child report at 8 and 10 years; mother report at 7, 8, and 9 years; teacher report at 7 and 10 years) were specified as latent variables. Sex, adolescent psychopa-thology (BPD and depression symptoms). and self-harm were specified as observed variables. Figure 1
provides a representation of all potential pathways specified within the model. The weighted least squares with robust standard errors, mean, and variance adjusted (WLSMV) estimator was used because of its robustness when analyzing both continuous and categorical outcomes.37 Associations are reported as linear regression coefficients for latent dependent variables and probit coefficients for categorical observed dependent variables. Probit coefficients indicate the strength of association between predictor variables and the probability of group membership, and represent the difference that a 1-unit change in the predictor variable
FIGURE 1 Path diagram representing the pathways estimated between risk exposures and self-harm outcome. Note: Dotted lines indicate direct effects of sex on other variables.
makes in the cumulative normal probability of the outcome variable.36 Individuals with partially missing item-level data were included, and missing data were accommodated using a series of univariate and bivar-iate probit regressions that allow missingness to be a function of observed covariates.38 Finally, the "punaf" command in Stata (v12.1) was used to calculate the Population-Attributable Fraction (PAF) for self-harm based on being bullied (reported by child/adolescent, mother, or teacher).
results
Prevalence of Being Bullied and Self-Harm A total of 905 participants (18.8%; male, 180; female, 725) reported self-harm at any point in the past,4 and 792 (16.5%; male, 162; female, 630) reported harming themselves in the previous year. Of these 792 individuals, 306 (38.6%) harmed themselves once, 286 (36.1%) 2 to 5 times; 80 (10.1%) 6 to 10 times, and 120 (15.2%) more than 10 times. Although 579 adolescents (74.7%; male, 118; female, 461) self-harmed without an intention to die, 213 (26.9%; male, 44; female, 169) wanted to die. Cutting (n = 489; 61.8%) was the most commonly reported method of self-harm (details in Kidger et al.4). According to child report, 38% of children were bullied at 8 years and 22.9% at 10
years. According to mother report, 16% of children were bullied at 7 years, 20.5% at 8 years, and 21.5% at 9 years. According to teacher report, 8.7% of children were bullied at 7 years and 12.3% at 10 years. The relative prevalence according to informant is congruent with previous findings, suggesting that some instances of being bullied may go unnoticed by teachers.22 Among the 792 children who self-harmed, 514 (66%) were victims of bullying, according to child, mother, or teacher report. This yielded a Population-Attributable Fraction (PAF) of 19.9% (95% confidence interval = 12.3%-26.8%), indicating that if bullying could have been eliminated while other exposures remained constant, 20% of self-harm cases could potentially have been prevented.
Associations Between Being Bullied and Self-Harm, Controlling for Confounding Factors
In crude analysis (model A), there was a modest association between being bullied and self-harm, according to all respondents (Table 1). After controlling for sex, preschool domestic violence, preschool maladaptive parenting, and externalizing/internalizing behavior (model B), and after controlling for all potential confounders (model C), being bullied remained associated with
self-harm according to child reports at ages 8 and 10, mother reports at ages 7 and 8, and teacher report at age 7. Moreover, stable victimization remained strongly associated with self-harm according to child, mother, and teacher report.
BPD and Depression Symptoms as Mediators The direct and indirect pathways (via BPD and depression symptoms) between being bullied and self-harm are presented in Table 2. The odds ratios (OR) and 95% confidence intervals (CI) are shown for each pathway while controlling for sex, preschool domestic violence, preschool mal-adaptive parenting, and externalizing/internalizing behavior. Respectively, 13%, 13%, and 1% of the association between being a stable victim of bullying (according to child, mother, and teacher report) and self-harm was accounted for by both mediators. According to child and mother report,
there was a significant indirect pathway, from stable victimization to self-harm via depression symptoms.
Path Analyses
A single model was specified using bullying reports by child, mother and teacher. The root-mean square error of approximation (RMSEA) and the Comparative Fit Index (CFI) were used to assess model fit. RMSEA values less than 0.05,39 and CFI values greater than 0.90, indicate close fit.40 The data showed good fit to the model (RMSEA = 0.03; CFI = 0.97).
All direct associations between risk factors and self-harm outcome are shown in Table 3. All other direct pathways specified within the model (Figure 1) are presented in Table S2, available online. Being bullied between 7 and 10 years was directly associated with self-harm in late
Total OR (95% CI) Direct OR (95% CI) Indirect OR (95% CI) % Mediated
Child report victimization, n = 2,782
Unstable
All mediators 1.44 (1.14-1.81) 1.38 (1.10-1.74) 1.04 (.99-1.10) 12
BPD only 1.38 (1.10-1.74) 1.38 (1.10-1.74) 1.00 (.99-1.01) 0
Depression only 1.43 (1.14-1.81) 1.38 (1.10-1.74) 1.04 (.99-1.10) 11
Stable
All mediators 1.78 (1.29-2.46) 1.66 (1.20-2.30) 1.08 (.99-1.16) 13
BPD only 1.66(1.20-2.28) 1.66 (1.20-2.30) 1.00 (.94-1.06) 0
Depression only 1.76 (1.27-2.44) 1.66 (1.20-2.30) 1.06 (1.01-1.11) 10
Mother report victimization, n = 2,810
Unstable
All mediators 1.13 (.87-1.48) 1.13 (.87-1.47) 1.00 (.96-1.05) 2
BPD only 1.13 (.87-1.48) 1.13 (.87-1.47) 1.00 (.99-1.01) 0
Depression only 1.13 (.87-1.48) 1.13 (.87-1.47) 1.00 (.96-1.05) 1
Stable
All mediators 1.70(1.27-2.28) 1.59 (1.19-2.13) 1.07 (1.01 -1.13) 13
BPD only 1.60 (1.19-2.14) 1.59 (1.19-2.13) 1.00 (.98-1.03) 1
Depression only 1.68 (1.25-2.25) 1.59 (1.19-2.13) 1.06 (1.01-1.11) 10
Teacher report victimization, n = 2,087
Unstable
All mediators 1.45 (1.01-2.09) 1.40 (.97-2.01) 1.04 (.91-1.19) 10
BPD only 1.40 (.97-2.00) 1.40 (.97-2.01) 1.00 (.97-1.02) 0
Depression only 1.45 (1.02-2.08) 1.40 (.97-2.01) 1.03 (.91-1.18) 9
Stable
All mediators 4.57(1.66-12.54) 4.48 (1.63-12.33) 1.02 (.89-1.16) 1
BPD only 4.48 (1.63-12.30) 4.48 (1.63-12.33) 1.00 (.98-1.02) 0
Depression only 4.55 (1.66-12.52) 4.48 (1.63-12.33) 1.02 (.89-1.15) 1
Note: Boldface type indicates significant associations at p < .05. OR = odds ratio; Unstable = bullying reported only at 1 time point. Stable = bullying reported at both 2 or more time points. aAdjusting for sex, preschool maladaptive parenting, preschool domestic violence, internalizing/externalizing behavior, and borderline personality disorder (BPD) symptoms (for depression symptoms, only mediation model) and depression symptoms (for BPD symptoms, only mediation model).
TABLE 2 Associations Between Being Bullied and Self-Harm, Split Into Total Effects, Direct, and Indirect Pathways via Potential Mediators Together and Then via Each Individual Mediator on Its Own
TABLE 3 Nonstandardized Direct Associations Among Sex, Preschool Maladaptive Parenting, Domestic Violence, Internalizing/Externalizing Behavior, Being Bullied, Borderline Personality Disorder (BPD) and Depression Symptoms, and Self-Harm Outcome
Sex / Self-harmb,c Preschool domestic
violence / Self-harmb Preschool maladaptive
parenting / Self-harmb Being bullied / Self-harmb Internalizing/Externalizing
Behavior / Self-harmb BPD symptoms / Self-harmb Depression symptoms / Self-harmb
B SE p Value"
0.580 0.052 0.000
0.094 0.041 0.022
-0.046 0.054 0.392
0.235 0.063 0.000
-0.003 0.011 0.791
0.006 0.049 0.898
0.211 0.036 0.000
Note: Boldface type indicates significant associations at p < .05. B = nonstandardized probit coefficients. Model fit: Root mean square error of approximation (RMSEA) = 0.025 (0.023-0.027); confirmatory fit index (CFI) = 0.97. Values are given as non-standardized probit coefficients. aThe p value is 2-tailed. bProbit regression coefficient.
cA probit coefficient of 0.58 indicates that for each unit increase in sex (from male to female) there is an increase of 0.58 SDs in the predicted zscore ofthe cumulative normal distribution ofself-harm.
adolescence. For example, a probit coefficient of 0.24 indicates that a 1-unit increase in being bullied (i.e., going from not bullied to bullied) resulted in an increase of 0.24 SD in the predicted z score of self-harm (of the cumulative normal probability distribution) (Table 3). Female sex, preschool domestic violence, and depression symptoms directly related to self-harm. Being bullied directly increased the risk of BPD and depression symptoms in early adolescence (Table S2, available online).
Table 4 shows all indirect associations between risk factors and self-harm outcome. The association between being bullied and self-harm was partially mediated by depression symptoms. Being bullied partially mediated the relationships between domestic violence, maladaptive parenting, and subsequent self-harm. The indirect association between the sex of the child and self-harm via depression symptoms was significantly stronger for females. The indirect association from the sex of the child to self-harm via being bullied was significantly stronger for males (Table 4). Boys were more likely to be victims of bullying, to exhibit internalizing/externalizing behavior, and to be exposed to maladaptive parenting (Table S2, available online).
discussion
This study considers multiple etiological pathways from risk factors during early childhood onward, to self-harm during late adolescence. When assessing all potential pathways occurring before (maladaptive parenting, domestic violence, sex), concurrently with (internalizing/externalizing behavior), or after (depression and BPD symptoms) being bullied, we found that being a victim of bullying was associated with increased risk of self-harm. Furthermore, being bullied indirectly increased the risk of self-harm via depression, and mediated the association following preschool exposures, in 2 separate pathways.
An association between being bullied and self-harm has been reported in previous studies; however, most were cross-sectional.15,41-43 Our results extend findings from the few prospective studies10,11 by investigating the relationship between being bullied and self-harm while controlling for risk factors occurring before, during, and after exposure to being bullied simultaneously. Results showed that being bullied in mid to late childhood has serious consequences persisting into late adolescence. Moreover, for a significant proportion of youth who were bullied, self-harm was not a consequence of psychopathology such as depression or BPD symptoms. Self-harm during adolescence is usually precipitated by stressful life problems,44 and being bullied is independently associated with high levels of distress. Congruent with the experiential avoidance model, deliberately harming oneself may represent a maladaptive strategy, independent of recognized psychopa-thology, for escaping from distressing internal experiences.45 Our analyses revealed multiple etiological pathways to self-harm involving exposure to bullying. In 1 pathway, being bullied was associated with subsequent depression symp-toms,9,46 which in turn increased the risk of self-harm,47 suggesting that depression symptoms are a second mechanism via which the association between being bullied and subsequent self-harm manifests. In 2 further pathways, there were significant indirect associations from domestic violence and maladaptive parenting to self-harm via being bullied. The resulting stress of exposure to a negative family environment may have caused some children to become dysregulated in their behavior,48 thereby attracting negative attention from peers.49 Indeed, strong positive associations between internalizing/externalizing
TABLE 4 Nonstandardized Probit Coefficients (B) for Indirect Paths Between Sex, Being Bullied, Borderline Personality Disorder (BPD), and Depression Symptoms, Domestic Violence, Maladaptive Parenting, and Subsequent Self-Harm Outcome
Preschool Domestic Preschool Maladaptive Sex Being Bullied Violence Parenting
B SE p Value" B SE p Value" B SE p Value" B SE p Value"
Via being bullied Via BPD symptoms Via depression symptoms 0.04 0.01 .001 - - - 0.05 0.02 .001 0.06 0.02 .001 0.00 0.01 .898 0.04 0.03 .898 0.00 0.00 .902 0.00 0.00 .995 0.10 0.02 .000 0.07 0.02 .000 0.02 0.01 .107 0.01 0.01 .500
Note: Boldface type indicates significant associations at p < .05. Self-harm outcome is categorical (yes vs. no). A significant positive B value for sex demotes that the strength of association is significantly stronger for females than for males. aThe p value is 2-tailed.
behavior and being bullied were observed, and there were significant indirect associations from domestic violence and maladaptive parenting to self-harm via being bullied. This suggests that children exposed to maladaptive family experiences are more likely to become victims of bullying, subsequently increasing the risk of self-harm in late adolescence.
Boys were significantly more likely to be bullied, and the association between the sex of the child and self-harm via being bullied was stronger for boys; overall, girls were more likely to engage in self-harm and to develop depression symptoms.41,50'51 Furthermore, the indirect association between child sex and self-harm via depression symptoms was significantly stronger for girls. These findings are congruent with literature suggesting that girls are twice as likely as boys to experience depression during adolescence52 and are also more likely to turn their distress inward,53 that is, to engage in self-harm.
Contrary to previous research,19 we found no association between borderline personality disorder symptoms (BPD) and self-harm. Emotional and behavioral dysregulation are core features of BPD,54 and it is these features within the symptom constellation that appear to be most strongly associated with self-harm behavior.55 Therefore, controlling for both depression and internalizing/externalizing behavior within the path analyses may have occluded any relationship between BPD and self-harm. Furthermore, it has been suggested that the reported association between BPD and self-harm may have been inflated because of cultural bias within psychiatry.56 Although BPD is commonly diagnosed in persons who self-harm, these individuals may often display no other signs or symptoms of BPD.57
Our study has a number of strengths. We used a large prospective cohort, with multiple informants, providing converging evidence for an association between being bullied and self-harm. We controlled for a wide range of confounders associated with both bullying and self-harm. Taking a life-span approach, we considered pathways from negative exposures during toddler-hood to self-harm during late adolescence.
With regard to limitations, self-harm was measured via self-report rather than clinical examination; however, this method may have encouraged adolescents to be more honest in their answers than an interview situation.58 Not all participants completed the self-harm questionnaire. Those with greater family adversity were more likely to drop out. Nonetheless, empirical simulations demonstrate that, even when dropout is correlated with predictor/confounder variables, the relationship between predictors and outcome is unlikely to be substantially altered by selective dropout processes.59 Indeed, our initial analyses with weighted and unweighted data yielded very similar results.
As we were using path analysis to determine prospective and mediational associations between various risk exposures and self-harm, variables were entered at discrete time points (based on theoretical models) to ensure that certain exposures preceded others, that is, bullying was assessed before depression, which was assessed before self-harm. Therefore, it is possible that, in some cases, the significant indirect and prospective associations between risk factors and subsequent self-harm outcome were the result of continuing exposure to risk factors rather than long-term effects.
Because of the very low prevalence of mother-reported sexual abuse in this sample (0.05%), it
JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY
VOLUME 52 NUMBER 6 JUNE 2013
was excluded as a confounder in the analysis. However, meta-analysis reveals that the association between sexual abuse and self-harm is relatively small, and that when psychiatric risk factors are controlled for, sexual abuse, explains little or no unique variance in self-harm behavior.60
Our results indicate that being bullied is a potent risk factor for self-harm. As suggested by the Population-Attributable Fraction (PAF), if bullying could have been eliminated while other exposures remained constant, 20% of self-harm cases could potentially have been prevented. This level of attributable fraction is considerable when compared, for example, to being obese (body mass index [BMI] >30) which occurs in 15% of the population but accounts for only about 2.8% of myocardial infarctions.61 Therefore, prevention of bullying is important to reduce the risk of self-harm. Not all victims of bullying who engage in self-harm show the typical psycho-pathological profiles (depression, BPD symptoms) that have previously been linked with self-harm, but may present with physical symptoms such
Clinical Guidance
Being bullied by peers is highly distressing for children and adolescents, and substantially increases the risk of self-harm.
Many children and adolescents do not tell their teachers or parents and suffer in silence. Clinicians should routinely ask children and adolescents about experiences of being bullied.
Although many children and adolescents who are bullied do not show overt depression, they may present a range of nonspecific symptoms,23 such headaches, backaches, stomach aches, dizziness, sleep problems, may be reluctant to go to school, may withdraw from social activities, and may self-harm.
Targeted interventions may place emphasis on teaching children how to deal with bullying and how to cope with emotional distress arising from being bullied.
If bullying could be eliminated while other exposures remained constant, 20% of self-harm cases could potentially be prevented. This level of attributable fraction is considerable when compared, for example, to being obese (body mass index [BMI] >30) which occurs in 15% of the population but accounts for only about 2.8% of myocardial infarctions.61
as headaches, backaches, stomach aches, and dizziness.62 Subsequently, general practitioners should also be aware of the physical indicators of bullying to identify at-risk youths.63 Targeted interventions should focus on improving the ways in which children and adolescents cope with emotional distress arising from being bullied, and mental health practitioners evaluating self-harm should consider being bullied as a serious potential risk factor. &
Accepted March 27, 2013.
Drs. Lereya, Winsper, and Wolke are with the University of Warwick, Coventry. Dr. Wolke is also with Warwick Medical School. Drs. Heron, Lewis, and Gunnell are with the School of Social and Community Medicine at the University of Bristol. Dr. Fisher is with the Medical Research Council (MRC) Social, Genetic, and Developmental Psychiatry Centre at the Institute of Psychiatry, King's College London.
The UK Medical Research Council (grant ref: 74882), the Wellcome Trust (grant ref: 076467) and the University of Bristol provide core support for the Avon Longitudinal Study of Parents and Children (ALSPAC). This study was supported by grant ES/K003593/1 by the Economic and Social Research Council (ESRC) in the UK.
The authors are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. Special thanks to Andrea Waylen, Ph.D., and Jeremy Horwood, Ph.D., of the University of Bristol, who helped in the conduct of the study.
This article is the work of the authors, and Dieter Wolke and Suzet Tanya Lereya serve as guarantors for the content of the article.
Disclosure: Drs. Lereya, Winsper, Heron, Lewis, Gunnell, Fisher, and Wolke report no biomedical financial interests or potential conflicts of interest.
Correspondence to Dieter Wolke, Ph.D., Department of Psychology, University of Warwick, Coventry, CV4 7AL, UK; e-mail: D.Wolke@ warwick.ac.uk
0-8567/$36.00/©2013 American Academy of Child and Adolescent Psychiatry
http://dx.doi.org/10.1016/j.jaac.2013.03.012
references
1. Whitlock J, Eckenrode J, Silverman D. Self-injurious behaviors in a college population. Pediatrics. 2006;117:1939-1948.
2. Ross S, Heath N. A study of the frequency of self-mutilation in a community sample of adolescents. J Youth Adolesc. 2002; 31:67-77.
3. Corso PS, Mercy JA, Simon TR, Finkelstein EA, Miller TR. Medical costs and productivity losses due to interpersonal and self-directed violence in the United States. Am J Prev Med. 2007;32: 474-482.
4. Kidger J, Heron J, Lewis G, Evans J, Gunnell D. Adolescent self-harm and suicidal thoughts in the ALSPAC cohort: a self-report survey in England. BMC Psychiatry. 2012;12:69.
5. Hawton K, Rodham K, Evans E. By Their Own Hand. Philadelphia, PA: Jessica Kingsley Publishers; 2006.
6. Muehlenkamp JJ, Gutierrez PM. Risk for suicide attempts among adolescents who engage in non-suicidal self-injury. Arch Suicide Res. 2007;11:69-82.
7. Owens D, Horrocks J, House A. Fatal and non-fatal repetition of self-harm. Br J Psychiatry. 2002;181:193-199.
8. Hawton K, Houston K, Shepperd R. Suicide in young people. Study of 174 cases, aged under 25 years, based on coroners' and medical records. Br J Psychiatry. 1999;175:271-276.
9. Klomek AB, Marrocco F, Kleinman M, Schonfeld IS, Gould MS. Peer victimization, depression, and suicidiality in adolescents. Suicide Life Threat Behav. 2008;38:166-180.
10. Winsper C, Lereya T, Zanarini M, Wolke D. Involvement in bullying and suicide-related behavior at 11 years: a prospective birth cohort study. J Am Acad Child Adolesc Psychiatry. 2012;51: 271-282.
11. Fisher HL, Moffitt TE, Houts RM, Belsky DW, Arseneault L, Caspi A. Bullying victimisation and risk of self harm in early adolescence: longitudinal cohort study. BMJ. 2012;344:e2683.
12. Yates TM, Tracy AJ, Luthar SS. Nonsuicidal self-injury among "privileged" youths: longitudinal and cross-sectional approaches to developmental process. J Consult Clin Psychol. 2008;76:52-62.
13. Wichstrom L. Predictors of non-suicidal self-injury versus attempted suicide: similar or different? Arch Suicide Res. 2009;13: 105-122.
14. Jutengren G, Kerr M, Stattin H. Adolescents' deliberate self-harm, interpersonal stress, and the moderating effects of self-regulation: a two-wave longitudinal analysis. J Sch Psychol. 2011;49:249-264.
15. Garisch JA, Wilson MS. Vulnerabilities to deliberate self-harm among adolescents: the role of alexithymia and victimization. Br J Clin Psychol. 2010;49:151-162.
16. McMahon EM, Reulbach U, Keeley H, Perry IJ, Arensman E. Bullying victimisation, self harm and associated factors in Irish adolescent boys. Soc Sci Med. 2010;71:1300-1307.
17. Wagner BM. Family risk factors for children and adolescents suicidal behavior. Psychol Bull. 1997;121:246-298.
18. Evans WP, Marte RM, Betts S, Silliman B. Adolescent suicide risk and peer-related violent behaviors and victimization. J Interpers Violence. 2001;16:1330-1348.
19. Klonsky ED, Oltmanns TF, Turkheimer E. Deliberate self-harm in a nonclinical population: prevalence and psychological correlates. Am J Psychiatry. 2003;160:1501-1508.
20. Wolke D, Schreier A, Zanarini MC, Winsper C. Bullied by peers in childhood and borderline personality symptoms at 11 years of age: a prospective study. J Child Psychol Psychiatry. 2012;53:846-855.
21. Reijntjes A, Kamphuis JH, Prinzie P, Telch MJ. Peer victimization and internalizing problems in children: a meta-analysis of longitudinal studies. Child Abuse Negl. 2010;34:244-252.
22. Ranning J, Sourander A, Kumpulainen K, et al. Cross-informant agreement about bullying and victimization among eight-year-olds: whose information best predicts psychiatric caseness 10-15 years later? Soc Psychiatry Psychiatr Epidemiol. 2009;44:15-22.
23. Srabstein J, Piazza T. Is there a syndrome of bullying? Int J Adolesc Med Health. 2012;24:91-96.
24. Madge N, Hewitt A, Hawton K, et al. Deliberate self-harm within an international community sample of young people: comparative findings from the Child and Adolescent Self-harm in Europe (CASE) Study. J Child Psychol Psychiatry. 2008;49:667-677.
25. Schreier A, Wolke D, Thomas K, et al. Prospective study of peer victimization in childhood and psychotic symptoms in a nonclin-ical population at age 12 years. Arch Gen Psychiatry. 2009;66: 527-536.
26. Goodman R. The strengths and difficulties questionnaire: a research note. J Child Psychol Psychiatry. 1997;38:581-586.
27. Waylen A, Stallard N, Stewart-Brown S. Parenting and health in mid-childhood: a longitudinal study. Eur J Public Health. 2008;18: 300-305.
28. Winsper C, Zanarini M, Wolke D. Prospective study of family adversity and maladaptive parenting in childhood and borderline personality disorder symptoms in a non-clinical population at 11 years. Psychol Med. 2012;42:2405-2420.
29. Zanarini M, Frakenburg F, Sickel A, Yong L. The Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV). Belmont, MA: McLean Hospital; 1996.
30. Zanarini M, Frankenberg F. Attainment and maintenance of reliability of axis I and II disorders over the course of a longitudinal study. Comp Psychiatry. 2001;42:369-374.
31. Angold A, Costello EJ, Messer SC, Pickles A, Winder F, Silver D. The development of a short questionnaire for use in epidemio-logical studies of depression in children and adolescents. Int J Method Psychol. 1995;5:237-249.
32. Patton GC, Olsson C, Bond L, et al. Predicting female depression across puberty: a two-nation longitudinal study. J Am Acad Child Adolesc Psychiatry. 2008;47:1424-1432.
33. Angold A, Erkanli A, Silberg J, Eaves L, Costello EJ. Depression scale scores in 8-17-year-olds: effects of age and gender. J Child Psychol Psychiatry. 2002;43:1052-1063.
34. Kinner SA, Alati R, Najman JM, Williams GM. Do paternal arrest and imprisonment lead to child behaviour problems and substance use? A longitudinal analysis. J Child Psychol Psychiatry. 2007;48:1148-1156.
35. Karlson KB, Holm A. Decomposing primary and secondary effects: a new decomposition method. Res Soc Strat Mobil. 2011;29: 221-237.
36. Muthen LK, Muthen BO. Mplus Statistical Analysis with Latent Variables User's Guide. Sixth ed. Los Angeles, CA: Muthen and Mutheen; 2010.
37. Flora DB, Curran P. An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychol Methods. 2004;9:466-491.
38. Mutheen BO, du Toit SHC, Spisic D. Robust inference using weighted least squares and quadratic estimating equations in latent variable modeling with categorical and continuous outcomes. 1997. Available at: http://gseis.ucla.edu/faculty/ muthen/articles/Article_075.pdf. Accessed February 1, 2013.
39. Browne MW, Cudeck R. Alternative ways of assessing model fit. In: Bollen KA, Long JS, editors. Testing Structural Equation Models. Newbury Park, CA: Sage; 1993:136-162.
40. Hu L-t, Bentler PM, Kano Y. Can test statistics in covariance structure analysis be trusted? Psychol Bull. 1992;112:351-362.
41. Hawton K, Rodham K, Evans DE, Weatherall R. Deliberate self harm in adolescents: self report survey in schools in England. BMJ. 2002;325:1207-1211.
42. Hay C, Meldrum R. Bullying victimization and adolescent self-harm: testing hypotheses from general strain theory. J Youth Adolesc. 2010;39:446-459.
43. Hilt LM, Cha CB, Nolen-Hoeksema S. Non-suicidal self-injury in young adolescent girls: moderators of the distress-function relationship. J Consult Clin Psychol. 2008;76:63-71.
44. Harrington R. Depression, suicide and deliberate self-harm in adolescence. Br Med Bull. 2001;57:47-60.
45. Chapman AL, Gratz KL, Brown MZ. Solving the puzzle of deliberate self-harm: the experiential avoidance model. Behav Res Ther. 2006;44:371-394.
46. Hawker DSJ, Boulton MJ. Twenty years' research on peer victimization and psychosocial maladjustment: a meta-analytic review of cross-sectional studies. J Child Psychol Psychiatry. 2000;41:441-455.
47. Hawton K, Kingsbury S, Steinhardt K, James A, Fagg J. Repetition of deliberate self-harm by adolescents: the role of psychological factors. J Adolesc. 1999;22:369-378.
48. Griffin MG, Resick PA, Yehuda R. Enhanced cortisol suppression following dexamethasone administration in domestic violence survivors. Am J Psychiatry. 2005;162:1192-1199.
49. Shields A, Cicchetti D. Parental maltreatment and emotion dys-regulation as risk factors for bullying and victimization in middle childhood. J Clin Child Psychol. 2001;30:349-363.
50. Kim YS, Leventhal B, Koh YJ, Boyce W. Bullying increased suicide risk: prospective study of Korean adolescents. Arch Suicide Res. 2009;13:15-30.
51. Klomek AB, Sourander A, Niemela S, et al. Childhood bullying behaviors as a risk for suicide attempts and completed suicides: a population-based birth cohort study. J Am Acad Child Adolesc Psychiatry. 2009;48:254-261.
52. Nolen-Hoeksema S. Gender differences in depression. Curr Direct Psychol Sci. 2001;10:173-176.
53. Crawford TN, Cohen P, Brook JS. Dramatic-erratic personality disorder symptoms: I. Continuity from early adolescence into adulthood. J Personal Disord. 2001;15:319-335.
54. Crowell SE, Beauchaine TP, Linehan MM. A biosocial developmental model of borderline personality: elaborating and extending linehan's theory. Psychol Bull. 2009;135:495.
55. Andieon Oe , Ferrer M, Calvo N, et al. Exploring the clinical validity of borderline personality disorder components. Compr Psychiatry. 2013;54:34-40.
56. McAllister M. Multiple meanings of self harm: a critical review. Int J Adolesc Med Health. 2003;12:177-185.
57. Bunclark MCJ. Repeated self-injury and its management. Int Rev Psychiatry. 2000;12:48-53.
58. Safer DJ. Self-reported suicide attempts by adolescents. Ann Clin Psychiatry. 1997;9:263-269.
59. Wolke D, Waylen A, Samara M, et al. Selective drop-out in longitudinal studies and non-biased prediction of behaviour disorders. Br J Psychiatry. 2009;195:249-256.
60. Klonsky ED, Moyer A. Childhood sexual abuse and non-suicidal self-injury: meta-analysis. Br J Psychiatry. 2008;192: 166-170.
61. Yusuf S, Hawken S, Ounpuu S, et al. Obesity and the risk of myocardial infarction in 27000 participants from 52 countries: a case-control study. Lancet. 2005;366:1640-1649.
62. Wolke D, Woods S, Bloomfield L, Karstadt L. Bullying involvement in primary school and common health problems. Arch Dis Child. 2001;85:197-201.
63. Carr-Gregg M, Manocha R. Bullying. Aust Fam Physician. 2011;40: 98-102.
TABLE S1 Dropout Analysis With Regard to Availability of Self-Harm Questionnaire
Self-Harm Questionnaire Status Associations
Questionnaire Not Available Questionnaire Available Available vs. Not Available
Male 5,248 (57.3) 1,972 (41.0) [Reference]
Female 3,918 (42.7) 2,838 (59.0) 1.93 (1.80-2.07)
Ethnicity
White 7,036 (94.2) 4,438 (96.1) [Reference]
Non-white 431 (5.8) 178 (3.9) .66 (.55 .78)
Birth weight
> 2,499 g 8,514 (94.1) 4,537 (95.5) [Reference]
< 2,500 g 535 (5.9) 212 (4.5) .74 (.63-.88)
Marital status
Single 2,469 (29.5) 812 (17.2) [Reference]
Married 5,905 (70.5) 3,901 (82.8) 2.01 (1.84-2.20)
Home ownership
Mortgaged 5,581 (66.8) 3,978 (85.2) [Reference]
Rent 2,775 (33.2) 693 (14.8) .35 (.32-.38)
Educational level mother
Below O level 2,854 (36.9) 874 (18.7) [Reference]
O level or above 4,881 (63.1) 3,809 (81.3) 2.55 (2.34-2.78)
None 3,560 (42.3) 2,765 (58.2) [Reference]
1 or more adversities 4,852 (57.7) 1,989 (41.8) .53 (.49-.57)
Preschool domestic violence
No 4,723 (63.7) 3,264 (68.9) [Reference]
Yes 2,693 (36.3) 1,473 (31.1) .79 (.73-.86)
Preschool maladaptive parenting
No 2,287 (38.0) 1,806 (39.6) [Reference]
1 3,285 (54.6) 2,408 (52.8) .93 (.86-1.01)
2 444 (7.4) 350 (7.7) 1.00 (.86-1.16)
Being bullied (child report)0
No 1,997 (52.8) 2,306 (55.2) [Reference]
Yes 1,787 (47.2) 1,875 (44.8) .91 (.83.99)
Being bullied (mother report)b
No 3,261 (65.9) 2,985 (65.5) [Reference]
Yes 1,691 (34.1) 1,572 (34.5) 1.02 (.94-1.11)
Being bullied (teacher report)c
No 4,851 (81.8) 3,014 (85.8) [Reference]
Yes 1,082 (18.2) 499 (14.2) .74 (.66-.83)
BPD Symptoms
No 2,341 (92.9) 3,289 (93.6) [Reference]
Yes 179 (7.1) 224 (6.4) .89 (.73-1.09)
Depression Symptomsd
No 3,407 (90.7) 4,119 (90.8) [Reference]
Yes 351 (9.3) 418 (9.2) .99 (.85-1.14)
Note: Boldface type indicates significant associations. BPD = borderline personality disorder; FAI = Family Adversity Index.
aOvert or relational bullying at 8 or 10 years.
bBeing bullied at 7, 8, or 9 years.
cBeing bullied at 7 or 10 years.
dDepression symptoms at 12, 13, ir 14 years.
TABLE S2 Direct Associations Among All Predicted Pathways
B SE p Value"
Sex / Self-harmb,c 0.580 0.052 .000
Preschool domestic violence / Self-harmb 0.094 0.041 .022
Preschool maladaptive parenting / Self-harmb -0.046 0.054 .392
Being bullied / Self-harmb 0.235 0.063 .000
Internalizing/Externalizing Behavior / Self-harmb -0.003 0.011 .791
BPD symptoms / Self-harmb 0.006 0.049 .898
Depression symptoms / Self-harmb 0.211 0.036 .000
Sex / Preschool domestic violenced 0.062 0.038 .099
Sex / Preschool maladaptive parentingd 0.133 0.030 .000
Sex / Being bulliedd 0.180 0.032 .000
Sex / Internalizing/Externalizing behaviord 0.285 0.105 .007
Sex / BPD symptomsb 0.175 0.069 .011
Sex / Depression symptomsb 0.462 0.056 .000
Preschool domestic violence / Being bulliedd 0.212 0.029 .000
Preschool domestic violence / Internalizing/Externalizing behaviord 0.608 0.092 .000
Preschool domestic violence / BPD symptomsb -0.029 0.060 .627
Preschool domestic violence / Depression symptomsb 0.074 0.045 .098
Preschool maladaptive parenting / Being bulliedd 0.237 0.032 .000
Preschool maladaptive parenting / Internalizing/Externalizing Behaviord 1.994 0.119 .000
Preschool maladaptive parenting / BPD symptomsb 0.000 0.079 .995
Preschool maladaptive parenting / Depression symptomsb 0.040 0.058 .496
Being bullied / BPD symptomsb 0.585 0.075 .000
Being bullied / Depression symptomsb 0.335 0.057 .000
Internalizing/Externalizing Behavior / BPD symptomsb -0.008 0.016 .602
Internalizing/Externalizing Behavior / Depression symptomsb 0.050 0.011 .000
Covariance between preschool domestic violence and preschool 0.137 0.018 .000
maladaptive parenting
Covariance between being bullied and internalizing/externalizing behavior 0.758 0.056 .000
Correlation between BPD symptoms and depression symptoms 0.203 0.047 .000
Note: Boldface type indicates significant associations at p < .05. Model fit: Root mean square error of approximation (RMSEA) = 0.025 (0.023 —0.027); confirmatory fit index (CFI) = 0.97. Values are given as nonstandardized linear regression coefficients and probit coefficients, covariance and correlations. B = nonstandardized coefficients; BPD = borderline personality disorder. aThe p value is 2-tailed. bProbit regression coefficient. cA probit coefficient of 0.58 indicates that, for each unit increase in sex (from male to female), there is an increase of 0.58 standard deviation in the predicted z score of the cumulative normal distribution of self-harm. dLinear regression coefficient.
