eTable. Attrition and Bullying–Exposure to Bullying at 8 Years of Age
Sourander A, Gyllenberg D, Brunstein Klomek A, Sillanmäki L, Ilola A, Kumpulainen K. Association of Bullying Behavior at 8 Years of Age and Use of Specialized Services for Psychiatric Disorders by 29 Years of Age. JAMA Psychiatry. 2016;73(2):159-165. doi:10.1001/jamapsychiatry.2015.2419
Bullying and being exposed to bullying among children is prevalent, especially among children with psychiatric symptoms, and constitutes a major concern worldwide. Whether childhood bullying or exposure to bullying in the absence of childhood psychiatric symptoms is associated with psychiatric outcomes in adulthood remains unclear.
To study the associations between bullying behavior at 8 years of age and adult psychiatric outcomes by 29 years of age.
Design, Setting, and Participants
Nationwide birth cohort study of 5034 Finnish children with complete information about childhood bullying behavior was followed up from 8 to 29 years of age. Follow-up was completed on December 31, 2009, and data were analyzed from January 15, 2013, to February 15, 2015.
Main Outcomes and Measures
Information about bullying, exposure to bullying, and psychiatric symptoms were obtained from parents, teachers, and child self-reports when children were 8 years of age. Use of specialized services for psychiatric disorders from 16 to 29 years of age was obtained from a nationwide hospital register, including outpatient and inpatient treatment.
Among the 5034 study participants, 4540 (90.2%) did not engage in bullying behavior; of these, 520 (11.5%) had received a psychiatric diagnosis at follow-up; 33 of 166 (19.9%) who engaged in frequent bullying, 58 of 251 (23.1%) frequently exposed to bullying, and 24 of 77 (31.2%) who both frequently engaged in and were frequently exposed to bullying had received psychiatric diagnoses at follow-up. When analyses were adjusted by sex, family factors, and child psychiatric symptoms at 8 years of age, we found independent associations of treatment of any psychiatric disorder with frequent exposure to bullying (hazard ratio [HR], 1.9; 95% CI, 1.4-2.5) and being a bully and exposed to bullying (HR, 2.1; 95% CI, 1.3-3.4). Exposure to bullying was specifically associated with depression (HR, 1.9; 95% CI, 1.2-2.9). Bullying was associated with psychiatric outcomes only in the presence of psychiatric problems at 8 years of age. Participants who were bullies and exposed to bullying at 8 years of age had a high risk for several psychiatric disorders requiring treatment in adulthood. However, the associations with specific psychiatric disorders did not remain significant after controlling for concurrent psychiatric symptoms.
Conclusions and Relevance
Exposure to bullying, even in the absence of childhood psychiatric symptoms, is associated with severe adulthood psychiatric outcomes that require treatment in specialized services. Early intervention among those involved in bullying can prevent long-term consequences.
Bullying can be defined as a repetitive aggressive act embodying an imbalance of power in which those exposed cannot defend themselves.1 The involvement in bullying is generally distinguished between those who bully (exposing others to bullying), those who are exposed to bullying, and those who are bullies and exposed to bullying (bullying–exposed to bullying status). Involvement in bullying may often be part of other types of exposure to violence.2 Frequent involvement in bullying has been shown to have more negative consequences compared with infrequent involvement.3 Cross-informant agreement about bullying and exposure to bullying is found to be low; however, all informant sources have been found to be associated with long-term adversities.4 Boys are usually involved in more physical bullying, whereas girls are more involved in relational types.5,6 In addition, boys and girls tend to respond differently to bullying.7,8
An increasing amount of evidence suggests that bullying and exposure to bullying contribute to later mental health problems.3,9- 14 However, only a few large-scale longitudinal prospective studies examine whether the effects of bullying behavior in childhood extend into adulthood and whether preexisting behavioral and emotional problems or bullying per se explain adult outcomes. The first large study about adult psychiatric outcomes of childhood bullying8,15 is the Finnish prospective nationwide birth cohort study that used nationwide registers to assess adversities in adulthood. However, information about psychiatric outpatient diagnoses was based on information from military call-up at ages 18 to 23 years and was therefore restricted to men.16 Three other large-scale longitudinal prospective studies examined long-term outcomes of exposure to bullying, including the Great Smokey Mountain Study,17 the British National Child Development Study,18 and the Avon Longitudinal Study of Parents and Children in the UK,19 but only the first of these17 examined bullies. However, the Great Smokey Mountain Study17 relied on a relatively smaller sample, was based on reports from children and parents, and followed up the children to 26 years of age. In addition, the participants in the Great Smokey Mountain Study were considerably older at baseline than the participants in the present study (8 vs 9-16 years). These age differences are important to consider because the frequency and types of bullying behavior among both sexes change from childhood to adolescence.20- 22
The present study is the largest, to our knowledge, to examine bullying and exposure to bullying among boys and girls and includes the longest follow-up of their long-term psychiatric outcomes. Specifically, in the current nationwide study, the participants were assessed for bullying and exposure to bullying at 8 years of age, and the outcome was derived from prospectively acquired contacts to outpatient and inpatient services between 16 and 29 years of age. This study is the only one, to our knowledge, to assess bullying and exposure to bullying based on information from the children at 8 years of age and their parents and teachers. The 3 informants enable an ecological examination of the social and contextual phenomena of bullying behavior and are important to understand the available antibullying interventions that target children, schools, and families. We expected that frequent bullying and exposure to bullying are associated with psychiatric disorders in adulthood. In particular, we examined whether bullying and/or exposure to bullying status at 8 years of age in the absence of psychiatric problems predicts adult psychiatric outcomes.
The study is part of the multicenter Finnish Nationwide 1981 Birth Cohort Study.23 The study population included all 60 007 Finnish children born from January 1 to December 31, 1981, and alive in 1989. A representative sample of 6017 children was invited to take part in the study in 1989. This nationwide representative sample of 8-year-old children came from all 5 university hospital areas in Finland. The sociodemographic characteristics of the sample are representative of the population.23 The Joint Commission on Ethics of Turku University and Turku University Central Hospital approved the study. Written informed consent was obtained from the parents of participants at enrollment, but the combined information from questionnaires and registry data for the present study was analyzed in such a way that no participant could be identified.
Data collection in 1989 was organized through teachers when the participants attended grade 2 in elementary school. The parents returned the questionnaire in a sealed envelope to the teacher and consented to participation. The children filled in a questionnaire in the classroom. After the teacher had completed teacher questionnaires, all the material was sent to the researchers.
In Finland, every citizen has a unique personal identification number (PIN), consisting of their date of birth and a 4-digit suffix. The study sample from January 1, 1998, to December 31, 2009, included 5400 participants whose PINs could be linked with the Population Register and the Finnish Hospital Discharge Register (FHDR) and who lived in Finland in 1998. The attrition at follow-up (617 of 6017 [10.3%]) was owing to random error, such as inappropriately documented PINs. In addition, children who had died or moved from Finland from the time of the baseline assessment in 1989 to the start of the follow-up in 1998 were excluded (n = 16). Complete information about bullying and exposure to bullying from the 3 informants (children, parents, and teachers) at 8 years of age and outcome information were obtained from 5034 participants. As shown in the eTable in the Supplement, attrition was not related to bullying status at 8 years of age.
Children were asked about bullying by giving them the following 3 alternatives to choose from: (1) “I bully other children almost every day”; (2) “I bully sometimes”; and (3) “Usually I do not bully.” Exposure to bullying was assessed by the following 3 alternatives: “Other children bully me (1) almost every day, (2) sometimes, and (3) do not usually bully me.” Similar questions focusing on bullying and exposure to bullying were included in the parent and teacher questionnaires, with probe and response items worded as follows: “The child bullies other children: (1) does not apply, (2) applies somewhat, and (3) certainly applies.”
We classified the sample into the following groups: (1) those who never or only sometimes bully and are not exposed to bullying according to parent and teacher reports and self-reports; (2) those who frequently bully (but are not exposed to bullying) according to at least 1 informant; (3) those who are frequently only exposed to bullying according to at least 1 informant; and (4) those who frequently bully and are exposed to bullying using pooled information from all 3 informants. For example, if a child frequently bullied according to teachers and was frequently exposed to bullying according to self-reports, he or she was classified in the bullying–exposed to bullying group. Only participants with complete information about bullying and exposure to bullying from all 3 informants were included in the analysis. We combined the parent, teacher, and child reports of information about bullying and exposure to bullying by using the either-or rule, justified by the finding that the interrater agreement was low (weighted κ, 0.11-0.22), and all 3 informant groups’ reports have been shown to be associated with long-term psychiatric symptoms.4
Three measures of psychiatric symptoms were constructed by pooling together information from parent and teacher Rutter Behavior Scales24,25 at 8 years of age. The scores from the Rutter Behavior Scales for parents and teachers were united to generate pooled Conduct, Hyperactivity, and Emotional subscales. The subscales were standardized separately as described in a prior report for boys and girls.26 The Conduct, Hyperactivity, and Emotional subscales were given equal weight from each informant, and questions about bullying and exposure to bullying were excluded. If information from only the parent or the teacher but not both was available, the pooled scale was not generated for the participant concerned. Finally, the results of the 3 subscales were categorized as less than or greater than the 90th percentile because these psychiatric measures have been found to be associated with subsequent adverse outcomes.26- 31 The cutoff points were sex specific and based on the distribution of scores in the present sample. The child was considered to have a positive screen for psychiatric problems if he or she scored at greater than the 90th percentile in 1 or more of the 3 subscales.
To examine whether bullying or exposure to bullying at 8 years of age, with or without co-occurring psychiatric symptoms, was related to adult outcomes, we stratified the whole sample based on information about bullying and/or exposure to bullying and psychiatric symptoms. Pooling information about psychiatric problems (negative or positive findings based on the 3 psychiatric measures) and bullying status (no frequent bullying or exposure to bullying, frequent bullying, frequent exposure to bullying, or bullying–exposure to bullying), we stratified the whole sample into 8 groups by screen finding and bullying status.
Family background variables at 8 years of age shown to be associated with bullying behavior and later psychiatric outcomes26,27,29- 32 included information about parental educational level based on the father’s or the mother’s completion of at least 12 years of education. The family structure was based on whether the child lived in a family with 2 biological parents or whether the family had some other constellation.
The PIN codes of the participants were linked to data from the FHDR from January 1, 1998, through December 31, 2009, when the participants were 16 to 29 years of age. The follow-up started in 1998 because, since 1998, the FHDR has included information from all visits in public outpatient clinics in Finland in addition to inpatient data. Health care is free in Finland, and most patients in need of specialist treatment are referred to public outpatient clinics, whereas the private sector is primarily complementary to the public sector. The FHDR is maintained by the Institute of Health and Welfare and extensively documented in psychiatric research.8,33,34 The FHDR includes the PIN of the patient, the date of the visit or admission, and the diagnosis according to the International Statistical Classification of Diseases, Tenth Revision (ICD-10). In this study, only main diagnoses were taken into account. All outpatient visits or registered inpatient admissions, that is, all services in psychiatric and general hospitals, were included. The following diagnostic groups were studied: any psychiatric diagnosis (ICD-10 codes F00-F99); schizophrenia and other psychotic disorders (ICD-10 codes F20-F29); depressive disorders (ICD-10 codes F32-F39); anxiety, stress-related, adjustment, and somatoform disorders (ICD-10 codes F40-F48; abbreviated anxiety); eating disorders (ICD-10 code F50); and substance use disorders (ICD-10 codes F10-F19).
Data were analyzed from January 15, 2013, to February 15, 2015. To take into account the possibility of emigration and death during the follow-up, we used Cox proportional hazards regression models, which are a class of survival models. The age of the participants in days defined the survival time. The event was defined as the first visit or admission to specialized services with the studied diagnosis. The time to moving from the country (n = 5), death (n = 11), or the end of the follow-up on December 31, 2009, was treated as the censored time observation. The associations between the predictors and the outcomes were quantified with hazards ratios and their 95% CIs from the Cox proportional hazard regression models. P values were calculated using the Wald χ2 test. In unadjusted analyses, each predictor was analyzed separately in a model. In adjusted analyses, all defined predictors were entered at once into the model. Statistical analysis used SAS software (version 9.3; SAS Institute Inc).
Table 1 shows the frequencies of bullying and exposure to bullying at 8 years of age in relation to psychiatric disorders diagnosed in specialized services from 16 to 29 years of age among 5034 study participants. In the total sample, among the 4540 participants who did not engage in bullying behavior, 520 (11.5%) had received a psychiatric diagnosis at follow-up. Among those who bullied frequently, 33 of 166 (19.9%) received a psychiatric diagnosis; among those frequently exposed to bullying, 58 of 251 (23.1%); and among those with bullying–exposure to bullying status, 24 of 77 (31.2%).
Table 2 shows the associations between frequent bullying, exposure to bullying, and bullying–exposure to bullying at 8 years of age and psychiatric disorders diagnosed in specialized services from 16 to 29 years of age. All analyses were adjusted for sex, parental educational level, and family structure (2 biological parents vs other). We found no statistically significant interactions between sex and the bullying–exposure to bullying variable with regard to the psychiatric disorders (P > .10). Bullying–exposure to bullying status was associated with any psychiatric disorder, psychoses, depression, and substance use disorders. Frequent exposure to bullying status was associated with any psychiatric disorder, psychoses, anxiety, and depression. Frequent bully status was associated with any psychiatric disorder and depression. Table 2 also shows the results when all analyses were additionally controlled with psychiatric symptoms at 8 years of age. Bullying–exposure to bullying and frequent exposure to bullying remained significantly associated with any psychiatric disorder outcome. Furthermore, frequent exposure to bullying was independently associated with depression.
Table 3 presents the results when the sample was stratified to groups based on bullying–exposure to bullying status and psychiatric symptoms in at least 1 domain (ie, Conduct, Hyperactivity, or Emotional subscales) at 8 years of age. The outcomes in this analysis were the presence of any psychiatric disorder and depression from 16 to 29 years of age. The reference group consisted of those without frequent bullying involvement or exposure to bullying and negative findings on all 3 psychiatric subscales. Frequent exposure to bullying with and without co-occurring psychiatric symptoms was associated with later psychiatric disorders. Frequent bullying and bullying–exposure to bullying predicted psychiatric outcome only when co-occurring with psychiatric symptoms.
The findings of the study are important for mental health research, prevention, assessment, and intervention. Frequent exposure to bullying at 8 years of age was associated with later adult psychiatric disorders, even in the absence of childhood psychiatric symptoms. This association indicates that frequent exposure to bullying in childhood is a significant stand-alone indicator for later psychiatric disorders requiring treatment in specialized services. Bullying was associated with later psychiatric disorders only when occurring with psychiatric symptoms at 8 years of age. In particular, bullying–exposure to bullying status was associated with a wide range of adult psychiatric outcomes, including schizophrenia and nonaffective psychoses, depression, and substance use disorders. Most participants with bullying–exposure to bullying status were male and had comorbid psychiatric symptoms at 8 years of age, which explains the findings. This result is in accordance with those of previous longitudinal studies showing that individuals with bullying–exposure to bullying behavior often have poor outcomes.16,17,35 Bullying–exposure to bullying behavior serves as an important red flag that intensive preventive or ameliorative interventions are warranted.
Our findings provide strong evidence by its prospective nature; a long period of data collection; a large sample size; pooled information about bullying and exposure to bullying from the child, parents, and teachers; the use of psychiatric outcomes based on a nationwide register; controlling for parents’ educational level, family structure, and childhood psychiatric symptoms; and the stratification of the sample based on information about bullying–exposure to bullying and psychiatric disorders. Our results are in line with those of previous longitudinal findings that the effects of exposure to bullying remained significant after controlling for known correlates of exposure to bullying,15,18,19 whereas bullying others was not associated with adverse outcomes in adulthood when other baseline variables were controlled for.3,36 In the Great Smokey Mountain Study—the only previous longitudinal study from childhood or adolescence to adulthood17 that included baseline information about bullying and exposure to bullying—bullying–exposure to bullying at 9 to 16 years of age was associated with psychiatric outcomes at 26 years of age, and the effects were maintained even after accounting for preexisting psychiatric problems or family hardships.
To our knowledge, the present study provides the strongest evidence to date that frequent exposure to bullying in early childhood increases the risk for depressive disorder later in life. The findings remained significant after controlling for psychiatric symptoms, indicating that exposure to bullying puts children at high risk for later depression, requiring treatment in specialized services regardless of the comorbid psychiatric symptoms at 8 years of age. Accordingly, previous studies have found that exposure to bullying by peers in adolescence is associated with an increase in the risk for depression,2,18,37 and childhood exposure to bullying is associated with later use of antidepressants.8 The association between exposure to bullying and depression is in line with the interpersonal theory of depression, indicating that negative interpersonal events in childhood may lead to the development of later clinical depression.38,39 In this sense, exposure to bullying by peers in childhood is similar to other types of trauma, humiliation, and/or neglect, which lead to similar consequences.40 Another possible option is the existence of a common factor (eg, genetic, personality) for being exposed to bullying and depressed. Media reports often link childhood exposure to bullying with suicide.41 The possible link between early childhood exposure to bullying by peers and later clinical depression and possible suicidal tendency has high public health importance and deserves future research.
The primary limitation in the present study is the lack of understanding on how bullying–exposure to bullying leads to psychiatric disorders. Future studies containing more nuanced information about the mediating factors that occur between childhood bullying and adulthood disorders will be needed to shed light on this important question. Of note, in the present study bullying, bullying exposure, and bullying–exposure to bullying were assessed, but no information about bystander behavior was obtained. In addition, other factors that were not assessed include parenting style, family functioning, and biological, genetic, and other environmental factors. Bullying was assessed only at 8 years of age, which prevented us from assessing the chronicity of the bullying behavior. Bullying questions did not include the definition of or information about the type of bullying. However, providing a definition of bullying in assessments does not yield a more rigorous measure.42 An additional limitation is that childhood psychiatric symptoms were based on screening instruments, and no diagnostic assessment was available. Finally, the outcome was based on service use, and the use of registers allowed us to capture only those participants who enter treatment. In Finland, health care services are primarily public; for example, almost all adults with a psychotic disorder can be found in the registers.33 However, participants with less severe disorders may not enter treatment and therefore are not included in the registers.
Bullying should be conceptualized within the framework of mental health. From a research perspective, prospective longitudinal studies should examine the various mechanisms and pathways that play a role in the severe mental health outcomes of the children involved in bullying. With regard to prevention and treatment, school authorities should consider integrating preventive efforts for bullying and psychiatric problems. Parents, teachers, and pediatricians need to be educated about the association between bullying involvement and psychiatric outcomes so as to screen and intervene if needed. In some cases, exposure to bullying, even in the absence of childhood psychiatric symptoms, is associated with severe adulthood psychiatric outcomes. Exposure to bullying is a form of abuse that should be addressed effectively even when other problems do not exist. Policy makers and health care professionals should be aware of the complex nature between bullying and psychiatric outcomes when they implement prevention and treatment interventions.
Corresponding Author: Andre Sourander, MD, PhD, Department of Child Psychiatry, University of Turku, Lemminkäisenkatu 3/Teutori (Third Floor), 20014 Turku, Finland (email@example.com).
Submitted for Publication: June 8, 2015; final revision received October 7, 2015; accepted October 8, 2015.
Published Online: December 9, 2015. doi:10.1001/jamapsychiatry.2015.2419.
Author Contributions: Dr Sourander had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sourander, Gyllenberg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Sourander, Klomek, Ilola.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sillanmäki.
Obtained funding: Gyllenberg.
Study supervision: Sourander.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Sigrid Juselius Foundation and Finnish Academy.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Jukka Huttunen, BM, Sebastian Jäntti, CTE, Jarna Lindroos, MA, and Saara Santanen, BA, Department of Child Psychiatry, University of Turku, helped with the manuscript. None received compensation for these contributions.