Health Care Utilisation by Bullying Victims: A Cross-Sectional Study of A 9-Year-Old Cohort in Ireland

Children frequently refrain from disclosing being bullied. Early identification of bullying by healthcare professionals in children may prevent adverse health consequences. The aim of our study was to determine whether Health Care Utilisation (HCU) is higher in 9-year-olds who report being bullied and factors influencing type of HCU. The study consists of cross-sectional surveys of Child Cohort of Irish National Longitudinal Study of Children (Wave 1), 8,568 9-year-olds, and their carers. Being bullied was assessed by a self-reported questionnaire completed by children at home. HCU outcomes consisted of the following: visits to GP, Mental Health Practitioner (MHP), Emergency Department (ED), and nights in hospital by parent interview. Bivariate logistic regression and gender-stratified Poisson models were used to determine association. Victimisation by bullying independently increased visits to GP (OR 1.13, 95% confidence interval (CI): 1.03 to 1.25; p = 0.02), MHP (OR 1.31, 95% CI: 1.05 to 1.63; p = 0.02), though not ED visits (OR 0.99, 95% CI: 0.87 to 1.13; p = 0.8) or nights in hospital (OR 1.07 95% CI: 0.97 to 1.18; p = 0.2), adjusting for underlying chronic condition(s) and socio-demographic confounders. Victimised girls made higher GP visits (RR 1.14, 95% CI: 1.06 to 1.23; p < 0.001) and spent more nights in hospital (RR 1.10, 95% CI: 1.04 to 1.15; p < 0.001). Victimised boys were more likely to contact MHPs (RR 1.21, 95% CI: 1.02 to 1.44; p = 0.03). 9-year-old bullied subjects were more likely to utilise primary care services than non-bullied 9-year-olds. Different HCU patterns were observed according to gender and gender differences in the presentation of victimisation. Our findings may lead to the development of clinical practice guidelines for early detection and appropriate management of bullied children.


Introduction
Bullying may be defined as a recurrent, intentional, and unprovoked form of aggressive behaviour to inflict pain on, or cause distress to, another individual [1][2][3]. Bullying may be subcategorised as direct, e.g., physical or verbal, or indirect or relational bullying, e.g., social isolation or cyber bullying. The prevalence of bullying and being bullied in early and late adolescence varies widely depending on the definition of bullying used, age, gender, social class, and the country being assessed [2][3][4].

Study Design
Data were provided from the first wave of data from the Child (9-year-old) cohort of the Irish National Longitudinal Study of Children, Growing Up in Ireland (GUI). Inclusion in the study was on an opt-in basis with consent and assent forms signed by parent(s)/guardian(s) and study child.
Participants were selected using a two-stage sampling method [25,26]. In stage one, 1105 primary schools from a national total of 3200 were randomly selected using probability proportionate to size (PPS) sampling [25], of which 910 primary schools participated (response rate, 82%). The schools provided the sampling frame for stage two. The final Growing up in Ireland cohort sample size comprised 8568 children (4404 girls and 4164 boys), or approximately 1 in 7 (14%) of resident 9-year-olds [27]. A subset of 8183 children was used in our study, as those with missing information on being bullied in either self or parent questionnaire were excluded.
The study design was approved by the Scientific and Policy Advisory and the Research Ethics Committees of the Irish Health Research Board. Questionnaires were designed by a study team using a Delphi process involving a wide range of stakeholders. Questionnaires were piloted firstly with a Children's Advisory Forum (84 children) and then with a random sample of 145 families recruited via nine randomly selected schools (response rate 136/145, 94%). A second pilot was conducted with a further 62 children and their families. Minor modifications were made to the questionnaires as a result of pilot testing.

Survey Design and Data Collection
Study participants were recruited between September 2007 and June 2008. Bully victimisation data were obtained from directly from children who completed a paper Child Sensitive Questionnaire (CSQ) provided by the interviewer on their own at home. The CSQ was designed with input from children themselves and included pre-pilot and pilot studies to test validity, which are described in more detail elsewhere [26].
The child was asked: "Thinking back over the last year would you say that anyone (either a child or an adult) picked on you". A positive answer was followed by a prompt to specify the forms of bullying experienced; shoving, pushing, hitting, name calling, slagging (teasing), text messaging, emails, written messages/notes, cyber bullying, deliberate exclusion from games/conversations, or other forms of bullying.
In addition to the child survey, a survey was carried out with the primary care givers by trained interviewers. Pre-defined healthcare utilisation (HCU) measures of frequency of contact with primary health care professionals were obtained (adpated from the Canadian National Longitudinal survey of Children and Youth): number of contacts with GP, another medical doctor or mental health professional (MHP), i.e., psychologist, psychiatrist, or counsellor over the previous 12 months. Child's use of hospital and Emergency Department visits-number of nights the child spent in hospital over his/her lifetime (excluding time of birth) and number of visits to ED over the previous 12 months-were sought as an indicator of severe morbidity. Parent report of bullying prevalence was also sought in an interview question: "Has (insert name of Study Child) been a victim of bullying in the past year".
The following information was sought on potential confounders; child factors included gender, presence/absence of chronic condition, and measured BMI; parental/family factors included parental age household composition, education level, household occupational class, general medical services entitlement, and social welfare entitlement [28].
The primary caregiver was asked whether or not the study child had a chronic condition: defined as "any on-going chronic physical or mental health problem, illness or disability?" If this question was answered in the affirmative, an open-ended question was asked: "What is the nature of this problem, illness or disability? Please describe as fully as possible." Responses to this item were coded using International Classification of Diseases 10 [29].
The primary caregiver was also asked about specific learning disability (LD), communication, or coordination disorders. For ease of interpretation, all subsequent references to "ongoing chronic condition" refer to ongoing chronic illness or disability, including LD, unless specifically mentioned.

Statistical Analyses
All results are based on statistically reweighted data to ensure representation of all 9-year-olds in Ireland and to account for the complex sampling design. Specific details of sampling frames and methodology, weighting strategies, questionnaires, and response rates have been reported elsewhere [26].
Chi-Square tests summarised the proportion of children being victimised and type of bullying experienced, and tested gender differences in victimisation prevalence. Multivariable logistic regression models were used to determine association between HCU and bully victimisation: Any visits to GP vs. None; Any visits to MHP vs. None; Any visits to ED vs. None; Nights in Hospital vs. None. Confounding variables were selected using a backwards stepwise method. Odds Ratios (OR) with 95% CIs described the strength of the association. Gender stratified Poisson models examined specific associations by gender. All models were adjusted for potential confounders. The cut-off of the p-value for statistical significance was 0.05. Analysis was carried out in SPSS V24.
The STROBE checklist for reporting of cross-sectional studies was followed in the writing of this article [30]. Table 1 summarises the proportion of the total cohort who experienced bully victimisation in the past year according to self and parent report, the form(s) of bullying experienced, and breakdown by gender. Four in 10 children reported being a victim of bullying, with no gender difference in prevalence (χ 2 = 0.09; p = 0.8). Boys were more likely to report being physically (χ 2 = 157.1; p < 0.001) or verbally bullied (χ 2 = 41.3; p < 0.001). Girls were more likely to report being bullied by exclusion (χ 2 = 23.0; p < 0.001) and by written messages (χ 2 = 4.37; p = 0.04). While 39.9% of the study children disclosed having been bullied, 23.5% of parents thought their child was/had been bullied (Table 1). All data from Table 2 onwards refer to child-reported bullying. Table 1. Proportion of being vicitimised by bullying and type of bullying in the past year by gender (total cohort N = 8183).

Self-Report (Child) Parent Report
All N (%)   Table 2 summarises the healthcare utilisation, child health and socio-demographic characteristics of bullied and non-bullied subjects in our cohort on univariate analysis. Results demonstrated significantly higher GP use and contact with Mental Health Professionals (MHP) among bullied subjects. The logistic regression model comparing "Any visits" with "No visits" (Table 3) for each type of HCU showed that victimised children were 13% more likely to visit GPs (OR 1.13, 95% CI:1.03 to 1.25, p = 0.02) and 31% more likely to visit MHPs (OR 1.31, 95% CI: 1.05 to 1.63; p = 0.02), but were not likely to visit EDs (OR 0.99, 95% CI: 0.87to 1.13; p = 0.8) or spend nights in hospital (OR 1.07; 95% CI: 0.97 to 1.18; p = 0.18), following adjustment for ongoing chronic condition and socio-demographic factors.
Other factors associated with increased HCU irrespective of bullying status in the bivariate model were presence of chronic illness (higher GP and ED visits), learning disability (higher GP and ED visits), full entitlement to free medical care (higher GP visits), and children of lone parent families with a sibling (higher MHP visits). Children of parents with lower educational attainment were less likely to visit GPs and MHPs (Table 3).
Victimised boys were less likely to attend GPs than girls (OR 0.84; 95% CI: 0.76 to 0.93; p = 0.001). Although victimised boys were more likely to seek MHP assistance than girls (OR 1.25; 95% CI: 0.99 to 1.56), this was not statistically significant in the regression model (p = 0.057) ( Table 3). In summary, Table 3 showed the increased likelihood of having visited a professional among victimised children compared to non-victimised children, adjusting for confounding personal and socioeconomic characteristics.

Summary
Our study sought to determine an association between being bullied as a 9-year-old and frequency of primary health service attendance within a 12 month period and ever use of secondary healthcare. It sought to establish if HCU is higher in 9-year-olds who report being bullied compared to non-victims and factors influencing HCU. Our results suggest that self-reported bully victimisation is associated with higher HCU, irrespective of having a pre-existing chronic condition, socio-demographic background, or free GP care. Although gender differences in forms of bullying experienced are well documented, our study also found significant differences in type of health service used, with girls more likely to present to GPs and spend nights in hospital, and boys more likely to present to MHPs.

Strengths and Limitations
A major strength of this study is the large national representative sample, which equates to approximately 1/7 of all 1997 births in Ireland [26,32]. The CSQ is likely to have improved the extent and accuracy of reporting, as it was completed privately by children in their own homes. We have adjusted our analysis of HCU for a wide range of confounders associated with exposure (bullying) and outcome (HCU).
Our analysis is limited in that data are self-reported, which, though it may facilitate disclosure, may also bias responses of 9-year-olds depending on their understanding of the questions asked. However, the CSQ was developed with input from children themselves and piloted extensively to address this issue. With regard to the definition of bullying used, the child was asked to "think back over the past year" in relation to occurrence of being bullied which implied a "period prevalence" rather than a "point prevalence" of victimisation. The prompts were given to elicit the form of bullying experienced, e.g., hitting, name-calling, deliberate exclusion, and conveying deliberate intention. However, data on the number of bullying episodes over the past year were not sought, which is a limitation of the study. In addition, data were not available on severity and intensity of victimisation that may have elicited a more nuanced understanding of HCU according to bullying experiences. However, difficulty in obtaining this type and quantity of data from a 9-year-old should not be underestimated. It will not encourage all young people who are being bullied to disclose and is still likely to be subject to an under-reporting bias. However, prevalence rates from child self-report were much higher than parent-reported rates. The principal limitation of the study relates to the cross-sectional nature of the data, which precludes determination of causal relationships between bullying and HCU; hence, our findings are limited to association.

Comparison with Existing Literature
The negative impact of bullying on children's health has been well researched, particularly in those with underlying chronic illness or disability [33]. It is therefore unsurprising that victimisation is associated with HCU and, especially, GP utilisation. Furthermore, having a long-term condition or LD may contribute not only to HCU but also to the person being bullied in the first place.
Many children who see a MHP and/or spent nights in hospital will have been referred via general practice. Gender differences in GP attendance by adolescents with depressive symptoms have been noted previously [34]. Adolescent boys were reported to find it difficult to discuss psychological problems with GPs but were more likely to attend to physical problems. Female attendance was predicted in part by a belief that the GP's role included an interest in psychological symptoms. Vila et al. reported that, in addition to physical health problems, social factors and psychiatric difficulty are significant in young people who frequently attend primary care [35].
Several theories have emerged to explain gender differences in health resource use in different contexts, biological and/or behavioural reasons, and bias in the referral process [36]. It may be that distress associated with victimisation manifests differently in boys and girls. Girls may be more likely to present with physical manifestations of psychosomatic symptoms and boys with behavioural problems, thus leading to perceived differences in healthcare needs. The higher rates of physical and verbal bullying experienced, and higher prevalence of LD, are likely to contribute to greater contact with MHPs by boys. Previous studies suggest that boys with behavioural difficulties attract most attention and resources [37,38]. Hartley et al. found that students in special education reported higher levels of physical victimisation, as well as more physical and emotional harm in addition to increased psychological distress as a result of their victimisation [39].

Implications for Research and/or Practice
Previous studies indicate that families are often unaware of the child's victimisation due to a reluctance to disclose [8]. Our study confirms a gap between parents' perception of their child being bullied and a child's self-reported experience of victimisation [40]. How parental knowledge and engagement in recognising bullying may be improved is a subject for further research. Although the dataset was published in 2008, there have been negligible public health interventions targeting bullying, so we believe the data is still relevant to the current experience of children. Parents, teachers, and healthcare professionals all need to be aware of the high prevalence of victimisation in children and the different ways that this may manifest. Regarding health professionals, there has been some preliminary work on the acceptability of routine questioning by clinicians, which found over half of young people would be happy for their GP to ask about bullying, and over 80% would be happy to complete a screening questionnaire before seeing their GP [41]. This mirrors the UK's National Society for the Prevention of Cruelty to Children report, which found that young people wanted healthcare professionals to ask about their problems and spend more time getting to the root of their problems [42]. Our study suggests that increased vigilance is needed when dealing with medically unexplained symptoms that may be psychosomatic manifestations of being bullied, particularly in girls, and in managing unexplained behavioural difficulties, especially in boys. In these instances, sensitive questioning about bullying may lead to disclosure and appropriate management, especially if referral is being considered. Our findings need to be replicated in the cohort as it moves forward and in other longitudinal studies. Further research is required to understand gender differences in health impacts of victimisation.

Conclusions
This study found that bully victimization as reported by children themselves is itself a significant factor in increasing the usage of healthcare services by children, having adjusted for the obvious effect of having a chronic condition and having free GP access. Different HCU patterns were observed according to gender and gender differences in the presentation of victimisation. We believe increased vigilance among health professionals is especially required when dealing with medically unexplained symptoms that may be psychosomatic manifestations of being bullied, particularly in girls, and in managing unexplained behavioural difficulties, especially in boys. At a wider population level, the development of clinical practice guidelines for the early detection and appropriate management of bullied children would help raise awareness among health professionals and parents about the impact of victimization on children's health.