Traumatic and Psychosomatic Symptoms in Children Exposed to Peer Violence: A Cross-Sectional Study in Bosnia and Herzegovina

This cross-sectional study was conducted between February and March 2023 in 18 primary schools from seven of the ten cantons in the FBiH. The number of schools was determined based on available approvals and the aim of including schools from diverse geographical areas, within the limits of administrative feasibility. A stratified cluster sampling design was applied: cantons served as strata, while within each stratum the cantonal ministries of education provided lists of eligible schools. Schools were then randomly selected, and in each selected school, all available classes from grades seven to nine (ages 13 to 15) were included in the study, ensuring full participation across the targeted age range. All students present in the selected classes on the day of data collection were invited to participate.

Although the sampling plan initially targeted all ten cantons of the FBiH, three cantons did not participate because approval from the relevant ministries could not be obtained. Therefore, the final sample included seven cantons.

Cochran’s formula was used to calculate the minimum required sample size (n=1,537), assuming a 20% prevalence rate of bullying, a 95% confidence level, and a margin of error of ±2%. A total of 2,822 students completed the questionnaire. After excluding 491 incomplete or invalid responses, the final sample consisted of 2,331 students (1,127 boys, 48.3%; 1,204 girls, 51.7%).

Most students were in the eighth (33.6%) or ninth grades (33.5%), while 32.9% were in the seventh grade. The majority of students lived with both parents (88.2%), had at least one sibling (90.5%), and came from an urban area (64.4%). Regarding academic performance, 70.5% of students reported having very good or excellent grades, while only 6.6% had failing or poor grades.

Validated and standardised instruments were used to assess school violence, trauma symptoms, and psychosomatic complaints, with authorisation from the respective authors or institutions (20,21,22,23).

All instruments were administered in Croatian, which is mutually intelligible with Bosnian and Serbian and widely understood in the FBIH. A detailed overview of the instruments is provided in Table 1.

Demographic and educational variables were collected using the introductory section of the School Violence Questionnaire (SVQ 2003), which includes items on school name, grade level, gender, age, place of residence, number of siblings, family structure, and academic performance in the previous school year.

Categorisation into bullying roles was based on responses to the School Violence Questionnaire (SVQ 2003) (20). Students who reported experiencing at least one form of peer violence or admitted to perpetrating it were classified as victims or bullies, respectively. Those who reported both experiences were categorised as bully/victims. Students who indicated no involvement in any form of violence were classified as neutral (20).

Academic performance was assessed by asking participants to indicate their final achievement in the previous school year using a five-point categorical scale (1 = Fail to 5 = Excellent).

The Cronbach’s alpha coefficients for the instruments used in this study were satisfactory. For the School Violence Questionnaire, internal consistency was 0.81 for experiencing violence and 0.78 for committing violence. The Trauma Symptom Checklist for Children showed coefficients ranging from 0.77 to 0.87 across its subscales. For the Psychosomatic Symptom Scale, reliability values across its seven domains ranged from 0.70 to 0.85.

Data were collected with the cooperation of school principals, teachers, and counsellors. Parents received written information about the study and provided informed consent. Students filled in anonymous paper questionnaires during class time, placing them in sealed boxes after completion to ensure confidentiality.

The study was approved by the Ethics Committee of the University of Mostar Faculty of Medicine (number: 01-I-2110/22, date: December 5, 2022), as well as the federal and cantonal ministries of education and science. Participation was voluntary and anonymous. The study followed national ethical guidelines for research involving children (3).

Data analysis was performed using IBM SPSS Statistics, Version 25.0 (Armonk, NY: IBM Corp.). The normality of continuous variables was tested using the Shapiro-Wilk test. Categorical variables were presented as counts and percentages, with the Chi-square (χ²) test with a comparison of proportions with the Bonferroni correction used to assess statistical significance. Numerical variables were presented as medians and interquartile ranges, with the Mann-Whitney and Kruskal-Wallis tests with Dunn’s post-hoc test and the Bonferroni correction for multiple tests used for significance testing. The significance level was set at α=0.05. P-values that could not be expressed to three decimal places were reported as p<0.001.

Out of the total sample, 13.3% of students (n=310) identified as victims, 3.1% (n=72) as bullies, and 4.4% (n=102) as both victims and bullies. The remaining 79.2% of students were not involved in bullying and are referred to as the neutral group.

Statistically significant differences were found between groups in relation to gender, family structure, grade level, and academic performance (Table 2). No significant differences were observed for the number of siblings or area of residence.

Post-hoc pairwise comparisons showed that the bully group included a significantly higher proportion of male students and a lower proportion of female students compared to the neutral and victim groups. No significant gender differences were observed between the bully and bully/victim groups.

Students living with both parents were significantly more represented in the neutral group, while those from other family structures were more frequently found in the bully/victim group.

Regarding grade level, seventh-grade students were significantly less represented in the bully and bully/victim groups, whereas ninth-grade students were more commonly found in these roles. Eighth-grade students showed a relatively balanced distribution across all groups. In terms of academic performance, students with excellent grades were significantly less represented in the bully group compared to the neutral and victim groups. No significant differences were observed among students with lower levels of academic achievement.

Significant differences were observed among students categorised by their role in bullying (neutral, victim, bully, bully/victim) for both the frequency and impact of psychosomatic symptoms (Table 3).

Post-hoc pairwise comparisons showed that, for pseudoneurological symptoms, victims and bullies both reported significantly higher frequencies than neutral students. No significant differences were found between victims and bullies. Bully/victims also scored significantly higher than neutral students, but did not differ from bullies in this regard.

In the cardiovascular domain, victims and bully/victims reported significantly higher frequencies than neutral students. Bullies did not differ significantly from any other group.

Musculoskeletal symptoms were significantly more frequent among victims and bully/victims compared to neutral students. Bullies did not differ significantly from neutral students, but scored significantly lower than victims.

For respiratory symptoms, victims and bully/victims scored higher than both bullies and neutral students, while bullies did not differ from neutrals.

Gastrointestinal symptoms were significantly more frequent among victims and bully/victims compared to neutral students. Bullies did not differ significantly from neutral students or bully/victims, but scored significantly lower than victims.

Dermatological symptoms were significantly more frequent among victims and bully/victims compared to neutral students. Bullies did not differ significantly from any other group.

Pain and fatigue symptoms were significantly more frequent among victims, bullies, and bully/victims compared to neutral students. No significant differences were found among the three involved groups.

In terms of the interference of symptoms with daily functioning, victims and bully/victims consistently reported greater interference than neutral students across all symptom domains, while bullies showed intermediate levels. Victims and bully/victims differed significantly from neutrals in pseudoneurological symptoms and pain and fatigue, but not in other domains. The most pronounced impacts were observed in pain and fatigue, pseudoneurological, and gastrointestinal symptoms.

In addition to the statistical results presented in Table 3, Figures 1 and 2 provide visual representations of the distribution of psychosomatic symptoms across bullying roles. These box plots illustrate both the frequency and interference of symptoms, highlighting group differences and variability within each domain.

Figure 1.

Frequency of psychosomatic symptoms by bullying role.

Figure 2.

Interference of psychosomatic symptoms with daily functioning due to bullying role.

Figure 3 presents the distribution of trauma-related symptoms across bullying role groups. Students who identified as bully/victims show consistently higher levels of all measured symptoms, including anxiety, depression, anger, posttraumatic symptoms, dissociation, overt dissociation, and fantasy dissociation, compared to neutral students. Their box plots display the highest median scores and broader ranges, suggesting greater severity and variability in symptom expression. Neutral students consistently show the lowest symptom levels across all domains (Figure 3).

Figure 3.

Distribution of traumatic symptoms according to bullying involvement.

Both bully/victims and victims demonstrate elevated trauma-related symptoms, with bully/victims tending to have higher median values and wider distributions, particularly for anger, posttraumatic symptoms, and dissociative indicators (Figure 3). Anxiety and depression are elevated in both groups, with slightly higher scores among victims. Bullies generally report lower symptom levels than victims and bully/victims, but higher than neutral students in domains such as anger and posttraumatic symptoms (Figure 3).

Overall, the group of bully/victims appears most affected, followed by victims, while bullies show intermediate scores. Neutral students consistently report the lowest levels of traumatic symptoms (Figure 3).

This study has several limitations. First, since it is based on cross-sectional data, it is not possible to draw firm conclusions about causal relationships between victimisation, violent behaviour, and the examined factors. Although the study covered various aspects of victimisation, it did not include information on exposure to other potentially traumatic events, such as witnessing violence or the loss of a loved one, which may limit the comprehensiveness of understanding children’s experiences. Additionally, polytraumatisation and its connection to traumatic symptoms were not specifically explored.

Second, all data were collected through self-report questionnaires, which, although common, carries the risk of subjectivity and socially desirable responses. Future studies could benefit from the inclusion of multiple informants, such as peers, parents, or teachers, to enhance data reliability.

Third, this study focused exclusively on students in grades seven to nine, leaving out students of younger and older ages, which may limit the ability to generalise the findings to the wider school-age population. Additionally, the sample did not include students who may have witnessed bullying without direct involvement (bystanders), although this group is known to experience emotional and psychosomatic consequences. Their exclusion may thus underestimate the broader impact of peer violence.

Fourth, the study did not include control variables such as socioeconomic status (SES), parental education, parental employment, school-level characteristics, or broader contextual factors. Without these, it is not possible to fully account for potential confounding effects, which is especially important in cross-sectional research. Given the established link between SES, exposure to violence, and health outcomes, this represents a relevant limitation that should be addressed in future research.

Finally, the sample did not cover all cantons in the FBiH. Although the study included schools from seven cantons and thus ensured broad geographic coverage, it cannot be considered fully representative of the entire federation. The findings should therefore be interpreted with caution when generalising to the whole population.

These findings should be interpreted in the context of BiH as a post-conflict society marked by war-related trauma, economic challenges, and constraints within institutional support systems. Such factors likely contribute to the persistence of school violence and increased emotional distress among children. Considering the limited regional research examining bullying within the socio-political context of the region, this study offers valuable insights.

The large sample and detailed assessment of psychosomatic and traumatic symptoms strengthen the robustness of the results. While previous regional studies have largely focused on the prevalence rates of general well-being indicators, this study contributes a more detailed analysis of psychological functioning. These insights underscore the need for school programmes that not only address violence, but also incorporate systematic screening and early identification of psychosomatic and trauma-related symptoms.

The relevance of peer networks in sustaining bullying behaviour has been well-documented in international research. Prior studies have demonstrated that bullies often occupy central positions within peer groups, where social status and peer reinforcement contribute to the persistence of aggression (9, 18, 19). In light of these findings, school-based prevention programmes should incorporate strategies that address group-level dynamics, including peer-led interventions, social network mapping, and staff training to identify and respond to harmful peer structures.

Training school staff and integrating psychosocial screening within school and primary healthcare settings are essential steps toward timely identification of and support for vulnerable children. This is particularly important in environments where stigma toward mental illness may hinder recognition and responses to psychological distress. In addition, the observed differences across grade levels suggest that prevention efforts should be developmentally tailored, with early interventions targeting younger students at higher risk of victimisation and peer-focused strategies for older students who more often assume aggressive or dual roles.