Bullying in adolescents: role, type of violence and determinants

1 Universidad de Santander, Facultad de Ciencias de la Salud, Grupo de Investigación Salud Pública UDES, Bucaramanga, Colombia. 2 Universidad de Santander, Facultad de Ciencias de la Salud, Grupo de Investigación de Enfermería EVEREST, Bucaramanga, Colombia. 3 Universidad de Santander, Facultad de Ciencias de la Salud, Bucaramanga, Colombia. ABSTRACT Objective: To determine the prevalence of bullying and its determinants among adolescents attending school. Method: An analytical cross-sectional study was conducted with a sample of adolescents, who were selected using a multistage probability sampling, from 20 public educational institutions in a Colombian city, in which a simple and multivariate binomial regression was carried out. Results: A total of 500 adolescents participated from which 50.4% were women and 53.2% in their middle adolescence. Verbal violence prevailed in 66.5%, followed by physical violence in 32.0% and social bullying in 30.6%. 69.5% of the adolescents have been witnesses, 35.8% victims, and 14.2% aggressors. 80.5% of the aggressors were victims of bullying. In the multivariate analysis, an association was found between being a victim and having disabilities (PR:2.4; CI: 1.6-3.7), verbal aggression in the home (PR: 1.7; CI: 1.2-2.3) and drug abuse (PR: 1.7; CI 1.1-2.8). Being an aggressor was also associated with being a victim (PR: 7.2; CI 3.6-14.3) and alcohol abuse (PR: 2.2; CI: 1.3-3.8). Conclusion: The frequency of bullying observed and the associated determinants demonstrate the persistence of this problem in adolescents and the need to develop a culture of appropriate and inclusive coexistence that goes beyond the school setting.

In this vein, bullying manifests itself as a broad and gradient spectrum that involves physical, psychological, social bullying and property-related violence (1) .Traditionally, bullying has been subdivided into two types, direct and indirect.The first type is related to physical aggression (hitting, taking objects without consent) and verbal aggression (insults, threats), while the second type is characterized by social bullying (ignoring and preventing involvement), rumor-spreading and manipulation (5) .Recently, various forms of this type of violence have been described, including physical harassment, harassment using weapons, insults or mockery, race or ethnicity-based harassment, sexual orientation or gender identity-based harassment, appearance-related harassment, cyberbullying, asking for personal information online, and feeling unsafe with someone on the Internet (6)(7) .
Based on its high prevalence and short-and long-term consequences, bullying is considered a public health problem, which is estimated to affect around 25-30% of adolescents (2) .In Bucaramanga, Colombia, this prevalence was calculated at 8.1%, being verbal violence the most frequent at 61.3% (3) .However, there may be variations in prevalence based on the frequency of violent episodes, the period assessed, the type of violence and the measuring instrument used (8)(9) .
In addition, the consequences of bullying range from physical, psychological and social symptoms or manifestations, and suicidal ideation in extreme cases (5,10) , which are reflected in high rates of school absenteeism, low academic performance and negative effects on physical and mental health (6) .
Furthermore, bullying has been associated with the use of psychoactive substances, male gender, past history of violence or aggression in the home and the environment in which adolescents live (4,11) .Therefore, this study aimed at determining the prevalence of bullying, the types of violence used, and the roles assumed by adolescents, as well as the analysis of the associated determining factors.

Study deSign
An analytical cross-sectional study was conducted.

PoPulation
Adolescents aged 10-19 based on a population of 31,385 adolescents attending school in Bucaramanga, which corresponds to 15,138 men and 16,247 women registered in the database of the Departmental Secretary of Education in 2015.The sample size was calculated using a formula with a finite population correction factor for which an expected global prevalence of bullying was estimated at 50±5%.In addition, associated factors showed a 95% confidence level, 90% power, and prevalence ratio of 1.3 (12) , apart from adding 10% for possible losses due to incomplete information, in order to have a total sample size of 500 adolescents.
Adolescents from 20 public educational institutions in Bucaramanga were selected through a multistage probability sampling.In the first stage, institutions were chosen proportional to the number of students per school; then, school years were randomly selected in the second stage and finally, students were selected through a simple random sampling in the third stage.

data ColleCtion
Information was collected in the second half of 2015 for which one asked for permission to the legal representative of the educational institutions, followed by the informed consent given by students' parents and the selected adolescents.Subsequently, the information was collected through a self-administered questionnaire in the school classrooms, after providing general instructions and clearing up some questions made by adolescents.
The subscale used was the School Coexistence Questionnaire for Students, which evaluates general violence based on the roles of witnesses, victims, and aggressors, being elaborated and validated in 856 Chilean children in 2010 (12) and using five Likert-type response options in ascending order based on their frequency.In addition, the questionnaire included questions related to sociodemographic, family, social and educational aspects (13) .

data analySiS and ProCeSSing
The information was retrieved and entered twice independently into the EpiData database, which was later validated and exported to Stata 14 for its analysis.Population parameters were estimated through proportions and their confidence intervals for qualitative variables.Additionally, a simple and multivariate binomial regression analysis was conducted to calculate crude and adjusted prevalence ratios, respectively.The goodness of fit of the model was also evaluated.All calculations were made considering the sampling plan, that is, according to the weights of each sampling unit at each stage.

ethiCal aSPeCtS
The project was approved and funded through the internal call of the Universidad de Santander under the act number 013-2013.It was also approved by the Ethics Committee of the Healthcare Institute of Bucaramanga through file 00006817 of August 4 th , 2015 and had the support of the Municipal Secretary of Education, which later issued a letter to request support for the process of collecting information from each principal of the schools included in the sample.
Furthermore, one followed the guidelines for human-related research established by the Council for International Organizations of Medical Sciences (CIOMS).In addition, based on Resolution 8430 of 1993 of the Ministry of Health in Colombia, this study was classified as minimal risk research.Therefore, written informed consent was provided to parents or legal tutors and consent to adolescents.

RESULTS
Table 1 shows the socio-demographic characteristics of adolescents with a prevalence of females (50.4%) in their middle adolescence stage (53.2%), who are from urban areas (86.2%), non-displaced people (82.3%), classified as two (34.0%) and three (25.2%) in the socio-economical strata, some of them as part of contributory healthcare system (43.2%)while others are part of the System for the Identification of Potential Social Programme Beneficiaries (SISBEN -Sistema de Selección de Beneficiarios Para Programas Sociales) (66.1%).In addition, most of them state that they practice a religion (85.2%), being the Catholic religion the most frequent (49.6%), followed by the Protestant religion (9.9%).6.0% of adolescents self-reported to have disabilities.Table 2 describes some characteristics of the relationship between adolescents attending school in Bucaramanga and their parents, neighbors and at their schools concerning bullying.It is observed that approximately one-fifth of adolescents do not share time with their parents, experiencing verbal aggression at 28.5% in the home and physical aggression at 5.5%.Regarding the relationship with neighbors, there was a high frequency of verbal aggression (49.9%), physical aggression (28.4%) and alcohol abuse (40.0%).Regarding school violence, adolescents reported that their classmates had brought knives (22.8%) and firearms (1%) into their educational institution, had witnessed fights between gangs (24.0%) and some even stated that they were afraid of attending class (3.8%).In addition, alcohol and drug abuse are currently reported in 15.6% and 8.1% of adolescents, respectively.As for violence and the roles based on which adolescents perceive it, Table 3 shows that 69.5% have witnessed some type of violence, 35.8% have been victims and 14.2% have been perpetrators or aggressors.In descending order, the type of violence that prevails in the role of witnesses is verbal violence (65.0%), physical violence (37.8%), social bullying (30.4%) and violence against teachers (22.9%).The role of victims is predominated by verbal violence (33.4%) and violence against the victim's belongings (23.7%).Lastly, adolescents are recognized as aggressors of verbal violence (13.8%), against teachers (9.6%) and of physical violence (9.3%).In relation to the determining factors associated with bullying both in the roles of victims and aggressors, crude prevalence ratios were calculated for some sociodemographic factors, psychoactive substance misuse and factors of the relationship with parents and/or in the home, neighbors, friends and at school, as shown in Table 4.
Thus, a risk association was found between victims of bullying and having disabilities, experiencing physical aggression by parents, physical and verbal aggression in the home, physical aggression by neighbors, friends smoking and drinking alcohol, bringing knives and firearms into schools, having belonged to a gang and drug abuse, with p-values less than 0.05.
Similarly, the factors associated with aggressors were having disabilities, verbal aggression in the home, physical aggression by neighbors, smoking friends, bringing knives into school, having belonged to a gang, and alcohol and drug abuse, with p-values less than 0.05.Sharing time with parents was the only protective determinant for bullying both in the role of victims and aggressors.
In addition, when evaluating the relationship between victims and aggressors, it was found that 32.1% (n=56) of the total number of bullying victims (n=173) are also considered aggressors.Likewise, 80.5% (n=56) were victims of bullying from the total number of aggressors (n=72), p-value of <0.01, X 2 test.Subsequently, a multivariate binomial regression analysis was conducted with the potential determinants that had been previously selected, as shown in Table 5.
Thus, a risk association between being a bullying victim and having some disabilities was found with a prevalence ratio (PR) of 2.4 (95% CI 1.6-3.7),verbal aggression in the home (PR of 1.7 (95% CI 1.2-2.3))and drug abuse (PR of 1.7 (95% CI 1.1-2.8)).Similarly, a risk association between being an aggressor and being a victim of bullying was found with a prevalence ratio (PR) of 7.2 (95% CI 3.6-14.3)and alcohol abuse with a PR of 2.2 (95% CI 1.3-3.8).

DISCUSSION
In this study, the prevalence of being victims of bullying was 35.8%.Additionally, the results showed a higher frequency of verbal violence with respect to other types of violence and the role of witnesses (69.5%) versus victims (35.8%) and aggressors (14.2%).In addition, violence against teachers was found in 26.5% and 32.1% of the total number of victims of bullying are considered aggressors and vice versa in 80.5%.Subsequently, an association was found between being a victim and having disabilities, experiencing verbal aggression in the home and drug abuse in the multivariate analysis.There was also an association between being an aggressor and being a victim and alcohol abuse.
Various studies conducted at the local, national and international levels have reported prevalence in victims similar to those calculated in this research (2)(3)(4) .In this sense, a prevalence of 30% was found in a study with a sample of 342,312 adolescents in 79 countries (14) .In Colombia, questionnaires conducted during the 2005 SABER tests showed that 28% of students reported themselves as victims (9) .In Bucaramanga, it was found that 8.1% of students enrolled in public educational institutions were victims of bullying, a lower value than that found in this research (3) .Moreover, a higher prevalence in Canada (58.3%) has been reported in a sample of 64,174 adolescents (6) .Although some studies have reported a similar prevalence, their statistics vary according to the population analyzed and the type of instrument used.
than that found in this study (66.5%).The prevalence of physical harassment is similar to that previously published but presents a greater variability (from 18.5% to 47%).Unlike this research, other authors analyzed variables such as cyberbullying, sexual abuse, other types of violence, and polyvictimization, this latter having results of up to 8.6% (15)(16)(17) .
Within the analysis of bullying, one sees how competitive relationships are revealed in the school context.This could be justified by the attempt to be accepted by other groups which are considered more popular, placing young people -who are victims -in lower positions, generating dependency and vulnerability.These young people may have a social skills deficit that makes it difficult for them to have friends and to be socially accepted (18) .
Besides, the risk factors related to bullying are multiple.There have been associations between bullying and age group, ethnicity, parent violence, and psychoactive substance misuse among others (16) .Some previous studies (15) indicate that the support provided by parents, in terms of the time shared with their children, is protective to become an aggressor and victim of bullying, which is similar to the present finding in the bivariate analysis.However, this association was not the same in the multivariable analysis.
Additionally, several authors show a positive association between witnessing verbal and physical violence in parents and becoming victims and aggressors, as well as physical punishment in the home is positively related to being bullied (15,19) , consistent with this findings.However, a study reveals a connection between such punishments and being an aggressor (20) .This study also confirms the previous data which established that a risk factor for being a victim of bullying is having any type of disability (21)(22) .
Furthermore, similar prevalences have been reported in other studies that analyze the use of knives and firearms in the context of bullying.However, there are no studies that show an association between the use of knives and victims or aggressors (23) .Additionally, firearms are 5.4 to 35.6 more times more likely to be used in bullying victims versus non-victims (24)(25) .In addition, aggressors are 5.6 times more likely to belong to a gang, victims 2.3 times and those who have both roles 12.1 times more (26) as bullying is considered as one of the reasons to join gangs, which may be a protective factor when being assaulted (27) .
Moreover, the consumption of alcohol, cigarettes and other psychoactive substances are associated with being a victim and in some cases an aggressor (10,(28)(29) , consistent with what has been reported in this work.These results differ from a study conducted in Barcelona in which there was a relationship between age and being a regular smoker, alcohol risk consumer, experimental cannabis user and nightlife activities, all of which are associated with a lower probability of suffering bullying (30) .
Within the limitations of this study, it is important to mention that the data was collected through a one-time-only, self-administered questionnaire using data from the very statement of the questions answered regarding bullying behaviors, consumption, and family behavior statements, which can lead to information bias.However, some strengths were confidentiality and anonymity of the information collection instruments as these were maintained to promote and encourage honest responses.On top of this, previously validated instruments were used in a similar population to diminish the information bias, in addition to a probabilistic multistage population-based sampling in order to avoid selection bias and have external validity.

CONCLUSION
The high prevalence of bullying, which has been identified in the role of victims, witnesses, and aggressors, demonstrates the persistence of this problem in the school context.In addition to the associated factors identified in the role of victims such as the presence of disabilities, aggression in the home and the consumption of psychoactive substances, and the strong relationship between victims and aggressors, there is an evident need to improve coexistence, tolerance, respect, dialogue and inclusive strategies of disabled persons, not only in the school environment but also in the family and social environment since interaction in one of these scenarios may influence the behavior of adolescents in others.Therefore, parents, teachers, school board of directors, community, territorial and control entities should be involved to promote a culture of healthy coexistence in the home, educational institutions, public spaces and social context, supplemented by strategies for the prevention of the consumption of psychoactive substances and the development of social skills in adolescents.In other words, actions aimed at preventing bullying must be of a population nature that go beyond the educational setting.Moreover, it is necessary to provide comprehensive care to victims of bullying in order to prevent him/her from becoming an aggressor, and thus avoid perpetuating these patterns of school violence.

Table 2 -
Description of the relationship between adolescents and their families, neighbors, and classmates -Bucaramanga, Colombia, 2015.

Table 3 -
Distribution of the type of violence based on the role assumed by adolescents -Bucaramanga, Colombia, 2015.

Table 4 -
Bullying determinants according to the role assumed by adolescents -Bucaramanga, Colombia, 2015.
Note: Bivariate analysis was used.

Table 5 -
Bullying determinants according to the role assumed by adolescents -Bucaramanga, Colombia, 2015.