INTRODUCTION

In addition to being considered as a violation of rights, violence is characterized as a public health issue due to the consequences to the physical and psychologica health that it causes on the people involved. It may be defined as the use of physical force, power or threats, which may lead to lesions, death, deprivation or psychological damages against oneself or against another person (World Health Organization [WHO], 2002). Complementing this concept, ^{Koller and De Antoni (2004}) suggest that violence is characterized by actions or omissions that may interrupt, prevent or delay the healthy development of humans.

Although the phenomenon may be observed in several contexts and age ranges in the population, certain social, cultural and developmental characteristics may make some people more vulnerable to the exposure to violence. According to ^{Souza and Lima (2006}), violence is more prevalent in some populations, generating different risks depending on schooling, age, gender and ethnicity. For example, communities with a lower purchasing power and cultures where violence is seen as a legitimate way of solving conflicts may be more exposed to violence (Brazilian Institute of Geography and Statistics [IBGE], 2009; ^{Sanchez & Minayo, 2006}).

In terms of development, current data have identified that adolescence is a period of greater vulnerability to the exposure to violence (^{Waiselfisz, 2014}). This greater exposure may be connected to the quick and intense biopsychosocial changes, in which experimentation processes of new environments and behaviors occur in the search for autonomy and identity (^{Barbosa & Wagner, 2013}; ^{Pratta & Santos, 2007}). Behaviors and situations that, up to then, had not yet been experienced become part of the adolescent's life and are associated to the need for greater independence from parents and identification with other groups. In addition, differences of opinions and values, as well as conflicts, may arise within the family, requiring the flexibility from the parents in order not to lead to violent situations (^{Cerveny & Berthoud, 2009}). In the search for autonomy and identity, the group of friends also influences the life of the adoles cent, and it may serve both as a protective or a risk factor. Regarding the latter, it may be considered that, since they feel stronger and more omnipotent, adolescents are exposed to risk situations (^{Huculak, McLennan & Bordin, 2011}) such as, for example, when they get involved in violence situations against other adolescents or groups.

In fact, data from national and international studies have shown the high exposure to violence during adolescence (^{Benetti et al., 2006}; Benetti, Pizetta, Schwartz, Hass & Melo, 2010; Benetti, Ramires, Schneider, Rodrigues & Tremarin, 2007; ^{Haynie, Petts, Maimon & Piquero, 2009}; ^{Margolin, Vickerman, Oliver & Gordis, 2010}; ^{Mrug & Windle, 2010}; ^{Souza et al., 2014}; ^{Williams, D'Affonseca, Correia & Albuquerque, 2011}; ^{Waiselfisz, 2014}). For ex ample, data from the Map of violence against children and adolescents in Brazil indicate that deaths due to external causes (accidents and violence, usually homicides) have increased if compared to the deaths due to natural causes (diseases, for example). In 2011, over a million adolescents were victims of homicides, and their incidence, within this age range, is higher than for the rest of the Brazilian population (Waiselfisz, 2014).

^{Benetti et al. (2007}), in a review of the Brazilian literature on mental health during adolescence, highlight the vulnerability of this age range and the consequences to its mental health. Regarding violence, the authors indicate that, although the national production on the theme has increased in volume, most papers focus on identifying the cases and producing theoretical models that explain the theme.

Another study, conducted in the south region of Brazil, investigated how the exposure to violence (whether direct or indirect) was related to parental socialization, sociode-mographic characteristics and stressful events. The authors indicated that violence was the condition that was more intensively associated to internalization behaviors (such as depression, for example) and externalization problems (aggressive behaviors) within the group of investigated adolescents. It was also observed that the association between violence and these behaviors does not depend on the context of exposure to violence (community or family) (^{Benetti, Schwartz, Soares, Macarena, & Patussi, 2014}).

In fact, the exposure to violence in several contexts impacts human development (^{Benetti et al., 2006}; Benetti et al., 2010; ^{Hardaway, McLoyd, & Wood, 2012}; ^{Haynie et al., 2009}; ^{Murray, Cerqueira, & Kahn, 2013}). Although most studies focus on the consequences of the direct exposure to violence, researchers have indicated that not only being the victim, but also observing, witnessing or hearing about violence cases - indirect exposure - is also a risk factor for development (^{Almeida, Miranda, & Lourenço, 2013}; ^{Ho & Cheung, 2010}; ^{Mrug & Windle, 2010}).

In the international sphere, researches have been conducted with the aim of creating visibility to the indirect exposure and its consequences for adolescents, a theme that has been scarcely studied in Brazil For example, a study conducted with Chinese adolescents confirmed that the indirect exposure to violence in the school and family contexts was associated to emotional problems (^{Ho & Cheung, 2010}). Another research conducted in the United States showed that the exposure to indirect violence in the family, school or community was associated to internalization disorders (such as depression) and externalization ones (such as aggression). In addition, it was also observed that violence exposure at home and school has a more negative impact on the development of adolescents than the exposure to violence in the community (^{Mrug & Windle, 2010}). In another study, also conducted with adolescents from the United States, it was observed that the exposure to direct and indirect violence is related to a greater risk of running from home, dropping out of school, adolescent pregnancy, suicide and problems with justice (^{Haynie et al., 2009}).

In the south of Brazil, ^{Benetti et al. (2006}) confirmed that 90.2% of the adolescents got involved in at least one episode of violence as the direct victim, and 91.6% of them were exposed to at least one episode of indirect violence. Both exposure methods were positively correlated to the use of drugs and involvement with the police.

Still in the Brazilian scenario, it is observed that researches on violence focus mainly on the direct exposure (being the victim), and the exposure to indirect violence is scarcely investigated, although data indicate that this method of exposure affects the development of adolescents. In general, researches on indirect exposure in Brazil have focused on marital violence and violence among peers. In relation to the first type of exposure, children are the ones who usually witness marital violence, which causes a short-term and long-term impact (^{Fantinato & Cia, 2015}; ^{Maldonado & Williams, 2005}; ^{Sani & Cunha, 2011}). Studies on violence among peers, mainly on bullying, have emphasized the roles of the victim, of the aggressor, but also of the peers that witness or experience the aggressions (^{Bandeira & Hutz, 2012}; ^{Binsfeld & Lisboa, 2010}; ^{Oliveira, Silva, Yoshinaga, & Silva, 2015}; Williams et al., 2011). In addition, in the Brazilian context, few investigations consider both types of exposure to violence - the direct and the indirect one -jointly (^{Benetti et al., 2006}; Benetti et al., 2010; ^{Zavaschi et al., 2002}). The family context seems to be privileged in these researches, and there is a lower number of studies on extra-family violence (^{Braga & Dell'Aglio, 2012}; ^{Hildebrand, Celeri, Morcillo, & Zanolli, 2015}).

On the variables associated to the exposure to violence during adolescence, national and international studies have indicated that older adolescents (^{Benetti et al., 2006}; ^{Fowler, Tompsett, Braciszewski, Jacques-Tiura, & Baltes 2009}; ^{Weintraub, Vasconcellos, Bastos, Fonseca, & Reis, 2013}; ^{Zavaschi et al., 2002}) and male adolescents (Benetti et al., 2010; ^{Moreira et al., 2013}; ^{Souza et al., 2014}; ^{Waiselfisz, 2014}; Weintraub et al., 2013; Zavaschi et al., 2002) are more vulnerable to the exposure to violence when compared to younger female adolescents. However, when the different exposure contexts (intra or extra-family exposure) and types of violence (negligence, physical, psychological and sexual violence) are analyzed, the research results may be different among age ranges and genders. Some studies conducted in Brazil, with samples of adolescents from public schools, showed that boys are more exposed to extra-family violence, and girls, to intra-family violence (Benetti et al., 2006; ^{Braga & Dell'Aglio, 2012}).

Another factor that is related to the exposure to violence is the performance at school, although divergences have been observed in the research results. For example, international studies have indicated the association between the exposure to violence (whether physical, sexual or to negligence) and the performance at school (measured by the school grades, approval or failure, etc.) during childhood (^{Coohey, Renner, Hua, Zhang, & Whitney, 2011}) and on young adulthood (^{Tanaka, Georgiades, Boyle, & MacMillan, 2015}). In Brazil, ^{Brancalhone, Fogo and Williams (2004}) did not find statistically significant differences on the school performance of children from 7 to 11 years old, when comparing the group exposed to marital violence with another group of students that had not been exposed to it. Another Brazilian study indicate that, on the perception of the teachers, in addition to a lower performance, the students exposed to violence showed more behavioral problems (^{Pereira & Williams, 2008}). When comparing children directly or indirectly exposed to domestic violence with children without this type of exposure, Pereira, Santos and Williams (2009) observed differences on school performance, considering that the former showed a lower performance. The authors highlight the need for new investigations on the association between violence and school performance.

The way the family is organized also seems to influence the greater or lower exposure of adolescents to violence. A study conducted with adolescents from Rio Grande do Sul, in Brazil, indicated that those who did not live with both parents (the mother and the father) showed a greater exposure to violence (^{Zavaschi et al., 2002}). Another research, however, indicated that living with both parents or only with the father represented a protective factor only in relation to the exposure to physical violence in girls (^{Andrade et al., 2012}), although only a few studies observed the association between living or not with both parents and violence (^{Horta, Horta, Pinheiro, & Krindges, 2010}).

In order to provide visibility to a public health problem, which is the exposure to violence during adolescence, the aim of this study was to investigate the direct and indirect exposure to violence, as well as the associated variables (gender, age range, school failure and family configuration), in adolescents from public schools in the city of Porto Alegre, in the south of Brazil. This paper is part of a larger research project entitled "Exposure to violence during adolescence: Welfare and mental health relation", whose general objective was to investigate the relationship between the exposure to different types of violence, welfare and mental health in adolescents, through a quantitative design.

METHOD

RESULTS

Most adolescents that participated in the study were females, with ages between 12 and 15 years old, with no history of school failure, and belonging to the nuclear family configuration. The sociodemographic characteristics of the sample are described in Table 1.

Table 1 Sociodemographic characteristics of the adolescents 

The scores on the instrument Evaluation of children's Exposure to Violence in the Community (^{Zavaschi et al., 2002}) varied from 0 to 39, with a mean of 14.58 (SD=9,12) and internal consistency of 0.91. Direct violence varied from 0 to 8, with mean of 1.64 (SD= 1,77), while indirect violence varied from 0 to 34, with mean of 12.94 (SD=7,96). It may be observed that 65% of the adolescents were exposed to at least one episode of direct violence, while 97% were exposed to at least one form of indirect violence.

Table 2 shows the data related to the exposure to direct and indirect violence per gender, age range (12 to 15 years old and 16 to 18 years old), history of school failure and family configuration (nuclear, monoparental, or reconstituted configuration). Statistically significant differences were not observed across genders and different types of family configuration. However, differences were observed between age ranges, considering that the older adolescents showed greater exposure to direct and indirect violence; as well as between the group with and without a history school failure, considering that the adolescents that had failed at least once showed greater exposure, both to direct and indirect violence.

Table 2 Means and medians of the exposure to direct and indirect violence per gender, age range, school failure and family configuration 

Frequency analyses were conducted in order to verify, separately, the rates of exposure to direct and indirect violence, according to each affirmation from the instrument. Table 3 shows the five affirmations with higher frequency in terms of direct and indirect violence, considering that the latter shows more frequent items.

Table 3 Percentage of the most frequent items regarding the exposure to direct and indirect violence 

DISCUSSION

The results confirm the high exposure of adolescents to direct violence, a problem that has already been described and discussed by other Brazilian studies (^{Almeida et al., 2013}; ^{Braga & Dell'Aglio, 2012}; ^{Souza et al., 2014}; ^{Waiselfisz, 2014}). However, this research included mainly aspects related to indirect violence (witnessing or hearing about it) in extra-family contexts - such as robberies, attacks, shooting. The results indicate a high percentage of exposure to indirect violence, since almost all adolescents have been exposed to at least one episode of violence. This aspect is an important differential of the study, since in Brazil, the focus has been on researches related to direct violence occurred mainly in the family (Braga & Dell'Aglio, 2012; ^{Hildebrand et al., 2015}) and there are only a few investigations on the indirect exposure and in extra-family contexts (^{Benetti et al., 2006}; Benetti et al., 2010).

The exposure to both forms of violence may be a result of violence in the urban areas of the country, which has expanded in terms of occurrence and contexts (^{Souza & Lima, 2006}; ^{Waiselfisz, 2014}; ^{Zavaschi et al., 2002}). According to ^{Marín-León, Oliveira, Barros, Dalgalarrondo and Botega (2007}), the reality is constituted by the frequent exposure to violent images both on the media (virtually) and in real life, by witnessing violence in the community, for example. Therefore, it is not surprising that the adolescents have witnessed or heard about violent situations at least once. However, it is noteworthy that, on the instrument used, violence situations on the television, news or the internet are not considered, although studies state that this type of exposure also impacts the development, affecting mainly on the aggressive behavior of children and adolescents (^{Vieira, Mendes, & Guimarães, 2010}). Therefore, perhaps the indirect exposure to violence is even greater, since it is common for the media to daily point out episodes of violence and deaths (Marín-León et al., 2007).

Therefore, this study highlights that, in addition to being exposed to several types of violence as direct victims - as observed by other researchers (^{Braga & Dell'Aglio, 2012}; ^{Moreira et al., 2013}; ^{Waiselfisz, 2014}), adolescents are also exposed to several forms of indirect violence, as witnesses. In fact, the percentages indicate that violence is inserted in all contexts and in several ways, and the exposure and daily coexistence with violence is common (^{Benetti et al., 2006}; ^{Zavaschi et al., 2002}). On this study, significant relationships were observed between exposure to violence and age range, as well as between violence and school failure. No associations were observed between gender and family configuration.

Although with a slight to moderate effect size (d =0,39 for direct violence and d=0.46 for indirect violence), significant differences were observed in relation to the age range, with an exposure mean to both types of violence more frequent in older adolescents (16 to 18 years old), thus confirming findings of previous studies (^{Benetti et al., 2006}; ^{Fowler et al., 2009}). Most exposure to violence in older adolescents may be explained by their greater experience - considering their age, with a larger number of situations where there is no direct supervision of adults, and therefore, the exposure percentages and means are higher than for younger adolescents. On the other hand, it may be inferred that younger adolescents remain for a longer time within the family environment than older adolescents and, therefore, they are less exposed to the risk situations in the street context.

Regarding school performance, adolescents who had failed at least once at school were more exposed both to direct and indirect violence, with a moderate effect size (d =0,49) for direct violence and small effect size for indirect violence (d =0,36). It may be inferred that the exposure to violence, both directly and indirectly, affects the school performance, and it may lead to failing (^{Margolin & Gordis, 2000}). On the other hand, adolescents with a low performance at school may be exposed to higher risk situations, such as violence, since studies have indicated that school can be both a protective and a risk factor, although it may also be a risk factor (^{Andrade & Acle-Tomasini, 2012}; ^{Nunes, Pontes, Silva, & Dell'Aglio, 2014}). International studies found an association between exposure to physical violence and performance at school, with lower performance and greater history of failing among young adults that suffered from violence during their childhood, when compared to the ones who did not suffer from this. For the authors, there are factors that mediate the relationships, such as, for example, the fact that the aggressor is a member of the family (^{Tanaka et al., 2015}). Therefore, it is considered that the relationships between exposure to violence and school failure, mainly regarding indirect violence, must be further investigated.

On the other hand, no differences were observed on the exposure to direct and indirect violence in relation to family configurations (monoparental, nuclear and reconstituted configuration). On a review of empirical articles, it was observed that family configurations does not differ in terms of developmental results in children and adolescents (^{Oliveira, Siqueira, Dell'Aglio, & Lopes, 2008}). In fact, results from other studies have indicated that the relationships established in the family are more important than its organization (^{Mota & Matos, 2009}; ^{Sbicigo & Dell'Aglio, 2012}). This result indicates that other factors, apart from family configuration and gender, may explain the higher or lower exposure of adolescents to direct and indirect violence, suggesting the need for studies that evaluate broader personal and contextual variables, from the understanding of the phenomenon of violence as a complex process.

The most frequent items of exposure to direct violence, according to Table 3, suggest that adolescents are exposed to the different forms of violence, mainly to physical violence, in the family context. By analyzing the most frequent affirmations regarding indirect violence, it is observed that most adolescents have witnessed drug trafficking, physical violence, robberies, invasions and deaths; aspects that confirm that adolescents are inserted in risk contexts for their development, and that these situations are increasingly present, mainly in Brazilian capital cities (^{Souza & Lima, 2006}). These data may indicate the growth of violence in Brazil, described by some studies (^{Murray et al., 2013}; ^{Porto, 2002}) and, consequently, its presence in all the developmental contexts regarding adolescents, whether in intra or extra-family contexts (^{Marín-León et al., 2007}; ^{Mrug, Madan, & Windle, 2015}; Porto, 2002). The consequences of these different forms of violence and exposure, during adolescence, must be investigated.

Some limitations of the study must be pointed out. For example, those regarding the instrument used to investigate the direct and indirect exposure to violence. The instrument deals mainly with the exposure to indirect violence, with 37 items related to this form of exposure, and only 12 related to direct exposure. In addition, most of the affirmations do not specify the context of the exposure, whether it is intra or extra-family; only one affirmation is explicitly characterized as intra-family ("I got slapped, punched or beat up by a member of my family" and the equivalent affirmations regarding indirect violence). Other affirmations mention forms of violence (attacks, robberies, invasions, shooting, death, chasing, drug trafficking, etc.) that may occur both in the intra and extrafamily context, although they usually occur in the community.

Also, in relation to the instrument, it is possible to point out that it did not include all forms of violence (physical, psychological, sexual and negligence), only physical and sexual violence, although studies indicate the co-occurrence of psychological violence associated to other forms of violence (^{Costa et al., 2007}). However, this is the only instrument that has been translated and adapted to Brazil, including both forms of exposure (direct and indirect vio lence), identified through a search on recent publications in the area.

In Brazil, there are underreports of violence against children and adolescents, and there is some difficulty to obtain data on the occurrence of violence due to several factors (^{Almeida et al., 2013}; ^{Costa et al., 2007}; ^{Pelisoli, Pires, Almeida, & Dell'Aglio, 2010}). This study may also have had limitations to access the cases of violence, since it was necessary to obtain the signature of the parents or tutors on the Consent Term. If, on one hand, many adolescents may have forgotten to bring the signed document, others may have not obtained the consent from the parents. Although obtaining the consent from the parents (or other tutors) is an ethical issue, this demand may make the process of collecting information more difficult, and even indicate cases of violence that need intervention (^{Sbicigo, Tronco, & Dell'Aglio, 2013}).

Despite the limitations, this study seems to be a warning indication for the situation of the exposure of adolescents to violence. It is important to point out that violence is a complex phenomenon, and that several factors are related to the greater or lower exposure to violence during adolescence. Therefore, it is suggested that other variables, such as socioeconomic status, cultural and religious aspects, parental educational level, among others, are considered on new investigations. It is also necessary to investigate the effects of these forms of exposure (direct and indirect violence) and, mainly, the effects of indirect exposure, on the biopsychosocial development on the short, medium and long term, through quantitative, but also through qualitative and longitudinal studies. Public policies in the area of violence must consider the indirect exposure to violence as an important aspect that deserves intervention.