Criminal victimization in childhood and adolescence according to official records : the Pelotas ( Brazil ) birth cohort study

1 Programa de Pós-graduação em Epidemiologia, Universidade Federal de Pelotas, Pelotas, Brasil. 2 Department of Psychiatry, University of Cambridge, Cambridge, U.K. 3 Universidade Católica de Pelotas, Pelotas, Brasil. Erika Alejandra Giraldo Gallo 1 Ana Maria B. Menezes 1 Joseph Murray 2 Luciana Anselmi Duarte da Silva 1 Fernando César Wehrmeister 1 Helen Gonçalves 1 Fernando Barros 1,3 ARTIGO ARTICLE


Introduction
Violence is defined as the use of force or power as a form of threat against other persons, groups, or communities, with the likelihood of causing injury, death, psychological harm, alterations in development, or deprivation 1 .The World Health Organization (WHO) 1 has declared violence a serious public health problem.According to the United Nations 2 , 468,000 homicides were committed in the world in 2011.In 2010 3 , 25.5 million years of healthy life were lost due to injuries resulting from interpersonal violence.
A person that suffers mental, physical, or economic harm, or harm by omission, and which violates the criminal code, is considered a "victim", individually or as a group 1 .According to global data, 16% of the general population has been a victim of at least one crime in the previous year, while 40 million children under 15 years of age suffer abuse and neglect 4 .Approximately 150 million girls and 73 million boys under 18 years were forced into sexual relations or suffered forms of victimization by sexual violence that involved physical contact 2 .Between 133 and 275 million children witness domestic violence every year 4 .The International Crime Victim Survey (ICVS) 5 , using information on individuals over 16 years of age from 54 countries from 2000 to 2010, showed that countries of Africa had the highest incidence of victimization by theft (15.2%), aggression (19.8%), and sexual crime (14.5%), with 5-year recall; Latin American and Caribbean countries recorded the highest incidence of victimization by robbery (22.2%), and developed countries the highest victimization from domestic violence (12.4%) 5 .
The incidence of victimization in American adolescents 12 to 20 years of age decreased from 175 to 47.5 per 1,000 adolescents between 1993 and 2011 6 .In Australia 5 , the incidence remained stable from 2000 to 2011 (5.1 and 5.9 per 1,000 adolescents, respectively).In Canada 7 , the childhood victimization rate was 9.7 victims per 1,000 children; victimization at 2 and 3 years of age was more common in boys, and after 5 years of age it was more common in girls.
In Brazil, the National Household Sample Survey (PNAD) 8 , conducted in 26 states in 2009, covering the population over 10 years of age and using a questionnaire with a one-year recall, showed that 7.4% of Brazilians self-reported victimization by robbery and 1.6% by physical aggression.According to the same survey, victimization by robbery was more common among individuals in the highest family income bracket, while victimization by physical aggression was more common in the lowest income bracket.
The Center for Studies on Violence conducted a joint study in 2010 9 in 11 cities of Brazil, interviewing 4,025 persons over 16 years of age: 4.2% of victimization in the previous 12 months was due to physical aggression and 0.6% due to physical injury with firearms.
The map of violence against children and adolescents in Brazil 10 , based on records from the Information System for Notifiable Diseases (SINAN), showed that victimization involving physical violence in 2011 had an incidence of 20.2 per 100,000 children and adolescents (less than 19 years of age).As for victimization from sexual violence 10 , the highest incidence was in the 10 to 14 year-old bracket, with 23.8 per 100,000 adolescents.
The aim of this article was to analyze the different types of victimization recorded in official sources among members of the 1993 birth cohort followed up to 18 years of age in the city of Pelotas, Rio Grande do Sul State, Brazil.

Methodology
The current study was based on information on members of the 1993 Pelotas Birth Cohort Study in the city of Pelotas.A unified database was constructed containing information from both sources.This information allowed identifying the following for each incident of victimization: date of the incident, aggressor's relationship to the victim (family: father, mother, sibling, grandparent, uncle/aunt, cousin, stepfather, stepmother, brother-in-law/ sister-in-law; community: boyfriend/girlfriend, friend, neighbor, classmate, or stranger), use of weapon (yes/no), and classification of the crime.The crimes identified in the different sources were classified according to the Brazilian legislation prevailing at the time of the incident 11,12,13 and grouped in the classes presented in Tables 1 and 2. Each article of the law was classified in turn as violent versus non-violent victimization, according to the WHO definition 1 .The analysis excluded victimization from traffic accidents, not considered crimes per se.
To ensure standardization of the classification system and adequate data quality control, three undergraduate law students conducted classification independently of each other, after orientation by co-author Joseph Murray, expert in criminology from the University of Cambridge (United Kingdom).The three students later compared the information to assess the classificatory agreement; when there were differences, a dis-

237
Removal of a child or adolescent from the custody of others as provided by law or court order for the purposes of foster home placement.
cussion was held among the three, supervised by the expert in criminology (Joseph Murray) until a consensus was reached.Mothers of children born from January 1 to December 31, 1993, in the city's five maternity hospitals were interviewed by trained personnel to obtain demographic, socioeconomic, and health data.These children were later followed at different time intervals.Details on the 1993 cohort study have been published in previous articles 14,15,16 .
The statistical analyses used Stata (StataCorp LP, College Station, USA).Incidence rates were obtained per 1,000 person-years; time exposed to risk was calculated with: date of birth as the start date, and end date as last date until death, the occurrence of victimization, or one of the cohort follow-ups.For those for whom no information was available from a given period, we added half of the time at risk until reaching 18 years, assuming that these losses were distributed uniformly between the groups.For example, for those followed at 11 years, but without subsequent followups and without having died or suffered some incident of victimization, we added 3.5 years to their time at risk.This addition was only applied to 0.8% of the sample.Incidence of victimization according to socioeconomic characteristics was estimated using the ir command in the Stata statistical package.The 1993 birth cohort study was approved by the Ethical Research Committee of the Federal University of Pelotas.Until the cohort members reached 18 years of age, their mothers or legal guardians signed a free and informed consent form.From 18 years on, the form was signed by participants themselves.

Results
Of the total of 5,249 live born children, onefourth had mothers with four years of schooling or less, and 17.4% of the mothers were under 19 years of age.77.3% of the children's mothers reported having white skin color and 12.4% lived with a partner.Concerning cohort members, two-thirds were the first child and 50.3% were females (Table 3).
According to official records, 1,150 cohort members appeared in 1,396 police records as victims as of December 31, 2012.The total incidence of victimization was 15.7 per 1,000 person-years.Table 3 shows the incidence rate for victimization according to socioeconomic characteristics and the child's sex.Incidence was higher in girls (p < 0.05), in children of families in the lowest family income quintile (p < 0.05), with mothers less than 19 years of age (p < 0.001), and with single mothers (p < 0.05).
Table 4 shows the incidence of violent and non-violent victimization according to sex.The highest incidence rates for violent victimization were: physical injuries, robbery, blackmail, and crimes against personal liberty.Females had higher incidence rates than males for violent victimization involving crimes against personal liberty (p < 0.001), against sexual dignity, and misdemeanors (p < 0.01), while males had higher incidence rates than females for violent victimization involving robbery (p < 0.001) and crimes against life (p < 0.01).For non-violent victimization, the highest incidence rate was for theft (1.7 per 1,000 person-years).Girls had higher incidences of crimes against honor and misdemeanors, when compared to boys (significant differences, p < 0.001).Table 5 shows the incidence rate for victimization according to the aggressor's relationship to the victim and the use of weapons.Community victimization showed a higher incidence (12.6 per 1,000 person-years) when compared to family victimization (3.1 per 1,000 person-years).Among crimes perpetrated by family members, neglect and crimes against family were the most common.When the aggressor was someone from the community, the most common crimes were physical injuries, robbery, blackmail, and crimes against personal liberty.The non-violent crime with the highest incidence was theft.Victims' aggressors used some kind of weapon in 3.5 per 1,000 person-years.

As for age at the time of victimization (Figures 1a and 1b
), there was a slow upward trend in victimization up to 11 years of age, and a sharp increase from 11 to 18 years.Violent and nonviolent victimization showed similar patterns.Victimization at the hands of community aggressors was more common than family victimization, except in the 5 to 9 year age-bracket.

Discussion
The principal conclusion from this prospective study with more than 5,000 children in a city in southern Brazil is that officially recorded criminal victimization increases gradually during childhood and then rapidly throughout adolescence, and that violent crimes showed the highest incidence.The highest incidence rates for violent victimization were physical injuries, robbery and blackmail, and crimes against personal liberty, while the highest incidence rate for non-violent victimization involved theft.To our knowledge, this is the first prospective study on the incidence of officially recorded victimization, and it is also a pioneering study on criminal victimization in childhood and adolescence in a low or middleincome country.
Various studies on perpetration of crime (unlike those on victimization) have used official records and questionnaires to assess crime 17 .Criminal records tend to measure more serious crimes, while questionnaires measure more frequent and less serious crimes 18 .However, previous studies on victimization have only used questionnaires or interviews to measure incidents of victimization retrospectively.In Brazil, official records begin when a police report is filed, while in many countries these records only include data on arrests and convictions, which may explain why many other studies of victimization do not include official law enforcement records.
A clear advantage of official records on victimization is that they provide detailed information gathered when the incident occurred, for example, they date of the alleged crime and its legal classification.The obvious disadvantage of official records is that a significant proportion of crimes are not reported to the police.According to the British Crime Survey 19 , only 39% of crimes are reported to law enforcement agencies in the United Kingdom.Instituto Futuro Brasil 20 revealed 68% underreporting in the city of São Paulo, consistent with the findings from the British survey.The main reasons identified by international studies for victims not to report crimes 19 have to do with the lack of severity of some crimes and a lack of trust in the police.Other reasons cited in the literature 21,22 attribute lack of reporting of violent victimization to a so-called "code of silence", whereby family, friends, neighbors, schoolteachers and school staff, and even victims themselves refrain from filing police complaints due to fear of subsequent violent reprisals.The current study detected a large increase in officially recorded victimization over the course of adolescence.Adolescence is a time of important changes in cognition, behavior, and social orientation 23 which can increase the risk of victimization 24 , especially in the community.Involvement in crime also increases during adolescence 25 , and there is a strong correlation between perpetration and victimization 26 .Part of the increase in officially reported victimization during adolescence may be explained by an increase in the rate of reporting by victims themselves, as opposed to childhood, when reporting is done mostly by the victim's parents.
There were important differences in officially recorded victimization according to the child's sex and some of the family's social and demographic characteristics.The study showed significant differences in victimization between boys and girls, consistent with other studies 8,21,27,28,29 .Women are more prone to report violent victimization involving crimes against their sexual dignity, against their personal freedom, and misdemeanors against their person 29,30,31 .Meanwhile, men mainly report violent victimization involving robbery, physical aggression, and crimes against life 7,21,32,33,34,35,36 .Non-violent victim-ization involving crimes against honor (false incrimination, slander, libel) was more common in women, corroborating other studies 29,30,31,32,33,3 4,35,36,37,38,39 .Studies in many countries show that men are more involved in violence and crime when compared to women 40 .Therefore, the greater propensity of males towards being victims of robbery, aggression, and crimes against life might be explained by greater male participation in crime.Since the majority of sexual crimes are committed against the opposite sex 29 , the higher rate of male perpetrators may also explain the higher incidence of sexual victimization in women.
As documented by multiple studies 20,21,37,41,42,43,44 , low family income is an important factor in violent victimization.The current study's results showed an inversely proportional relationship between family income and victimization.Lower maternal education was also inversely associated with victimization, as in other studies 20,37 .The fact that children of single mothers showed higher incidence of victimization may reflect that single mothers have fewer resources to watch over and protect the children in families without a father present, or it may reflect other socio-demographic factors associated with the mother's conjugal status, such as family income.Birth cohort members born to teenage mothers had a higher incidence of victimization 45,46,47,48 .Teenage pregnancy is an important risk factor to be contemplated in public health strategies, aimed at preventing violence against children and adolescents and its consequences.
Direct comparisons of incidence rates in different Brazilian and international studies are difficult, due to differences in the procedures for recording, characterization, and classification of victimization, diversity of ages in the target population, and diverse data collection methods.Studies in the literature on victimization 14,36,49,50,51,52 used different types of questionnaires with different recall periods for the incidents.Many studies also focus on high-risk target groups 53,54,55,56,57 .
Studies in developing countries are scarce, and most focus on the adult population or indi-Cad.Saúde Pública, Rio de Janeiro, 32(8):e00072915, ago, 2016 viduals older than eleven years 20,21,32,33,37,44,48 .Risk factors are investigated during the interview, leading to recall bias if the events or exposures occurred at early ages.Many studies in the literature include any form of victimization, like fights between siblings, school bullying, or witnessing violence in the neighborhood or school, which hinders comparison between different studies.
The study's principal limitation is that no data were obtained on self-reported victimization, to compare with official records.In addition, 26 older court cases could not be located in the Circuit Court archives.Another 39 cases of letters rogatory were located that indicate that the respective cases were filed with other circuit courts, not Pelotas.After contacting the various circuit courts, 67% of these cases were clarified.Notwithstanding the limitation involved in locating some cases, the study adopted a conservative approach, i.e., the results may actually be underestimated.We preferred this more conservative approach rather than potentially overestimating the results.
Demographic and socioeconomic factors collected at the child's time of birth may influence different forms of victimization of adolescents later in life, as demonstrated in other studies 4,8,21,22,33,48,52 .Official crime data are known to potentially reflect only the "tip of the iceberg", but it is important for future epidemiological studies to adopt a longitudinal design, collecting detailed information on the victimization recorded in official databases and self-reported by participants, in order to better elucidate the problem.Such research would also enable careful profiling of victimization, identifying risk and protective factors and possible consequences over the individual's life.
More efforts at surveillance and detection of violence need to be implemented, controlled, and prioritized through efficient communication networks, with active participation by teachers, school officials, community workers, health professionals, law enforcement agencies, public ombudsman's offices, community leaders, and other stakeholders.

The 5 ,
249 live births in the cohort study were searched systematically in the databases of both the Secretariat for Public Security and the Special Court for Children and Youth.In 2013, information on victimization in the cohort was consulted in the archives of the Rio Grande do Sul State Secretariat for Public Security and the Special Court for Children and Adolescents in Pelotas.The Secretariat for Public Security consists of the Military Brigade, the General Forensics Institute, the Civil Police, and the Division of Prison Services (SUSEPE).Police records are reported by these agencies and stored by the Secretariat for Public Security in the Integrated Consultations System.The city of Pelotas has only one circuit court for minors, which hears cases on denial of parental custody, child abuse (< 18 years), and criminal cases involving children (0-11 years) or adolescents (12-18 years) 11 as perpetrators, among other proceedings.

Figure 1 Victimization
Figure 1 Victimization according to age, type of victimization, and victim's relationship to aggressor.1993 Pelotas Birth Cohort, Rio Grande do Sul State, Brazil.
Contributors E. A. G. Gallo, A. M. B. Menezes, J. Murray, and F. C. Wehrmeister participated in the data analysis, writing of the manuscript, and approval of the final version.L. A. D. Silva, H. Gonçalves, and F. Barros collaborated in writing the article, critical revision of the intellectual content, and approval of the final version.Cad.Saúde Pública, Rio de Janeiro, 32(8):e00072915, ago, 2016 Víctimas de Crimen; Violencia; Niño; Adolescente Submitted on 07/May/2015 Final version resubmitted on 11/Aug/2015 Approved on 25/Aug/2015

Table 1
Types of victimization according to Brazilian legislation, classified as violent.

Table 1 (
continued) Material neglect: without just cause, withholding or failure to provide support to spouse or to child under 18 (eighteen) years of age or unfit for work or to parent or grandparent over 60 (sixty) years of age, not providing them with the necessary resources or failing to provide legally established or adjusted child support; without just cause, to fail to assist a seriously ill child, parent, or grandparent.246Intellectualneglect:failure, without just cause, to provide for primary schooling for school-age child.249Abduction of minor: to abduct a minor (under 18 years of age) or remove from the power of someone who has legal custody 15Firing a firearm or otherwise triggering ammunition in an inhabited place or vicinity, in a public byway or in its direction, without involving the aim of committing another crime: sentence -imprisonment for 2 (two) to 4 (four) years, and fine.2,848/1940 Penal Code 137 Brawl: to engage in a brawl, except to separate the contenders 250 Fire: to cause a fire, thereby endangering the lives, physical integrity, or property of others 344 Coercion in the course of proceedings: use of violence or serious threat to favor one's interests or those of others, against an authority, party, or any other person acting in or called on to participate in court, law enforcement, administrative, or arbitration proceedings 4,898/1965 Regulates the Right to Representation and due process in Civil Administrative and Criminal Liability, in cases of abuse of authority 6 Abuse of authority 8,069/1990 Rules on the Statute of Children and Adolescents and other provisions.

Table 2
Types of victimization according to Brazilian legislation, classified as non-violent.
Damage: to damage, disable, or degrade someone's property 168 Unlawful appropriation: to appropriate things from someone who possesses or holds them 171 Embezzlement: to obtain, for oneself or others, illegal advantage, to the detriment of others, inducing or maintaining someone in error, through deceit, trickery, or any other fraudulent means 359 Contempt of a court injunction on loss or suspension of rights: to exercise a role, activity, right, authority, or function that has been suspended or rescinded by court injunction 3,688/1941 Misdemeanors Act 31 To release a dangerous animal, or to entrust same to the custody of an inexperienced person, or to fail to properly restrain said animal

Table 3
Incidence of total victimization and socioeconomic characteristics at birth.1993 Pelotas Birth Cohort, Rio Grande do Sul State, Brazil.

Table 4
Incidence of different types of victimization according to sex.1993 Pelotas Birth Cohort, Rio Grande do Sul State, Brazil.
* Number of incidents of victimization;** Incidence per 1,000 person-years at risk.

Table 5
Number of incidents of victimization and incidence rates for different types of victimization, according to aggressor's relationship to victim and use of weapons.1993Pelotas Birth Cohort, Rio Grande do Sul State, Brazil.
* Number of incidents of victimization;** Incidence per 1,000 person-years at risk.