Introduction

The long-lasting health consequences of child maltreatment (any experience of physical, emotional, or sexual abuse, or neglect), as well as adverse socioeconomic outcomes in adulthood, have been well-documented [1,2,3,4,5]. Both biological and psychological mechanisms have been implicated in the adverse effects of neglect and abuse, particularly in relation to the stress response system, and the person’s sense of afety and trust in interpersonal relationships [6]. Recent research suggests that experience of early adversity can interact with later stressful experiences to produce even more toxic effects on health [7]. For example, in a nationwide study of adults in the USA, the impact of stressful events in the prior 12 months on mental disorders was about twice as large among individuals who had previously experienced multiple adversities in childhood compared to none [7]. However, evidence on the link between childhood maltreatment and later violence, a particularly severe form of stress, is limited and new studies are needed, particularly in low-and middle-income countries where child support services are under resourced and rates of violence are high [8].

Childhood maltreatment might predict risk for further violent victimisation, and thus continuity of stress across the life-course” of stress across the life-course, partly because of continuity in adverse environments conducive to victimisation [9]. Child maltreatment itself also might contribute to risk of future victimisation because of its impact on risk behaviours (such as alcohol or drug use, and aggressive behaviour) or because it predicts partnering, or interacting with more antisocial individuals [10, 11]. Later victimisation after maltreatment might occur in either the domestic or community context. In a large Canadian study, child maltreatment associated with increased risk for intimate partner violence, with larger effects for women than men [12]; however, like most studies of victimisation after child maltreatment, this survey had a cross-sectional design, and focused on only one type of violent outcome. Longitudinal studies are scarce and tend to include only women when examining risk of intimate partner violence (IPV) after child maltreatment. According to a systematic review in 2018 [13], only five longitudinal studies, all conducted in the United States, including a total of 1,516 women, had tested the association between childhood abuse and intimate partner violence against women, with a non-significant average effect (odds ratio = 1.3, 95%CI 0.93–1.80). Beyond IPV, there is a dearth of longitudinal evidence on how child maltreatment associates with family violence more generally in adulthood - committed by other family members as well as intimate partners.

Considering victimisation in the community, associations with prior child maltreatment may differ for women and men, particularly as males commit most community violence, and perpetration and victimisation are strongly associated [14].Although the “cycle of violence” literature has focused on the effects of maltreatment on the risk of perpetrating violence [15], less is known about risk for victimisation in the community. In the current study we present the epidemiology of maltreatment until age 15, and subsequent risk for experiences of violence in the family and community, in young adulthood at age 22 years, in a Brazilian longitudinal study. Prior evidence suggests that maltreatment tends to occur in an especially severe form in Brazil [16], that rates of violence against women are high [17], and serious community violence is an enormous social and public health problem [18]. Identifying links between these different forms of violence across the life-course is thus of great importance in this context.

Methods

Data from the 1993 Pelotas Birth Cohort were used, which has a prospective longitudinal design. In 1993, all live births in the city’s maternity hospitals taking place between January 1 and December 31 were identified, and the mothers were invited to participate in the study. A total of 5249 individuals (response rate of 99.7%) were included, and they were followed from birth to 22 years of age. For analyses on maltreatment and later violent victimisation, information from the following follow-ups was used: perinatal visit (n = 5249), 15 years (n = 4349; 85.7% of eligible individuals were followed) and 22 years (n = 3810; 76.3% of eligible individuals were followed). At 22 years, REDcap software was used to collect data (https://doi.org/10.1016/j.jbi.2008.08.010). Details about data collection are available in other published articles [19, 20]. See also the supplementary table.

Measurements

Child maltreatment up to age 15 years

Experiences of child maltreatment were assessed at the 15-year follow-up through questions from the Childhood Trauma Questionnaire (CTQ) [21, 22]. Questions applied in this study were the following: (a) Have there ever been fights with physical aggression in your home between adults or has an adult ever assaulted a child or adolescent?; (b) Has it ever happened that you did not had enough food at home or that you put on dirty or torn clothes because you had no others?; (c) Have you ever thought or felt that your father or mother never wanted you to have been born?; (d) Have you ever thought or felt that someone in your family hates you?; (e) Has it ever happened that an adult in your family or someone who was taking care of you hit you in a way that left you hurt or with marks?; (f) Has anyone ever tried to do sexual things with you against your will, for which they threatened you or hurt you? (g) Have you ever been separated from your parents to be cared for by someone else? A dichotomous maltreatment variable was created by coding “1” if any answer to the above questions was yes, and coding “0” if no answer was given as yes. All questions refer temporally from the participants’ childhood to the time of the interview, i.e. when they were at the age of 15 years.

Victimisation at age 22 years

At the 22-year follow-up, family victimisation and community victimisation were measured in relation to the preceding one-year period. Family victimisation violence was recorded as aggression perpetrated by a family member (father, mother, brother, sister, uncle, aunt, companion, or other person) against the participant, and victimisation due to community violence was considered to consist of aggression perpetrated by someone who was not a family member. Questions about violent victimisation were taken from previous victimisation studies conducted in Brazil [23, 24] and were adapted to identify whether the act of violence was committed by a person within the family or in the community. These questions were then tested in a pilot study among male and female undergraduate students prior to application in the current study. The five questions, first asked about in relation to family members, and then in relation to people in the community were as follows: (a) How many times has someone made a serious threat to hurt you physically?; (b) How many times has someone hit you, pushed you, kicked you or physically assaulted you without a weapon?; (c) How many times has someone attacked you with a knife, firearm or other weapon?; (d) How many times has someone stolen an object from you, using violence or threats?; (e) How many times has someone grabbed, touched or assaulted your private sexual parts against your will? Affirmative answers (“yes”) to any of these questions were computed as a positive event regarding family victimisation (where the act had been perpetrated by a family member) and community victimisation (where the act had been perpetrated by a non-family member). A final 4-category variable was created regarding victimisation at age 22 years: no victimisation, victimisation in the family only, victimisation in the community only, and co-occurrence of family and community victimisation.

Sociodemographic characteristics

Sociodemographic characteristics of the participants are reported based on measures assessed in the perinatal period and at 15 years of age. The following variables were used from the perinatal period: sex of the child (male/female); self-reported maternal age, which was then categorized into age groups (< 20; 20 to 29; or 30 years or more); maternal education in complete years of study, categorized as 0–4, 5–8, 9–11 or 12 years or more; family income in minimum monthly wages, categorized into wealth quintiles as Q5 (richest), Q4, Q3, Q2 and Q1 (poorest); and maternal marital status (with partner/without partner). At the age of 15 years, self-reported skin colour was recorded: white, black, brown or others (indigenous and East Asian), considered in this study as a socially relevant variable, rather than a biological entity.

Statistical analysis

First, the prevalence of young adult violent victimisation (considering separately: family violence only, community violence only, and both family and community violence) was presented according to sample sociodemographic characteristics and experience of maltreatment up to age 15 years. All these results are shown stratified by participant sex, because of different rates of exposure to violence for females and males, and because of potentially different consequences of maltreatment between by sex, as has been considered in the literature on mental health [25]. Associations between sociodemographic characteristics and violence outcomes were examined using Pearson’s chi-square test for dichotomous exposure variables and the heterogeneity test for categorical variables, and p < 0.05 was considered statistically significant.

In a second set of analyses, the association between maltreatment and young adult victimisation (considering separately: family violence only, community violence only, and both family and community violence) was examined in multinomial logistic regression (with no young adult victimisation as the reference category), adjusting for maternal and participant sociodemographic characteristics (used in continuous form for maternal age, family income, and maternal education). Adjusted odds ratios and 95% confidence intervals are reported, stratified by sex. One additional model using the whole sample was run to test for a possible interaction between sex and maltreatment in predicting later experiences of violence. The analyses were carried out in the Stata statistical software, version 15.1 (Stata Corporation, College Station, USA).

Results

Tables 1 and 2 show the sociodemographic profile of females and males in the study, respectively, as well as experiences of maltreatment up to age 15, and violence experienced at age 22. Up to age 15 years, 39% (95%CI = 37.02; 41.46) of females and 27% (95%CI = 25.09; 29.49) of males reported maltreatment (p < 0.001 for sex difference). At 22 years, 17.5% (95%CI = 15.8; 19.4) of females reported violent victimisation in the previous year (8.7% in the family only, 5.3% in the community only, and 3.6% in both contexts), compared to 20.2% (95%CI = 18.3; 22.3) of males (12.8% in the family only, 3.0% in the community only, and 4.4% in both contexts); p = 0.054 for sex difference in the overall prevalence of violence at age 22.

Table 1 Maltreatment and young adult victimisation prevalence among females, according to sociodemographic factors in the 1993 Pelotas Birth cohort study
Full size table
Table 2 Maltreatment and young adult victimisation prevalence among males, according to sociodemographic factors in the 1993 Pelotas Birth cohort study
Full size table

Maltreatment of females was more common among children with mothers who were young, less well educated, with lower family income, without a partner, and participants with brown or black skin colour (Table 1). Experiences of family and community violence in young adulthood were associated with skin colour (mostly higher rates for people of brown and black skin colour), and maltreatment by age 15 for females. For males, Table 2 shows that maltreatment was more common among children whose mothers did not have a partner, and participants with brown or other skin colour. Experiences of family and community violence in young adulthood were associated with maternal schooling, family income, skin colour, and maltreatment by age 15 for males. Experiencing both family and community violence at age 22 years was reported at a higher rate by youth with lower maternal schooling, lower family income, brown and black skin colour, and maltreatment experiences.

Table 3 shows the results of multinomial logistic regression, examining the association between maltreatment and young adult violent victimisation, adjusting for sociodemographic factors. For both females and males, maltreatment was associated with over twice the odds of experiencing both family and community violence in young adulthood. Associations between maltreatment and community victimisation only were also large for both females and males. However, associations between maltreatment and victimisation in the family were weak, and only significant for females. Across these three categories of young adult victimisation, there was no evidence that associations with earlier maltreatment significantly differed for females and males (all p values for interaction terms > 0.40).

Table 3 Unadjusted and adjusted associations between maltreatment and young adult victimisation, among (A) Females, and (B) Males
Full size table

Discussion

In a large population-based, prospective cohort study in Brazil, child maltreatment was strongly associated with later victimisation in both the family and community environments, and associations were similar for females and males. Although childhood socioeconomic conditions were associated with various forms of violence, childhood maltreatment predicted young adult victimisation in the family and community, independently of socioeconomic factors, demonstrating important continuity between child maltreatment and young adult victimisation independent of socioeconomic background.

Despite decades of studies demonstrating adverse biopsychosocial consequences of maltreatment [26], surprisingly little evidence is available on the longitudinal links between maltreatment and further victimisation, especially in the transition to adulthood. In the current study, experiencing maltreatment up to age 15 was associated with over double the odds of suffering community violence on its own, or both community violence and family violence in young adulthood. Although this increased vulnerability could reflect continuity in risky environments, the association remained adjusting for several important socioeconomic factors, suggesting that child maltreatment itself might also involve pathways of psychosocial and behavioural change, leading towards further victimisation.

Cascading experiences of victimisation through the life-course may put individuals at particular risk for poor health and social outcomes. According to the stress sensitisation hypothesis, individuals with stressful experiences such as maltreatment in early life may be particularly vulnerable to poor health when confronted by new experiences of stress later in the life-course [7]. As such, the risk of further victimisation after maltreatment in the current study points to experiences of multiple forms of stress through the life-course that are potentially particularly threatening to health, and may have clinical implications for people suffering violence across these multiple contexts [27].

In the current study, females were more exposed to any child maltreatment, as well as family violence in young adulthood, consistent with studies across many other populations [26, 28,29,30]. Violence against girls and women, especially in the family environment, is a major public health and human rights problem [31]. Worldwide, it is estimated 30% of women are exposed to lifetime violence perpetrated by intimate partners, or sexual abuse, although rates vary according to the country and the methodology applied [32]. A culture of machismo in Latin America is still a critical issue to address in preventing violence against women in this region, as well as empowering women, and reinforcing laws to combat violence against women [33]. Like in other studies [34,35,36], young men in this population were at greater risk of exposure to community victimisation, which is consistent with Brazilian official statistics showing that the highest rates of mortality caused by violence (homicides) are among young men aged 15–29 years [37]. While sex differences in health and social exposures across the life-course are deeply complex phenomena, relating to cultural, social and biological processes, the higher rate of exposure to community violence among young men may be understood partly in relation to a higher likelihood that young men are involved in risky activities in the community environment, including perpetration of crime and violence, as local epidemiological studies have reported [38, 39]. Gender studies also reveal processes of violence socialization among boys as they seek to confirm a masculine identity, leading them to become more frequently involved in risky situations leading to higher rates of community victimisation and early mortality, and that these issues are exacerbated by social inequalities in Brazil [40, 41].

Among the main strengths of this study is its large sample size based on a total population, its prospective design, and consideration of multiple contexts of violence, for both females and males. Among study limitations, the following should be considered. First, although maltreatment was measured at age 15, and many studies use retrospective reports taken later in adulthood [42]. The current study measure is still retrospective across childhood and may have been subject to memory bias, reducing reporting of maltreatment. Second, this study used an abbreviated set of questions on maltreatment that did not permit sub-analyses by type of maltreatment, and a questionnaire on violence in adulthood that was specific to the Brazilian population, which limits comparisons with other studies. In a large study, losses over a 22-year period of this study are inevitable, and may have biased the results. Those excluded from the analysis are similar to those included according to gender, maternal age and mother’s marital status and different according to maternal education, family income and skin color of the young person, compared to those included in the sample.

Among the limitations of the study, it is important to recognise that causal inference about the effects of maltreatment on later violence exposure was not the aim of the study, and cannot be made on the basis of the associations that were tested. In particular, absence of paternal data and data on other factors in the complex etiology of victimisation, such as the neighbourhood context were not included. Also, the socioeconomic conditions that were associated with violence in this study were examined in a purely descriptive form. We recognise that these socioeconomic factors are highly interdependent, and causal inference about any single factor is not intended. Another important limitation is that the measure of maltreatment at age 15 years is retrospective regarding childhood exposure and is subject to memory bias.

In conclusion, this study shows significant continuity in violent victimisation through the life-course, in a Brazilian population. Specifically, child maltreatment was strongly associated with later experiences of violent victimization in the family and in the community in young adulthood. Although health impacts of victimisation were not investigated in the current study, the association between maltreatment and later violence highlights clinical issues of concern. When maltreatment occurs, vulnerability towards further victimisation, not only in the immediate context, but in later life phases, requires attention. When violent victimisation occurs in early adult life, a history of maltreatment may be particularly significant to consider regarding concurrent traumatic responses.