The prevalence and factors associated with ever perpetrating intimate partner violence by men receiving substance use treatment in Brazil and England: A cross‐cultural comparison

INTRODUCTION AND AIMS
Intimate partner violence (IPV) perpetration is common among men who use substances. Substance use is a contributing factor for IPV perpetration. This cross-sectional study determined lifetime prevalence and factors associated with ever perpetrating IPV by men receiving substance use treatment in Brazil (n = 281) and England (n = 223).


DESIGN AND METHODS
IPV, adverse childhood experiences, attitudes towards gender relations and roles, current health state, substance use, depressive symptoms and anger expression were assessed. Logistic regression determined factors associated with ever perpetrating any (emotional, physical and/or sexual) IPV. Multinomial logistic regression determined factors associated with ever perpetrating different types of IPV.


RESULTS
74.6% (373/500) reported ever perpetrating IPV: 16.5% (82/498) emotional IPV only, 46.4% (231/498) physical IPV (with/without emotional IPV) and 11.6% (58/498) sexual IPV (with/without emotional and/or physical IPV). Higher anger expression, higher depressive symptoms, fighting physically with another man in the past year (Brazil only), experiencing a greater number of adverse childhood experiences and a higher hazardous drinking score (England only) predicted ever perpetrating IPV. Compared to never perpetrating any IPV, anger expression was associated with emotional and physical IPV perpetration; fighting physically with another man in the past year was associated with physical IPV perpetration and experiencing a greater number of adverse childhood experiences and a higher hazardous drinking score were associated with both physical and sexual IPV perpetration.


DISCUSSION AND CONCLUSIONS
Integrated interventions that address IPV and substance use delivered in substance use treatment could improve outcomes for perpetrators and victims.[Gilchrist G, Radcliffe P, Noto AR, d'Oliveira AFPL. The prevalence and factors associated with ever perpetrating intimate partner violence by men receiving substance use treatment in Brazil and England: A cross-cultural comparison. Drug Alcohol Rev 2017;36:34-51].


Introduction
Intimate partner violence (IPV), that is, controlling, coercive or threatening behaviour, violence or abuse between ex/current-partners, is a leading contributor to disease burden [1], impacting negatively on victims' mental, physical and reproductive health [2][3][4][5], and resulting in high societal costs [6]. No single factor explains IPV [7]. Lower socio-economic status, adverse childhood experiences (ACE), substance (alcohol and/or drug) use, psychological problems, anger expression, perpetrating other forms of violence, having inequitable gender attitudes, support of gender-specific roles and permissive attitudes towards violence against women are associated with IPV perpetration [8][9][10][11][12][13][14][15]. Men receiving treatment for substance use [16][17][18] report higher rates of IPV perpetration (34-60% in past year) than men in the general population [15]. Around half of men in perpetrator programs have substance use problems [19]. Alcohol, cocaine and methamphetamine use are associated with IPV perpetration [20][21][22][23][24][25]. There are several explanations for the correlation between substance use and IPV [14,22,23]. Impaired cognitive processing as a result of the pharmacological properties of substances could result in IPV perpetration [26], substance use causes marital conflict that could lead to IPV perpetration [27] or that the relationship is because of risk factors common to both substance use and violence [28]. Alternatively, substance use may be the mechanism for reducing the threshold at which a perceived provocation results in IPV for people who do not usually behave aggressively, but not for those who are aggressive regardless of whether they are under the influence of substances [29].
While IPV is common in all cultures and countries [15,30], research from general practice populations suggests that men in Brazil may be more likely to perpetrate IPV (52%) [31] than men in England (16%) [32]. Brazil has higher gender inequality (Gender Inequality Index in Brazil was 0.457 compared to 0.177 in UK) [33] and higher rates of general violence (intentional homicide rate 23.4/per 100 000 population in Brazil compared to 1.0 in UK) [34], which may contribute to this higher prevalence [15].
This study determined: (i) the prevalence of ever perpetrating IPV by men receiving substance use treatment; (ii) the risk factors for IPV perpetration in Brazil and England, countries with different cultures and drug use profiles (e.g. men in England predominantly attend substance use treatment facilities for alcohol, heroin and/or crack use [35], while in Brazil alcohol, cocaine and crack cocaine are most commonly used [36,37]); and (iii) the factors associated with ever perpetrating different types of IPV (emotional, physical and sexual) compared to never perpetrating IPV.

Method
Procedure A convenience sample of 519 participants were recruited during November 2014 to June 2015 by researchers in six public health system funded outpatient community substance use services in São Paulo, Brazil (one provided by direct government administration and five provided by a social organisation), three in London (two provided by the National Health Service and one provided by a third sector organisation) and three in South East England (provided by a third sector organisation). Services were representative of available substance use treatment provision in both countries and provided free of charge. Researchers verbally explained the study to potential participants and gave them a study information sheet prior to gaining informed consent. Participants received a £10 gift voucher or monetary equivalent for their time in England only, as this was not usual research practice in Brazil. As previous studies reported a high proportion of incomplete questionnaires among this client group [18] and similar disclosure rates of sensitive or stigmatising information (including IPV and substance use) have been reported across face-to-face interview and paper-and-pencil questionnaire [38], 17 (five females in England, and six females and six males in Brazil) experienced interviewers (all received 8 h training on interview administration and study protocol) administered the interviews in a private room to enhance completion rates. Men aged 18 or older, who were engaged in outpatient substance use treatment and were able to give informed consent, were eligible to participate. Researchers approached all men during the treatment opening hours. However, when researchers were interviewing, potential participants may have been missed. Six hundred and thirty-seven eligible men were invited to participate; 86.7% (288/332) in Brazil and 75 [30]. Questions on participants' perception of theirs and their partners' IPV perpetration (e.g. it was wrong but it was not a crime, why they/their partner behaved that way) were recorded [39,40]. Age, relationship status, living arrangements, highest level of education attained (no schooling/primary education (i.e. attended primary school or left high school without qualifications) and secondary education (i.e. left high school with qualifications and/or completed further education/university)), current employment status, how the participant managed on their available income [41] and self-reported Hepatitis C and HIV seroprevalence were collected. Adverse childhood experiences. Ten ACE were summed to calculate a mean score (childhood sexual and physical abuse [42], witnessing inter-parental violence [43], father never/rarely at home, mother never/rarely at home [44], being looked after or adopted, neglect, parental death, separation/divorce and being told you were weak or lazy). Total ACE score was calculated only for participants that responded to all 10 ACE (470/504; 93.3%).
Substance use. Hazardous drinking in the previous 12 months was assessed [45]. Participants were asked how many days in the past 30 they had used a list of illicit drugs, whether they thought their current/most recent partner had a problem with alcohol and/or drug use, what substances they had sought treatment for and the length of time they had been receiving treatment.
Mental health, anger and health state. Depressive symptoms were assessed [46]. Participants were also asked whether they had ever been told by a health professional that they had manic depressive illness or bipolar disorder [47]. Anger expression and control were measured [48]. Current health state was assessed using a Visual Analogue Scale [49].
Criminality. Questions on arrests and imprisonment for the following crimes were recorded: crimes against property or fraud (burglary, larceny, shoplifting, fraud, forgery, extortion, receiving stolen goods); possession or dealing drugs; domestic violence; crimes of violence other than domestic violence (robbery, assault, arson, rape, homicide, manslaughter) and possession of a weapon [50].
Participants were also asked about physical violence towards other men outside of their intimate relationship.
Culture. Participants were asked whether they practised religion and what religion they practiced. Attitudes about relations between men and women [44] and attitudes towards gender roles were assessed [30].

Analysis
Descriptive statistics were calculated using frequencies and percentages for categorical data and means and standard deviations (SD) for continuous data. Odds ratios (OR) and 95% confidence intervals (CI) were calculated using logistic regression. Differences in sample characteristics are presented in Table 2. Table 3 describes variables associated with any IPV perpetration by country. Variables with cell counts of ≥10 and P ≤0.2 in the univariate analyses were entered into backward stepwise multivariate logistic regression analyses to ascertain variables associated with any IPV perpetration for each country (Table 4). There was no evidence of multicollinearity among the independent variables included in the multiple logistic regression analyses [51]. Using backward stepwise multinomial logistic regression and a reference group consisting of participants who reported never perpetrating any IPV (no IPV), factors associated with the following outcomes were examined: perpetrated emotional IPV only [emotional IPV], perpetrated physical IPV (with/without emotional IPV) [physical IPV] or perpetrated sexual IPV (with/without emotional and/or physical IPV) [sexual IPV] ( Table 5). Main effects and interactions between main effects and country were considered.

Results
The mean age of participants was 43 years (SD 10.6, range 19-73 years), the majority were heterosexual (96.6%) and lived in their country of birth (93.3%). Only 37.5% of participants were currently married/had an intimate partner. Participants from England were more likely to be homeless or be unemployed/receiving benefits, and less likely to have no/primary schooling only, live in their country of birth, practice a religion or to have been unfaithful in their current/most recent relationship than participants from Brazil (Table 2).

Substance use
Participants were more likely to be receiving treatment for drugs and less likely to be receiving treatment for alcohol in England, with 65.3% of men in England compared to 74.0% of men in Brazil meeting criteria for hazardous drinking in the past 12 months. The most commonly used drugs in the past 30 days in Brazil were cocaine, cannabis and crack; and heroin, crack and cannabis in England. Participants in England were more likely to be poly drug users and to report that their current/most recent partner has/had a problem with alcohol and/or drug use (Table 2).

Health
Almost half the sample (48.7%) met criteria for probable depression and 17.4% had ever been told by a health professional that they had manic-depressive illness or bipolar disorder. Participants from England reported poorer current health state and were more likely to self-report being Hepatitis C seropositive (Table 3), potentially because of having greater numbers of current injectors in England (n = 61) than Brazil (n = 0).

Adverse childhood experiences
ACE were common, 65.5% reported any (physical or sexual) childhood abuse and 70.0% had witnessed inter-parental violence. Participants from England had experienced a greater number of ACE (Table 2).

Attitudes towards gender relations/roles
Participants from England were more likely to support gender equitable relations and less gender stereotyped attitudes towards gender roles (Table 3).

Criminality
Participants from England were more likely to have committed and been arrested for crimes against property/fraud, possession or dealing drugs, IPV, other violent crimes, possession of a weapon and were almost        three times as likely to have fought physically with another man than participants from Brazil ( Table 2).

Intimate partner violence
Three quarters reported ever perpetrating any (emotional, physical and/or sexual) IPV (373/500). Participants from England were more likely to have perpetrated physical IPV and less likely to have perpetrated sexual IPV than participants from Brazil (  Variable(s) entered on step 1: manage on available income, at least one child living with participant, current/most recent partner was a substance user, been in physical fight with another man in past 12 months, number of adverse childhood experiences, anger expression index, AUDIT total score, PHQ-9 total score, cocaine use in past 30 days, crack use in past 30 days, GEM score, had sex with someone else during current/most recent relationship a . c Variable(s) entered on step 1: education, practiced a religion, been in physical fight with another man in past 12 months, number of adverse childhood experiences, anger expression index, AUDIT total score, ever arrested for violent crime, GEM score to Attitudes to gender relations score, Gender roles score to Attitudes to gender roles score. AUDIT, Alcohol Use Disorders Identification Test; CI, confidence interval; IPV, intimate partner violence; OR, odds ratio; PHQ-9, Patient Health Questionnaire. Perpetrated physical IPV (with or without emotional IPV, but sexual IPV not perpetrated). c Perpetrated sexual IPV (with or without emotional and/or physical IPV). d Variable(s) entered on step 1: country (forced), been in physical fight with another man in past 12 months, number of adverse childhood events, anger expression index, AUDIT total score, PHQ-9 total score, and interaction between country and each of these variables. AUDIT, Alcohol Use Disorders Identification Test; CI, confidence interval; OR, odds ratio; PHQ-9, Patient Health Questionnaire. Only 6.9% (14/202) of participants in Brazil who reported perpetrating IPV had ever been arrested for this, and none had ever been imprisoned. In England, 24.7% (42/170) of those who reported perpetrating IPV had ever been arrested for it and 12.1% (20/165) had been imprisoned as a result. Despite similar proportions of men meeting criteria for ever perpetrating IPV only 15.7% (44/280) of participants in Brazil and 27.5% (60/ 218) considered they had ever been in a relationship where they could be described as 'domestically violent or abusive' towards their partner (OR 2.04, 95% CI 1.31, 3.16). When those who had perpetrated IPV were asked to reflect on their behaviour: only 31.8% (87/274) of those who perpetrated physical IPV and 18.5% (10/ 54) of those who perpetrated sexual IPV considered their behaviour was 'a crime'. The majority believed they had been using alcohol and/or drugs at the time they last perpetrated emotional (221/306, 72.2%), physical (204/ 275, 74.2%) or sexual (42/54, 77.8%) IPV. Participants also reported high levels of lifetime IPV victimisation (425/501, 84.8%).

Factors associated with ever perpetrating any IPV by country
Variables associated with ever perpetrating any IPV in bivariate analysis by country are reported in Table 3.
The following variables remained significant in the multiple logistic regression model predicting any IPV perpetration for participants from Brazil: had a physical fight with another man in the past 12 months, experiencing a greater number of ACE, higher depressive symptomatology and higher anger expression. The following variables remained significant in the model predicting any IPV perpetration for participants from England: higher anger expression and a higher Alcohol Use Disorders Identification Test (AUDIT) score (Table 4).

Factors associated with type of IPV ever perpetrated
In the multinomial logistic regression analysis, compared to participants who had never perpetrated any IPV: the perpetration of emotional IPV was associated with having higher anger expression; the perpetration of physical IPV was associated with having a physical fight with another man in the past 12 months, higher anger expression, experiencing a greater number of ACE and a higher AUDIT score; and the perpetration of sexual IPV was associated with a greater number of ACE and a higher AUDIT score.

Discussion
The prevalence of IPV perpetration is high among men receiving treatment for substance use, far higher than among general population or general practice samples from the same countries [15,31,32]. Our findings are comparable to lifetime rates of psychological (77%) and physical IPV (49%-54%) perpetration reported in other studies of men receiving substance use treatment [17,52]. We found a lower rate of sexual IPV perpetration than other studies (32%) [17], perhaps because of different methodologies used. Participants from England were more likely to have perpetrated physical IPV and less likely to have perpetrated sexual IPV than those from Brazil. That participants from England were more likely to perpetrate physical IPV contrasts with general violence levels reported in each country [34]; therefore, the differences reported in our study are more likely to be a result of the differences in the treatment samples recruited and the profile of clients (e.g. different drugs used and methods of drug administration) attending these substance use treatment services in each country (e.g. 86% and 57% of participants in Brazil were receiving treatment for alcohol and drugs respectively compared to 35% and 76% in England; and 23% of participants in Brazil compared to 52% in England were poly drug users), rather than country or cultural differences per se. Participants from England reported greater deprivation (homelessness, unemployment/receiving benefits) and complex needs (greater number of ACE, poly drug use/injecting, higher criminality/violence outside their relationships) that may have contributed to the higher prevalence of physical IPV perpetration [9,11,14,15]. Participants from Brazil reported greater infidelity, were less likely to support gender equitable relations and more likely to hold gender stereotyped attitudes towards gender roles [15,33,34]. For example in our study, a greater proportion of participants from Brazil than England agreed or strongly agreed with the statements 'if a woman doesn't physically fight back its not rape' (18% vs. 4%) and 'that a woman cannot refuse to have sex with her husband' (29% vs. 4%) which may account for the higher lifetime prevalence of sexual IPV perpetration reported.
Similar risk factors for IPV perpetration found in our study have been reported in studies among males receiving substance use treatment: higher anger expression, higher depression symptomatology, physical fight with another man (Brazil only), greater number of ACE (Brazil and England) and a higher AUDIT score (England only) [17,[53][54][55]. Compared to never perpetrating any IPV, anger expression was associated with emotional and physical IPV perpetration; fighting physically with another man in the past 12 months, experiencing a greater number of ACE and a higher hazardous drinking score were associated with physical IPV perpetration; and experiencing a greater number of ACE and a higher hazardous drinking score were associated with sexual IPV perpetration.
Similar to other studies [18], lifetime IPV victimisation reported by participants was also high. This may be because of participants' partners responding in selfdefence, conflicts around substance use [56] or it may also be influenced by reporting bias.

Adverse childhood experiences
We found a significant association between ACE and IPV perpetration in both countries and also for physical and sexual IPV perpetration [18,[53][54][55]. Experiencing ACE influences IPV perpetration partially through psychosocial characteristics including depression, anxiety, and impulsivity [57]. This relationship could be mediated through substance use, with more severe dependence reported by substance users who have experienced childhood abuse [58]. Forty percent of male IPV perpetrators are also violent towards their children [59] which is a significant risk factor for IPV perpetration, mental health and substance use problems in adulthood [11,60,61]. As many men in substance use treatment are fathers, and may still have access rights to their children or co-parent; it is important that they are offered interventions that address the father-child relationship [62] or family-based interventions that address both partner and parent-child aggression [63], to reduce the 'intergenerational' transfer of IPV as a result of learned behaviours or acceptance of such behaviours in adult relationships [64]. Such interventions have shown promising results in reducing IPV perpetration, substance use and improving parent-child relationships [62,63].

Alcohol
Similar to other studies, we found that hazardous drinking predicted any IPV perpetration and physical and sexual IPV perpetration [10,13,14,18,22]. Alcohol expectancies or beliefs about the cognitive and behavioural effects of alcohol contribute to IPV perpetration in addition to alcohol use [65].

Anger
Anger expression was associated with both emotional and physical IPV perpetration. A recent meta-analysis found that IPV perpetration was associated with anger, hostility, and internalising negative emotions [66]. Similar to other studies we found that a history of IPV victimisation was a strong predictor of IPV perpetration [67], and that the majority of participants reported a history of being both perpetrators and victims of IPV [68]. ACE and anger remained significant (or marginally significant) in the models predicting emotional, physical and sexual IPV perpetration. Emotion dysregulation resulting from ACE include problems with understanding, responding to, expressing and managing emotional responses and is associated with IPV perpetration [69]. ACE have been associated with anger-related dysregulation, which has been found to mediate the relationship with IPV perpetration [70]. There is also evidence of altered brain functioning as a result of ACE to parts of the brain responsible for emotion and anger, and those that have a role in functions that relate to adult behaviours including attention, inhibition, emotion, expression of personality and moderation of learned social behaviour [71], that could contribute to IPV perpetration.

Masculinity
Societies and individuals that support stronger ideologies of male dominance have elevated rates of IPV perpetration [72]. Few participants acknowledged their IPV perpetration as a crime. While gender equitable relations and attitudes towards gender roles were not predictive of IPV perpetration in this study, masculine roles (including physical fight with another man, infidelity and violent crime) were [15,[72][73][74]. It has been argued that holding traditional ideas about masculinity that support dominance over women often result in 'exaggerated displays of male heterosexuality' [73]. Moreover, it has been suggested that 'masculinity is embodied via alcohol-related violence that is perpetrated against intimates, acquaintances, and strangers alike' [75] (pp. 404).

Mental health
Depression is associated with IPV perpetration [13,32,44]. However, in our study higher depression symptomatology was associated with any IPV perpetration only for participants in Brazil. It is not clear why this finding was not reported in the English sample, as approximately half the sample in each country met criteria for probable major depressive disorder. The higher prevalence of hazardous drinking in Brazil may explain the association between depression and any IPV perpetration [76].

Treatment implications
Participants reported a high prevalence of ACE and IPV victimisation, suggesting that trauma-informed substance use treatment where 'service delivery is influenced by an understanding of the impact of interpersonal violence and victimization on an individual's life and development' [77] (pp. 462) may be beneficial to men.
Several studies have found that reductions in substance use results in reductions in emotional and physical IPV perpetration [78]. People in relationships where mutual violence is common may 'face frequent and intense provocation toward aggressive behaviour, with other impelling, inhibiting and disinhibiting factors [alcohol] moderating the likelihood that aggressive urges result in IPV perpetration' [67] (pp. 274).
Few perpetrator intervention studies have been conducted among men receiving substance use treatment. Integrated interventions, that address both substance use and IPV, appear promising [79][80][81][82].
Men receiving substance use treatment who perpetrate IPV are rarely referred to perpetrator programs [29,83]. Moreover, when they are, treatment completion is low and uptake is poor [29,84]. Perpetrator programs are traditionally offered through criminal justice settings; however, only 15% of IPV perpetrators in our study (7% in Brazil and 25% in England) had ever been arrested for IPV. While the Maria de Penha law introduced in Brazil in 2006 has increased penalties for perpetrators and support for victims, research suggests that there are still inefficiencies in the implementation of the law which may account for lower arrest rates in Brazil [85]. Community perpetrator programs out with the criminal justice system are estimated to meet around 10% of existing demand from referring agencies [86]. Substance users may be less likely to attend treatment in parallel systems in general [87] and for IPV perpetration more specifically [84]. Integrated interventions, that address both IPV and substance use, delivered in substance use treatment could ensure more perpetrators are reached and better outcomes achieved for perpetrators and victims [88] and their children.

Strengths and weaknesses
As this cross-sectional study recruited a convenience sample from outpatient services, findings may not be generalisable to men from other substance use treatment services and causal associations cannot be implied. Conflicting evidence surrounds the impact of interviewer gender (12/18 interviewers in our study were female) on disclosing sensitive issues [89]. Participants may be more likely to respond in a socially desirable way to questions about gender attitudes. Therefore, the prevalence of IPV reported and the support for gender norms/attitudes may be under-reported.

Conclusions
The prevalence of IPV perpetration is high among men receiving substance use treatment, far higher than among general population or general practice samples from the same countries [15,31,32,90]. Findings highlight the multiple and complex risk factors for IPV perpetration across both cultures. Similar to the ecological model of IPV [3,30], we have identified risk factors for IPV perpetration at the individual, relationship, community and societal levels that could inform population health prevention of IPV and integrated treatment approaches for perpetrators receiving substance use treatment.