Violence against women : prevalence and associated factors in patients attending a public healthcare service in the Northeast of Brazil

The prevalence of and factors associated with violence against women attended as outpatients between October 2005 and January 2006 by the Instituto Materno Infantil Professor Fernando Figueira (IMIP), Recife, Pernambuco State, Brazil, were investigated using a cross-sectional type study. 619 women over the age of 18 were included in a systematic probabilistic sample. The modified Abuse Assessment Screen (AAS) was applied and the data were evaluated statistically by way of univariate and bivariate analyses, using the χ2 or Fischer’s exact test and an adjusted multivariate logistic regression model. The prevalence of violence against women was 27.5% (95%CI: 24.0%-31.2%) in the twelve-month period prior to the consultation. The associated factors were low level of schooling (OR = 2.34), a history of domestic violence (OR = 2.21) the woman being mentally disturbed (OR = 2.35), and the partner’s consumption of alcohol (OR = 1.77). The prevalence of violence was high in the group of women studied, indicating the need to broaden preventive measures and all-round health care for women. Violence Against Women; Women’s Health; Health Services Recognized as a public health problem for over ten years, violent crime has contributed to an increase in mortality and morbidity rates among the most vulnerable population groups. According to the Pan-American Health Organization (PAHO) 1, violence is not a new health problem, although it has only recently been accorded the status of a human rights violation. The multiple causes of violence against women, according to the World Health Organization (WHO) 2,3 stem from various factors, ranging from those relating to the norms that rigidly define the social roles of men, giving them control and domination over women, to the acceptance of violence as a conflict-resolution strategy 2. Poverty, low socioeconomic status, unemployment, association with delinquent partners, isolation of the woman and family are community factors, and marital discord and the partner’s control of property and decision-making regarding family affairs are factors linked to personal relationships. Factors associated with the aggressor are being male and young, having witnessed domestic violence in childhood, being an absentee father or father who rejects the child, a lower level of schooling, depression, personality disorders, having suffered abuse in childhood and alcohol consumption 2. Beyond the emotional cost for women and their families, which are incalculable, violence against women has an economic impact on the health sector, as the women who are subjected 264 ARTIGO ARTICLE VIOLENCE AGAINST WOMEN: PREVALENCE AND ASSOCIATED FACTORS 265 Cad. Saúde Pública, Rio de Janeiro, 26(2):264-272, fev, 2010 to violence tend to be those that are most at risk, and therefore more frequently have recourse to the health service. This burden could be reduced if further preventive measures were taken in the sector 4. Violence against women has been on the WHO’s list of priorities since 1996 1 and has been the subject of various studies around the world. These studies have revealed the scale of the problem: a prevalence of 41% was found among women over 16 years of age in London 5, United Kingdom, and, in a representative sample of the adult population in China 6, 34% of women and 18% of men investigated had already been subjected to aggression during their relationships. A multi-country study of women’s health and domestic violence 7 found a 3.7 to 53.7% prevalence of physical or sexual violence perpetrated by an intimate partner in the previous 12 months in Serbia and Montenegro and in rural parts of Ethiopia. In Brazil, the study evidenced a prevalence of 9.3% in the city of São Paulo and 14.8% in the sugarcane plantation region in the State of Pernambuco. Schraiber et al. 8, in a literature review of studies of violence in Brazil, found a greater emphasis on the issue of gender in publications between 2000 and 2005 and studies in services indicated a higher frequency of violence committed by the partner with rates ranging from 36% to 45% for women reporting physical violence at least once in a lifetime and from 9% to 19% for sexual violence. Partner violence was surveyed in 15 capital cities in Brazil and in the Federal District as part of a wider study. The highest prevalence for physical abuse was 34.7% in the city of Belém, Pará State, in the North region of the country and the lowest was 12.8% in the city of João Pessoa, Paraíba State, in the Northeast region 9. At emergency referral services in the city of Salvador, Bahia State, also in the Northeast region, the results also showed a high prevalence of violence against women occurring at least once in a lifetime: 46% for general violence, 36.5% for physical violence and 18.5% for sexual violence 10. Studies conducted in Brazil 11,12,13 indicate a high prevalence of violence against women and suggest the need for improved knowledge of its causes to aid the design of public policy and the organization of healthcare services. The aim of this study is to estimate the prevalence of and factors associated with violence against women, among those attending a tertiary healthcare referral center, 100% of whose patients depend on the Brazilian public health system, the SUS. Subjects and methods A cross-sectional study was carried out to estimate the prevalence of and main factors associated with violence against women. The study was conducted at the Gynecology Clinic of the Women’s Care Center at the Instituto Materno Infantil Professor Fernando Figueira (WCC-IMIP), currently known as the Instituto de Medicina Integral Prof. Fernando Figueira (IMIP), located in the city of Recife, the capital of the State of Pernambuco, in the Northeast region of Brazil. The IMIP attends approximately 700,000 outpatients a year from the Metropolitan area of Recife and from other cities in the rural areas of Pernambuco, all of whom depend on the public healthcare service (SUS). A systematic sample procedure was carried out, with a sample interval equal to four, considering that approximately 2,400 women would have been attended at the gynecology outpatients clinic of the institution during these three months, had data been collected. The sample interval was calculated by dividing the number of expected patients (2,400) by N to give the prevalence estimate (600 including loss prevention). Women aged ≥ 19 years old, who were not pregnant and received care during the data collection period were included. The final sample size was 619 women, with an estimated error no greater than 3%, a 95% confidence interval and an expected prevalence of 20%. The Abuse Assessment Screen (AAS) 14,15 was used to collect data, with the omission of the question on violence during pregnancy, and closed questions on other variables were investigated and registered in a specific questionnaire. To investigate violence against women, women were asked whether anyone had in the last 12 months offended them with words, beaten, slapped, punched, pushed or hurt them physically in any way and who this person was. The variables investigated were: had suffered physical and sexual violence (AAS) 14, place of residence (Recife, Metropolitan area of Recife, rural parts of the State of Pernambuco and other States); color and race (established during the interview and categorized as white or non-white), age (completed years), schooling (completed school years), martial status (married/cohabiting or non-married); occupation (waged or not); children (have children or not) and household income (the total monthly income of the interviewee’s household members, the minimum wage in Brazil being R$ 300 ≈ US$ 132 at the time of the study); sexual violence (if, in the previous year, a woman had been forced to have sex against her will and who forced


The prevalence of and factors associated with violence against women attended as outpatients between October 2005 and January 2006 by the Instituto Materno Infantil Professor Fernando
Figueira (IMIP), Recife, Pernambuco State, Brazil, were investigated using a cross-sectional type study.619 women over the age of 18 were included in a systematic probabilistic sample.The modified Abuse Assessment Screen (AAS) was applied and the data were evaluated statistically by way of univariate and bivariate analyses, using the χ 2 or Fischer's exact test and an adjusted multivariate logistic regression model.The prevalence of violence against women was 27.5% (95%CI: 24.0%-31.2%) in the twelve-month period prior to the consultation.The associated factors were low level of schooling (OR = 2.34), a history of domestic violence (OR = 2.21) the woman being mentally disturbed (OR = 2. 35), and the partner's consumption of alcohol (OR = 1.77).The prevalence of violence was high in the group of women studied, indicating the need to broaden preventive measures and all-round health care for women.

Violence Against Women; Women's Health; Health Services
Recognized as a public health problem for over ten years, violent crime has contributed to an increase in mortality and morbidity rates among the most vulnerable population groups.According to the Pan-American Health Organization (PAHO) 1 , violence is not a new health problem, although it has only recently been accorded the status of a human rights violation.
The multiple causes of violence against women, according to the World Health Organization (WHO) 2,3 stem from various factors, ranging from those relating to the norms that rigidly define the social roles of men, giving them control and domination over women, to the acceptance of violence as a conflict-resolution strategy 2 .Poverty, low socioeconomic status, unemployment, association with delinquent partners, isolation of the woman and family are community factors, and marital discord and the partner's control of property and decision-making regarding family affairs are factors linked to personal relationships.Factors associated with the aggressor are being male and young, having witnessed domestic violence in childhood, being an absentee father or father who rejects the child, a lower level of schooling, depression, personality disorders, having suffered abuse in childhood and alcohol consumption 2 .
Beyond the emotional cost for women and their families, which are incalculable, violence against women has an economic impact on the health sector, as the women who are subjected to violence tend to be those that are most at risk, and therefore more frequently have recourse to the health service.This burden could be reduced if further preventive measures were taken in the sector 4 .
Violence against women has been on the WHO's list of priorities since 1996 1 and has been the subject of various studies around the world.These studies have revealed the scale of the problem: a prevalence of 41% was found among women over 16 years of age in London 5 , United Kingdom, and, in a representative sample of the adult population in China 6 , 34% of women and 18% of men investigated had already been subjected to aggression during their relationships.
A multi-country study of women's health and domestic violence 7 found a 3.7 to 53.7% prevalence of physical or sexual violence perpetrated by an intimate partner in the previous 12 months in Serbia and Montenegro and in rural parts of Ethiopia.In Brazil, the study evidenced a prevalence of 9.3% in the city of São Paulo and 14.8% in the sugarcane plantation region in the State of Pernambuco.
Schraiber et al. 8 , in a literature review of studies of violence in Brazil, found a greater emphasis on the issue of gender in publications between 2000 and 2005 and studies in services indicated a higher frequency of violence committed by the partner with rates ranging from 36% to 45% for women reporting physical violence at least once in a lifetime and from 9% to 19% for sexual violence.
Partner violence was surveyed in 15 capital cities in Brazil and in the Federal District as part of a wider study.The highest prevalence for physical abuse was 34.7% in the city of Belém, Pará State, in the North region of the country and the lowest was 12.8% in the city of João Pessoa, Paraíba State, in the Northeast region 9 .At emergency referral services in the city of Salvador, Bahia State, also in the Northeast region, the results also showed a high prevalence of violence against women occurring at least once in a lifetime: 46% for general violence, 36.5% for physical violence and 18.5% for sexual violence 10 .
Studies conducted in Brazil 11,12,13 indicate a high prevalence of violence against women and suggest the need for improved knowledge of its causes to aid the design of public policy and the organization of healthcare services.The aim of this study is to estimate the prevalence of and factors associated with violence against women, among those attending a tertiary healthcare referral center, 100% of whose patients depend on the Brazilian public health system, the SUS.

Subjects and methods
A cross-sectional study was carried out to estimate the prevalence of and main factors associated with violence against women.The study was conducted at the Gynecology Clinic of the Women's Care Center at the Instituto Materno Infantil Professor Fernando Figueira (WCC-IMIP), currently known as the Instituto de Medicina Integral Prof. Fernando Figueira (IMIP), located in the city of Recife, the capital of the State of Pernambuco, in the Northeast region of Brazil.The IMIP attends approximately 700,000 outpatients a year from the Metropolitan area of Recife and from other cities in the rural areas of Pernambuco, all of whom depend on the public healthcare service (SUS).
A systematic sample procedure was carried out, with a sample interval equal to four, considering that approximately 2,400 women would have been attended at the gynecology outpatients clinic of the institution during these three months, had data been collected.The sample interval was calculated by dividing the number of expected patients (2,400) by N to give the prevalence estimate (600 including loss prevention).Women aged ≥ 19 years old, who were not pregnant and received care during the data collection period were included.
The final sample size was 619 women, with an estimated error no greater than 3%, a 95% confidence interval and an expected prevalence of 20%.
The Abuse Assessment Screen (AAS) 14,15 was used to collect data, with the omission of the question on violence during pregnancy, and closed questions on other variables were investigated and registered in a specific questionnaire.
To investigate violence against women, women were asked whether anyone had in the last 12 months offended them with words, beaten, slapped, punched, pushed or hurt them physically in any way and who this person was.The variables investigated were: had suffered physical and sexual violence (AAS) 14 , place of residence (Recife, Metropolitan area of Recife, rural parts of the State of Pernambuco and other States); color and race (established during the interview and categorized as white or non-white), age (completed years), schooling (completed school years), martial status (married/cohabiting or non-married); occupation (waged or not); children (have children or not) and household income (the total monthly income of the interviewee's household members, the minimum wage in Brazil being R$ 300 ≈ US$ 132 at the time of the study); sexual violence (if, in the previous year, a woman had been forced to have sex against her will and who forced her to); history of family violence (witnessed someone who was abused in her own residence during her childhood or adolescence) and which types of aggression witnessed (insults, beating or slapping, verbal threats, hit or spanked a lot, threats involving weapons or pushing) investigated in a non-exclusive fashion.
The behavioral risk factors investigated included alcohol consumption (yes or no and does not drink or occasionally drinks or drinks once or more than once a week), use of other drugs (the woman was asked if drugs were used, which drugs were used and who used them) and mental disorders (with further investigation of the morbidity) 16,17,18 .
The research assistants were trained using an instruction manual designed for this purpose, which followed WHO ethical and security recommendations for studies of violence against women 19 .The consistency of the questionnaire was verified in a pilot study (47 interviews) and the interviews were conducted in a private room with appropriate conditions to ensure compliance with ethical principles and confidentiality.All participants signed informed consent forms.
In order to determine the association between the explanatory variables and the response variable, the chi-square test or Fisher's exact test was employed preliminarily with p < 0.05 (Epi Info 6.04; Centers for Disease Control and Prevention, Atlanta, U.S.A.).The simultaneous effect of the explanatory variables on the response variable was assessed using a multivariate logistic regression model, carried out by the selection of all variables with p < 0.
A total of 29 (4.7%) women reported having been subjected to physical aggression, as classified by the AAS.In 17 (58.6%) of the cases, the women suffered the following types of violence: slaps, pushing without hurting or wounding or lasting pain; and 24.1% suffered punches, kicks, bruises/black eyes, cuts and/or continuous pain.
The frequency of sexual violence reported was 4.7%; in 89.7% of these cases, the husband/partner was the perpetrator.
The demographic and socioeconomic profile of the women (Table 1) reveals that most, (320 or 51.7%), were from the Metropolitan area of Recife, and 223 (36%) from Recife city itself.The frequency of violence against women in these two regions was 28.45% and 29.9%, respectively.
Almost one third of the women, (187; 30.2%), were in their fifth decade of life and 460 (74.3%) designated themselves as black or of mixed race.The majority of the women, (373; 60.3%), had between one and eight years of schooling, 505 (81.6%) had a partner and 541 (87.4%) had children.236 (38.1%) declared having a low income of double the minimum wage or less (< R$ 600 ≈ US$ 264 at the time of the study).The associations between violence against women and age, color, schooling, and children were positive (p < 0.05).
The frequency of violence was higher among younger women (p = 0.039) and among black women and those of mixed race (p = 0.027) compared with white women.With regard to schooling, violence against women was more frequent among those who had from zero to eight full years of schooling (p < 0.001).Among women who had a partner, the frequency of violence was 28.5% and violence was more prevalent among those without offspring compared to those with children (p = 0.039).
Of the women's partners (Table 1), 121 (25.3%) were in their fifth decade of life, and most (68.3%) were brown and black; 212 (46.8%) had between four and eight full years of schooling and 386 (68.3%) were in paid employment.The frequency of violence against women was higher among women who had a younger partner and a low level of schooling, and there was a similar distribution of violence against women, regardless of the partner's skin color and income.
Eighty-five (39.7%) of the women who had been subjected to violence also reported having witnessed domestic violence in childhood and/ or adolescence (p < 0.001) (Table 2) and the kinds of violence reported were: 30.4% insults, 25.5% punches, 21.8% verbal threats, 19.2% beatings, 6.8% threats involving weapons and 22.6% pushing.In 148 cases (69.8%), a person of male gender was reported to have been the perpetrator of this violence.In the group studied, a high frequency of violence against women was found among women who reported having a mental disorder (p < 0.001) (Table 2).
A significant association was found (p = 0.001) between the pattern of alcohol consumption of the partner and violence against women, with an  increase in violence when the partner's pattern of alcoholic use was more frequent (Table 2).
The adjusted multivariate logistic regression model included the variables: age, skin color, schooling, children, having witnessed violence in childhood and/or adolescence and mental disorders, in the case of the woman; and schooling, alcohol consumption and mental disorders, in the case of the partner.
The results revealed the following to be factors associated with violence against women: lower level of schooling among women (OR = 2.34), the woman having witnessed violence in childhood and/or adolescence (OR = 2.21), alcohol consumption on the part of the partner once or more than once a week (OR = 1.77) and mental disorders (OR = 2.21) (Table 3).

Discussion
The present study found a high prevalence of physical and emotional violence (27.5%).A result similar to that found by Schraiber et al. 11 (34,1%) at a primary care center in the city of São Paulo and to other studies carried out in Brazil 9,10,13 .However, Reichenheim et al. 9 found a higher prevalence (73.9% for verbal aggression and 18.8% for physical violence in the city of Recife).These authors exclusively studied acts of violence committed by the partner among women with a current partner, taking into consideration age, and used different instruments from ours, which may explain the different results 6,20,21,22 .
Violence against women has been widely studied in Brazil, although there are relatively few Cad.Saúde Pública, Rio de Janeiro, 26(2):264-272, fev, 2010 strongly associated with violence against women, supporting the results of Gómez-Dantés et al. 24 , Klevens 36 , Rivera-Rivera et al. 29 , and Menezes et al. 13 .Several investigations of violence against women 13,23,36 including those conducted by the WHO 2 , present this variable as a risk factor for violence against women.Low levels of schooling are almost always associated with poverty, although it should be pointed out that most studies of violence against women involve women who use the public health service, a fact which suggests a lack of information on the extent to which the problem exists among women from more affluent social classes, who therefore probably have more educational opportunities 24 .
Mental disorders among the women were found to be a factor strongly associated with violence against women.In the literature 2,20,39 , mental disorders such as depression, phobias, post-traumatic stress and others are referred to as a consequence of the violence.The etiology of mental disorder, in some cases, is attributed to a combination of factors -genetic, organic, functional and/or psychological.A mental disorder may produce some degree of negative impact on interpersonal relationships and an individual's inner world.The negative impact on contact with the outside world may lead to an exacerbation of the violent impulse against oneself or towards other people.These disorders may be associated with co-morbidities, such as alcohol use, or may have emerged as a consequence of having been subjected to violence 40 .
Although they produce weaker evidence regarding causality, cross-sectional studies in the area of health may reveal the true magnitude of the problem and the results may suggest the need to take the efficient and effective measures necessary to ensure a broader impact 41 .
There is also a need to address the difficulty of investigating violence against women arising from the private nature of the phenomenon.It is possible that some of the women interviewed withheld information, either as a result of feelings of shame and humiliation or for fear of exposing their private lives and the fact that they have been subjected to violence in their own home.
Furthermore, a gold standard could have been used for variables such as alcohol consumption and mental disorders.Despite these limitations, the results of the present study could be used to aid the development of measures to identify and deal with hitherto overlooked cases of violence against women and to help educate health workers and the community alike.studies that include risk factors, most of them population-based studies, using a different design.This diversity represents a rich source of research for what is a serious public health problem, but makes it difficult to compare data.

Resumo
Our findings on sexual violence are similar to those described by Vizcarra et al. 23 and McCloskey et al. 21, who found a prevalence of 3.4% in the city of Temuco, in Chile, and 3% in Tanzania, respectively.Our findings also came close to the 7.5% described by Gómez-Dantés et al. 24 in a study carried out in Mexico.The similarity of the results could be attributable to the cultural similarities of these countries, where power relations between men and women are very unequal.The social power conferred on the male gender recognizes and legitimizes violence against women and reinforces women's submissiveness, even with regard to sex.
Violence against women was observed to be constant in all age groups studied here, being more frequent among younger women, with a possible decline among those of a more advanced age.Kronbauer & Meneghel 25 report results of a higher frequency of psychological violence among older women (> 30 years old), and a higher frequency of psychological (59%) and physical (40%) violence among black women when compared with white women, even though the authors report higher frequencies in general.Our findings also indicate a higher frequency of violence against black women and those of mixed race.Racial discrimination is an important historical factor from a cultural and social point of view, and is a relevant factor in gender violence.As gender, race and ethnic background are structural aspects of society, the similarity of the results can be taken as a manifestation of the social representation of racial prejudice.However, this factor alone is not sufficient to explain the levels of violence encountered.
These results concur with those described by Menezes et al. 13 , Zalvídar et al. 26 and Kyriacou et al. 27 , who found that, in the United States, unemployment or intermittent employment and the partner's low level of schooling were risk factors for violence against women, although our study did not show a statistically significant association between this and violence against women.However, the frequency of violence against women was constant for various levels of schooling of the partner.The different results may be attributed to the different socioeconomic conditions of the populations studied and the distinction between the types of violence and the individual characteristics of women investigated, which have been proven to be important risk factors for violence against women.
Alcohol consumption by the partner once or more than once a week presented a positive association in the bi-variable analysis and remained a strong associated factor for violence against women after a multi-variable analysis, corroborating data collected by Boyle & Todd 28 in the United Kingdom and other studies 23,27,29,30,31 .Alcohol consumption by the partner has, in a variety of studies 27,28,31,32 , been considered to be a significant and serious factor associated with violence against women.An abusive user or alcohol-dependent partner tends to develop violent behavior, putting others at risk 33,34,35 .Domestic violence witnessed in childhood and/or adolescence was positively associated with violence against women and remained significant in the adjusted logistic regression model.Our findings support those of Menezes et al. 13 , Vizcarra et al. 23 and Klevens 36 .The data corroborate the literature 37,38 , confirming the psychoanalytical theory that argues that experiencing or witnessing violent behavior as a common and normal pattern of conflict resolution leads to such behavior-patterns being transmitted from one generation to the next.
Low levels of schooling among women remained significant in the adjusted logistic regression model, indicating that this is a factor

Table 1
Distribution frequency of women victims of violence according to socioeconomic and demographic characteristics of the woman and her partner, attended in the period between October 2005 and January 2006.Instituto de Medicina Integral Prof. Fernando Figueira, Recife, Pernambuco State, Brazil.

Table 2
Frequency distribution of women victims of violence according to risk behavior.Family violence witnessed during childhood and/or adolescence and woman's and partner's mental disorder.Attended in the period between October 2005 and January 2006.Instituto de Medicina Integral Prof. Fernando Figueira, Recife, Pernambuco State, Brazil.

Table 3
Multivariable analysis of factors associated with violence against women attended.In the period between October 2005 and January 2006.Instituto de Medicina Integral Prof. Fernando Figueira, Recife, Pernambuco State, Brazil.Contributors M. A. Silva, G. H. Falbo Neto and J. E. Cabral Filho participated in the conception of the project, analysis and interpretation of the data, and writing the article.J. N. Figueroa contributed with the analysis and interpretation of the data.