SciELO - Scientific Electronic Library Online

vol.33 issue1Traditional and Complementary Medicine in Brazil: inclusion in the Brazilian Unified National Health System and integration with primary careHealthy lifestyle in São Paulo, Brazil author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand



  • text new page (beta)
  • English (pdf)
  • Article in xml format
  • How to cite this article
  • SciELO Analytics
  • Curriculum ScienTI
  • Automatic translation


Related links


Cadernos de Saúde Pública

On-line version ISSN 1678-4464

Cad. Saúde Pública vol.33 no.1 Rio de Janeiro  2017  Epub Jan 23, 2017 


Factors associated with physical violence against pregnant women from São Luís, Maranhão State, Brazil: an approach using structural equation modeling

Fatores associados à agressão física contra gestantes em São Luís, Maranhão, Brasil: uma abordagem com modelagem de equações estruturais

Factores asociados a la agresión física contra embarazadas en São Luís, Maranhão, Brasil: un enfoque con modelos de ecuaciones estructurales

Danielle Cristina Silva Costa1  * 

Marizélia Rodrigues Costa Ribeiro2 

Rosângela Fernandes Lucena Batista1 

Camila Maia Valente1 

João Victor Fonseca Ribeiro1 

Laysa Andrade Almeida1 

Ludmilla Emilia Martins Costa1 

Maria Teresa Seabra Soares de Britto e Alves1 

Antônio Augusto Moura da Silva1 

1 Programa de Pós-graduação em Saúde Coletiva, Universidade Federal do Maranhão, São Luís, Brasil.

2 Departamento de Medicina III, Universidade Federal do Maranhão, São Luís, Brasil.


The factors associated with physical violence against pregnant women were analyzed in a cross-sectional study of 1,446 pregnant women from a prenatal cohort who were interviewed in 2010 and 2011 in São Luís, Brazil. In the initial model, socioeconomic status occupied the most distal position, determining sociodemographic factors, social support and the behavioral factors that ultimately determined physical violence, which was investigated as a latent variable. Structural equation modeling was used in the analysis. Pregnant women who were from more disadvantaged backgrounds (p = 0.027), did not reside with intimate partners (p = 0.005), had low social support (p < 0.001) and had a high number of lifetime intimate partners (p = 0.001) reported more episodes of physical violence. Low social support was the primary mediator of the effect of socioeconomic status on physical violence. The effect of marital status was mainly mediated by a high number of lifetime intimate partners.

Keywords: Pregnant Women; Prenatal Care; Violence Against Women


Foram analisados fatores associados à agressão física contra gestantes, em um estudo transversal com uma amostra de 1.446 mulheres de uma coorte pré-natal, entrevistadas em 2010 e 2011 em São Luís, Maranhão, Brasil. No modelo inicial, o nível socioeconômico ocupou a posição mais distal, determinando os fatores sociodemográficos, de apoio social e comportamentais, que por vez determinavam a violência física, investigada enquanto variável latente. A análise usou modelagem de equações estruturais. O relato de mais episódios de violência física esteve associado estatisticamente ao nível socioeconômico mais baixo (p = 0,027), não residir com parceiro (p = 0,005), apoio social baixo (p < 0,001) e alto número de parceiros na vida (p = 0,001). Apoio social baixo apareceu como o principal mediador do efeito do nível socioeconômico sobre a violência física. O efeito do estado conjugal foi mediado principalmente pelo número de parceiros na vida.

Palavras-chave: Gestantes; Cuidado Pré-Natal; Violência contra a Mulher


Se analizaron factores asociados a la agresión física contra embarazadas, en un estudio transversal con una muestra de 1.446 mujeres de una cohorte prenatal, entrevistadas en 2010 y 2011 en São Luís, Maranhão, Brasil. En el modelo inicial, el nivel socioeconómico ocupó la posición más distal, determinando los factores sociodemográficos, de apoyo social y comportamentales, que a su vez determinaban la violencia física, investigada como variable latente. El análisis usó modelos de ecuaciones estructurales. El relato de más episodios de violencia física estuvo asociado estadísticamente al nivel socioeconómico más bajo (p = 0,027), no residir con pareja (p = 0,005), apoyo social bajo (p < 0,001) y alto número de parejas en la vida (p = 0,001). El apoyo social bajo apareció como el principal factor del efecto del nivel socioeconómico sobre la violencia física. El efecto del estado conyugal fue medido principalmente por el número de parejas en la vida.

Palabras-clave: Mujeres Embarazadas; Atención Prenatal; Violencia contra la Mujer


The expression violence against women expresses gender violence associated with unequal power relations that reflect the dominance of the male over the female 1.

Considered as a violation of human rights and a public health problem 1), (2), (3, gender violence is conceived as any threat or act of violence based on gender that was found to or that appeared to cause injuries or physical, sexual or psychological suffering to women, such as coercion or arbitrary deprivation of freedom, occurring either in a public or private place 2), (3.

Violence perpetrated against pregnant women has been highlighted as being more frequent than the complications that are routinely investigated before birth, such as pre-eclampsia and diabetes 4. Its prevalence varies from 0.9% to 57.1%, depending on the methods used to measure it and on sociocultural characteristics 4), (5), (6), (7. In Brazil, researchers of the WHO Multi-Country Study on Women's Health and Domestic Violence Against Women found an 8% prevalence of abuse of pregnant women in the municipality of São Paulo and an 11.1% prevalence in Zona da Mata of Pernambuco State 1. The prevalences of physical violence against pregnant women, the most investigated type of violence, varied from 1.2% to 40% 4) , (5), (6), (7.

Violence during pregnancy has been associated with socioeconomic, demographic and behavioral factors. The results regarding the associations between age, minority status, education, employment status, income level and parity with physical violence against pregnant women have been inconclusive. In addition, most studies have only used bivariate analyses and were performed in health services with homogeneous samples, which are prone to selection bias 4.

In a multi-centre study performed in sixteen states of the United States, physical abuse was more frequent among pregnant women who were young, unmarried, had less than 12 years of education, were non-white, received Medicaid benefits, and had an unintended pregnancy and in women with stressful experiences during pregnancy. In that study, only bivariate analysis results were reported 8.

A study in China, which used multiple logistic regression and the Abuse Assessment Screen to assess violence, showed a greater risk of physical maltreatment in women who had less than 9 years of education, had unemployed partners and were in a patriarchal family situation and those who smoked cigarettes and used alcohol and non-prescription drugs 9.

In Brazil, in São Paulo, no association was found between schooling and social class of pregnant women and physical abuse by an intimate partner 10. Another study conducted in Rio de Janeiro revealed that physical violence was more frequent against adolescent pregnant women who had a low educational level, did not work outside the home, had fewer prenatal appointments, had low levels of social support, were from low-income families, had three or more children aged less than five and who reported use of alcohol 11. In Campinas (São Paulo), pregnant women with low schooling and women who were heads of their household were more exposed to physical and sexual violence 12. In another study in Rio de Janeiro, physical violence was more frequent among pregnant adolescents who reported use of alcohol 13.

Most studies have used bivariate or multivariable analyses by means of logistic regression 4), (5), (11), (12), (13. There are criticisms of this type of analysis because it only investigates direct relationships between explanatory variables and an outcome, not allowing for the evaluation of intermediate paths, i.e., indirect or mediation effects 14), (15.

From this perspective, investigations that view violence as a phenomenon with multiple causes are still required, and statistical analyses that consider the factors associated with the causes as a complex interconnected structure should be conducted 1), (16.

This article had the objective of analyzing the factors associated with physical violence perpetrated against pregnant women who used prenatal services in the municipality of São Luís (Maranhão State). For this purpose, we used structural equation modeling, which allowed for simultaneous evaluation of direct and indirect effects of several variables on physical violence against pregnant women 14), (15.


This cross-sectional study used data collected from the BRISA prenatal cohort, which investigated new etiologic factors for preterm birth in São Luís from 2010 to 2011.

Participants and sample

Pregnant women who used prenatal services in the municipality of São Luís were enrolled in the study from 2010 onwards to be interviewed from the 22nd to the 25th weeks of their pregnancy. Having their first ultrasound performed at less than 20 weeks of pregnancy and intending to give birth at one of the maternity hospitals in the municipality were criteria for inclusion. Women with multiple pregnancies were excluded.

To analyze the factors associated with physical violence, a minimum number of 948 women would be necessary based on a type I error of 5%, a power of 80% and a prevalence of violence against pregnant women of 11.1%1 (estimate of the WHO Multi-Country Study in the Zona da Mata region of Pernambuco for violence during pregnancy).

From February 2010 to June 2011, 1,447 pregnant women participated in the study, which was conducted at the Clinical Research Centre (CEPEC) of the Federal University of Maranhão (UFMA). One interviewed woman was not included in this study because she failed to complete the questions on violence during pregnancy; thus, there was a total of 1,446 observations.

A convenience sample was used because it was not feasible to obtain a random sample representative of the population of pregnant women of São Luís.

Data collection and storage

Two questionnaires were used to collect data in the BRISA prenatal cohort. The Prenatal Interview Questionnaire, applied by interviewers, obtained information on the socioeconomic status, sociodemographic characteristics and behavioral factors of the pregnant women. From the Self-Applied Prenatal Questionnaire, which was answered individually by the pregnant women, data were extracted on physical violence during pregnancy and social support. Questions about perpetrators and episodes were asked.

Constructs and indicator variables

All variables were declared categorical in Mplus version 7.31 (Muthén & Muthén, Los Angeles, U.S.A.). Socioeconomic status, social support, and physical violence were treated as latent variables. Sociodemographic characteristics and pregnant women's behavioral factors were considered observed variables. The latent variable physical violence against pregnant women was a first-order construct that was considered the outcome.

Physical violence against pregnant women was obtained from the Brazilian version of the WHO Multi-Country Study questionnaire through six questions used to identify physical violence during pregnancy perpetrated by different subjects, not only by the women's intimate partners. The pregnant women replied to the following situations: since you became pregnant has someone (V5) slapped you or thrown something at you that could hurt you?; (V6) pushed or shoved you, hit you with a fist or something else that could hurt?; (V7) hit you with his/her fist or with some other object that could have hurt you?; (V8) kicked, dragged or beaten you up?; (V9) choked or burnt you on purpose?; and (V10) threatened you with, or actually used, a gun, knife or other weapon against you? Each of these questions on physical violence had the following possible response options: (a) never, (b) once, (c) rarely and (d) frequently 17.

The instrument Scale of Social Support from the Medical Outcomes Study (MOS) was used to investigate the tangible (four questions), positive social interaction (four questions), affectionate (three questions) and emotional/informational (eight questions, of which four were on emotional support and another four on informational support) dimensions of social support. The interviewed women replied with the frequency with which they could rely on someone in the following scenarios: (A9) to help you if you were confined to bed; (A10) you can count on to listen to you when you need to talk; (A11) to give you good advice about a crisis; (A12) to take you to the doctor if you needed it; (A13) who shows you love and affection; A14) to have a good time with; (A15) to give you information to help you understand a situation; (A16) to confide in or talk to about yourself or your problems; (A17) who hugs you; (A18) to get together with for relaxation; (A19) to prepare your meals if you were unable to do it yourself; (A20) whose advice you really want; (A21) to do things with to help you get your mind off things; (A22) to help with daily chores if you were sick; (A23) to share your most private worries and fears with; (A24) to turn to for suggestions about how to deal with a personal problem; (A25) to do something enjoyable with; (A26) who understands your problems; and (A27) to love and make you feel wanted. The social support questions had the following response options: (a) never, (b) rarely, (c) sometimes, (d) almost always and (e) always 18. Social support was analyzed as a second-order latent construct consisting of the tangible, positive social interaction, affectionate and emotional/informational dimensions.

The instruments used to investigate physical violence 17 and social support 19 were validated in Brazil. The World Health Organization Violence Against Women (WHO VAW) instrument was validated for this cohort of pregnant women 20.

Socioeconomic status, a more distal latent variable, was investigated as a first-order construct and consisted of the following observed variables: (a) occupation of the head of family (unskilled manual laborer, semi-specialized manual laborer, specialized manual laborer, office worker, higher level professional and administrator/manager/director/owner); (b) years of study of the pregnant woman (0 to 4, 5 to 8, 9 to 11, 12 or more); (c) family income in multiples of the monthly Brazilian minimum wage (less than 1, 1 to less than 3, 3 to less than 5 and 5 or more); and (d) Brazilian economic class (D/E, C and A/B). The instrument used to measure economic class was designed by the Brazilian Association of Research Companies (ABEP) 21. The monthly minimum wage in 2010 was BRL 510.00.

All of the observed variables were categorized in increasing order: age group of the pregnant woman and of the resident intimate partner within the household (up to 19, 20-24, 25-29 and 30 years or over); number of resident children of the pregnant woman (no children, 1 child, 2 children and 3 or more children); years of study of the resident intimate partner (0 to 4, 5 to 8, 9 to 11, 12 or more); and number of male intimate partners with whom the pregnant woman has had a sexual relationship in her life (1 partner, 2 or 3 partners, 4 or 5 partners and 6 or more). The marital status of the pregnant woman was categorized as married, consensual union, unmarried/widow or divorcee/separated. Abuse of alcohol by the pregnant woman, considered as consumption of four or more alcohol units on a single occasion, was categorized as "no alcohol consumption", "no abuse" or "abuse".

The interviewers or field coordinators reviewed the responses of the pregnant women before the data were entered. Inconsistencies were corrected whenever possible. The data were recorded in an Access 2007 (Microsoft Corp.) spreadsheet through double independent entry. Errors and inconsistencies were verified. After final correction, the data were transferred to Stata version 12.0 (StataCorp LP, College Station, U.S.A.), and subsequently to Mplus, version 7.31, to perform the statistical analyses.

Descriptive analysis and structural equation modeling

In the descriptive analysis, physical violence was considered to have occurred when the interviewee answered yes to at least one of the six questions. Frequencies and percentages were calculated in Stata.

In the proposed initial theoretical model (Figure 1), socioeconomic status occupied the most distal position, determining sociodemographic characteristics, social support and women's behavioural factors, which ultimately determined the outcome of physical violence.

Figure 1: Initial hypothesized structural equation model of direct and indirect effects of social support, sociodemographic and behavioral factors on physical violence against pregnant women. São Luís, Maranhão State, Brazil, 2010-2011. 

Structural equation modeling, a multivariate statistical analysis technique, allows the researcher to analyze patterns of correlations between observed variables (or indicators) and latent variables and to test hypotheses in addition to proposing alternative models to the initial one14 was used. The analysis was conducted using Mplus software. Because all variables were categorical, the mean and variance adjusted weighted least squares estimator (WLSMV) were used. Theta parameterization was used to control for residual differences in variances.

To determine whether the models had a good fit, the following indices were considered: (a) a p-value (p) greater than 0.05 for the chi squared test (χ2) 14; (b) a p-value less than 0.05 and an upper limit of the 90% confidence interval (90%CI) lower than 0.08 for the Root Mean Square Error of Approximation (RMSEA); (c) values greater than 0.95 for the Comparative Fit Index and the Tucker Lewis Index (CFI/TLI); and (d) a Weighted Root Mean Square Residual (WRMR) value lower than 115.

To obtain suggestions of modifications to the initial hypotheses, the modindices command was used to calculate the modification indices. When the proposed modifications were considered acceptable from a theoretical viewpoint, a new model was elaborated and analyzed.

Ethical aspects

The research that led to this article was compliant with the requirements of Resolution n. 196/96 of the Brazilian National Health Council and its supplementary regulations and was approved by the Ethics Research Committee of the University Hospital of the UFMA (Opinion n. 4771/2008-30). All women signed an Informed Consent Form. The researchers declare that there were no conflicts of interest.


The results of the descriptive analysis are shown in Table 1. Approximately 12% of the interviewees were less than or equal to 19 years of age, 75% had 9-11 years of study, 3% lived with three or more children, 22% were married and 57% lived in a consensual union. Approximately 88% of the resident intimate partners were 25 years or older, and 73% of them had 9-11 years of study. The percentage of pregnant women who were part of families in economy class C was 67.7%; of those, the proportion of women who survived on less than one minimum wage was 5%. Approximately 75% of the heads of households had manual occupations. The percentage of pregnant women who abused alcohol was 10%. Approximately 3% of the pregnant women had six or more lifetime intimate partners.

Table 1: Characteristics of pregnant women, intimate partners and head of family. São Luís, Maranhão State, Brazil, 2010-2011. 

The prevalence of physical violence against pregnant women was 12.4%, and 66% suffered abuse on a single occasion. Physical intimate partner violence was involved in 66% of these cases.

The initial model (Model 1), shown in Figure 1, did not show a good fit (CFI = 0.935 and TLI = 0.917; WRMR = 1.310). The modification suggestion with the highest modification index (94.710) for model 1 was to add a path from the abuse of alcohol by the pregnant woman to the number of intimate partners in the woman's life. For model 2, which also did not have a good fit (TLI = 0.942; WRMR = 1.144), the highest modification index (96.436) suggested adding a path from the number of intimate partners in the woman's life to the marital status of the pregnant woman. Model 3 showed a good fit, but the suggestion to include a path from abuse of alcohol by the pregnant woman to the marital status of the pregnant woman (modification index of 17.117) was considered plausible. After adding this last modification, the model fit improved, and no further suggested modifications were considered plausible; thus model 4 was chosen as the final model (Table 2).

Table 2: Fit indices of models 1 to 4. São Luís, Maranhão State, Brazil, 2010-2011. 

In the final model, the construct physical violence and the latent variables socioeconomic status and social support had factor loadings exceeding 0.5, with p-values < 0.001 for all of their components (Table 3).

Table 3: Factor loadings, standard errors and p-values of direct and indirect effects for indicator variables and constructs. São Luís, Maranhão State, Brazil, 2010-2011. 

Low social support (standardized coefficient SC = -0.210 and p < 0.001) and a high number of lifetime intimate partners (SC = 0.166 and p = 0.001) had significant total and direct effects on physical violence (Table 3 and Figure 2).

Figure 2: Final structural equation model estimates of direct and indirect effects of social support, sociodemographic and behavioral factors on physical violence against pregnant women. São Luís, Maranhao State, Brazil, 2010-2011. 

Low socioeconomic status had a significant total effect (SC = -0.114 and p = 0.027) on physical violence. Its effect on violence was only indirect (SC = -0.159 and p = 0.048), as it was mediated by low social support (SC = -0.050 and p < 0.001) and living with a partner (estimate = -0.039 and p = 0.008) (Table 3).

Not living with a partner had only an indirect effect on physical violence (SC = 0.057 and p = 0.009); it was mediated by a high number of lifetime intimate partners (SC = 0.051 and p = 0.001) and social support (SC = 0.020 and p = 0.024) (Table 3).


In the São Luís BRISA prenatal cohort, physical violence was more common among pregnant women who were from more disadvantaged backgrounds, did not reside with their intimate partners, had low social support and had a high number of lifetime resident intimate partners. Physical abuse occurred indiscriminately among pregnant women of different age groups. There was also no association between the number of resident children of the pregnant women, alcohol abuse by the pregnant women, or the age and education of the resident intimate partners and physical violence.

The effect of socioeconomic status was only indirect and was completely mediated by low social support and having a companion. In the BRISA prenatal cohort, current and previous intimate partners were the main perpetrators of physical violence. It is possible to explain the effect of socioeconomic status from male domination standards. From this perspective, social and structural contexts contribute to shaping values and norms in the domestic-family environment, including gender relations. Intimate partners from families with low socioeconomic status could experience high levels of stress because of their lack of financial success, which is valued in the patriarchal culture. They could assert their domination by showing strength and power through violence when they felt challenged by women 22), (23. In this context, low social support would be an aggravating factor for women in unfavorable socioeconomic situations 6. Low socioeconomic status together with low social support have been related to poorer health 24.

Low socioeconomic status is frequently associated with violence against pregnant women 4), (5), (6), (11), (22), (25. Researchers nevertheless use one or more variables representing this condition, and the conclusions are mostly based on bivariate analyses, which is a limitation of those studies 4), (11. With regard to intimate partner violence, it has been suggested that it would be more closely related to gender inequalities than to low socioeconomic status 26.

Low social support, represented by the material, emotional/informational, affective and positive social interaction dimensions, was associated with physical violence during pregnancy in São Luís and mediated the indirect effects of socioeconomic status and marital status on physical violence against pregnant women. Low levels of social support may be related to high levels of stress and may thus contribute to increased risk of violence 6. Social support, evaluated by the MOS instrument, has also been associated with violence in a bivariate analysis of a study conducted in the municipality of Rio de Janeiro 11.

Having had six or more lifetime intimate male partners was associated with physical abuse against pregnant women in the BRISA cohort. A similar result was found for psychological violence against pregnant women in the same cohort 27. The effect of the number of intimate male partners on violence was direct and positive, and it was the main mediator of the effect of marital status on physical violence. This is a variable that also reflects gender violence. A possible explanation for this finding could be that when a male partner knows that his wife/companion/girlfriend has had other male partners, he may perceive a lack of control of the woman's body that results in violence 5. A systematic review revealed a higher risk of violence for pregnant women who had had more than five intimate partners in their lives 6. Another review of the literature also concluded that pregnant women with 5 or more lifetime intimate partners tended to suffer from a greater number of episodes of violence 7.

With regard to marital status, it was found in this study that not living with a partner had only an indirect effect on physical violence, as it was mediated by a high number of lifetime intimate partners and social support. Women without resident partners (single, widowed, and separated/divorced women) had more lifetime intimate partners and high social support and were more exposed to physical violence during pregnancy. This situation also seems to indicate gender conflicts that result in violence 4. A study in sixteen states of different countries found a higher frequency of physical violence during pregnancy in unmarried women through a bivariate analysis 8. A literature review showed a greater risk of violence for women who are separated or divorced during the gestational period 4.

The lack of associations between physical violence and the age of the pregnant women, the age of the partner, the partner's education and alcohol abuse by the woman, which was found in other publications, could be due to the sociocultural diversity of the samples or because different types of violence (and not only physical violence) were investigated. Furthermore, data were collected in different periods of the women's life (pregnancy, postpartum or other situation), and the methodological choices regarding sample selection, measurement of violence and methods of adjustment for confounding factors also differed between these studies. Some studies only assessed intimate partner physical violence 4), (5. In a review of African studies, most researchers developed their own questionnaire to measure violence 5, making it difficult to perform comparisons. Most studies used logistic regression to analyse data 4), (5, which is not the most appropriate statistical method to test associations in complex phenomena such as violence 14), (15.

The use of structural equation modeling in this analysis is one of the main strengths of this study. It allowed the construction of latent variables to study violence, social support and socioeconomic status, phenomena that are difficult to measure; improvement of the originally proposed model by adding paths suggested by the modification indices; and assessment of the direct and indirect effects of socioeconomic status, social support and other variables on violence 4), (15. In the literature consulted, no articles were found that had used socioeconomic status, social support and violence as latent constructs or that had used structural equation modeling in its data analyses 4), (5), (6), (7.

A limitation of this study was that it was not representative of the population of pregnant women of the municipality of São Luís, as a convenience sample was used. Additionally, this was a cross-sectional study, thus it was difficult to assess the temporality of the associations, and they were prone to reverse causality. Another limitation is that only data regarding resident partners were collected. Finally, we must consider that the authors investigated physical violence perpetrated by different subjects (intimate partners, other family members or known and unknown persons), not just physical violence by the intimate partner.


Knowledge of the factors associated with physical maltreatment of pregnant women enables prevention and assistance to women in situations of violence during the prenatal period. Healthcare professionals may thus provide more effective guidance to pregnant women who have a greater risk of suffering violent episodes during pregnancy.

Structural equations modeling allowed for more explicit knowledge of how the factors included were related to each other by demonstrating the direct and indirect influences of the factors on physical violence against pregnant women, as well as their intercorrelations. Low social support and a high number of lifetime intimate partners had effects on physical violence during pregnancy. Low socioeconomic status was associated with physical violence against pregnant women; this effect was only indirect and was wholly mediated by low social support and living with a partner. Single mothers, widows and women who were separated/divorced were more battered when they had more lifetime intimate partners and high social support. The effect of not living with a companion on violence was only indirect and was mediated by high social support and having more lifetime intimate partners.

In conclusion, physical violence was a common phenomenon during pregnancy among women interviewed in the BRISA study, and low social support, low socioeconomic status, a high number of lifetime intimate partners and living without a companion increased the risk of physical violence, which was mostly manifested as gender violence.


Pregnant women from the BRISA cohort. Maranhão State Research Foundation (FAPEMA), Brazilian National Research Council (CNPq), Brazilian Ministry of Health and São Paulo State Research Foundation (FAPESP).


1. World Health Organization. WHO Multi-Country Study of Women's Health and Domestic Violence against Women: summary report of initial results on prevalence, health outcomes and women's responses. Geneva: World Health Organization; 2005. [ Links ]

2. Asamblea General, Naciones Unidas. Intensificación de los esfuerzos para eliminar todas las formas de violencia contra la mujer. New York: Naciones Unidas; 2010. [ Links ]

3. Secretaria de Políticas para as Mulheres, Presidência da República. Política Nacional de Enfrentamento à Violência contra as Mulheres. Brasília: Secretaria de Políticas para as Mulheres, Presidência da República; 2011. [ Links ]

4. Taillieu TL, Brownridge DA. Violence against pregnant women: prevalence, patterns, risk factors, theories, and directions for future research. Aggress Violent Behav 2010; 15:14-35. [ Links ]

5. Shamu S, Abrahams N, Temmerman M, Musekiwa A, Zarowsky C. A systematic review of African studies on intimate partner violence against pregnant women: prevalence and risk factors. PLoS One 2011; 6:e17591. [ Links ]

6. Jasinski JL. Pregnancy and domestic violence: a review of the literature. Trauma Violence Abuse 2004; 5:47-64. [ Links ]

7. Campbell J, García-Moreno C, Sharps P. Abuse during pregnancy in industrialized and developing countries. Violence Against Women 2004; 10:770-89. [ Links ]

8. Saltzman LE, Johnson CH, Gilbert BC, Goodwin MM. Physical abuse around the time of pregnancy: an examination of prevalence and risk factors in 16 States. Maternal Child Health J 2003; 7:31-43. [ Links ]

9. Yang MS, Yang MJ, Chou FH, Yang HM, Wei SL, Lin JR. Physical abuse against pregnant aborigines in Taiwan: prevalence and risk factors. Int J Nurs Stud 2006; 43:21-7. [ Links ]

10. Durand JG, Schraiber LB. Violência na gestação entre usuárias de serviços públicos de saúde da Grande São Paulo: prevalência e fatores associados. Rev Bras Epidemiol 2007; 10:310-22. [ Links ]

11. Moraes CL, Reichenheim ME. Domestic violence during pregnancy in Rio de Janeiro, Brazil. Int J Gynaecol Obstet 2002; 79:269-77. [ Links ]

12. Audi CAF, Segall-Corrêa AM, Santiago SM, Andrade MGG, Pérez-Escamilla R. Violência doméstica na gravidez: prevalência e fatores associados. Rev Saúde Pública 2008; 42:877-85. [ Links ]

13. Viellas EF, Gama SG, Carvalho ML, Pinto LW. Factors associated with physical aggression in pregnant women and adverse outcomes for the newborn. J Pediatr (Rio J.) 2013; 89:83-90. [ Links ]

14. Kline RB. Principles and practice of structural equation modeling. New York: The Guilford Press; 2011. [ Links ]

15. Wang J, Wang X. Structural equation modeling: applications using Mplus. Noida: Thomson Digital; 2012. [ Links ]

16. Ellsberg M, Heise L. Researching violence against women: a practical guide for researchers and activists. Washington DC: World Health Organization; 2005. [ Links ]

17. Schraiber LB, Latorre MRDO, França Júnior I, Lucas AFP. Validade do instrumento WHO VAW para estimar violência de gênero contra a mulher. Rev Saúde Pública 2010; 44:658-66. [ Links ]

18. Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med 1991; 38:705-14. [ Links ]

19. Griep RH, Chor D, Faerstein E, Werneck GL, Lopes CS. Validade de constructo de escala de apoio social do Medical Outcomes Study adaptada para o português no Estudo Pró-Saúde. Cad Saúde Pública 2005; 21:703-14. [ Links ]

20. Ribeiro MRC, Alves MTSSB, Batista, RFL, Ribeiro CCC, Schraiber LB, Barbieri MA, et al. Confirmatory factor analysis of the WHO Violence Against Women instrument in pregnant women: results from the BRISA prenatal cohort. PLoS One 2014; 9:e115382. [ Links ]

21. Associação Brasileira de Empresas de Pesquisas. Critério de classificação econômica Brasil. São Paulo: Associação Brasileira de Empresas de Pesquisas; 2010. [ Links ]

22. Schraiber LB, d'Oliveira AF, Couto MT. Violence and health: theoretical, methodological, and ethical contributions from studies on violence against women. Cad Saúde Pública 2009; 25 Suppl 2:S205-16. [ Links ]

23. Renzetti CM, Larkin VM. Economic stress and domestic violence. Harrisburg: National Resource Center on Domestic Violence; 2011. [ Links ]

24. National Institutes of Health. Health and behavior: the interplay of biological, behavioral, and societal influences. Washington DC: National Academies Press; 2001. [ Links ]

25. Benson ML, Fox GL. When violence hits home: how economics and neighborhood play a role. Washington DC: Department of Justice, National Institute of Justice; 2004. [ Links ]

26. d'Oliveira AF, Schraiber LB, França-Junior I, Ludemir AB, Portella AP, Diniz CS, et al. Fatores associados à violência por parceiro íntimo. Rev Saúde Pública 2009; 43:299-310. [ Links ]

27. Ribeiro MRC, Silva AAM, Alves MTSSB, Batista RFL, Rocha LMLN, Schraiber LB, et al. Psychological violence against pregnant women in a prenatal care cohort: rates and associated factors in São Luís, Brazil. BMC Pregnancy Childbirth 2014;14:66. [ Links ]

Received: May 15, 2015; Revised: March 11, 2016; Accepted: March 22, 2016

* Correspondence D. C. S. Costa Programa de Pós-graduação em Saúde Coletiva, Universidade Federal do Maranhão. Rua Barão de Itapary 155, São Luís, MA 65020-070, Brasil.

Contributors: D. C. S. Costa, M. R. C. Ribeiro, R. F. L. Batista, C. M. Valente, J. V. F. Ribeiro, L. A. Almeida, L. E. M. Costa, M. T. S. S. B. Alves and A. A. M. Silva participated in the conception and design, acquisition of data, analysis and interpretation of data, drafting and revising critically the article, final approval of the version to be published and are responsible for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License