Violence against women increases the risk of infant and child mortality: a case-referent study in Nicaragua.

Violence against women increases the risk of infant and child mortality: a case–referent study in Nicaragua Kajsa A˚sling-Monemi,1 Rodolfo Pen˜a,2 Mary Carroll Ellsberg,3 & Lars A˚ke Persson4 Objective To investigate the impact of violence against mothers on mortality risks for their offspring before 5 years of age in Nicaragua. Methods From a demographic database covering a random sample of urban and rural households in Leo´n, Nicaragua, we identified all live births among women aged 15–49 years. Cases were defined as those who had died before the age of 5 years, between January 1993 and June 1996. For each case, two referents, matched for sex and age at death, were selected from the database. A total of 110 mothers of the cases and 203 mothers of the referents were interviewed using a standard questionnaire covering mothers’ experience of physical and sexual violence. The data were analysed for the risk associated with maternal experience of violence of infant and under-5 mortality. Findings A total of 61% of mothers of cases had a lifetime experience of physical and/or sexual violence compared with 37% of mothers of referents, with a significant association being found between such experiences and mortality among their offspring. Other factors associated with higher infant and under-5 mortality were mother’s education (no formal education), age (older), and parity (multiparity). Conclusions The results suggest an association between physical and sexual violence against mothers, either before or during pregnancy, and an increased risk of under-5 mortality of their offspring. The type and severity of violence was probably more relevant to the risk than the timing, and violence may impact child health through maternal stress or care-giving behaviours rather than through direct trauma itself. Keywords Domestic violence; Infant mortality; Maternal welfare; Sex offenses; Sexual partners; Pregnancy complications; Cause of death; Risk factors; Socioeconomic factors; Odds ratio; Case-control studies; Nicaragua (source: MeSH, NLM ). Mots cle´s Violence familiale; Mortalite´ nourrisson; Protection maternelle; Abus sexuel; Partenaire sexuel; Grossesse complique´e; Cause de´ce`s; Facteur risque; Facteur socio-e´conomique; Odds ratio; Etude cas-te´moins; Nicaragua (source: MeSH, INSERM). Palabras clave Violencia dome´stica; Mortalidad infantil; Bienestar materno; Delitos sexuales; Parejas sexuales; Complicaciones del embarazo; Causa de muerte; Factores de riesgo; Factores socioecono´micos; Razo´n de diferencia; Estudios de casos y controles; Nicaragua (fuente: DeCS, BIREME ). Bulletin of the World Health Organization 2003;81:10-18. Voir page 15 le re´sume´ en franc¸ais. En la pa´gina 15 figura un resumen en espan˜ol. Introduction Violence against women has serious consequences for their physical (1, 2) as well as mental health (3–5). Physical violence against women is a major public health problem in many settings, with a lifetime prevalence varying from 20% to 50% (6–10). During pregnancy 1–20 % of women are exposed to violence (11), and there are indications that the severity of violence may increase during pregnancy (12). Unemployment, strained economic resources, a history of family violence, and alcohol abuse have been reported to increase the occurrence of physical violence against women (13, 14). A few studies, mostly in high-income countries, have suggested that physical violence against pregnant women increases the risk of preterm labour (15) or delivery (16), fetal distress or death (16–18), and low-birth-weight off- spring (19–23). So far, little is known about the possible effect of violence against women on the survival of their offspring. However, low birth weight is an important risk factor for increased infant mortality (24, 25), and an abused and chronically stressed mother may experience difficulties in coping with the multiple needs of her small child (26). A recent population-based study in Leo´n, Nicaragua, indicated that 40% of women of reproductive age (n= 488) had been exposed to physical violence by a partner (27). Among ever-married women (n = 360), the lifetime prevalence of physical violence by a current or former intimate partner was 52%, and 27% of women reported having been exposed to violence in the 12 months prior to being interviewed. 1 Paediatrician, Division of Epidemiology, Department of Public Health and Clinical Medicine, Umea˚ University, SE-901 85 Umea˚, Sweden (email: kajsa.asling@epiph. umu.se); and Department of Pediatrics, Umea˚ University, Umea˚, Sweden. Correspondence should be sent to this author at the former address. 2 Epidemiologist, Department of Preventive Medicine, Universidad Nacional Auto´noma, Leo´n, Nicaragua; and Division of Epidemiology, Department of Public Health and Clinical Medicine, Umea˚ University, Umea˚, Sweden. 3 Senior Program Officer, Program for Appropriate Technology in Health (PATH), Washington, DC, USA; and Division of Epidemiology, Department of Public Health and Clinical Medicine, Umea˚ University, Umea˚, Sweden. 4 Professor, Division of Epidemiology, Department of Public Health and Clinical Medicine, Umea˚ University, Umea˚, Sweden; and International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR), Center for Health and Population Research, Dhaka, Bangladesh. Ref. No. 01-1017 Research 10 Bulletin of the World Health Organization 2003, 81 (1) Furthermore, 70% of cases of violence were classified as severe. Violence was associated with poverty, high parity, and a history of marital violence in the partner’s family (27). A total of 31% of women exposed to violence were beaten during one or more pregnancies, and 33% reported that beatings were commonly accompanied by forced sex (28). Physical violence from partners also increased the risk of the woman suffering from emotional distress (29), and the children of mothers who had experienced violence were more than twice as likely to suffer from learning, emotional, or behavioural problems compared with children whose mothers had never been so exposed (28). Using the same population-based sampling frame that was employed in this population-based study in Nicaragua (27), we report here the results of a case–referent study on mortality among under-5-year-olds. The aim was to assess the effect of physical and sexual violence against mothers on the mortality risks of children in this age group. Methods A case–referent study was nested into a demographic database consisting of 9500 households, covering 50 out of the 208 geographical clusters in urban and rural areas of the municipality of Leo´n, Nicaragua. The database was established in 1993 by Leo´n University andUmea˚ University bymeans of a population survey performed on a random sample of house- holds, representing nearly 25% of the population of Leo´n. All women aged 15–49 years in the sample were interviewed and detailed information regarding their migration history, birth history, deaths of children, education, employment, and housing conditions was obtained (30). In mid-1996, all households were revisited and information on all the women of reproductive age was updated, including answers to specific questions identifying all births and any deaths of children aged <5 years. Cases were defined as children born alive to women in the database described above and who died before the age of 5 years, between January 1993 and June 1996. For each case, two referents (alive), matched for sex and age at death, were randomly selected from the database. Initially, 156 children, identified as potential cases, were matched with 312 referents. Themothers of all cases and selected referents were visited and invited to participate in the study. Upon completion of the interviews, it was ascertained that 24 of the cases had actually been stillbirths, and therefore did not meet the inclusion criteria. An additional 15 cases had migrated out of the study area, and three mothers of cases were unable to complete the interview because they were mentally retarded. A further four mothers of cases refused to participate in the complete interview. As a result, 46 of the initial cases and each of their two corresponding referents (92) were excluded from the study. It was not possible to trace 16 mothers of the referents, and onemother of a referent refused to be interviewed (refusal rate among mothers of cases and referents altogether was less than 2%). A total of 110 mothers of cases and 203 mothers of referents were interviewed for the study, resulting in 93 complete triplets (one case and two referents) and 17 pairs with only one referent. Interviews Four trained female Nicaraguan field workers interviewed the mothers in privacy, using a standardized, pretested ques- tionnaire. Information on the deaths of under-5-year-old children were ascertained by means of a verbal autopsy, including detailed standardized questions previously used in low-income settings (31). Mothers were also asked to provide a narrative account of the circumstances leading to their children’s deaths. A diagnosis of the principal cause of death was extracted from this information by two paediatricians through a consensus process. Dates of births and deaths were carefully registered, using a local events calendar. Physical and sexual violence against mothers was assessed through two groups of questions. The first group dealt with lifetime experiences of physical and sexual violence by any person, including sexual violence in childhood. Women experiencing violence were further questioned about the perpetrator, frequency of violent incidents, and how much they felt that the violence had affected their emotional well- being. The second group of questions was based on the abuse assessment screen (AAS), a five-question instrument that has been used successfully to screen for violence in pregnancy (32), but our instrument differed from the AAS in some important aspects, as follows. Only data on physical and sexual violence by a former or current intimate partner were included, and separate questions were used to determine the severity and temporal sequence of violence. Moderate violence was defined as slaps, pushes and shoves, whereas punches, kicks, bites or blows with objects were classified as severe violence. Forced sexual acts were considered as sexual violence and were classified as severe. Lifetime experiences of partner violence, and violence during the index pregnancy as well as in the year before the child’s death (for referents, 12 months prior to the interview) were also assessed. Mothers who experienced any type of violence were also asked to state, using a four-step scale from none to very much, the degree to which they felt that it had affected their emotional well-being. Information was collected about mother’s age, parity, educational level, occupation, and social network. Socioeconomic status was estimated using the unsatisfied basic needs assessment method, which measures household access to a series of basic services, such as sanitation, housing conditions, and educa- tional level. This method has been adapted and used for socioeconomic research in Nicaragua (33, 34). Low socio- economic status was defined as one or more unsatisfied basic needs. Women were also questioned about their breastfeeding habits, use of health services (antenatal care and delivery place), smoking, and alcohol use. Questions addressing child abuse or alcohol use by partners were not included. Data analysis All completed interview forms were reviewed by a field supervisor and inspected by one of the principal researchers. Forms with missing data or inconsistencies were returned to the interviewers for correction.Data were entered and checked by trained personnel under continuous supervision by a principal researcher. Odds ratios for infant and under- 5 mortality were calculated using matched analysis. Condi- tional logistic regression analyses were performed by use of EGRET software version 2.0 (Statistical and Epidemiological Research Corporation, Seattle, WA, USA). A model was developed to evaluate if physical and sexual violence from a current or former partner against the mother was indepen- dently associated with increased risk of death of an offspring during the first 5 years of age, adjusting for potential 11Bulletin of the World Health Organization 2003, 81 (1) Violence against women: a case–referent study in Nicaragua confounding factors including mother’s age, parity, educa- tional attainment, place of residency, and basic needs assessment. The proportion of child deaths attributable to violence was estimated from the frequency of physical and sexual violence among mothers of cases and the odds ratios obtained in the multivariate model. Population-attributable risk was calculated using the following expression: ((proportion exposed among all mothers of cases) 6 (odds ratio – 1))/odds ratio. Ethics Data were handled with strict confidentiality. Ethical review and clearance was obtained from the Medical Faculty, University of Leo´n, Nicaragua, and the Research Ethics Committee of the Medical Faculty, Umea˚ University, Sweden. Informed consent was obtained at the community level through meetings with local health organizations, community representatives, and by the participating women. All the women and children who were included in the study were offered free medical or mental health services at the local hospital and psychiatric outpatient clinic. Results Causes of deaths A total of 92 (84%) of the 110 deaths identified occurred during the first year of life (Table 1). The commonest causes of death among neonates were complications arising from preterm delivery and low birth weight, while for the older age groups the major causes were infectious diseases, mainly diarrhoea. Five mothers (4%) reported that their children had experienced some kind of trauma, and subsequent evaluations of the circumstances surrounding those deaths indicated that insufficient care or neglect were more likely causes than child abuse. Patterns of violence A total of 61% of the mothers of children who had died (cases) had ever experienced any physical or sexual violence by any person, compared with 37% of referent mothers (Table 2, available only on the online version at: URL: http:// www.who.int/bulletin/). Sexual violence had been experi- enced by 26% of the mothers of the cases and 10% of referent mothers. The vast majority of all violence was from a current or former intimate male partner (51% of cases, 33% of referents). Aside from the male partner, there was a wide range of other offenders (non-partner violence), including fathers (7% of mothers of cases and 3% of referents), mothers (5% of mothers of cases and 3% of referents), and other family members, as well as friends or strangers. A total of 17% of mothers of cases, compared with 6% of mothers of referents, had been abused by partners as well as non- partners. Among the women reporting physical violence, 90% classified the violence as severe. Only five mothers had experienced sexual partner violence but no physical partner violence, whereas 20% of mothers of cases and 6% of those of referents had experienced both physical and sexual violence. Violence during the index pregnancy was reported by 21% of the mothers of cases compared with 12% of those of the referents. One mother had adopted her child (a case) and could therefore not give any information regarding the pregnancy. Furthermore, 29% of the mothers of cases and 18% of those of the referents had been exposed to violence during the 12 months preceding a child death (cases) or the interview (referents). Only one woman (the mother of a case) reported violence during pregnancy but not during the previous 12 months. Almost all mothers (92%) who had ever experienced any kind of violence reported that it had greatly affected their emotional well-being, and all who had experienced a combination of physical and sexual violence were deeply affected by it. All mothers reporting both sexual and physical violence reported that the latter was severe. A significant association was found between lifetime experiences of physical and sexual violence towards mothers and mortality among their children (see Table 2, available online). In addition, mother’s educational level (no formal education), age (older), parity (multiparity) and area of residence (rural) were associated with higher infant and under-5 mortality (Table 3). Low socioeconomic status was strongly associated with mothers who had no formal education and those living in rural areas. However, no association was found between socioeconomic status and infant or under-5 mortality. Smoking and alcohol consump- tion were rare (3% and 4%, respectively), and did not differ significantly between mothers of cases and mothers of referents. Lifetime exposure to violence was more often reported by multiparous mothers and by those of low educational level. No significant association was found between violence and age of mother, employment status of mother or father, place of delivery, breastfeeding practices, basic needs assessment level, or area of residency. Mortality risks The risk of death in infancy or before 5 years of age was more than six times greater if the mother had been exposed to both physical and sexual violence by a current or former partner at any point in her life, even after adjusting for educational, parity, area of residency, and basic needs assessment level (Table 4). No significant interactions were found between violence and mothers’ educational attainment, violence and parity, violence and area of residency, or between violence and basic needs assessment in relation to mortality. Mother’s age was not Table 1. Causes and symptoms prior to death of 110 children who died before the age of 5 years in Leo´n, Nicaragua, 1993–6 % of deaths at age: Causes/symptoms 0–29 days 0–11 months 0–59 months (n = 58) (n = 92) (n = 110) Diarrhoea 2 22 25 Prematurity, low birth weight 43 27 23 Respiratory infections 5 10 12 Fever of unknown origin 14 14 14 Malformations 17 12 10 Asphyxia 14 10 8 Trauma 2 2 4 Unknown 3 3 4 Total 100 100 100 12 Bulletin of the World Health Organization 2003, 81 (1) Research included in the final multivariate model due to collinearity between parity and age. Given a causal link between the demonstrated association of violence andmortality, asmuch as one-fourth (27%) of the under-5 deaths could be attributed to physical or sexual violence by a partner (frequency among cases, 51%; odds ratio, 2.1). Discussion The central finding of the present study was the increased risk of infant and under-5 mortality that was found to be associated with partner violence. Any history of violence was associated with a twofold increase in risk, and children of women who experienced both sexual and physical violence had a sixfold greater risk of death. This association has, to our knowledge, not previously been reported. Nevertheless, recent findings in a survey in rural India have suggested a relation between wife beating and infant death (35). Selection bias It is unlikely that our findings can be attributed to selection bias. Cases were recruited from a representative sample of the community by means of a demographic database that involved home visits to all households. Great efforts were made to identify all deaths in the study population during the reference period. Mothers of cases who did not participate in the study (mostly due to recent migration out of the region) and non- participatingmothers of referents had the same socioeconomic characteristics as those who did participate, thereby minimiz- ing the possibility of selection bias. Furthermore, there were no differences in the socioeconomic characteristics of the mothers of referents and those of women in the overall study population, indicating that the referent sample was represen- tative with respect to those characteristics. Underreporting versus overreporting In general, underreporting of violence ismuch commoner than overreporting, largely because of the stigma attached to victimization as well as to the fear of reprisals (2, 4, 6, 36). Althoughmothers who have experienced the trauma of a child death may be more likely to report violence, it seems unlikely that selective overreporting would account for the strength of the associations that we found. Selective underreporting by the mothers of referents is theoretically possible, but does not seem likely, since the level is similar to that previously reported from the study area (37). Recall bias amongmothers of cases is also possible, since they were asked to remember events taking place during the 12 months before the child’s death, which could have occurred anytime from January 1993 until the study was performed in January to June 1996. In contrast, the mothers of referents were asked about the 12 months prior to the interview. However, any recall bias would most probably lead to an underestimation of abuse among the mothers of cases due to the longer recall period. Table 3. Social and demographic characteristics of mothers of 110 cases and 203 referents as risk factors of infant and under-5 mortality in Leo´n, Nicaragua, 1993–6 Risk factor Odds ratio for deaths at:a % of % of 0–11 months 0–59 months casesb referentsc (n = 110) (n = 203) Educational level Formal educationd 67 86 1.0 1.0 No formal educatione 33 14 2.3 (1.3–4.3)f 3.0 (1.7–5.3) Parityg 1 birth 28 36 1.0 1.0 2–5 51 56 1.1 (0.7–2.0) 1.1 (0.7–1.8) >5 21 8 3.9 (1.7–9.4) 3.0 (1.4–6.4) Age (years) 15–19 18 21 1.0 1.0 20–34 60 72 1.0 (0.5–2.0) 0.9 (0.5–1.8) 35–49 22 7 3.0 (1.3–7.3) 3.4 (1.4–8.2) Area of residency Urban 57 80 1.0 1.0 Rural 43 20 3.1 (1.7–5.8) 3.3 (1.9–5.8) Basic needs assessment Satisfied 26 28 1.0 1.0 Not satisfied 74 72 1.1 (0.6–2.0) 1.2 (0.7–2.1) a Bivariate odds ratio based on matched analysis. b For basic needs, n =106 for cases. c For basic needs, n = 200 for referents. d 3 years of schooling or more, completed primary school. e 2 years of schooling or less, not completed primary school. f Figures in parentheses are 95% confidence levels. g No. of live births. Table 4. Effect of physical and sexual violence against mothers on infant and under-5 mortality, in Leo´n, Nicaragua, 1993–6a Risk factor Odds ratio for deaths at: 0–11 months 0–59 months Experience of violence No physical or sexual violence by anyone 1.0 1.0 Physical or sexual violence by a non-partner only 3.1 (0.9–11.0)b 4.1 (1.2–13.7) Physical or sexual partner violence 2.2 (1.0–4.5) 2.1 (1.1–4.1) Both physical and sexual partner violence 7.8 (2.5–24.8) 6.3 (2.3–17.1) Education Formal educationc 1.0 1.0 No formal educationd 1.4 (0.6–3.0) 2.0 (1.0–3.9) Paritye 1 birth 1.0 1.0 2–5 births 0.8 (0.41.6) 0.9 (0.5–1.6) >5 births 2.1 (0.7–6.2) 2.0 (0.8–4.8) Area of residency Urban 1.0 1.0 Rural 3.5 (1.7–7.2) 3.4 (1.8–6.7) Basic needs assessment Satisfied 1.0 1.0 Not satisfied 0.9 (0.4–1.8) 0.9 (0.5–1.8) a Based on a multivariate conditional (age and sex matched) logistic regression analyses for infant mortality outcome (92 cases, 169 referents) and under-5 mortality (110 cases, 203 referents). b Figures in parentheses are 95% confidence levels. c 3 years of schooling or more, completed primary school. d 2 years of schooling or less, not completed primary school. e No. of live births. 13Bulletin of the World Health Organization 2003, 81 (1) Violence against women: a case–referent study in Nicaragua Basic needs assessment We did not find any significant association between the basic needs assessment level (poverty) and infant or under-5 mortality level, although there was an association between rural mothers and those with low educational attainment and mortality among their offspring. It is possible, however, that the instrument used to estimate unsatisfied basic needs was not sensitive enough to pick up poverty differences. Violence against mothers and risk of mortality among their offspring There are several explanations for the association between physical and sexual violence against mothers and the increased risk of infant and under-5 mortality. First, violence during pregnancy increases the risk of low-birth-weight infants, a well- known risk factor for increased infant mortality (24, 25). Low birth weight may be a direct consequence of violence, for example, in the case of preterm delivery provoked by direct abdominal trauma (17, 38, 39). However, violence may also affect birth weight indirectly, through changes to physiology (increased levels of stress hormones (40–42) and in immuno- logical factors (40, 41)) and behavioural mechanisms (43, 44). This is supported by results from a recent hospital-based case– referent study in Leo´n, Nicaragua, which found that, after adjusting for other known risk factors of low birth weight, partner violence against pregnant women increased the risk of low birth weight by a factor of three (45). Maternal stress due to violence may increase women’s likelihood of engaging in negative health or coping behaviours, such as smoking and substance abuse (46). However, as only 3% of the mothers reported smoking during pregnancy, it is unlikely that smoking is a factor affecting birth weight in our study. Violence may also act as a stressor in itself, affecting women’s ability to obtain adequate nutrition, rest, exercise and medical care. Several studies have indicated that women experiencing violence during pregnancy aremore likely to enter antenatal care late in the pregnancy, and to report having unintended pregnancies (47, 48). Second, violence may impact child health by diminishing women’s access tomaterial as well as internal resources necessary for safeguarding their children’s health. Women experiencing physical or sexual violence are likely to suffer from a variety of mental health disorders, including depression, anxiety, and post- traumatic stress syndrome (3, 4). In addition, physical violence is often accompanied by feelings of powerlessness, social isolation, and economic dependency (43, 49, 50). In Nicaragua violence against women has been reported to be closely associated with controlling behaviour on the part of partners (28). Violence may interfere with the caring capacity of mothers through emotional distress or because they are physically prevented from obtaining care for their children. A study conducted in India found that the children of battered womenweremore likely to bemalnourished and to receive less food than those of women who were not beaten (51). Third, the child deaths may have been due to direct trauma. Ellsberg et al. found that the children of mothers who had experienced violence were almost seven times more likely to be physically and sexually abused themselves (28). In our study only five (4%) out of 110 deaths were explained by trauma, and subsequent evaluations of the circumstances surrounding those deaths indicated that insufficient care or neglect were more likely causes than direct trauma. However, we cannot rule out an underreporting of trauma as a cause of death since no questions addressing child abuse were included in order not to blame or distress the respondent. Our findings indicate that the type and severity of violence were more relevant to the risk of child death than the timing of the abuse or the relationship between themother and the perpetrator. The highest risk of child death was found among mothers who were victimized by both partners and non-partners, as well as among women who had experienced both physical and sexual partner violence at any time, even compared with women who had experienced severe physical partner violence during the index pregnancy or previous 12 months. This finding lends support to the view that violence impacts child health through maternal stress or care- giving behaviours rather than direct trauma. Previous research in Nicaragua indicates that sexual coercion by partners is generally associated with greater severity of physical as well as emotional violence (28). Therefore it is not known whether the increased risk of child death when sexual and physical violence are combined is due to the specific effect of sexual assault, or whether this represents a more severe level of violence overall. Our findings underscore the extent to which the traumatic effects of violence may persist long after the violence itself has ended. There are no reasons to assume that the reported association between violence against mothers and increased risks of child mortality are unique for the study area. In any setting, the biological consequences of violence during pregnancy could have a negative impact on pregnancy outcome, although that might be compensated for through better economic resources andwidespread health care services. Our findings indicate that violence against women represents an important public health concern not only for women’s health but also for children’s survival. This under- scores the need for further research to confirm our results and to understand the mechanisms whereby physical and sexual assault increases the risk of child mortality. n Acknowledgements This study was jointly supported by the Swedish Agency for Research Co-operation with Developing Countries (SAREC) and by the National Autonomous University of Leo´n, Nicaragua. We would like to thank the four fieldworkers who performed the interviews and all the participating mothers. We are also grateful to Jacqueline Campbell and Lori Heise for their valuable comments on earlier drafts of the manuscript. Conflicts of interest: none declared. 14 Bulletin of the World Health Organization 2003, 81 (1) Research Re´sume´ La violence a` l’encontre des femmes augmente le risque de mortalite´ infantile : une e´tude cas-te´moin au Nicaragua Objectif Etudier les effets de la violence a` l’encontre des me`res sur les risques de mortalite´ que pre´sentent les enfants de moins de 5 ans au Nicaragua. Me´thodes A partir d’une base de donne´es de´mographiques couvrant un e´chantillon ale´atoire de foyers urbains et ruraux de la re´gion de Leon, au Nicaragua, nous avons recense´ toutes les naissances vivantes chez les femmes aˆge´es de 15 a` 49 ans. Les cas ont e´te´ de´finis comme e´tant les enfants de´ce´de´s avant l’aˆge de 5 ans, entre janvier 1993 et juin 1996. Pour chaque cas, deux te´moins, apparie´s pour le sexe et l’aˆge au moment du de´ce`s, ont e´te´ choisis dans la base de donne´es. Au total, on a interroge´ 110 me`res de cas et 203 me`res de te´moins au moyen d’un questionnaire standard portant sur l’expe´rience qu’elles avaient de la violence physique et sexuelle. Les donne´es ont e´te´ analyse´es afin de de´terminer chez les nourrissons et les enfants de moins de 5 ans le risque de mortalite´ associe´ a` l’expe´rience maternelle de la violence. Re´sultats Au total, 61 % des me`res de cas avaient eu l’expe´rience de la violence physique et/ou sexuelle au cours de leur vie contre 37 % des me`res de te´moins, et l’on a trouve´ une association significative entre ces expe´riences et la mortalite´ observe´e chez leurs enfants. Les autres facteurs associe´s a` une mortalite´ plus e´leve´e chez les nourrissons et les moins de 5 ans e´taient le niveau d’instruction (pas d’instruction), l’aˆge (avance´) et la parite´ (multiparite´) de la me`re. Conclusion Ces re´sultats laissent a` penser qu’il existe une association entre la violence physique et sexuelle a` l’encontre des me`res, avant ou pendant la grossesse, et un risque accru de mortalite´ avant 5 ans chez leurs enfants. Ce risque est probablement davantage lie´ au type et au degre´ de la violence qu’au moment ou` elle s’exerce, et celle-ci peut influer sur la sante´ des enfants du fait du stress maternel ou des comportements qu’elle engendre au niveau des soins aux enfants, plus que du fait du traumatisme direct. Resumen La violencia contra las mujeres aumenta el riesgo de defuncio´n infantil: estudio de casos y testigos en Nicaragua Objetivo Investigar las repercusiones de la violencia contra las madres en la mortalidad de sus hijos hasta los 5 an˜os de edad en Nicaragua. Me´todos A partir de una base de datos demogra´ficos que abarcaba una muestra aleatoria de hogares urbanos y rurales en Leon (Nicaragua), identificamos a todos los nacidos vivos de mujeres de 15 a 49 an˜os. Se consideraron casos los nin˜os fallecidos antes de alcanzar los 5 an˜os entre enero de 1993 y junio de 1996, y para cada caso se seleccionaron a partir de la base de datos dos testigos, emparejados por sexo y edad en el momento de la muerte. Se entrevisto´ en total a 110 madres de casos y 203 madres de testigos, utilizando un cuestionario esta´ndar en el que se les preguntaba a las madres si habı´an sufrido violencia fı´sica y sexual. Se analizaron los datos para determinar el riesgo asociado a la experiencia materna de violencia en el caso de la mortalidad de lactantes y menores de 5 an˜os. Resultados Un 61% de las madres de casos habı´an sufrido a lo largo de su vida violencia fı´sica y/o sexual, en comparacio´n con el 37% de las madres de testigos, y la relacio´n entre esa experiencia y la mortalidad de su descendencia era significativa. Otros factores asociados a una mayor mortalidad de lactantes y menores de 5 an˜os fueron la educacio´n de la madre (carencia de estudios escolares), la edad (mayor) y el nu´mero de partos (multiparidad). Conclusio´n Los resultados indican que la violencia fı´sica y sexual contra las madres, antes del embarazo o durante el mismo, se asocia a un mayor riesgo de defuncio´n de sus hijos menores de 5 an˜os. Probablemente la naturaleza y la gravedad de los actos violentos contribuyeron al riesgo en mayor medida que el momento en que tuvieron lugar, y tales actos podrı´an repercutir en la salud del nin˜o no tanto de forma directa como a trave´s del estre´s materno o de cambios en el comportamiento de cuidado de los nin˜os. References 1. Abbot J, Johnson R, Koziol-McLain J, Lowenstein SR. Domestic violence against women: incidence and prevalence in an emergency department population. JAMA 1995;273:1763-7. 2. Grisso JA, Schwarz DF, Miles CG, Holmes JH. Injuries among inner-city minority women: a population-based longitudinal study. American Journal of Public Health 1996;86:67-70. 3. Koss MP. The women’s mental health research agenda: violence against women. American Psychologist 1990;45:374-80. 4. Walker L. Abused women and survivor therapy: a practical guide for the psychotherapist. Washington (DC):American Psychological Association;1996. 5. Stark E, Flitcraft A. Women at risk: domestic violence and women’s health. Thousand Oaks (CA): Sage Publications;1996. 6. Heise LL, Raikes A, Watts CH, Zwi AB. Violence against women: a neglected public health issue in less developed countries. Social Science and Medicine 1994;39:1165-79. 7. Johnson H. Dangerous domains: violence against women in Canada. Toronto: International Thomson Publishing Co.;1996. 8. Violence against women. Geneva; World Health Organization. Fact sheets. No. 239, June 2000. Available from:URL: http://www.who.int/inf-fs/en/fact.239.html. 9. Heise L, Ellsberg M, Gottemoeller M. Ending violence against women. Baltimore (MD):Johns Hopkins University School of Public Health, Population Information Program, Population Reports Series L. No. 11; December 1999. 10. Watts C, Zimmerman C. Violence against women: global scope and magnitude. Lancet 2002;359:1232-7. 11. Gazmararian JA, Lazorick S, Spitz AM, Ballard TJ, Saltzman LE, Marks JS. Prevalence of violence against pregnant women. JAMA 1996;275:1915-20. 12. Campbell JC. Abuse during pregnancy: progress, policy and potential. American Journal of Public Health 1998;88:185-7. 13. Hotaling GT, Sugarman DB. An analysis of risk markers in husband to wife violence: the current state of knowledge. Violence and Victims 1986;1:101-25. 14. Jewkes R. Intimate partner violence: causes and prevention. Lancet 2002; 359:1423-9. 15Bulletin of the World Health Organization 2003, 81 (1) Violence against women: a case–referent study in Nicaragua 15. Berenson AB, Wiemann CM, Wilkinson GS, Jones WA, Anderson GD. Perinatal morbidity associated with violence experienced by pregnant women. American Journal of Obstetrics and Gynecology 1994;170:1760-9. 16. Connolly AM, Katz VL, Bash KL, McMahon MJ, Hansen WF. Trauma and pregnancy. American Journal of Perinatology 1997;14:331-6. 17. Pearlman MD, Tintinalli JE, Lorenz RP. Blunt trauma during pregnancy. New England Journal of Medicine 1990;323:1609-13. 18. Dye TD, Tolliver NJ, Lee RV, Kenney CJ. Violence, pregnancy and birth outcome in Appalachia. Paediatric and Perinatal Epidemiology 1995;9:35-47. 19. Bullock LF, McFarlane J. The birth-weight/ battering connection. American Journal of Nursing 1989;1153-5. 20. Parker B, McFarlane J, Soeken K. Abuse during pregnancy: effects on maternal complications and birth weight in adult and teenage women. Obstetrics and Gynecology 1994;84:323-8. 21. McFarlane J, Parker B, Soeken K. Abuse during pregnancy: associations with maternal health and infant birth weight. Nursing Research 1996; 45:37-42. 22. Valdez-Santiago R, Sanin-Aguirre LH. La violencia dome´stica durante el embarazo y su relacio´n con el peso al nacer. [Domestic violence during pregnancy and its relation to birth weight]. Salud Pu´blica de Mexico 1996;38:352-62. In Spanish. 23. Murphy CC, Schei B, Myhr TL, DuMont J. Abuse: a risk factor for low birth weight? A systematic review and meta-analysis. Canadian Medical Association Journal 2001;164:1578-9. 24. Behrman R. Prematurity and intrauterine growth retardation. In: Behrman R, editor. Nelson’s textbook of pediatrics. Philadelphia (PA):WB Saunders & Co.; 1992:441-9. 25. Kliegman R. Intrauterine growth retardation. In:Fanaroff A, Martin R, editors. Neonatal–perinatal medicine: diseases of the fetus and infant. St. Louis (MO), Mosby; 1997. Chapter 12. 26. Engle PL, Menon P, Haddad L. Care and nutrition: concepts and measurement. Washington(DC): International Food Policy Research Institute; 1997. 27. Ellsberg M, Pen˜a R, Herrera A, Liljestrand J, Winkvist A. Wife abuse among women of childbearing age in Nicaragua. American Journal of Public Health 1999;89:241-4. 28. Ellsberg M, Pen˜a R, Herrera A, Liljestrand J, Winkvist A. Candies in hell: women’s experiences of violence in Nicaragua. Social Science and Medicine 2000;51:1595-1610. 29. Ellsberg M, Caldera T, Herrera A, Winkvist A, Kullgren G. Domestic violence and emotional distress among Nicaraguan women: results from a population based study. American Psychologist 1999; 54:30-6. 30. Pen˜a R, Liljestrand J, Zelaya E, Persson LA˚. Fertility and infant mortality trends in Nicaragua 1964–1993. The role of women’s education. Journal of Epidemiology and Community Health 1999;53:132-7. 31. Bang AT, Bang RA and the SEARCH team. Diagnosis of causes of childhood deaths in developing countries by verbal autopsy: suggested criteria. Bulletin of the World Health Organization 1992;70:499-507. 32. McFarlane J, Parker B, Soeken K, Bullock L. Assessing for abuse during pregnancy. Severity and frequency of injuries and ssociated entry to prenatal care. JAMA 1992;267:3176-8. 33. Renzi MR, Agurto S. La esperanza tiene nombre de mujer [Hope is a woman’s name]. Managua, Nicaragua: Fundacio´n Para el Desafio Global; 1998. In Spanish. 34. Zelaya E, Pen˜a R, Garcı´a J, Berglund S, Persson LA˚, Liljestrand J. Contraceptive patterns among women and men in Leo´n, Nicaragua. Contraception 1996; 54:359-65. 35. Jejeebhoy SJ. Associations between wife-beating and fetal and infant death: Impressions from a survey in rural India. Studies in Family Planning 1998; 29:300-8. 36. Ellsberg M, Heise L, Pen˜a R, Agurto S, Winkvist A. Researching domestic violence against women: methodological and ethical considerations. Studies in Family Planning 2001; 32:1-16. 37. Rosales J, Loaiza E, Primante D. Encuesta Nicaraguense de Demografia y Salud, 1998. [Nicaraguan Demographic and Health Survey 1998]. Managua, Nicaragua: Instituto Nacional de Estadisticas y Censos; 1999. In Spanish. 38. Ribe JK, Teggatz JR, Harvey CM. Blows to the maternal abdomen causing fetal demise: report of three cases and a review of the literature. Journal of Forensic Sciences 1993;38:1092-6. 39. Williams JK, McClain L, Rosemurgy AS, Colorado NM. Evaluation of blunt trauma in the third trimester of pregnancy: maternal and fetal considerations. Obstetrics and Gynecology 1990;75:33-7. 40. Omer H, Everly GS. Psychological factors in preterm labor: critical review and theoretical synthesis. American Journal of Psychiatry 1988;145:1507-13. 41. Paarlberg MK, Vingerhoets JJM, Passchier J, Dekker GA, Van Geijn HP. Psychosocial factors and pregnancy outcome: a review with emphasis on methodological issues. Journal of Psychosomatic Research 1995;39:563-95. 42. Wadhwa PD, Dunkel-Schetter C, Chicz-DeMet A, Porto M, Sandman CA. Prenatal psychosocial factors and the neuroendocrine axis in human pregnancy. Psychosomatic Medicine 1996;58:432-46. 43. Newberger EH, Barkan SE, Lieberman ES, Mc Cormick MC, Yllo K, Gary LT, et al. Commentary: abuse of pregnant women and adverse birth outcome: current knowledge and implications for practice. JAMA 1992;267:2370-2. 44. Petersen R, Gazmararian JA, Spitz AM, Rowley DL, Goodwin MM, Saltzman LE, et al. Violence and adverse pregnancy outcomes: a review of the literature and directions for future research. American Journal of Preventive Medicine 1997;13:366-73. 45. Valladares Cardoza ME, Ellsberg M, Pen˜a R, Ho¨gberg U, Persson LA˚. Physical abuse during pregnancy: a risk factor of low-birth weight. American Journal of Obstetrics and Gynecology (forthcoming). 46. Amaro H, Fried LE, Cabral H, Zuckerman B. Violence during pregnancy and substance use. American Journal of Public Health 1990;80:575-9. 47. Gazmararian JA, Adams MM, Saltzman LE, Johnson CH, Bruce FC, Marks JS, et al. The relationship between pregnancy intendedness and physical violence in mothers of newborns. Obstetrics and Gynecology 1995;85:1031-8. 48. Dietz PM, Gazmararian JA, Goodwin MM, Bruce FC, Johnson CH, Rochat RW. Delayed entry into prenatal care: effect of physical violence.Obstetrics and Gynecology 1997;90:221-4. 49. Forte JA, Franks DD, Forte JA, Rigsby D. Asymmetrical role-taking: comparing battered and non-battered women. Social Work 1996;41:59-73. 50. Smith PH, Earp JA, DeVellis R. Measuring battering: development of the women’s experience with battering (WEB) scale. Women’s Health: Research on Gender, Behaviour and Policy 1995;4:273-88. 51. Rao V, Bloch F. Wife-beating, its causes and its implications for nutrition allocations to children: an economic and anthropological case study of a rural South Indian community. Washington (DC): The World Bank, Policy Research Department, Poverty and Human Resources Division; 1993. 16 Bulletin of the World Health Organization 2003, 81 (1) Research Commentary Violence against women and the risk of infant and child mortality Alexander Butchart1 & Andre´s Villaveces2 The preceding paper by A˚sling-Monemi et al. tests the hypothesis that violence against women, either before, during, or after pregnancy, increases the risk of child and infant mortality among their offspring. Their findings are startling— after controlling for other possibly confounding factors, the children of womenwhowere physically and sexually abused by a partner were six times more likely to die before the age of 5 years than children of non-abused women, and partner abuse was found to account for around one-third of all deaths of under-5-year-olds in the study region of Le´on, Nicaragua. Among neonates the main proximal causes of death were preterm delivery and low birth weight, and among older age groups infectious diseases. The authors speculate that this relationship between violence and child mortality is mediated through chronic stress of the mother, possibly affecting both the fetus and the care provided to the child. The paper by A˚sling-Monemi et al. is timely and important in view of the recently published World report on violence and health (1), which found that interpersonal violence may produce a wide spectrum of health consequences quite apart from deaths and injuries. This means that violence, in addition to warranting preventive attention as a direct cause of physical and psychosocial damage, is also a significant risk factor for other health outcomes, and the stronger the evidence for the size and strength of these risk effects, the stronger the arguments for preventing violence. To claim, albeit in the specific setting of A˚sling-Monemi et al.’s study, that one-fourth of all under-5mortality is causally related to violence against the mothers of the deceased is to claim a massive health risk-effect for violence against women. If this claim is validated for Nicaragua and shown to apply in other countries, it would be a powerful additional argument for increased investment in developing and implementing effective violence-prevention strategies. The findings reported by A˚sling-Monemi et al. are consistent with those of a handful of other studies that have associated violence during pregnancy with premature labour and birth, fetal injury, and low birth weight (2–7). They are also consistent with one other study linking violence during pregnancy to infant death (8). Where A˚sling-Monemi et al. go beyond these existing studies is in looking at violence not only during but also after pregnancy, and in attempting to estimate the proportion of all under-5 mortality that could be attributed to violence against mothers. A strength of the study is the fact that it adjusted for several important confounding variables such as the mothers’ age and parity, educational level, employment status, and the degree to which basic needs were satisfied. However, like all pioneering studies of relatively under-researched areas of public health concern, it suffers from some limitations, as discussed below. Because the exposure data were obtained retrospec- tively, the possibility that recall bias accounted for some or all of the reported associations cannot be excluded. Such bias may have arisen because women who were in abusive relationships may have remembered their experiences differently fromother women (9). Physical and sexual abuse are widely believed to be detrimental to family functioning and health. Mothers who experienced the loss of a child may have consciously or unconsciously exaggerated their prior level of exposure to abuse if they believed it was related to the child’s death, which would artificially inflate the estimate of the risk to infant and child mortality represented by violence towards the mother. As mentioned by the authors, the study neither collected data on the extent to which the pregnancies were intended nor on child abuse. Without such data it is difficult to test the hypothesis that offspring born to mothers who experience violence before during and after pregnancy will be more likely to die than those borne to mothers who do not experience violence. The population-attributable risk (33% of all under-5 mortality in the study area) reported by the authors must therefore be treated with caution and is most likely to overestimate substantially the true association. Data on child abuse and pregnancy intendedness would possibly have helped to explain better the relationship between violence towards the mother and subsequent child death. Such informationmay have also provided useful indications of when to intervene to prevent more effectively abuse and neglect. Examples of early interventions for the prevention of abuse and neglect within the family include the prevention of unintended pregnancies (10) and home visitation in the first 3 years after birth (11, 12). Information on the added risk of death for infants due to abuse by their mothers could potentially and more cost-effectively target violence-preven- tion interventions. This is an important study which replicates findings about the relationship between violence against women and infant mortality, and which for the first time attempts to measure the scale of the impact on infant mortality. However, its methodological weaknesses raise major uncertainties about the size of the estimated impact of violence towards mothers on infant mortality. These uncertainties must be resolved through further studies in different settings that use improved methods to test the hypothesis. 1 Team Leader, Prevention of Violence, Department of Injuries and Violence Prevention, World Health Organization, 1211 Geneva 27, Switzerland (email: butcharta@who.int). Correspondence should be addressed to this author. 2 Medical Officer, Prevention of Violence, Department of Injuries and Violence Prevention, World Health Organization, Geneva, Switzerland. Ref. No. 02805 17Bulletin of the World Health Organization 2003, 81 (1) Violence against women: a case–referent study in Nicaragua To date, the role of violence as a risk factor for health consequences other than injuries has been studied extensively in respect of the distal consequences of infant and child physical and sexual abuse on adult-onset psychiatric disorders (e.g. depression, anxiety disorders), behavioural problems (e.g. smoking and substance abuse, unsafe sexual practices), and the perpetration of violence (e.g. child maltreatment, youth violence, self-directed violence). Indeed, such is the weight of evidence for child sexual abuse as a risk factor that it is included as such in the World health report 2002. Prevention of child abuse through home visitation and parent support are among the most effective violence prevention interventions known to date (11, 12). Coupled with findings showing the proportion of adult-onset disease attributable to child sexual abuse, the results of such studies are strong arguments for investing in the prevention of child sexual abuse. It is hoped that, if the findings of the study by A˚sling- Monemi et al. are confirmed, and that as similar evidence accumulates for the non-injury health consequences of violence against women, prevention of violence towards women during and beyond pregnancy will be brought even further into the mainstream of public health interventions. n Conflicts of interest: none declared. References 1. Krug E, Dahlberg LL, Mercy JA, Zwi AB, Lozano R., editors. World report on violence and health. Geneva: World Health Organization; 2002. 2. Valdez-Santiago R, Sanin-Aguirre LH. Domestic violence during pregnancy and its relation to birth weight. Salud Pu´blica de Me´xico 1996;38:352-62. In Spanish. 3. Valladares E. Physical abuse during pregnancy: a risk factor for low birth weight. Umea˚, Umea˚ University, Department of Epidemiology and Public Health; 1999. 4. Parker B, McFarlane J, Soeken K. Abuse during pregnancy: effects on maternal complications and birth weight in adult and teenage women. Obstetrics and Gynecology 1994;84:323-8. 5. Murphy CC, Schei B, Myhr TL, Du Mont J. Abuse: a risk factor for low birth weight? A systematic review and meta- analysis. Canadian Medical Association Journal 2001;164:1567-72. 6. Curry MA, Perrin N, Wall E. Effects of abuse on maternal complications and birth weight in adult and adolescent women. Obstetrics and Gynecology 1998;92 (4 Pt 1):530-4. 7. Bullock LF, McFarlane J. The birth-weight / battering connection. American Journal of Nursing 1989;89:1153-5. 8. Jejeebhoy SJ. Associations between wife-beating and fetal and infant death: impressions from a survey in rural India. Studies in Family Planning 1998; 29:300-8. 9. Hennekens CH, Buring JE, Mayrent SL. Epidemiology in medicine. Boston (MA): Little Brown; 1987. 10. Kellermann AL, Fuqua-Whitley DS, Rivara FP, Mercy J. Preventing youth violence: what works? Annual Review of Public Health 1998;19:271-92. 11. Kitzman H, Olds DL, Henderson CR, Jr., Hanks C, Cole R, Tatelbaum R, et al. Effect of prenatal and infancy home visitation by nurses on pregnancy outcomes, childhood injuries, and repeated childbearing. A randomized controlled trial. JAMA 1997;278:644-52. 12. Olds DL, Henderson CR, Jr., Chamberlin R, Tatelbaum R. Preventing child abuse and neglect: a randomized trial of nurse home visitation. Pediatrics 1986; 78:65-78. 18 Bulletin of the World Health Organization 2003, 81 (1) Research Table 2. Physical and sexual violence against mothers as risk factors of infant and under-5 mortality in Leo´n, Nicaragua, 1993–6 Risk factor Odds ratio for deaths at:a % of cases % of referents 0–11 months 0–59 months (n = 110) (n = 203) Experience of physical violence(not including sexual violence) No physical violence by anyone 40 64 1.0 1.0 Physical non-partner violence 11 5 2.9 (1.0–8.4)b 3.2 (1.2–8.5) Physical partner violence 32 25 2.1 (1.1–4.0) 2.2 (1.2–3.9) Both non-partner and partner violence 17 6 5.0 (2.0–12.7) 4.3 (1.9–9.7) Experience of sexual violence (not including physical violence) No sexual violence 74 90 1.0 1.0 Sexual non-partner violencec 4 2 3.7 (0.8–16.7) 2.6 (0.7–9.9) Sexual partner violence 22 8 4.3 (1.9–9.5) 3.2 (1.6–6.3) Violence by type No physical or sexual violence by anyone 39 63 1.0 1.0 Non-partner violence onlyc 10 4 2.9 (1.0–8.6) 3.7 (1.3–10.1) Physical partner violence 29 25 1.9 (1.0–3.9) 2.1 (1.2–3.8) Sexual partner violence 2 2 3.6 (0.5–27.6) 2.3 (0.4–15.1) Both physical and sexual partner violence 20 6 6.5 (2.5–16.7) 5.5 (2.4–12.6) Physical partner violence by temporal sequence No physical violence by partner 51 68 1.0 1.0 Former physical violenced 20 13 2.2 (1.1–4.6) 2.1 (1.1–4.1) Physical violence during last 12 months only 8 6 1.6 (0.5–4.7) 1.9 (0.7–4.8) Physical violence during index pregnancy and last 12 monthse 21 12 2.8 (1.3–6.2) 2.5 (1.2–5.0) Partner violence by severity No partner violence 49 67 1.0 1.0 Moderate partner violencef 4 3 1.8 (0.5–7.0) 1.9 (0.5–6.8) Severe partner violenceg 46 30 2.5 (1.4–4.5) 2.3 (1.4–3.9) Violence by severity, offender, and temporal sequence No severe violence 44 66 1.0 1.0 Non-partner violence onlyh 10 4 2.6 (1.0–7.4) 3.3 (1.3–8.9) Former severe violence by partnerd, g 19 13 2.4 (1.1–5.1) 2.4 (1.2–4.6) Recent severe violence by partnerg, h 27 17 3.0 (1.4–6.2) 2.8 (1.5–5.4) a Bivariate odds ratios based on matched analysis. b Figures in parentheses are 95% confidence levels. c Includes physical or sexual non-partner abuse; women having experience of both partner and non-partner abuse are included in the partner abuse group. d Prior to index pregnancy and to the last 12 months before child’s death (in cases) or to 12 months before interview (referents). e Only one mother (of a case) had experience of violence during pregnancy but not during the last 12 months before child death. f Includes pushes, slaps or throwing objects. g Includes hits, kicks, bites, beating up or forced sex. h During index pregnancy or during the last 12 months before child’s death (in cases) or 12 months before interview (referents). Violence against women: a case–referent study in Nicaragua ABulletin of the World Health Organization 2003, 81 (1)