BIRTH 40:1 March 2013
Intimate Partner Violence and the Association with Very Preterm Birth Lyndsey F. Watson, BSc, MSc, PhD and Angela J. Taft, BA, MPH, PhD ABSTRACT: Background: Intimate partner violence is a major public health problem. It
occurs commonly in pregnancy, resulting in adverse events for women and their fetus or children. The objective of this study was to examine the association between intimate partner violence and very preterm birth. Methods: This population-based, case-control study was conducted in Victoria, Australia, from 2002 to 2004. Interviews were conducted with 603 women who had a singleton very preterm birth (20–31 weeks’ gestation), 770 women who had a singleton term birth (37 or more completed weeks’ gestation), 139 women who had a very preterm twin birth, and 214 women who had a term twin birth. Intimate partner violence was measured using the Composite Abuse Scale, and questions were also asked about fear of partners and violence from others. Results: Prevalence of intimate partner violence in the past 12 months was 14.9 percent in singleton case women, 11.7 percent in singleton control women, 9.5 percent in twin case women, and 14.7 percent in twin control women. Fear of a previous partner and reporting similar violence experience with someone else were more likely in singleton births (AOR = 1.36; 95% CI 1.03, 1.79) and (AOR = 1.44; 95% CI 1.12, 1.86), respectively. No differences between twin case women and twin control women were observed. When the precipitating cause of very preterm birth was investigated, antepartum hemorrhage was signiﬁcantly associated with intimate partner violence and all its subscales. Conclusions: The heterogeneity of causes of very preterm birth may explain the lack of association found with intimate partner violence in pregnancy. Pregnant women have a signiﬁcant risk of intimate partner violence, which should be a serious concern for all care providers. (BIRTH 40:1 March 2013) Key words: case-control study, intimate partner violence, singleton birth, twin birth, very preterm birth
Intimate partner violence is a major public health problem with signiﬁcant consequences for women’s mental, emotional, and physical health (1). Globally, lifetime prevalence has been found to be between 10 and 52 percent (2) and even as high as 71 percent in developing countries (1). Reported 12-month prevalence of
intimate partner violence is between 15 and 30 percent in the community (1). Its prevalence in pregnant women has been found to be between 4 and 8 percent (3), but this ﬁnding depends on the deﬁnition, population, timing, and method of assessing violence (4). Several studies have reported a prevalence of about 6
Lyndsey F. Watson is a Senior Research Fellow and Biostatistician at Mother and Child Health Research; and Angela J. Taft is Associate Professor and Principal Research Fellow at Mother and Child Health Research, La Trobe University, Melbourne, Australia.
Address correspondence to Lyndsey F. Watson, BSc, MSc, PhD, Mother and Child Health Research, La Trobe University, 215 Franklin Street, Melbourne, Vic. 3000, Australia.
The study received a Faculty of Health Sciences Grant, La Trobe University, Bundoora, Victoria (1998, 2003); a National Health and Medical Research Council Project Grant, Canberra, ACT (2001– 2003); a SIDS and Kids Victoria Grant, Malvern, Victoria (2002); and Telstra Community Development Fund Grant, Melbourne, Victoria (2003).
Accepted June 14, 2012
© 2013, Copyright the Authors Journal compilation © 2013, Wiley Periodicals, Inc..
BIRTH 40:1 March 2013
18 percent (5–8), although it has been found to be as high as 20 percent (9). Pregnant women who have experienced intimate partner violence face the added potential risk of femicide (10); possible adverse outcomes for the fetus include low birthweight or preterm birth (3). In a case-control study of very preterm birth, we aimed to quantify the prevalence of intimate partner violence of women in pregnancy and its association with very preterm birth in singleton and twin births.
Methods The study design and data collection have been described in detail elsewhere (11). In brief, the casecontrol study recruited cases from all women having a singleton or twin preterm birth from 20 to less than 32 completed weeks’ gestation between April 2002 and March 2004 in Victoria, Australia. All women having a twin birth in Victoria of 37 or more completed weeks’ gestation from May 2002 to April 2004 were eligible to be the twin control participants. Women who could not speak English, had an intellectual disability, illness, or extreme anxiety were excluded. Ethics approval was sought from all hospitals providing maternity care in Victoria and was obtained for 95 percent (n = 73) (12) of those where the recruitment for both case and control participants occurred. Women were interviewed within a few weeks of the birth by either telephone or faceto-face. Twenty percent more case women than control women were interviewed face-to-face. The research interviewers all had nursing or midwifery backgrounds and were trained in the study protocols and interview skills and monitored for these throughout the study (13,14). Intimate partner violence was assessed using the Composite Abuse Scale (15), a 30-item, well-validated measure of whether women were probably abused in the past 12 months. Women who scored a total of between 3 and 6 were categorized as probable intimate partner violence victims and greater than 6 as experiencing intimate partner violence. We also calculated subscales which included the following (15): 1. Severe combined abuse: an 8-item factor with severe physical abuse items, all sexual abuse items, and the physical isolation aspects of emotional abuse (cutoff 1). 2. Emotional abuse: an 11-item factor including verbal, psychological, dominance, and social isolation abuse items (cutoff 3). 3. Physical abuse: a 7-item factor of less severe physical abuse (cutoff 1). 4. Harassment: a 4-item factor (cutoff 2).
One recruiting hospital did not allow the Composite Abuse Scale to be administered, and women recruited in that hospital were excluded from the intimate partner abuse analysis.
Data Analysis The analysis was undertaken using unconditional logistic regression—unadjusted and adjusted for sociodemographic factors including maternal age, country of birth, education, marital status, and parity; in pregnancy, factors included smoking, alcohol, and illicit drug use all previously found to be risk factors for very preterm birth (11) and having a strong correlation with intimate partner violence (16,17). Interaction effects between intimate partner violence measures and sociodemographic factors were investigated. Singleton case women were divided into a hierarchy of precipitating causes of very preterm birth (in order: hypertension, conditions associated with antepartum hemorrhage, maternal factors, fetal growth restriction, prolonged preterm rupture of membranes, and spontaneous labor), as done elsewhere (18). Conditions associated with hemorrhage included placenta previa, placental abruption, and blood loss before delivery; maternal factors included perinatal infection, diabetes, uterine abnormalities, cytopenia, ﬁbroids, and other speciﬁed maternal factors. Analysis for association of intimate partner violence with these precipitating causes of very preterm birth was undertaken using multinomial logistic regression. Sensitivity analysis was undertaken to see if the method of interview (by telephone or face-to-face) affected the ﬁndings. Comparison with violence recorded in medical records was also conducted. Data were analyzed using the data analysis program, Stata version 11 (19). Multivariable logistic regression was used to examine the association between abuse and very preterm birth. Results are presented as prevalence of responses, odds ratios, 95 percent conﬁdence intervals (95% CIs), and p values. Post hoc calculations for the association between very preterm birth and intimate partner violence indicated 80 percent power to determine odds ratios for intimate partner violence of 1.6 or more.
Results The sociodemographic characteristics of this study sample are shown in Table 1, based on the entire study sample of 603 singleton case births, 796 singleton control births, 139 twin case births, and 224 twin control births. Signiﬁcant differences (p < 0.05) were
BIRTH 40:1 March 2013
seen between singleton births and control births in place of birth, marital status, parity, smoking, and use of alcohol and illicit drugs in pregnancy; for twins, a signiﬁcant difference only occurred in maternal age. Thirty-four women having term births (27 singleton and 7 twin births) were not asked intimate partner violence questions because of our agreement with one recruiting hospital. An additional 20 women (8 with very preterm singleton births, 6 with singleton term births, 2 with very preterm twin births, and 4 with term twin births) declined to answer all intimate partner violence questions or were not asked them as their partner was present. This process left 595 singleton case women, 763 singleton control women, 137 twin case women, and 217 twin control women for the intimate partner violence analysis (Table 2). The prevalence of intimate partner violence measured by positive scoring on the Composite Abuse Scale and on its four subscales is shown in Table 2. Of singleton case women, 14.9 percent (n = 88) reported probable intimate partner violence of whom 44 scored 7 or above. Of singleton control women, 11.7 percent (n = 89) reported intimate partner violence, with a similar number (n = 46) scoring 7 or above. Of twin case women, 9.5 percent (n = 13) reported probable intimate partner violence of whom six scored 7 or above. Of twin control women, 14.1 percent (n = 30) reported probable intimate partner abuse, of whom 20 scored 7 or more. None of the changes was statistically signiﬁcant. Singleton case women were more likely to report all types of intimate partner violence compared with singleton control women, with emotional abuse and physical abuse nearly reaching statistical signiﬁcance
(p = 0.06 and 0.08, respectively). In the twin arm of the study, no pattern of difference occurred between case women and control women. These odds ratios decreased, and all p values were 0.3 or higher when adjusted for maternal age, country of birth, education, marital status, parity, and use in pregnancy of smoking, alcohol, and illicit drugs. To understand the decrease, interactions were investigated. Two interactions were found to be signiﬁcant in the singleton arm. Women experiencing physical abuse and drinking alcohol in pregnancy were more likely to have very preterm births (OR for interaction 2.24; 95% CI 1.01, 4.97; p = 0.05). Women experiencing severe combined abuse and starting antenatal care in the second trimester were also more likely to have very preterm births (OR for interaction 15.7; 95% CI 1.70, 144; p = 0.02). Women with very preterm singleton babies were signiﬁcantly more likely (p = 0.03) to report being afraid of a previous partner than control women and to report that they had experienced violence from someone else (p = 0.005). Analysis of the precipitating cause of preterm birth showed that singleton case women having conditions associated with antepartum hemorrhage were signiﬁcantly more likely than control women to have experienced intimate partner violence overall, and each of the subgroups of the Composite Abuse Scale. In addition, they were more likely to be afraid of their current or a previous partner (Table 3). Other preterm birth pregnancy complications were not associated with intimate partner abuse with the exception of singleton case women with prolonged preterm rupture of membranes who were signiﬁcantly more likely to report experiencing violence
Table 1. Descriptive Analysis of Sociodemographic Factors Singleton and Twin Births in the Very Preterm Birth CaseControl Study
Singleton Births Sociodemographic Factors Total Maternal age 35 yr Education: degree or diploma Income > AUD 50K per annum Born overseas Single/separated/divorced Parous in index pregnancy Smoking in pregnancy Alcohol use in pregnancy Illicit drug use in pregnancy Starting antenatal care in the second trimester
Cases No. (%) 603 140 281 312 125 45 291 136 188 33 44
(100) (23.3) (47.0) (53.9) (20.8) (7.5) (48.3) (22.6) (31.2) (5.5) (7.4)
Controls No. (%) 796 161 386 451 126 29 478 134 335 23 55
(100) (20.2) (48.7) (58.7) (15.9) (3.6) (60.1) (16.8) (42.1) (2.9) (7.0)
*p value associated with chi-square test for independence between case and control women.
Twin Births p* 0.2 0.5 0.08 0.02 0.002