Intimate partner violence during pregnancy: victim or perpetrator? Does it make a difference?

Authors

  • Y Shneyderman,

    1. Borough of Manhattan Community College, City University of New York, New York, NY, USA
    Search for more papers by this author
  • M Kiely

    Corresponding author
    1. Eunice Kennedy Shriver National Institute of Child Health and Human Development/NIH/DHHS, Rockville, MD, USA
    • Correspondence: Dr M Kiely, Division of Epidemiology, Statistics and Prevention Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, 6100 Executive Blvd, Rockville, MD, USA. Email kielym@nih.gov

    Search for more papers by this author

Abstract

Objectives

To differentiate between forms of intimate partner violence (IPV) (victim only, perpetrator only, or participating in reciprocal violence) and examine risk profiles and pregnancy outcomes.

Design

Prospective.

Setting

Washington, DC, July 2001 to October 2003.

Sample

A total of 1044 high-risk African-American pregnant women who participated in a randomised controlled trial to address IPV, depression, smoking and environmental tobacco smoke exposure.

Methods

Multivariable linear and logistic regression.

Main outcome measures

Low and very low birthweight, preterm and very preterm birth.

Results

Five percent of women were victims only, 12% were perpetrators only, 27% participated in reciprocal violence and 55% reported no IPV. Women reporting reciprocal violence in the past year were more likely to drink, use illicit drugs and experience environmental tobacco smoke exposure and were less likely to be very happy about their pregnancies. Women reporting any type of IPV were more likely to be depressed than those reporting no IPV. Women experiencing reciprocal violence reported the highest levels of depression. Women who were victims of IPV were more likely to give birth prematurely and deliver low-birthweight and very-low-birthweight infants.

Conclusions

We conclude that women were at highest risk for pregnancy risk factors when they participated in reciprocal violence and so might be at higher risk for long-term consequences, but women who were victims of IPV were more likely to show proximal negative outcomes like preterm birth and low birthweight infants. Different types of interventions may be needed for these two forms of IPV.

Introduction

The Centers for Disease Control and Prevention define intimate partner violence (IPV) as physical, sexual or psychological harm by a current or former spouse or partner,[1] with serious psychosocial and physical sequelae. The National Violence against Women Survey found that 22.1% of women and 7.4% of men report any violence by an intimate partner during their lifetimes. Annually, in the USA, approximately 1.3 million women and 835 000 men report physical assault by an intimate partner.[2] Using the Behavior Risk Factor Surveillance System, Breiding and colleagues[3] found a lifetime prevalence of IPV of 26.4% in women and 15.9% in men. They also found that the lifetime prevalence of IPV was similar for non-Hispanic African-American and non-Hispanic white women, whereas the rate for the 12-month period preceding the survey was almost twice as high among African-Americans.[3]

There are many factors associated with or causally linked to IPV. Women who were younger, had less education and lower income, and those who were single mothers reported more lifetime IPV than their counterparts.[4, 5] A prevalence study from Canada used data from the Canadian Perinatal Surveillance System, which questioned women on abuse before, during and after pregnancy. Overall, the prevalence was 10.9% of women reporting abuse during the 2 years preceding interviews. Women who had low income (21.2% abused), were single, divorced, separated or widowed (35.3% abused), were <19 years old (40.7% abused) and were Aboriginal mothers (30.6% abused) had a higher prevalence of abuse.[6] These findings reinforce those by Bhandari and colleagues[7] who reported that family stressors such as financial issues, lack of social support, and legal and transportation issues put women at increased risk for abuse.

There are conflicting reports in the literature about whether pregnancy raises or lowers the risk of IPV.[8, 9] Alcohol use has consistently been associated with IPV. A World Health Organization multi-country study found that when one or both partners abused alcohol, there were significantly higher rates of IPV experienced by women.[5] Kiely and colleagues[10] reported that women with continued IPV during pregnancy were significantly more likely to use alcohol. Breiding and colleagues[11] reported that IPV victimisation in women was associated with heavy or binge drinking and cigarette smoking. Illicit drug use has been associated with physical partner violence.[12, 13] Physical, sexual or psychological IPV have been associated with depressive symptoms.[10, 14, 15] Mistimed or unintended pregnancies were linked to higher rates of IPV in the year before conception or during pregnancy.[16] Alcohol,[9] tobacco and drug use,[17] depression[18] and unintended pregnancy[9] are not only associated with IPV but are also considered as known risk factors for IPV during pregnancy. Hence, it is important to note if these risk factors are present in women who experience IPV to ensure that they receive proper care during pregnancy and postpartum.

Intimate partner violence increases both pregnancy complications (e.g. inadequate weight gain, maternal infections and bleeding) as well as adverse pregnancy outcomes (low birthweight [LBW], preterm birth [PTB] and neonatal death).[19-21] Kiely and colleagues[10] report that among women experiencing IPV victimisation throughout pregnancy and postpartum, those randomised to the intervention compared with usual care had significantly fewer very preterm births (VPTB) (<33 weeks of gestation) and significantly longer mean gestational age at delivery. A Pregnancy Risk Assessment Monitoring System study found that women reporting IPV in the year before pregnancy were more likely to deliver prematurely and to have LBW infants.[22] Other studies have found similar associations between abuse during pregnancy and LBW, PTB and maternal infections, low gestational weight gain, smoking, alcohol and illicit drug use.[19, 21, 23]

Although the link between victimisation and negative outcomes is established, there may also be an association between women's aggression and physical and psychological sequelae. Girls who are aggressive in adolescence have higher rates of early pregnancy, have higher rates of obstetric and delivery complications, and score higher for depression and anxiety than their nonaggressive counterparts.[24]

Partners experience perpetration and victimisation differently. The types of violence reported tended to differ, with men reporting more verbal or psychological abuse and women reporting more physical or sexual abuse.[25] It has been posited that women react to violence in their relationships, whereas men initiate it.[26] Women are injured more[27-29] and have more psychological consequences.[3, 15]

Few, if any, studies have examined the different ways in which pregnant women experience IPV and the risk profiles of each form of IPV (victim, perpetrator or reciprocal). The purpose of this study was to examine the different forms of IPV present in a sample of high-risk pregnant women. We examine the risk factors (alcohol, illicit drug and tobacco use, whether the pregnancy was wanted or not [‘pregnancy wantedness’] and depression) by the form of IPV and pregnancy outcome.

Methods

This study uses data from the National Institutes of Health—District of Columbia Initiative to Reduce Infant Mortality in Minority Populations, a congressionally mandated project to improve maternal and child health outcomes in African-Americans living in the District. The data presented here are from the Healthy Outcomes of Pregnancy Education study, or DC-HOPE, a randomised controlled trial designed to address smoking, exposure to environmental tobacco smoke (ETSE), IPV and depression, by providing an integrated behavioural intervention and following the women throughout pregnancy and postpartum. The methods and intervention have been previously described.[30]

Participants

Women were recruited from six prenatal care clinics in DC from July 2001 to October 2003. Women were screened for eligibility in two stages, first based on demographic characteristics (self-identifying as black, African-American or Latina, 18 years old or older, ≤28 weeks of gestation, DC resident and English speaking). Those who were demographically eligible gave consent and were screened by audio-computer-assisted self-interview for one of the following risks: smoking, ETSE, IPV and/or depression. An average of 9 days after screening, they completed the baseline interview and then gave consent to participate in the randomised controlled trial. A total of 2913 women were screened for eligibility. Of those, 1191 women consented to participate in the study and 1070 (90%) completed baseline phone interviews and were further randomised into intervention and usual-care groups by using site-specific and risk-specific block randomisation. Follow-up data collection by telephone, conducted by interviewers blinded to care group, occurred during the second and third trimesters of pregnancy (22–26 and 34–38 weeks of gestation, respectively) and 8–10 weeks postpartum. Data on maternal and infant outcomes were abstracted from medical records. Figure 1 displays the eligibility, consent and randomisation process of DC-HOPE. The current analysis includes the 1044 African-American women who were still pregnant at the time of the baseline questionnaire. These analyses include the data on women in the intervention and control groups.

Figure 1.

Profile of project DC-HOPE randomized.

Measures

Intimate partner violence

Intimate partner violence was measured by the Revised Conflict Tactics Scale.[31] During the baseline interview, women reported IPV they had experienced or had perpetrated during the previous year. During the follow-up interviews, the period of IPV was since the previous interview. For each item on the Revised Conflict Tactics Scale, the women rated the frequency with which a particular event happened to them and the frequency with which the women used violence on their partner. Women who reported only being the victim were classified as victims only; women who reported that they used violence on their partner, but their partner did not, were considered perpetrators; and women who were both victims and perpetrators were classified as participating in reciprocal violence. This study used the physical assault and sexual coercion subscales.[31] The minor items on the physical assault subscale include twisting a partner's arm or hair, pushing or shoving, or grabbing, while the severe items include punching, choking, kicking or using a knife or a gun with a partner. The sexual coercion scale includes asking about minor items such as insisting on oral, anal or vaginal sex, and more severe items include using force and threats to make a partner have oral, anal or vaginal sex. We report on severe and minor physical IPV, severe and minor sexual IPV, and three different forms of IPV: perpetration, victimisation and reciprocal.

Pregnancy risk factors

The pregnancy risk factors assessed at baseline and analysed include alcohol and illicit drug use during pregnancy, depression, smoking, ETSE, pregnancy wantedness and pregnancy happiness. These were chosen because previous studies found associations between these factors and negative pregnancy and infant outcomes, such as LBW and PTB.

Alcohol use questions asked about frequency of use during pregnancy of different alcoholic beverages (beer, wine, wine coolers and spirits). For this analysis, if women reported any type or quantity of drinking during pregnancy, alcohol use was considered to have occurred. Women were coded as using illicit drugs if they reported using marijuana, cocaine, heroin, LSD, amphetamines, sedatives or tranquillisers, or any other drugs since learning they were pregnant (yes/no).

Depression was assessed using the Hopkins Symptom Check List.[32] This scale consists of 20 questions asking participants about how they have felt in the past month and whether they were distressed by these symptoms. The symptoms include feeling hopeless about the future, poor appetite, trouble falling asleep, thoughts of death, feeling worthless and difficulty making decisions. The responses were on a five-point Likert scale ranging from ‘Not at all’ to ‘Extremely’. Depression was defined as a mean Hopkins score >0.75.

Smoking at baseline was considered to be present if the participant reported smoking currently or within the last 6 months and verified by salivary cotinine, and had smoked a total of more than 100 cigarettes in her lifetime. ETSE was marked as present if the participants reported being exposed to one or more cigarettes smoked by someone else inside the home or in other places in the past 7 days.

Pregnancy wantedness was determined by participants reporting having an intended pregnancy or one that was not intended currently but wanted eventually. All other women were considered to have an unwanted pregnancy. Finally, happiness about pregnancy was measured by one question that queried participants about the level of their happiness on a scale of 1–10. Those who reported happiness levels of 1–3 were categorised as unhappy, those who reported happiness of 4–7 were categorised as moderately happy, and a report of >7 was considered very happy.

Pregnancy and birth outcomes

We measured pregnancy and birth outcomes in the current study. PTB was defined as gestation <37 weeks and VPTB was defined as gestation <34 weeks. Birthweight was measured in grams; LBW was defined as <2500 g and very low birthweight (VLBW) as <1500 g at delivery. Finally, small for gestational age (SGA) was based on sex of the infant, birthweight and gestation at delivery. Infants who weighed less than the 10th centile of weight for gestational age were considered as SGA. These variables were coded as dichotomous for statistical analysis.

Demographic variables

We used demographic variables previously associated with both IPV and pregnancy outcomes to control the relationships that were tested in the analyses. Maternal education was used as a proxy of socio-economic status and trichotomised as less than high school, high school diploma or GED®, and some college. A woman's relationship status was dichotomised into single (which included divorced, separated and widowed) or partnered (which included married or having a significant other). We also used maternal age as a continuous control variable in the analyses.

Statistical analysis

We performed analyses using sas version 9.1 (SAS Institute, Cary, NC, USA). To examine the associations between different forms of IPV and pregnancy risk factors, we used linear and logistic regression procedures, depending on the type of variable. Adjusted odds ratios were estimated in multivariable logistic regressions for the relationship of interest while controlling for the effects of maternal age, maternal education and relationship status. Similarly, multivariable linear regressions were estimated with demographic variables and the IPV indicator variables. Finally, happiness to be pregnant is an ordinal variable; therefore, we used multinomial logistic regression.

Similar testing was performed with the pregnancy-related outcomes—all of the outcome variables were dichotomous. Because pregnancy outcomes were determined postpartum and some women were lost to follow up, only data for the women (n = 832) who remained in the study were used in these analyses. We analysed the baseline data collected for women who did and did not have complete data at follow up to determine if there were any differences, using analysis of variance and Mantel–Haenszel chi-square. Furthermore, as some women received the intervention meant to reduce risky behaviours and exposures, we controlled for care group in the analysis. Finally, we controlled for other risk factors in the fully adjusted models of pregnancy outcomes (PTB, LBW, VLBW and SGA) if those risk factors showed an association with the outcome at the  0.10 level of significance. Therefore, some fully adjusted models include smoking, depression and alcohol or illicit drug use.

Results

The women ranged in age from 17 to 51 years, with a mean age of 24.57 years. All of the participants included in these analyses were African-American. At the time of the baseline interview, women were on average 19 weeks pregnant. A large majority of the women were single. Table 1 presents the sociodemographic characteristics and psycho-behavioural risks at baseline between women who reported any IPV perpetration (n = 127), IPV victimisation (n = 51), reciprocal IPV (n = 285) and those who reported no IPV (n = 577). Women who reported any IPV at baseline had significantly higher rates of alcohol and illicit drug use, higher depression and reported more ETSE. Women who dropped out of the study reported higher depression at baseline (mean = 16.48 ± 13.42 for those who remained versus 19.28 ± 16.18 for those who dropped out, < 0.01) and were more likely to be single (< 0.05) (data not shown).

Table 1. Select descriptive characteristics of sample at baseline
VariablesWomen who reported perpetrating IPV (n = 127) Mean ± SD/% (n)Women who reported being victims of IPV (n = 51) Mean ± SD/% (n)Women who reported reciprocal IPV (n = 285) Mean ± SD/% (n)Women who reported no IPV (n = 577) Mean ± SD/% (n)Difference tests P-value
  1. Differences in means or proportions are denoted by bolded text.

  2. WIC, Women, Infants and Children program.

Maternal age (years) 23.94 ± 5.10 26.31 ± 6.67 24.17 ± 5.27 24.74 ± 5.40 0.03
Gestational age (weeks) 20.25 ± 6.9818.24 ± 8.8018.92 ± 6.9018.72 ± 6.800.13
Education
Less than high school33.86 (43)19.61 (10)34.04 (97)28.60 (165)0.16
High school or GED51.97 (66)43.14 (22)42.81 (122)47.49 (274) 
Some college14.17 (18)37.25 (19)23.16 (66)23.92 (138) 
Relationship status
Married/significant other28.35 (36)23.53 (12)24.56 (70)22.18 (128)0.14
Single/divorced/widowed/separated71.65 (91)76.47 (29)75.44 (215)77.82 (449) 
Medicaid 86.51 (109) 66.00 (33) 80.99 (230) 75.83 (436) 0.04
WIC 49.21 (62)41.18 (21)45.26 (129)41.59 (240)0.13
Currently employed31.50 (40)35.29 (18)37.68 (107)37.09 (214)0.29
Risk factors
Alcohol use during pregnancy 18.90 (24) 25.49 (13) 32.98 (94) 15.80 (91) <0.001
Illicit drug use during pregnancy 12.60 (16) 11.76 (6) 17.54 (50) 8.84 (51) <0.01
Active smoking18.11 (23)29.41 (15)20.70 (59)17.33 (100)0.15
ETSE 76.86 (93) 66.67 (34) 80.00 (224) 68.37 (389) <0.01
Depression 17.34 ± 13.36 19.37 ± 14.19 23.16 ± 14.68 13.76 ± 12.81 <0.001
In intervention group 13.68 (71)4.82 (25)27.75 (144)48.35 (279)0.15
Wanted pregnancy 76.98 (97)71.43 (35)74.11 (209)77.60 (447)0.58
Happiness about pregnancy
Unhappy 14.17 (18) 31.37 (16) 22.81 (65) 18.20 (105) 0.03
Moderately happy 44.09 (56) 33.33 (17) 42.81 (122) 38.30 (221)
Very happy 41.73 (53) 35.29 (18) 34.39 (98) 43.50 (251)

Table 2 displays the adjusted odds ratios for the associations between different IPV forms and other pregnancy risk factors. The no IPV category is the referent group in the analyses. The most consistent finding is that regardless of type of violence (any, minor, severe, physical or sexual) these women are depressed. There is a clear linear trend for increasingly higher levels of depression going from perpetrator only to victim only to reciprocal violence. For women with minor IPV, those participating in reciprocal violence were significantly more likely to use alcohol (odds ratio [OR] 2.76, 95% confidence interval [95% CI] 1.95–3.90) and illicit drugs (OR 2.01, 95% CI 1.31–3.07). Women experiencing severe IPV (all forms) were significantly more likely to use alcohol. Women who perpetrate only were significantly more likely to use illicit drugs (OR 1.89, 95% CI 1.02–3.50) as were women who participated in reciprocal violence (OR 3.05, 95% CI 1.83–5.06). For physical IPV, there is a clear linear trend for increasingly significant odds of alcohol use going from perpetrator only (OR 1.63, 95% CI 1.02–2.60) to victim only (OR 2.04, 95% CI 1.00–4.13) to reciprocal (OR 2.89, 95% CI 2.03–4.11). For women reporting sexual IPV, alcohol use was significant for victims (OR 2.17, 95% CI 1.33–3.53) and women participating in reciprocal violence (OR 3.06, 95% CI 1.79–5.23). The women who participate in reciprocal violence were significantly more likely to use illicit drugs (OR 2.11, 95% CI 1.11–4.01).

Table 2. Adjusted odds ratios (95% confidence intervals) associating risk factors of pregnancy and IPV forms
Risk factorsPerpetrator onlyVictim onlyReciprocal
  1. The odds ratios are adjusted for maternal age, maternal education, and relationships status.

  2. a

    These relationships are bivariate between age, relationship status and maternal education and the different forms of IPV.

  3. b

    Depression is a continuous variable; therefore, the coefficients presented are from a linear regression.

  4. c

    Moderately happy category is the reference group.

  5. d

    P < 0.05.

  6. e

    P < 0.01.

  7. f

    P < 0.001.

Any IPV
Maternal age—B (SE)a−0.80 (0.53)1.58 (0.79)§−0.57 (0.39)
Relationship statusa0.72 (0.47–1.11)0.93 (0.47, 1.82)0.88 (0.63, 1.22)
Maternal education a
Less than high school2.00 (1.103.62)§0.44 (0.20, 0.98)§1.23 (0.84, 1.81)
High school graduate1.85 (1.063.23)§0.58 (0.31, 1.11)0.93 (0.65, 1.34)
Alcohol use1.36 (0.82–2.26)1.63 (0.82, 3.25)2.80 (1.99, 3.95)§§§
Illicit drug use1.43 (0.78–2.62)1.56 (0.63, 3.88)2.12 (1.39, 3.25)§§§
Smoking1.05 (0.62–1.78)2.08 (1.04, 4.15)§1.27 (0.87, 1.85)
ETSE1.44 (0.91–2.29)1.02 (0.55–1.90)1.83 (1.302.59)§§§
Depressionb—B (SE)3.93 (1.31)§§5.06 (1.96)§§9.40 (0.97)§§§
Pregnancy wantedness0.96 (0.60–1.53)0.72 (0.37–1.40)0.80 (0.57–1.12)
Pregnancy happinessc   
Unhappy0.72 (0.40–1.30)1.85 (0.89–3.85)1.17 (0.79–1.71)
Very happy0.81 (0.53–1.23)0.92 (0.46–1.8.)0.71 (0.510.96)§
Minor IPV
Maternal age—B (SE) a−0.68 (0.53)2.42 (0.79)§§−0.63 (0.39)
Relationship statusa0.76 (0.50–1.17)0.93 (0.48–1.84)0.86 (0.62–1.20)
Maternal educationa
Less than high school1.69 (0.95–3.01)0.43 (0.190.96)§1.26 (0.85–1.86)
High school graduate1.67 (0.97–2.86)0.58 (0.31–1.12)0.97 (0.67–1.40)
Alcohol use1.41 (0.85–2.33)1.68 (0.85–3.31)2.76 (1.953.90)§§§
Illicit drug use1.39 (0.76–2.55)1.23 (0.46–3.28)2.01 (1.313.07)§§
Smoking1.06 (0.63–1.79)2.13 (1.074.22)§1.27 (0.87–1.85)
ETSE1.36 (0.86–2.15)1.01 (0.54–1.87)1.75 (1.232.47)§§
Depressionb—B (SE)3.84 (1.31)§§4.61 (1.96)§9.57 (0.97)§§§
Pregnancy wantedness1.02 (0.64–1.64)0.77 (0.39–1.50)0.82 (0.59–1.15)
Pregnancy happinessc
Unhappy0.81 (0.46–1.45)2.34 (1.124.88)§1.14 (0.78–1.69)
Very happy0.91 (0.60–1.39)1.04 (0.51–2.13)0.71 (0.510.98)§
Severe IPV
Maternal Age—B (SE)a−0.97 (0.60)0.22 (0.65)−0.58 (0.56)
Relationship statusa0.88 (0.53–1.45)1.00 (0.58–1.74)1.02 (0.63–1.64)
Maternal educationa
Less than high school2.00 (1.023.94)§0.92 (0.49–1.72)1.47 (0.86–2.52)
High school graduate1.74 (0.92–3.31)0.79 (0.45–1.40)0.88 (0.52–1.50)
Alcohol use3.05 (1.855.01)§§§2.53 (1.494.29)§§§4.03 (2.596.27)§§§
Illicit drug use1.89 (1.023.50)§1.47 (0.72–3.01)3.05 (1.835.06)§§§
Smoking1.38 (0.79–2.43)1.05 (0.56–1.98)1.78 (1.102.91)§
ETSE1.99 (1.113.57)§2.43 (1.284.64)§§1.39 (0.85–2.27)
Depressionb—B (SE)6.79 (1.51)§§§7.87 (1.61)§§§9.59 (1.39)§§§
Pregnancy wantedness0.89 (0.53–1.50)0.98 (0.55–1.74)0.71 (0.45–1.12)
Pregnancy happinessc
Unhappy0.82 (0.43–1.57)1.76 (0.99–3.16)1.21 (0.72–2.05)
Very happy0.85 (0.52–1.38)0.76 (0.43–1.34)0.69 (0.43–1.11)
Physical IPV
Maternal age—B (SE)a1.05 (0.50)§1.63 (0.84)−0.69 (0.40)
Relationship statusa0.67 (0.44–1.00)0.84 (0.42–1.71)0.89 (0.63–1.27)
Maternal educationa
Less than high school1.68 (0.98–2.88)0.35 (0.140.86)§1.34 (0.90–2.00)
High school graduate1.53 (0.92–2.52)0.59 (0.30–1.16)0.95 (0.65–1.39)
Alcohol use1.63 (1.022.60)§2.04 (1.004.13)§2.89 (2.034.11)§§§
Illicit drug use1.24 (0.69–2.23)1.15 (0.39–3.37)2.08 (1.353.19)§§§
Smoking0.98 (0.59–1.61)1.16 (0.51–2.63)1.26 (0.86–1.85)
ETSE1.57 (1.012.45)§1.81 (0.86–3.81)1.74 (1.212.48)§§
Depression b—B (SE)5.18 (1.25)§§§7.51 (2.10)§§§9.16 (1.01)§§§
Pregnancy wantedness0.82 (0.53–1.27)1.09 (0.50–2.36)0.76 (0.54–1.08)
Pregnancy happinessc
Unhappy0.81 (0.48–1.39)1.39 (0.59–3.25)1.23 (0.83–1.81)
Very happy0.78 (0.52–1.16)1.25 (0.61–2.55)0.67 (0.480.94)§§
Sexual IPV
Maternal age—B (SE) a−1.38 (0.88)1.15 (0.61)−0.99 (0.69)
Relationship statusa0.92 (0.44–1.92)0.87 (0.53–1.44)2.05 (1.00–4.21)
Maternal Educationa
Less than high school2.42 (0.87–6.73)0.57 (0.33–1.00)1.29 (0.65–2.58)
High school graduate1.67 (0.61–4.57)0.39 (0.230.67)§§0.96 (0.50–1.86)
Alcohol use1.69 (0.79–3.60)2.17 (1.333.53)§§3.06 (1.795.23)§§§
Illicit drug use1.99 (0.87–4.54)1.42 (0.73–2.76)2.11 (1.114.01)§
Smoking1.97 (0.92–4.24)1.80 (1.043.10)§1.94 (1.073.53)§
ETSE1.31 (0.58–2.93)1.25 (0.74–2.11)2.42 (1.165.02)§
Depressionb—B (SE)6.91 (2.23)§§7.18 (1.55)§§§8.83 (1.73)§§§
Pregnancy wantedness2.70 (0.94–7.76)0.69 (0.42–1.16)1.19 (0.64–2.21)
Pregnancy happinessc
Unhappy0.31 (0.07–1.36)1.65 (0.95–2.86)1.14 (0.58–2.25)
Very happy1.49 (0.76–2.94)0.77 (0.45–1.32)1.11 (0.63–1.95)

Table 3 presents the adjusted odds ratios from the models associating baseline reports of IPV types and forms with pregnancy and infant outcomes (PTB, VPTB, LBW, VLBW, SGA), while controlling for demographic and risk factors. Women who were perpetrators only, were not at significantly increased risk of any of the adverse infant outcomes. Women who reported physical IPV and participated in reciprocal violence were more likely to have a PTB (OR 1.60, 95% CI 1.00–2.57). Women who were victims only were the ones with significantly worse birth outcomes. Victims reporting any type of IPV were more likely to have an LBW infant (OR 2.21, 95% CI 1.04–4.72) or VLBW infant (OR 4.54, 95% CI 1.06–19.44). Victims were more likely to have a VPTB if they reported minor IPV (OR 3.66, 95% CI 1.22–10.97), severe IPV (OR 2.78, 95% CI 1.10–7.06) or physical IPV (OR 3.52, 95% CI 1.06–11.65). Victims reporting physical IPV had significantly more LBW (OR 2.49, 95% CI 1.13–5.52) and VLBW (OR 5.67, 95% CI 1.29–25.02) infants. Victims reporting sexual IPV had significantly more VLBW infants (OR 3.74, 95% CI 1.09–12.85).

Table 3. Adjusted odds ratios (95% confidence intervals) in the model associating pregnancy-related outcomes and IPV forms (any IPV)
Pregnancy outcomesPerpetrator onlyVictim onlyReciprocal
  1. The odds ratios are adjusted for maternal age, maternal education, relationship status, care group and baseline risk factors (the latter only if they attained a significant association with the outcome at the P ≤ 0.01 level). Bold: All odds ratios in this table are significant at the P < 0.05 level.

  2. a

    NE, Not estimated because no women participating in reciprocal violence delivered a VLBW infant.

Any IPV
PTB (<37 weeks)0.74 (0.36–1.53)1.52 (0.68–3.36)1.40 (0.87–2.23)
VPTB (<34 weeks)0.61 (0.13–2.76)2.93 (0.90–9.54)2.06 (0.92–4.58)
LBW (<2500 g)0.89 (0.44–1.72)2.21 (1.044.72)1.07 (0.49–1.48)
VLBW (<1500 g)NEa4.54 (1.0619.44)1.72 (0.51–5.81)
SGA0.87 (0.49–1.56)1.85 (0.92–3.72)0.62 (0.38–1.00)
Minor IPV
PTB (<37 weeks)0.89 (0.46–1.75)1.64 (0.76–3.55)1.33 (0.82–2.15)
VPTB (<34 weeks)0.61 (0.13–2.76)3.66 (1.2210.97)2.01 (0.89–4.56)
LBW (<2500 g)1.02 (0.53–1.94)1.83 (0.84–3.98)0.96 (0.57–1.61)
VLBW (<1500 g)NEa4.24 (0.98–18.30)1.87 (0.55–6.32)
SGA0.87 (0.49–1.55)1.81 (0.90–3.63)0.59 (0.360.96)
Severe IPV
PTB (<37 weeks)0.75 (0.63–2.62)1.28 (0.75–2.57)1.39 (0.57–1.26)
VPTB (<34 weeks)0.31 (0.04–2.39)2.78 (1.107.06)1.13 (0.37–3.46)
LBW (<2500 g)0.51 (0.21–1.25)1.66 (0.84–3.28)1.06 (0.54–2.07)
VLBW (<1500 g)1.07 (0.13–8.74)3.28 (0.85–12.65)0.81 (0.10–6.59)
SGA0.40 (0.170.95)0.92 (0.45–1.89)0.88 (0.47–1.65)
Physical IPV
PTB (<37 weeks)0.74 (0.37–1.47)1.41 (0.58–3.41)1.60 (1.002.57)
VPTB (<34 weeks)0.54 (0.12–2.43)3.52 (1.0611.65)2.15 (0.96–4.79)
LBW (<2500 g)0.72 (0.36–1.45)2.49 (1.135.52)1.28 (0.77–2.13)
VLBW (<1500 g)NEa5.67 (1.2925.02)2.06 (0.61–6.98)
SGA0.75 (0.43–1.34)1.62 (0.75–3.48)0.65 (0.39–1.05)
Sexual IPV
PTB (<37 weeks)1.30 (0.52–3.28)1.08 (0.55–2.13)0.98 (0.44–2.21)
VPTB (<34 weeks)0.67 (0.09–5.25)1.57 (0.56–4.36)1.30 (0.36–4.67)
LBW (<2500 g)1.21 (0.47–3.06)1.12 (0.55–2.26)0.61 (0.23–1.62)
VLBW (<1500 g)2.43 (0.29–20.54)3.74 (1.0912.85)NEa
SGA0.62 (0.21–1.81)1.32 (0.71–2.45)0.77 (0.33–1.77)

Discussion

Main findings

The current study was novel in that we analysed different forms of IPV—perpetrators, victims and women participating in reciprocal violence—in a sample targeted specifically at high-risk African-American pregnant women. Intimate partner violence affects millions of women regardless of economic status, race or ethnicity. The results point to different risk profiles for different kinds of violence. Women who reported reciprocal violence in the past year had higher odds of consuming alcohol and illicit drugs during pregnancy and of ETSE exposure, and were the most depressed, which supports and extends findings linked to reciprocal violence in couples.[33] Women who reported only victimisation in the past year were more likely to smoke and had elevated levels of depression. To our knowledge, few studies have linked various forms of IPV with different types of pregnancy risk.[34, 35] Although women who reported reciprocal violence had the worst risk profiles, their birth outcomes were similar to those of women not experiencing IPV. Previous research reported that reciprocal violence was associated with higher injury rates,[36] but no one has studied IPV forms as predictors of pregnancy outcomes.

There has been controversy in the literature regarding perpetration, with one side asserting that female perpetration has been ignored[37] and the other emphasising male perpetrators.[38] The current study accounts for female perpetration and finds clearly delineated risk profiles for different forms of IPV, especially for victims and reciprocal violence. Women were willing to and did report reciprocal violence, more than victimisation only. Previous studies have found that women engaging in violence often do so in the context of responding to partner violence. Swan and Snow[39] reported that 75% of women studied stated that their violence was in self-defence. Women also acknowledged fear,[40, 41] defence of their children,[42, 43] relationship control,[39, 40] and retribution, often for being emotionally hurt[39, 44] as their motivations for violence. Women experience coercive control, including sexual coercion, while rarely being coercively controlling themselves.[40] Future studies should endeavour to gain more insight into the context of IPV during pregnancy.

Studies of IPV generally do not report reciprocal violence because most scales do not ask about perpetration. This may be a reflection that it is not asked simultaneously and of researcher's own biases about the likelihood that women would perpetrate violence. A compendium by the CDC[45] reveals that only two scales ask about both victimisation and perpetration: the Revised Conflicts Tactics Scale,[31] and the Multidimensional Measure of Emotional Abuse.[46, 47]

Alcohol and drug use are often studied in association with IPV, as risk factors or coping mechanisms. Alcohol use has been linked to victimisation and perpetration of IPV. Some studies report drinking as a coping mechanism for dealing with IPV,[48, 49] while others posit drinking as a risk factor for victimisation,[34] and some find that alcohol use was directly related to perpetration.[35] Causality is unclear in our study, but the results point to alcohol use as a particular problem for women who participate in reciprocal violence. It is possible that alcohol use could be linked to more aggressive behaviour in this group of women; however, there is no way to determine if drinking led to aggressive behaviour or if it was used as a coping mechanism. Contrary to current literature, we found that women who were perpetrators only were less likely to acknowledge either alcohol or illicit drug use.[50-52] This may be partially explained by the fact that most researchers do not refine perpetration as we did. Our findings suggest that providers of services should question women on both perpetration and victimisation.

The literature supports our findings that women who experience IPV as victims are more likely to have negative birth outcomes.[19, 20] However, our study is the first, to our knowledge, to separate the forms of violence and point to victimised women as particularly at risk for LWB and PTB. Previous studies have found that psychosocial stress and stressful life events are linked with LBW specifically in African-Americans.[53, 54] If IPV increases stress, these episodes may exacerbate the risk of poor pregnancy outcomes. Victims may have been unable to marshal the resourcefulness needed to fight back when abused and may have internalised the stress caused by the abuse. Future studies should include the necessary measures to understand this phenomenon.

Our results emphasise the need to understand how these risk factors interact and act as mediating mechanisms between IPV and pregnancy outcomes.[55] Addressing health behaviours may require a deeper understanding of the temporality and reasons (i.e. coping mechanisms). Likewise, we found that depression during pregnancy was elevated for all forms of IPV and should be addressed in the context of abusive relationships.

Strengths and limitations

These results should be interpreted in light of study limitations. The sample population is high-risk African-American pregnant women residing in Washington, DC, so they are not necessarily generalisable to a wider population. However, the results inform the kind of elevated risk that affects pregnant women who are already vulnerable. These analyses are longitudinal, involving data collected during pregnancy and including birth outcomes. Although there were few differences between women retained and lost to follow up, the higher depression levels in women who were lost suggest that this group was at higher risk and that our results would have been stronger with complete follow-up. The data on risk factors and IPV were self-reported, which may have led to under-reporting due to social desirability. However, the data on pregnancy outcomes were collected through record abstraction, and so should be considered reliable. The randomised controlled trial was not originally powered to detect differences in birth outcomes, but rather risk resolution. The original study randomisation did not account for different IPV forms. Future research should include a broader population of pregnant women. Following women longer would facilitate understanding the detrimental effects of different forms of IPV on long-term maternal and child health outcomes.

Interpretation

Despite the limitations, the results provide important insight into differences in the risk profiles of pregnant women experiencing various forms of IPV. We confirm that women who participate in reciprocal violence tend to suffer many negative consequences[40] and have serious pregnancy risk profiles. We also add to the literature describing the relationship between IPV and negative birth outcomes, specifically in women who report victimisation only. We found clear delineations between forms of violence experienced and risk profiles and outcomes, which may have implications for future research on IPV and clinicians' practice.

Previous studies reported the benefits of screening for IPV in clinical settings.[56, 57] Expanding the identification of IPV to include the type and form(s) should be possible within the clinical setting with a similar time investment. This would allow beneficial interventions to alleviate a woman's behavioural and mental health issues. This knowledge can help providers to guide the woman's pregnancy and birth care decisions. Our results support the recent American Congress of Obstetricians and Gynecologists opinion that screening for IPV should be a routine part of preventive care for women.[58]

Conclusions

We conclude that women have the most pregnancy risk factors when they are participating in reciprocal violence and so might be at higher risk for long-term consequences, but women who are victims of IPV are more likely to show proximal negative outcomes like PTB and LBW. When women present for care, their provider should consider perpetration as well as victimisation. Different types of interventions may be needed for these two forms of IPV.

Disclosure of interests

Neither of the authors has any competing interests to declare.

Contribution to authorship

YS performed the statistical analyses, participated in the interpretation of the results and the writing of the manuscript. MK, as the NICHD Project Officer, oversaw all the activities of the study while it was in the field. She participated in the analysis and interpretation of the results. MK did a significant amount of the original writing of the manuscript, as well as revising it critically for important intellectual content. Both YS and MK have given final approval of the manuscript.

Details of ethics approval

This study was approved by the Human Subjects Committees at Howard University (for the clinical sites), RTI International (the data coordinating centre) and the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Copies of the Institutional Review Board approval have been archived and I no longer have access to them. The data were collected as part of a clinical trial, registered at ClinicalTrials.gov, www.clinicaltrials.gov, NCT00381823.

Funding

This study was supported by grants no. 3U18HD030445; 3U18HD030447; 5U18HD31206; 3U18HD031919 and 5U18HD036104, Eunice Kennedy Shriver National Institute of Child Health and Human Development and the National Center on Minority Health and Health Disparities. This research was supported, in part, by the intramural programme of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Acknowledgements

The authors would like to acknowledge our many collaborators. We would like to thank Dr John L. Kiely for his scientific and editorial review of the manuscript. We would also like to thank the participants who welcomed us into their lives in the hopes of helping themselves and their children.

Ancillary