Associations between Intimate Partner Violence and Termination of Pregnancy: A Systematic Review and Meta-Analysis

Lucy Chappell and colleagues conduct a systematic review and meta analysis to investigate a possible association between intimate partner violence and termination of pregnancy. Please see later in the article for the Editors' Summary


Introduction
Intimate partner violence (IPV) has been defined by the World Health Organization (WHO) as ''behaviour within an intimate relationship that causes physical, sexual or psychological harm, including acts of physical aggression, sexual coercion, psychological abuse and controlling behaviours'' encompassing both current and past intimate partners [1]. Estimated prevalence varies globally and within countries, and is partly dependent on definition and methodology; lifetime exposure has been found to range from 15% in Japan to 71% in Ethiopia (estimated by WHO multi-country studies), has been estimated at 24% in the UK based on UK Home Office crime statistics [2], and has been estimated to be around 35% (inclusive of stalking) in the US [3]. Rape within intimate relationships has been reported to be common across a number of continents, with lifetime prevalence of forced sex ranging from 5.9% to 42% [4].
Randomised trial evidence has shown that training primary care professionals in selective questioning of women about IPV increases disclosure and referral to specialist IPV services [14]. Antenatal routine questioning for IPV is recommended in both the US and UK [15,16], despite uncertainty over the harms and benefits of universal questioning and subsequent intervention [17]. Ongoing pregnancy is considered to be a time of increased risk of IPV, yet women seeking termination of pregnancy (TOP) are not such a focus of attention [18]. An evidence-based understanding of the association between IPV and TOP would directly inform the development of strategies for effective interventions for IPV. To our knowledge there has been no previous systematic review of the literature.
The aim of this study was to determine whether there is an association between IPV and TOP.

Selection Criteria
Studies were considered eligible for inclusion if they (1) included women who were seeking or had undergone a TOP and studied at least one aspect of IPV in this group; (2) were a randomised control trial, case-control study, cohort study, cross-sectional analysis, experimental study, or secondary study with data of interest; and (3) were peer reviewed. Studies focusing on violence by individuals other than current or former intimate partners were excluded. No restrictions were placed on the setting, time, or language of the studies. Quantitative data were not necessary for inclusion.

Search Strategy
The population of interest was women seeking or having undergone a TOP in any setting; the exposure was the presence or absence of IPV; the control group, where reported, was a separate cohort of comparable individuals (e.g., pregnant women not seeking TOP, pregnant women not reporting IPV, or women attending a gynaecology clinic); the statistic of interest was the association between IPV and TOP. We included articles where the relative timing of IPV and TOP could not be delineated. The search strategy was devised using a combination of Medical Subject Headings (MeSH terms) and free text terms with synonyms (see Table 1). Searches were carried out in Medline Nine authors were contacted regarding results, and they identified further articles. There was no restriction on language. If multiple  articles based on the same study were identified, duplication was avoided by only using the data reported for different subgroups.

Data Extraction
All titles were independently screened by two authors (M. H. and S. B.). If either considered a title relevant, both reviewers independently screened the abstract. All articles were included for full-text assessment if either author considered the abstract relevant or there was uncertainty. Full-text assessment to determine inclusion was independently carried out by two authors (M. H. and S. B.). Any disagreements were discussed, and any study whose inclusion remained ambivalent was referred to a third author (L. C. C.).
A standard form was devised (see Table S1) prior to data extraction and quality scoring. Data were extracted on study type; population; setting, country, and region; demographic and health factors; intervention; comparator population; definition of IPV; screening tools used; and incidence and prevalence of IPV among general populations and in relation to TOP. Data from the articles were independently extracted by two authors (M. H. and B. L. P.). Results addressing violence by individuals other than current or previous intimate partners were excluded. Any uncertainties were discussed, and referred to a third author (L. C. C.) if necessary. Data extraction and quality scoring of articles published in languages other than English (n = 3) were undertaken by one author (M. H.) and the translator.

Quality Appraisal
Quality appraisal of quantitative and qualitative studies was carried out using Critical Appraisal Skills Programme (CASP) scales [19] as modified by Oram and colleagues [20] (see Tables S2 and S3), consisting of 15 and ten criteria, respectively, each of which could be scored between zero and two (maximum scores 30 and 20). Two authors (M. H. and B. L. P.) carried out full quality appraisal of all articles. Any disagreements were discussed, and referred to a third author (L. C. C.) if necessary. Quantitative CASP scores of $25, $20, $15, and #14 and qualitative CASP scores of $17, $14, $12, and # 11 were considered high, medium, low, and very low quality, respectively.

Meta-Analysis and Regression
The prevalences of IPV (as percentages) were converted to log odds prior to combining using the DerSimonian and Laird random effects method of meta-analysis [21]. Resulting estimates and confidence intervals were reconverted to percentages prior to display. Forest plots show actual percentages on the log-odds scale and are displayed sub-grouped by gross national income (GNI) per capita, as previous reports (e.g., from WHO) have chosen similar economic groupings [22]. Comparisons between groups used odds ratios (ORs). Estimated heterogeneity (I 2 ) is displayed for all group and sub-group analyses. We investigated the possibility of a sub-group of recent or high-quality studies with consistent methods and consistent results that could be used to give generally applicable results.
Meta-analysis regression is a meta-analysis technique developed specifically to explain large and unexplained differences in results between studies (also known as heterogeneity). The method used assumes that the differences are in part random and in part explainable [23].
Potential sources of heterogeneity were investigated using metaanalysis regression: country's GNI per capita (in intervals of INT$10,000), study quality (measured by CASP score), date of study (decade), study design (cross-sectional, cohort, or casecontrol), setting (urban versus regional versus national), and study size (total number of participants as a continuous variable). Egger's test was performed in order to assess potential publication bias [24].
The review was performed according to protocol (Text S1) and in line with PRISMA guidelines (Table S4).

Study Selection Process
The study selection process is shown in Figure 1. Of a total of 438 articles identified for screening after removal of duplicates, 104 were considered eligible for full-text screening, and 74 were  Lifetime history of physical, emotional, and sexual abuse was significantly higher among the TOP-seeking population than the control group. The same was also found to be true of the past year history of physical or sexual violence, and sexual violence alone (all p,0.001). Past year history of physical violence alone was also significantly higher in the TOP group (p,0.01). There was no significant difference in numbers of women living in fear among the two groups. Boyfriends were more likely to carry out physical violence than husbands, although the two groups were equally likely to perpetrate sexual violence. included in the review . Tables 2-9 detail the studies, grouped into tables by design-and additionally by continent for cross-sectional studies-and listed within tables by reverse chronology.

Key Features of Studies
The publication dates of the studies ranged from 1985 to 2013, with the majority (67/74, 91%) having been reported since 2000. Sixty-eight studies included quantitative data, with six being exclusively qualitative. The majority of quantitative studies were cross-sectional (57), with the remaining being cohort (three) and case-control (eight, of which five used a pregnant and three used a non-pregnant comparator group). Geographically, the study locations spanned six continents: North America (35), Asia (12), Europe (10), Africa (8), Australasia (6), and South America (2), with one further study analysing data from several continents. Women were asked about IPV in a variety of settings, including home, gynaecology wards, termination clinics, and other specialist medical clinics, and in various ways, including telephone or written questionnaire and national scale surveys. Among the quantitative studies there were 10 high, 31 medium, 19 low, and eight very low quality reports (mean score 20.2, standard deviation 4.0). Of the qualitative studies two were medium, one low, and three very low quality (mean score 11.5, standard deviation 2.2). The majority of studies analysed exclusively women (69), with few analysing exclusively men (two) or both men and women (three). Sample sizes ranged from eight to 33,385 participants (eight, 30, and 36 studies had ,100, 100-999, and $1,000 participants, respectively; median number of participants 942, interquartile range 208 to 2,391). The exposures included physical violence (53), sexual violence (47), and emotional violence (19), with many articles looking at combinations (42). Table S5 shows the associations of clinical and demographic factors with IPV in women seeking TOP.

Prevalence and Meta-Analysis of Lifetime Prevalence
Among women who underwent TOP, reported rates of IPV in the preceding year ranged from 2.5% [43] to 30% [76], while lifetime rates of IPV in this population varied from 14% [88] to 40% [59]. Meta-analysis of lifetime prevalence of IPV among TOP-seeking populations was found to be 24.9% (95% CI 19.9% to 30.6%), as shown in Figure 2. Following systematic metaanalysis regression, we found that the high level of variation between studies was not explicable by study quality, type, or size, or country GNI per capita (Table S6). The association between IPV and TOP is shown in Figure 3; variation is not explained by GNI per capita. I 2 values for heterogeneity were high, as expected when pooling data with proportions.

Meta-Analysis of Risk Factors
Meta-analysis was undertaken for four factors where data were available (woman being single versus married, partner not knowing about the TOP, partner support for the TOP, and previous TOP), but was not possible for others, as the definitions of both IPV and the risk factors varied, or there was a lack of numerical data or only a single study in many cases. Figure 4 shows no association (pooled OR 2.97, 95% CI 2.39 to 3.69) between being single and IPV among a TOP-seeking population. There was an association (pooled OR 2.32, 95% CI 2.00 to 2.69) between partner not knowing about the TOP and IPV ( Figure 5), but no association between partner support for the TOP and IPV (pooled OR 1.37, 95% CI 0.82 to 2.30) ( Figure 6). Meta-analysis of three studies reporting on the association between IPV and previous TOP showed no significant association (pooled OR 1.42, 95% CI 0.85 to 2.36) (Figure 7), but five of six further studies reported individual significant associations, though with inadequate quantitative data to include in the meta-analysis. Table S5 shows the associations of clinical and demographic factors with IPV in women seeking TOP.

Associations with Reproductive and Pregnancy Factors
Past obstetric history. Nine studies showed that women who reported IPV were more likely than the comparator group to have a history of multiple TOPs (Tables 3-5) [27,28,38,41,42,44,51,60,65]. The highest quality study found that women presenting for a third TOP were over two and half times more likely to have a history of physical or sexual violence than women presenting for their first [51]. There was no significant association between number of pregnancies and IPV (gravidity in group of women reporting IPV: mean 3.2, standard deviation 2.0; among women not reporting IPV: mean 3.0, standard deviation 2.0) [41], although one medium quality study found a higher odds of history of previous miscarriage among women reporting IPV at a termination clinic, as compared to those not reporting IPV (OR 1.6, 95% CI 1.1-2.2) [41].
Coercion, contraception, and conception. One study investigating pregnancy intention found that women in violent relationships were more likely to say that the pregnancy ''had been imposed'' upon them by their partner (13% versus 2% for women Beaten or abused by partner.
1,246 women presenting for TOP who did not report a history of sexual coercion.
Women who reported sexual coercion were also more likely to report being abused or beaten (p,0.01 in both cases). 18 Yimin et al. 2001 [35] 667 women presenting for TOP who reported a history of sexual coercion; China.
Abused or experienced battery at hands of partner.
726 women presenting for TOP who did not report a history of sexual coercion.
Women who reported sexual coercion were also more likely to report abuse or battery (p,0.01 for both).    Victims of violence were more likely to report a history of TOP than those without violence in history (p,0.05).
Lifetime, recent, and current pregnancy physical violence. Forced sex in relation to conception of current pregnancy and first intercourse. 39.5% respondents identified themselves as having a history of IPV. Women reporting IPV history significantly more likely not to have told their partner about their pregnancy than control group (p = 0.02). They were also more likely not to involve the partner in the decision to have a TOP (p,0.01). 21 not in physically or sexually violent relationship) [29]. A further two studies found that pregnancy associated with sexual coercion and ending in TOP was also 1.8-to 3.8-fold more likely to be associated with IPV [34,35]. One study found that women in violent relationships were significantly more likely to report going without contraceptives compared to women not reporting IPV (Table 3) [28], with another study finding that they were more likely to use them (Table 4) [38]. In four studies, of which two were very low quality, women reported being actively prevented from obtaining contraception by their partner, or that their partner was refusing or deceiving them about birth control use (Tables 3 and 4) [28,38,45,52]. It was reported that a partner preventing access to contraception led to concealed use of contraceptives among some women [45]. One cross-sectional population telephone survey of over 4,000 women found that over 46% of 616 ''completed rapes'' were perpetrated by a husband or boyfriend, and that 50% of 20 raperelated pregnancies ended in TOP [61]. One low quality study of women who migrated as adults from the Indian subcontinent to the US and had a history of seeking sex selection services found that a third of 65 women described past physical abuse and neglect related specifically to not producing a male child [94].

Factors Relating to Termination of Pregnancy
Decision-making. Women reporting IPV were more likely to report an unwanted pregnancy (7.4% of 163 women not reporting IPV versus 23.3% of 44 women reporting IPV) [53]. Additionally, a high quality study, though with small numbers, found that women attending a termination clinic with a planned pregnancy were more likely to report IPV (50% of 12 women) than those who did not plan their pregnancy (5.6% of 337 women) [30]. There was evidence from very low quality studies that some women felt coerced by their violent partner into having a TOP (Table 4) [45,52]. One medium quality study found that 2% of 1,215 women in a termination clinic reported being forced into the decision by their partner [88]. One small, very low quality study of 38 women participating in a hospitalbased IPV programme found that 18% reported feeling ''pressured'' into TOP and 5% were forced into undergoing the procedure [52].
There was consistent evidence that women in violent relationships were more likely not to tell their partner about their decision to terminate (pooled OR 2.97, 95% CI 2.39 to 3.69; Figure 5) [38,39,54]. Women with an IPV history were less likely to have their partner fund their TOP (27% of 25 women reporting IPV had their partner pay for their TOP versus 63% of 155 women not reporting IPV) [38].
Gestation and method of termination of pregnancy. Two studies, one of high and one of low quality, failed to find any association between IPV and gestation at TOP [54,57], although one lower quality study reported that women having a second trimester TOP were more likely to report a history of forced sex than women having a first trimester TOP (35.3% of 410 versus 11.5% of 139), [84], and another found that women later in  [41]. Psychosocial problems. In women who had undergone TOP, there was a significant association between reported IPV and psychosocial problems including depression [38,47], suicidal ideation [47], stress [38,47], and disturbing thoughts [47], although the temporal relationships were unclear (Table 4) [38,47]. No studies identified the subsequent impact of IPV and TOP on the woman, her partner, or their relationship.
Disclosure and intervention. Five studies examined disclosure in termination clinics: IPV questionnaires were highly acceptable [90], although non-responding women differed from those who responded and had undergone more TOPs [41]. Women in violent relationships were as likely to attend for followup (52.2% of 413 non-abused women defaulted versus 46.6% of 88 abused women) [33] and more likely to know about community resources for IPV (80% of 16 women reporting IPV versus 67% of 35 women not reporting IPV) [60]. Some women reported events that would meet the definition of IPV but did not identify themselves as experiencing IPV [98]. Nevertheless, many women wished to talk about IPV with regard to further management or intervention [33], with some citing their doctor as the main source of information [60]. However, during a period of universal screening only 51% of 499 women were asked about IPV; certain sub-groups of women were more likely to be asked about IPV (e.g.,

Demographic Factors
Female factors. The majority of studies focused on female factors. The results relating to impact of age on the association between IPV and TOP were discrepant, and interpretation is hindered by different age groups being used. One paper reported that past year incidence of IPV was higher among women under the age of 20 y seeking TOP (50.0%) than among those 20 y or over (26.9%); however, the converse [29,30] or no significant association (Tables 4, 6, and 7) [41,57,75,88] has also been reported.
There was contradictory evidence for variance between women of different race or ethnicity, with two studies finding converse associations; one found that significantly fewer white Caucasian women (12% of 160 women) reported IPV as compared to nonwhite women (21% of 94 women, p = 0.003) [57]; another study of women seeking elective pregnancy termination reported that a greater proportion of white women (48% of 226) had experienced IPV compared to black women (31% of 223, p = 0.001) [59]. This finding may not be generalisable to other locations. No significant associations were found with women's level of education in two studies [41,88] or with women's income [41]. Three studies reported no significant association with employment status [41,69,88], whilst two suggested that unemployed women or non-skilled labourers were more likely to report both IPV and TOP (25.3% of 291 women seeking TOP and reporting IPV were unemployed versus 19.1% of 1,096 women seeking TOP and not reporting IPV) [28,33].
Three studies of similar medium quality assessed drug and alcohol use; one study found that, compared to women not reporting IPV, women in a termination clinic who reported IPV were also more likely to smoke (25.6% of 117 women experiencing IPV versus 11.0% of 1,497 non-abused women) and drink (12.8% of women experiencing IPV versus 4.9% of non-abused women) [87], but other studies found no significant association [88] or reported an association only anecdotally [59].
An association between negative physical quality of life scores and IPV and TOP was found in one medium quality study (Table 7) [87].
Relationship factors. Five studies, one of high and four of medium quality, found that, compared to married women, those who were single, separated, divorced, or widowed were more likely to have a history of IPV and TOP (Tables 3, 4, and 7) [30,41,43,87,91], though one medium quality study found no significant association [88], and one further study of women experiencing IPV found married women were more likely to report a TOP (89% of 23,909 married women compared to 11% of 9,408 unmarried women) [73]. Women who reported difficulties in their relationship were also more likely to report IPV when asked at a termination clinic in all studies (with one study finding that 7.7% of 350 women undergoing a TOP reported difficulties in their relationship, as opposed to 1.8% of 653 women continuing pregnancy) [30,38,59]. One study found no significant association between prevalence of IPV and household income (27.2% of 669 Physical and/or sexual IPV perpetrated by husband, a man that the woman had lived with, or current, regular male sexual partner. Controlling for other variables, women who had experienced IPV were 2.5 times more likely to report a TOP than those who had never experienced IPV (21.4% versus 9.9%, p,0.0001).

24
Whitehead and Fanslow 2005 [90] 125 women who agreed to see a social worker whilst attending an abortion clinic; New Zealand.
Lifetime and past year histories of physical abuse and sexual abuse (forced or pressured into having sex).
Reported lifetime prevalence of physical or sexual abuse was 50.8%. 69% of women who reported a lifetime history of physical abuse also reported that their partner/father of pregnancy was a perpetrator of their abuse. 42% reported that a family member was responsible.
18 Taft et al. 2004 [91] Cohort of Australian women aged 18-23 y who are part of a long-term longitudinal study (n = 14,784). Survey 1 from prospective cohort study; Australia.
1-y history of physical and/or sexual violence. Lifetime history of violent relationship.
Lifetime partner violence was strongly associated with the following pregnancy outcomes: miscarriage and TOP (p,0.001) and birth, miscarriage, and TOP (p = 0.05). Recent partner violence was associated with the second of these outcomes (p,0.001). Neither situation was significantly related to TOP alone. Women reporting IPV were less likely to be married than single (OR 0.70, 95% CI 0.59 to 0.83), and more likely to be separated, divorced, or widowed than single (OR 2.62, 95% CI 1.78 to 3.86). 20 Webster et al. 1996 [92] 1,014 women seeking pregnancy care in Brisbane hospital; Australia.
Historic abuse (victim was .16 y old but abuse ended before current pregnancy began) and current pregnancy abuse. Physical, emotional, and sexual abuse studied.
29.7% of women disclosed past or present abuse history. Women reporting abuse were significantly more likely to have had a previous TOP than women who reported no history of violence (p = 0.0034). women reporting TOP were in the lowest income bracket, as compared to 26.5% of 139 women not reporting TOP) [54]. Male factors. One article studied men attending community health centres and identified that men who admitted to IPV perpetration were more likely to report having ''been involved in a pregnancy'' that ended in TOP (48.9% of 188 men reporting IPV perpetration also reported involvement in TOP, as opposed to 22.7% of 402 men not reporting IPV perpetration) [44]. Another study of partners of women having a first or subsequent TOP found higher rates of IPV experienced by the men involved in a second or more TOP (12% of 188 men whose partner was having a second or subsequent TOP reported ''being a victim'' of IPV versus 6% of men whose partner was having a first TOP) [66]. One low quality study noted that 36% of 16 women reporting IPV stated that their perpetrating partner had been the victim of abuse as a child [60].
For IPV in women presenting for TOP, there was no evidence of bias relating to non-publication of small non-significant studies; Egger's test [24] for small-study effects was non-significant (p = 0.13). These associations, the uncertainties, and gaps in knowledge are shown diagrammatically in Figure 8.
There were no intervention studies.

Summary of Main Findings
The literature is extensive, but variable in quality, and largely focused on female factors. High rates of physical, sexual, and emotional IPV were found across six continents among women seeking a TOP. According to meta-analysis, partner not knowing about the TOP was associated with IPV among women seeking TOP. However, lack of partner support for TOP is not associated with IPV among women seeking TOP. The literature also suggests that women in abusive relationships were more likely to report inability to make autonomous contraceptive choices, partner contraceptive sabotage, and sexual violence, and they were less likely to have informed their partner about the pregnancy or involved him in decision-making about it. IPV was cited as a reason for wanting TOP, and rape-related pregnancy had a particularly high chance of leading to TOP. The meta-analysis did not find that multiple TOPs were associated with IPV, but its Forced sex, or persuasion into having sex.
Those who reported that their pregnancy was the result of an incident of forced sex often reported fear of, as well as real, family violence.
14 Puri et al. 2011 [94] 65 women who had migrated from the Indian subcontinent to the US at the age of 18 y or greater, and had a history of seeking sex selection services; US.
Any marital violence related to fetal sex and/or sex selective TOP.
62% of women described verbal abuse from their female in-laws or husband; one-third described past physical abuse and neglect related specifically to their failing to produce a male child. credibility is undermined by both the number of studies that did (albeit unquantified) and the association reported by men who stated they had been perpetrators of IPV. Although it was not always determined whether experiencing IPV was a determining factor in the decision to end, rather than continue, a pregnancy, the findings support the concept that violence can sometimes lead to an initial pregnancy (via coercion, rape, sexual assault, or contraceptive sabotage) and to a subsequent TOP (via coercion).
There was a lack of data regarding long-term outcomes for women in violent relationships who underwent TOP, but associations with repeat TOP (and possibly miscarriage) lend support to the notion of a repetitive cycle of abuse and pregnancy. Not informing the male partner may then be explicable as a reason to avoid partner involvement or further abuse.

Strengths and Limitations
Strengths of the review and meta-analysis include use of multiple databases, no language restriction, hand searching of reference lists, double data entry, and quality assessment of both quantitative and qualitative studies from a wide variety of settings, thus improving reliability and generalisability. Limitations include unexplained large differences in both prevalences and odds ratios between studies (heterogeneity), likelihood of underreporting of both IPV [99] and TOP (particularly when both are stigmatised) in the primary data sources, and inherent difficulties in validation. There is potential publication bias towards research showing a positive relationship. The meta-analysis may be biased, as only those studies quantitatively reporting a significant result could be included, whilst those stating an association (without providing data for inclusion) had to be excluded. It was not possible to determine the legality of, or access to, TOP for each country or state at the specific time point of study; thus, the analysis has not included evaluation of such barriers. It was not possible to determine temporal relationships and patterns of abuse, pregnancy, and TOP.

Comparison with Other Studies
Only one review has previously examined the association between IPV and TOP within a study of broader sexual health issues [100], concluding that TOP and repeat TOP were associated with IPV, but without reporting on other associations. The review was singleauthored, lacked systematic analysis, and included only eight studies. Other relevant systematic reviews and observational studies on IPV have included women with ongoing pregnancies, women with pregnancy loss, or all women of reproductive age, noting poorer mental, physical, and pregnancy outcomes for the women Figure 3. Associations between intimate partner violence and termination of pregnancy grouped by country's gross national income per capita (in intervals of Int$10,000) and with setting (urban, regional, or national) given. Weights are from random effects analysis. D+L, combined effects using the DerSimonian and Laird [21] random effects method; I-V, combined effects using the inverse variance fixed effects method. Countries grouped by GNI (shown in parentheses). doi:10.1371/journal.pmed.1001581.g003    experiencing IPV [101][102][103][104]. One review found an association between presence of depressive or anxiety disorders in women and increased likelihood of IPV compared to women without mental disorders, but the direction of causality could not be determined, as few studies were longitudinal [104]. IPV was also associated with physical injury: presentation to hospital accident and emergency departments with unwitnessed head, neck, or facial injuries was a significant marker for IPV [103]. A large observational study Figure 8. Matrix of associations between domestic violence and termination of pregnancy. Key to associations: red, associations metaanalysed; amber, associations not meta-analysed but shown in literature; green, no significant association described in the literature; grey, not studied. doi:10.1371/journal.pmed.1001581.g008 conducted by WHO found that women who reported a history of partner violence were more likely to report physical and mental health problems including emotional distress and suicidal thoughts, as well as difficulties with activities of daily living [101]. Comparison groups that might prove useful would be women with ongoing but unplanned pregnancy (no specific studies available) and women with ongoing wanted pregnancy. In ongoing pregnancies, IPV was associated with 1.5-fold increased risks of low birth weight and preterm birth [102]. A large observational study has reported poorer maternal outcomes, including hypertension, renal and urinary tract infection, and vaginal bleeding, with IPV [8].

Implications and Clinical Relevance
Health-care professionals should be aware of the high rates of physical, sexual, and emotional violence among women seeking TOP, and particularly the clinical factors associated with greatest risk: previous TOP, lack of contraception, initially planned pregnancy, ultrasound redating, and the partner not funding or not being told about the TOP. There are potential associations for IPV with young age, marital status, ethnicity, and low household income. IPV compromises both the safety and health of the woman requesting the TOP, and potentially that of her partner and any existing children if a woman retaliates or children witness or experience the violence directly. In attempting to prevent repeat TOP [105], a narrow focus, especially on long-acting contraception, that excludes addressing the wider needs of a woman in a violent relationship might leave a woman less likely to become pregnant but just as vulnerable to IPV. Good practice obligates that termination services should have robust policies for ensuring women's safety and confidentiality, providing information and referral pathways for those who disclose IPV, and exemplar guidance exists [16].
Some groups have evaluated whether screening for IPV is justified in selected populations of women. Theoretically, early identification and effective intervention for violence may reduce repeat unintended pregnancy and TOP, as well as improve longer term health outcomes. Three systematic reviews [106][107][108] have concluded that screening is warranted, leading the US Preventative Task Force to recommend that clinicians should screen women of childbearing age for interpersonal violence, such as IPV, and refer women who screen positive to intervention services [109]. The UK National Institute for Health and Care Excellence is currently undertaking a consultation regarding guidance on identification and prevention of IPV [110]. However, interventions in this area are inherently complex and difficult to research [111], and evidence for the effectiveness of counselling intervention programmes or other interventions remains limited [106,112,113]. Of all approaches evaluated, intensive advocacy (aiming to provide women with information and support to facilitate access to community resources) appears the most promising in reducing physical abuse 1-2 y after the intervention, but impact on quality of life or mental health is not proven [112]. The majority of interventions studied have focused on the female experiencing IPV, but some have undertaken theoretical analysis of models in which changes in the behaviours of the male perpetrators are included [114]. A comparative study of promising health-based IPV interventions in primary care and maternity services across Europe found that key implementation issues of IPV interventions included clinical champions, leadership roles, funded and coordinated multiagency partnerships with clear referral pathways, multidisciplinary and participant feedback, and evaluation of outcomes [115]. Interventions as part of the pathway for women seeking TOP require further consideration.

Future Research
No study we found set out to examine the association or temporal relationships between IPV and TOP, which would require (at a minimum) including women with and without an IPV history and with and without a history of TOP. There is extremely limited information about male partners of women seeking TOP, as perpetrators or as experiencing IPV, and which male-related factors contribute to increased likelihood of IPV. Greater information is required on long-term outcomes of violence and TOP on both partners. The findings of pregnancy ''concealment'' and higher rates of murder and suicide with IPV [116,117] mean that researchers must be cautious and aware of women's safety. Harms have been identified following healthbased IPV interventions, such as breaches of confidentiality [115]. Therefore, a public health approach that does not focus solely on the woman (either as ''problem'' or ''solution'') or health services should be considered, for example, using educational, social norm, and/or criminal justice interventions. Nevertheless, given the clear associations, termination services provide an appropriate setting in which to assess screening for, or give information about, IPV, whether pre-or post-TOP, and for offering an intervention that women desire, such as a ''one stop'' offer of referral to specialist IPV services, especially in view of low return to clinics for follow-up [118]. Given that routine identification of women experiencing IPV and provision of a standard intervention has recently been shown to have no impact on quality of life or mental well-being, there is now a need for considering new strategies, including alternative intervention models and targeting perpetrators as well as the women affected [119][120][121]. On the basis of this review, research into the suitability, acceptability, and design of an intervention programme is justified, and should be tested preferably in a randomised control trial. Any legal barriers to intervention and reporting, such as criminalisation of TOP, should also be investigated and described.

Conclusion
IPV is associated with TOP. Novel public health approaches are required to address IPV against women and repeat TOP. Termination services provide an opportune health-based setting in which to design and test interventions at the individual level.   Editors' Summary Background. Intimate partner violence (sometimes referred to as domestic violence) is one of the commonest forms of violence against women and is a global health problem. The World Health Organization defines intimate partner violence as any act of physical, psychological, or sexual aggression or any controlling behavior (for example, restriction of access to assistance) perpetrated by the woman's current or past intimate partner. Although men also experience it, intimate partner violence is overwhelmingly experienced by women, particularly when repeated or severe. Studies indicate that the prevalence (the percentage of a population affected by a condition) of intimate partner violence varies widely within and between countries: the prevalence of intimate partner violence among women ranges from 15% in Japan to 71% in Ethiopia, and the lifetime prevalence of rape (forced sex) within intimate relationships ranges from 5.9% to 42% across the world, for example. Overall, a third of women experience intimate partner violence at some time during their lifetimes. The health consequences of such violence include physical injury, depression, suicidal behavior, and gastrointestinal disorders.

Supporting Information
Why Was This Study Done? Intimate partner violence can also lead to gynecological disorders (conditions affecting the female reproductive organs), unwanted pregnancy, premature labour and birth, and sexually transmitted infections. Because violence may begin or intensify during pregnancy, some countries recommend routine questioning about intimate partner violence during antenatal care. However, women seeking termination of pregnancy (induced abortion) are not routinely asked about intimate partner violence. Every year, many women worldwide terminate a pregnancy. Nearly half of these terminations are unsafe, and complications arising from unsafe abortions are responsible for more than 10% of maternal deaths (deaths from pregnancy or childbirth-related complications). It is important to know whether intimate partner violence and termination of pregnancy are associated in order to develop effective strategies to deal with both these global health concerns. Here, the researchers conducted a systematic review and meta-analysis to investigate the associations between intimate partner violence and termination or pregnancy. A systematic review identifies all the research on a given topic using predefined criteria; meta-analysis combines the results of several studies.
What Did the Researchers Do and Find? The researchers identified 74 studies that provided information about experiences of intimate partner violence among women who had had a termination of pregnancy. Data in these studies indicated that, worldwide, intimate partner violence rates among women undergoing termination ranged from 2.5% to 30% in the preceding year and from 14% to 40% over their lifetime. In the meta-analysis, the lifetime prevalence of intimate partner violence was 24.9% among termination-seeking populations. The identified studies provided evidence that intimate partner violence was associated with termination and with repeat termination. In one study, for example, women presenting for a third termination were more than two and a half times more likely to have a history of physical or sexual violence than women presenting for their first termination. Moreover, according to the meta-analysis, women in violent relationships were three times as likely to conceal a termination from their partner as women in non-violent relationships. Finally, the studies indicated that women undergoing terminations of pregnancy welcomed the opportunity to disclose their experiences of intimate partner violence and to be offered help.
What Do These Findings Mean? These findings indicate that intimate partner violence is associated with termination of pregnancy and that a woman's partner not knowing about the termination is a risk factor for intimate partner violence among women seeking termination. Overall, the researchers' findings support the concept that violence can lead to pregnancy and to subsequent termination of pregnancy, and that there may be a repetitive cycle of abuse and pregnancy. The accuracy of these findings is limited by heterogeneity (variability) among the included studies, by the likelihood of underreporting of both intimate partner violence and termination in the included studies, and by lack of validation of reports of violence through, for example, police reports. Nevertheless, health-care professionals should consider the possibility that women seeking termination of pregnancy may be experiencing intimate partner violence. In trying to prevent repeat terminations, health-care professionals should be aware that while focusing on preventing conception may reduce the chances of a woman becoming pregnant, she may still be vulnerable to abuse. Finally, given the clear associations between intimate partner violence and termination of pregnancy, the researchers suggest that termination services represent an appropriate setting in which to test interventions designed to reduce intimate partner violence.