Intimate partner violence against adult women and its association with major depressive disorder, depressive symptoms and postpartum depression: systematic review and meta-analysis

Introduction

Intimate partner violence (IPV) against women is a major public health concern in the United States and worldwide. IPV is often described as a specific type of family or domestic violence, in which the perpetrator is a current or former intimate partner of the victim. The Centers for Disease Control and Prevention (CDC) defines IPV as “physical violence, sexual violence, threats of physical/sexual violence, and psychological/emotional abuse perpetrated by a current or former spouse, common-law spouse, non-marital dating partners, or boyfriends/girlfriends of the same or opposite sex” (Chang, Cluss, Ranieri, Hawker, Buranosky, Dado et al., 2005).

IPV is highly prevalent among women in the general population as well as those identified in clinical settings (Coker, Smith, & Fadden, 2005; Kramer, Lorenzon, & Mueller, 2004; Magnussen, Shoultz, Oneha, Hla, Brees-Saunders, Akamine et al., 2004; Peralta & Fleming, 2003; Ross, Walther, & Epstein, 2004). Based on national surveys, 25–30% of women in the United States have reported physical and/or sexual abuse by an intimate partner during their lifetime, whereas 2–12% report physical and/or sexual IPV over the past year (Haggerty & Goodman, 2003; Lipsky, Holt, Easterling, & Critchlow, 2004). Based on clinical studies, the prevalence estimates of physical, sexual and/or emotional abuse against women are 21–55% over a lifetime and 4–44% in the past year (Breiding, Black, & Ryan, 2008; P. Tjaden, Thoennes, N., 2000; P. Tjaden, Thoennes, N., 1998a; P. Tjaden, Thoennes, N., 1998b).

IPV is associated with a wide range of short-term and long-term physical and mental health sequelae. Victims of IPV were shown to be at an increased risk for injury (Burke, Thieman, Gielen, O'Campo, & McDonnell, 2005; Campbell, 2002), disability (Coker et al., 2000a; Coker et al., 2005), chronic pain (Burke et al., 2005; Campbell, 2002; Koopman, Ismailji, Holmes, Classen, Palesh, & Wales, 2005; Kramer et al., 2004), arthritis (Coker et al., 2000a), headaches or migraine (Coker et al., 2000a; Kramer et al., 2004), gastrointestinal signs (Burke et al., 2005; Campbell, 2002), vaginal bleeding and sexually transmitted infections (Burke et al., 2005; Campbell, 2002; Kramer et al., 2004), substance use and abuse (Campbell, 2002; Fals-Stewart & Kennedy, 2005), social dysfunction (Burke et al., 2005; Campbell, 2002), insomnia (Burke et al., 2005; Campbell, 2002), post-traumatic stress disorder (Bradley, Schwartz, & Kaslow, 2005; Koopman et al., 2005; Woods, Page, O'Campo, Pugh, Ford, & Campbell, 2005; Woods, 2005), anxiety (Burke et al., 2005; Campbell, 2002), depression (Al-Modallal, Peden, & Anderson, 2008; Golding, 1999) and suicidal thoughts (Kramer et al., 2004; Meadows, Kaslow, Thompson, & Jurkovic, 2005).

The salience of IPV as a risk factor for mental health problems, in general, and depression, in particular, requires further evaluation. To date, few systematic reviews of observational studies have been conducted to comprehensively evaluate the co-morbidity of IPV and depression in women (Al-Modallal, Peden, & Anderson, 2008; Golding, 1999). In addition, no recent meta-analyses have been conducted to estimate the magnitude of the association between IPV and specific types of depressive outcomes. In 1999, Golding published a meta-analysis to assess the prevalence of mental health problems among women with a history of intimate partner violence. The prevalence of mental health problems among women experiencing IPV was 47.6% in 18 studies of depression, 17.9% in 13 studies of suicidality, 63.8% in 11 studies of posttraumatic stress disorder, 18.5% in 10 studies of alcohol abuse, and 8.9% in 4 studies of drug abuse. Odds ratios (OR) representing associations of these mental health problems with IPV ranged from 3.55–5.62, and were generally consistent across studies (Golding, 1999). In Golding’s meta-analysis, no distinction was made regarding specific types of depression; the estimated OR (95% confidence interval) for the association between IPV and depression was 3.80 (3.16–4.57) (Golding, 1999). In this systematic review and meta-analysis, we conducted a PubMed (1980–2010) search of population-based and clinical studies focusing on pregnant and non-pregnant women to summarize the extant literature and estimate the magnitude of the association between IPV and key depressive outcomes (elevated depressive symptoms (EDS) detected by various screening instruments, diagnosed major depressive disorder (MDD) and postpartum depression (PPD)).

Materials and Methods

Results

Table 1 characterizes the 37 studies included in our systematic review and meta-analysis. Although our literature search spanned publication years of 1980 until 2010, relevant studies which we finally selected were published between the years 2000 and 2010. Twenty-four of the 37 studies were conducted in the United States; 32 were cross-sectional studies while only five were prospective or retrospective cohort studies. Sample sizes ranged between 101 and 7154 surveyed participants, with about 60% of selected studies having a sample size greater than 1000. The minimum in the age range reported for each study was 18 years or more in 27 studies and <18 years in 10 studies. Most studies were population-based (24 of 37), while the remaining 13 were conducted in a clinical setting (e.g. hospital waiting room). In 17 out of 37 selected studies, the primary aim was to examine the relationship between IPV and depression.

For the depression outcome, the majority of studies examined EDS (n=25), whereas only seven examined PPD and five examined MDD. Seven of the 25 studies with EDS as outcome used the 20-item CES-D as the screening tool with a cut-point of 16 commonly used to reflect minor depressive symptoms (Radloff, 1977). However, other screening tools were also used commonly including the shorter versions of the CES-D. Some examples were the Patient Health Questionnaire (PHQ-9) (Spitzer, Williams, Kroenke, Hornyak, & McMurray, 2000) and its shorter version (n=2) (Berg, 2002; Corson, Gerrity, & Dobscha, 2004; Li, Friedman, Conwell, & Fiscella, 2007; Lowe, Kroenke, & Grafe, 2005; Whooley, Avins, Miranda, & Browner, 1997), Beck Depression Inventory (BDI) (Beck, Steer and Brown, 2000) and its shorter version (n=5) (Shaver, 1991) and the General Health Questionnaire (GHQ) (n=1) (Goldberg, 1972). It is worth noting that the latter study used a 1-question item, which is problematic due to the high proportion of false positives identified using only this item (Goldberg, 1972). For PPD, the most commonly used instrument was the Edinburgh postpartum depression scale (EPDS) (Cox, Holden, & Sagovsky, 1987) (n=4), whereas diagnosis of MDD was conducted using criteria from the International Classification of Diseases (ICD-9 or 10) (WHO, 1992) (n=2) or a combination of Diagnostic and Statistical Manual (DSM-IV) (Kessler & Ustun, 2004) and Composite International Diagnostic Interview (CIDI) (APA, 1994) (n=3). The mean prevalence rates of MDD, EDS and PPD from the selected studies were 11.4% (range: 4.8%, 24.1%), 26.1% (range: 7.0%, 75.0%) and 30.7% (range: 7.5%, 89.8%), respectively.

Although IPV was measured differently across studies, there were commonalities in screening tools among the selected studies. For instance, the AAS (McFarlane et al., 1992), a commonly used tool for the assessment of IPV, was used in seven of the 37 studies, while the CTS (Form R) was used in six studies (Straus, 1990). It is worth noting, however, that even if several studies have relied on the same IPV screening tool, some studies may have used all questions, while others may have used sub-scales to reflect specific IPV sub-types, namely physical, sexual and/or emotional/psychological IPV. One or several data points for each study were entered mainly due to alternative and distinctive ways of defining IPV (e.g. emotional abuse vs. sexual abuse vs. physical abuse or combinations). Moreover, in studies where dose-response was assessed, all reported measures of association (RR) were considered as independent data points.

RR estimates were adjusted for socio-demographic and other covariates in most cases (25 out of 37 studies). Some of the most commonly controlled for covariates included age (28 out of 37), education (20 out of 37), income (16 out of 37), employment status (13 out of 37) and marital status (13 out of 37). Alcohol and drug dependence was also adjusted for in a number of studies (11 out of 37) and so was childhood sexual abuse (10 out of 37). Other forms of stressful events or types of violence were adjusted for in 9 out of 37 studies.

Discussion

In this study, we conducted a systematic review and a meta-analysis of published research articles (1980–2010) that examined the relationship between IPV and various outcomes related to depression, namely MDD, EDS and PPD. Of the selected 37 studies, most had a cross-sectional population-based design, were focused on EDS as the outcome of interest and were conducted in the United States. A large number of studies found a moderate or a strong positive association between IPV and depression. This meta-analysis suggested two to three-fold increased risk of MDD and 1.5 to 2-fold increased risk of EDS and PPD among women exposed to IPV as opposed to women not exposed to IPV. Finally, a sizable proportion (9–28%) of MDD, EDS and PPD may be attributed to lifetime exposure to IPV.

Careful interpretation of study findings should take into account the strengths and limitations of the studies selected as well as those of our meta-analysis. Whereas most of the selected studies had relatively large sample sizes, most had a cross-sectional design which does not allow for establishing a temporal relationship between IPV and depression. In fact, some studies (Lehrer et al., 2006) actually look at depression as a cause of IPV, not vice versa. Nevertheless, the few prospective cohort studies that were included in our meta-analysis yielded similar estimates of an IPV-depression association when compared to cross-sectional studies. The selected English-language studies were mostly relevant to Western societies and were heterogeneous in various respects, including study design, setting and measurement of exposure and outcome variables. Interpretation of results pertaining to EDS is complicated since the depression disease spectrum covered by different screening instruments varies from mild (not requiring intervention) to severe (requiring intervention). Although search, inclusion and exclusion criteria were clearly outlined, no further efforts were made to obtain additional research articles through cross-references, the “Related Articles” feature on PubMed or by consulting experts in the area. Finally, PubMed was the only database used and no terms synonymous to “intimate partner violence” and “depression” were applied.

To our knowledge, only two similar studies have been conducted so far (Al-Modallal et al., 2008; Golding, 1999). A recent study by El-Modallal and colleagues evaluated childhood and intimate partner physical abuse as risk factors for depressive symptoms among women (Al-Modallal et al., 2008). The scope of our study was greater for the types of exposures and outcomes that were examined (Al-Modallal et al., 2008). In addition, our study included both a systematic review and a meta-analysis. Consistent with our findings, the systematic review by El-Modallal and colleagues indicated an association between physical abuse experiences and depressive symptoms, although the association of other risk factors, including other types of abuse, with depressive symptoms confounded this relationship (Al-Modallal et al., 2008).

Given the global burden of depression on the female population, IPV prevention provides an opportunity for reducing the risk of MDD, EDS and PPD. Preventive efforts can be described as being either universal (population-based approach) or targeted (high-risk approach) interventions, depending on their scope and coverage (Stith, 2006b). Alternatively, primary, secondary and tertiary prevention programs have been designed to reduce the burden of IPV and its health sequelae on society by targeting the appropriate risk factors for IPV. The task of public health is mainly primary prevention (Last, 2001). In the context of IPV, primary prevention is a universal effort targeting entire communities and aimed at warding off maladaptive behaviors that could lead to IPV. Examples of primary prevention strategies include educational and media programs that target entire communities with the goal of preventing IPV before it is initiated (Stith, 2006a). School-based intervention programs initiated in teenage years have been shown to be effective at reducing dating violence, a precursor of IPV, even after four years of program implementation (Foshee, Benefield, Ennett, Bauman, & Suchindran, 2004). By contrast, secondary prevention is usually the task of preventive medicine (Last, 2001). When IPV is considered as the “disease” of interest, secondary prevention involves identification of “high-risk” populations to which future IPV interventions can be targeted. Examples include screening programs for the early detection and treatment of IPV within healthcare settings (Stith, 2006a). A systematic review by the United States Preventive Services Task Force concluded that there was insufficient evidence to warrant universal IPV screening in healthcare settings (Nelson et al., 2004), implying a need for further research in that area. Similar findings were reported by other researchers in a systematic review (Ramsay et al., 2002) and a randomized trial (MacMillan et al., 2009). Finally, tertiary prevention is mostly the task of rehabilitation (Last, 2001). In the realm of IPV, tertiary prevention involves various strategies that target individuals who have already experienced IPV in the past in order to reduce recurrence risk. Examples of tertiary prevention include “batterer treatment” and “shelter-stay” programs (Stith, 2006a).

In the United States, health care settings are used as venues for IPV screening even though evidence for or against this practice remains inconclusive (Nelson et al., 2004). Certifying organizations such as the Joint Commission on Accreditation on Healthcare Organizations and the CDC have recommendations on screening for IPV. Although there is no standard screening tool for IPV, recommendations have been made in educating all health care providers about identifying and discussing IPV with patients (CDC, 2007). Despite these recommendations, a recent study showed that the compliance for IPV screening in the healthcare setting is relatively poor (VDH, 2009). Acute care settings, such as emergency departments, will often employ a brief IPV screening in triage on all patients regardless of chief complaint. However, a recent study showed a lack of benefit in the use of a brief screening tool for IPV for preventing further violence (Koziol-McLain, Garrett, Fanslow, Hassall, Dobbs, Henare-Toka et al., 2010). Some researchers and agencies (e.g. Health Resources & Services Administration) have already noted a strong relationship of IPV prior to or during pregnancy with PPD, implying that screening during pregnancy can help identify patients with a history of IPV who may at high-risk for developing PPD. These patients may benefit from intervention and closer follow-up. In an effort to reduce the burden of depression and other IPV-related health consequences, continued research is recommended for evaluating various forms of IPV preventive strategies, at the primary, secondary and tertiary levels.

Supplementary Material