|Home | About | Journals | Submit | Contact Us | Français|
This study assessed the course of perinatal depression amongst 210 Latinas who were and were not affected by intimate partner violence (IPV) and identified associated psychosocial factors. Peak depression prevalence occurred prenatally among 45.7% of IPV-exposed and 24.6% of non-IPV-exposed Latinas. At each assessment, depression was significantly higher for IPV-exposed compared to non-IPV-exposed mothers. Mastery and social support were associated with lower depression, while history of IPV, perceived stress and avoidant coping behaviors were associated with higher depression. Findings support recommendations for routine depression and IPV screening of Latinas in perinatal clinical settings.
Intimate partner violence (IPV) is a public health problem that affects one in five couples in the U.S (Field & Caetano, 2005) and 3-13% of pregnancies (Campbell, 2002) IPV is associated with a wide array of physical and mental health consequences for women including injuries, adverse pregnancy outcomes, poor physical health, and depression (Campbell & Lewandoski, 1997; Bonomi et al., 2006; Woods, 2005; Coker, Sanderson, & Dong, 2004), in addition to mental and physical health problems among children exposed to IPV (Bair-Merritt, Blackstone, & Feudtner, 2006; Graham-Berman, 2003; McCloskey, Figueredo, & Koss, 1995; Rossman, 2001). Major Depressive Disorder (MDD) is one of the most prevalent mental health consequences for women with IPV histories. Compared to lifetime estimates for MDD of 17.1-21.3% among women (Blazer, Kessler, & McGonagle, 1994), approximately 47.6% of women with IPV histories have had MDD (Golding, 1999). While there is substantial evidence linking IPV and depression from cross-sectional studies, there is limited information on the persistence and longitudinal course of depression among pregnant and postpartum women exposed to IPV. Depression occurring during pregnancy and/or the postpartum period, referred to as “perinatal depression,” represents depressive symptoms that can occur prenatally and/or up to 12 months postpartum (Gaynes et al., 2005). Postpartum depression represents a continuum of depressive disorders appearing after delivery that range from “baby blues,” a temporary, self-limiting and generally untreated depressive disorder occurring in the first few weeks after delivery, to postpartum depression which can occur up to a year after delivery and generally requires treatment due to increased severity (Prevalence of self-reported postpartum, 2008).
Prenatal and postpartum depression have significant health consequences for both mother and offspring. Prenatal depression increases the risk of complications during pregnancy (Chung et al., 2001), and suicidality (Lindahl, Pearson, & Colpe, 2005), and postpartum depression (Yonkers et al., 2001; Lee et al., 2007). Postpartum depression has consequences for maternal role functioning, infant mental health, and longer term child development, manifesting its adverse effects over the child's life course and into adolescence. Maternal depression compromises the early interactions between mother and child and can lead to decreased sensitivity, attentiveness, and cognitive stimulation (Knitzer, Theberge, & Johnson, 2008; The NICHD Study of Early Child Care, 2006). Infants of these mothers may be difficult to calm, be more irritable, withdraw from caregivers, and are at heightened risk of developing behavioral problems (Campbell et al., 2007; Thomaseli, 2008). Depressed mothers are also less likely to seek preventive health services for their infants and to follow safety prevention practices such as using car seats and covering electrical outlets (Paulson, Dauber, & Leiferman, 2006; McLennan & Kotelchuck, 2000). Offspring of depressed mothers are at heightened risk for psychopathology (Goodman & Gotlib, 1999). Prenatal and recurrent depressive symptoms are predictors of low social functioning and high externalizing behaviors among young children (Luoma et al., 2001), and lead to decreased emotion regulation and higher levels of socioemotional difficulties in early childhood (Maughan et al., 2007). Longitudinal studies have also demonstrated that adolescent behavior problems are associated with early maternal depression and low maternal responsiveness (Aisenberg et al., 2007; Evans et al., 2007).
The links between maternal depression and child mental health on the one hand, and IPV and maternal mental health on the other, suggest that maternal depression may be an important causal pathway in the well-documented association between IPV and children's mental health (Dubowitz et al., 2001; Morrel et al., 2003; Koverola et al., 2005; McFarlane et al., 2003; Whitaker, Orzol, & Kahn, 2006). Thus, understanding the impact of IPV-exposure, other risk factors, and potential protective factors, on the stability and persistence of perinatal depression throughout the first year postpartum is important for understanding potential treatment needs of affected women and for promoting the well-being of both mothers and children.
Evidence for the link between depression affecting IPV exposed women and associated protective and risk factors is limited for the population of childbearing Latinas. As a result of the stress and socioeconomic context of immigration and cultural assimilation experienced by diverse populations (Flaskerud & Winslow, 1998) the impact of IPV on maternal mental health may be significantly confounded by poverty and acculturation, which independently or together are associated with increased depression (Dearing, Taylor, & McCartney, 2004; Heilemann, Lee, & Kury, 2002; Hobfoll et al., 1995). While acculturation has been associated with depression and IPV among Latinos, this relationship is inconsistent in the literature (Caetano, 2000; Jasinski, 1998). Poverty is associated with increased IPV and depression prevalence both in general, and specifically among Latinos (17.6% versus 11% among whites)(Garcia, Hurwitz, & Graus, 2005) There is a small but growing body of literature on the specific link between IPV and depression among Latinas, but the evidence is also mixed. One recent cross-sectional study of IPV among immigrant Latinas found no association of IPV with major depressive disorder after controlling for socio-demographic factors (Fedovskiy, Higgins, & Paranjape, 2007). In contrast, we recently reported a greater than two-fold incidence of depression during pregnancy among a sample of IPV-exposed pregnant Latinas compared to non-IPV exposed women, after controlling for poverty, acculturation and other factors (Rodriguez et al., 2008). Protective factors of increased availability of social support, active non-avoidant coping style, sense of mastery (belief in one's own personal efficacy), and resilience (positive attitude in the face of adversity) have been associated with lower levels of depression in Latinas (McFarlane, Parker, & Soeken, 1995), and we found these same protective factors to be associated with less prenatal depression after controlling for IPV and other variables (Rodriguez et al., 2008).
In this paper, we extend previous research on Latina mothers exposed to IPV by analyzing data from a cohort of pregnant Latinas followed through the first year after birth. The objective of the current study is to assess the impact of IPV on the course of perinatal depression, while controlling for protective and risk factors, at four time periods beginning in pregnancy and continuing at 3 time points up to 13 months postpartum. Specific hypotheses tested in this study are:
A longitudinal study of adult pregnant Latinas attending obstetric clinics in Los Angeles was designed to examine the impact of IPV exposure on Latinas during the perinatal period. Prenatal interviews after the first trimester of pregnancy were followed by interviews at 3, 7, and 13 months post-partum. The recruitment sites were obstetric clinics at private, non-profit, healthcare organizations – a local HMO organization and a medical center - where more than 80% of the population was Latina. Latina patients who screened positive for IPV (IPV-exposed) were eligible for the study, as were pregnant Latinas who reported no previous exposure to IPV (non-IPV-exposed); the latter served as a comparison group. To promote retention, only women who reported that they planned to raise their children in Los Angeles County during the children's first year of life were eligible for study inclusion. The study protocol was approved by the Institutional Review Boards (IRB) of each recruitment site and the University of California, Los Angeles (UCLA).
All women attending obstetric clinics at the HMO and at the medical center between January 2003 and January 2004 were approached by research staff and informed about the study while waiting for routine appointments (n=1,728). Baseline interviews took place between April 2003 and March 2004. The first follow-up interviews were conducted between September 2003 and July 2004, the second follow-up interviews were conducted December 2003 and January 2005, and the third follow-up interviews were conducted between June 2004 and April 2005. These interviews were scheduled to take place initially when the women were at least three months pregnant but had not yet given birth with the three follow up interviews occurring at three, seven, and 13 months post-partum. At the HMO, the clinic staff would let each woman know that someone from a research study either would be approaching them when they were shown into an exam room at the time of appointment check in or the staff would direct the woman to a private room where researchers would be screening them prior to being shown into an exam room. At the private medical center, the research assistant flagged the patient's registration paperwork informing the front desk staff that the assistant would be approaching the patient and taking her aside to a private confidential space to introduce her to the study. The research assistant would introduce herself as a researcher working on a research study to understand the needs of pregnant Latina women and their babies to help enhance healthcare services provided to them. To avoid stigmatization, the study was not labeled as an IPV study but it was made clear that the study would inquire into sensitive issues, including IPV. Potential participants were asked if they could be screened to determine eligibility and if eligible, they were invited to participate. They were told that participation consisted of responding to five questionnaires (one during pregnancy and then up to four during their infant's first two years) and that the interviews would improve our understanding of the needs of pregnant Latina women and their infants in order to enhance healthcare services provided to them.
Of the original 1,728 potentially eligible women who were originally to be approached for the study, 140 refused to be screened while 44 women were either approached twice or were lost before they could be approached, leaving 1,544 who were screened. Of these, 682 screened eligible while 862 screened ineligible. Of these 682 eligible women, nine IPV positive women and 46 IPV negative women refused enrollment while 417 IPV negative women were turned away due to the quota of IPV negative women being filled leaving 92 IPV negative and 118 IPV positive women enrolled in the study. Thus, while there may have been some selection bias among the women who did not decline screening, the overwhelming majority of women agreed to be screened and selection effects should be small.
An interview was administered to study participants by trained bilingual Latina research associates. The questionnaire included validated questions, assessment scales and screening tools already available in both English and Spanish. Participants received $25 for each interview completed.
Demographic questions included age, birthplace (foreign/US- born) and language of interview. Total family income was divided by the federal poverty level for number of persons in the household to create the Poverty Index, in which higher scores indicate higher levels of income above the poverty threshold. Other sociodemographic variables collected included parity, employment status, partner status and educational attainment.
IPV status, the main independent variable, was ascertained using questions from the four-question Abuse Assessment Screen (AAS) (Perlin & Schooler, 1978). Questions include lifetime adult physical and sexual abuse by an intimate partner. This assessment precludes the assessment of temporality and whether the abuse was by the woman's current intimate partner and/or father of the baby. We modified this measure to include psychological experiences of being made to feel fearful for their safety. Women with positive responses to any one of the items were classified as IPV-exposed.
Protective factors included mastery and social support. Mastery was measured with a five-item modified version of the Mastery Scale (Perlin & Schooler, 1978), which measures sense of control over forces that influence one's life. Item scores range from five to 20, with higher scores indicating higher mastery. Social support was measured with a modified version of the Medical Outcomes Study Social Support Survey related to instrumental (e.g., receiving transportation assistance) and emotional support (e.g., having someone to talk to) from formal (agencies) and informal sources (friends, family) (Sherbourne & Stewart, 1999). Scores range from nine to 45, with higher scores indicating higher social support.
Risk factors included avoidant coping style, stress, and non-IPV trauma history. Avoidant coping style was assessed with a subscale from the Medical Outcomes Study (Abbey, Abramos, & Caplan, 1985) which included various “avoidant” coping strategies. Higher scores indicated poorer coping strategies. Stress level was measured with the Perceived Stress Scale (PSS-4) to assess perceived ability to handle “personal problems” and “important things in life,” with scores ranging from four to 20 in which higher scores indicated higher perceived stress (Cohen, Kamarck, & Mermelstein, 1983). Trauma history was assessed with six items from the Trauma History Questionnaire (Green, 1996) and six items from the Adverse Childhood Experiences Study Questionnaire (Felitti et al., 1998). Non-IPV trauma history included past physical or verbal child abuse, witnessing of domestic violence, past experience of physical abuse that left marks (non-IPV,) sexual abuse (non-IPV,) or loss of parent before age 18.
Depression, the major outcome of interest, was measured by the Beck Depression Inventory Fast Screen (BDI-FS) for Medical Patients (Beck, Steer, & Brown, 2000; Steer et al., 1991). The BDI-FS has seven items rated for the past two weeks on a four point Likert scale. Consistent with the literature, a cut-off score of four or above indicated that a participant was depressed. The BDI-FS has sensitivity and specificity greater than 0.80 (Winter et al., 1999). Total scores on the BDI-FS range from zero to 21, with higher scores suggestive of greater severity. Study protocols were used to assure participants' and their children's safety during the study.
The prevalence of depression among IPV-exposed women was compared with prevalence among non-IPV-exposed women at each of the four assessment periods: prenatal, 3, 7 and 13 months after birth. The cumulative frequency of depression (0, 1, 2, or 3-4), the prevalence of risk factors (avoidant coping style, perceived stress, non-IPV trauma, acculturation, and poverty,) and protective factors (social support and mastery) were compared among IPV-exposed and non-exposed mothers at each of the four time periods. A correlation matrix was used to evaluate potential multicollinearity among any of the predictors. Chi-square tests or t-tests were used to assess differences in all bivariate analyses for a given time point. The binomial exact test was used to calculate 95% confidence intervals for point estimates of depression. Marginal repeated measures mixed effects models examined the statistical significance of all independent predictors over time. After systematically testing each variable for statistical significance and using the literature as a basis for determining what variables should be included in the model, final multivariate repeated measures mixed effects models identified variables that were independently correlated with depression, as a continuous outcome over time. SAS Statistical Software, Version 9.1 was used for all analyses.
The study sample is composed of 210 women living in the greater Los Angeles metropolitan area of whom 44% (n=92) were IPV-exposed and 56% (n=118) were non-IPV-exposed at the baseline interview (Table 1). Compared to non-IPV-exposed participants, IPV-exposed participants were significantly more likely to be older and born in the U.S. (30.4% versus 18.6%). In addition, IPV-exposed women were significantly more likely to have higher parity, be employed full or part-time, unmarried and to have graduated from high school.
Women in both groups experienced depression in the prenatal and post-natal periods (Figure 1). While depression was highest in the prenatal period for both IPV-exposed (45.7%) and non-exposed women (24.6%), it was lowest three months after birth (28.0% and 7.3% respectively). Then, depression went up at seven months for both groups (to 42.7% of IPV-exposed and to12.4% of non-exposed women). Finally, at the 13 month time point, depression decreased among IPV-exposed women (39.8%) but increased among non-exposed women (17.5%). At each time point (prenatally, 3, 7, and 13 months after birth), significantly more IPV-exposed women scored at or over the cut-off for depression (p<.001 at each time point) compared to non-exposed women. Mean depression scores for IPV-exposed women at each time point were 3.5 (prenatally), 2.13 (3 months), 3.36 (7 months) and 2.97 (13 months) which were significantly higher than those of the non-IPV exposed women, i.e., 1.49 (prenatally), 0.67 (3 month), 0.97 (7 month) and 1.11 (13 month)(t-score=-3.58, -3.58, -4.84, and -3.74 respectively, p<.001 for each time point).
Approximately 71% of IPV-exposed and 40% of non-IPV-exposed women met or exceeded the cut-off for depression at least once during the four assessment periods. Persistence of depression, as indicated by scores that exceeded the cut-off for depression at more than two of the four assessments, differed significantly by IPV status (Figure 2.) Persistent depression was greater than five times more likely within the IPV-exposed group than within the non-IPV-exposed group (27.3% vs. 5.2%, p<.0001.) A significantly higher proportion of IPV-exposed women scored at or above the cut-off score for depression at two of the four assessments, with a greater than two-fold difference between the exposed and non-exposed groups (17.1% vs. 7.2%, p<.05.). Approximately, 29.5% of IPV-exposed mothers scored at or above the cut-off for depression only once during the perinatal period, compared to 21.7% of non-IPV-exposed mothers. However, this difference was not significant (p=0.28.)
Over the four time points, avoidant coping and perceived stress were higher in IPV-exposed subjects than in non-exposed participants across all time points (Table 2.) In contrast, mastery scores did not significantly differ during the prenatal period, but did significantly differ at the seven and 13 month post-natal assessment time points. Likewise, perceived social support was significantly higher in non-IPV-exposed mothers at the prenatal, three and seven month post-natal time points but not at the 13 month post-natal time point. IPV-positive participants scored lower on social support at each of the four time points when compared with non-IPV-exposed participants. Furthermore, IPV-exposed participants scored lower on mastery at all of the three post-natal time points, significantly higher on avoidant coping style at all four time points, and significantly higher on the perceived stress measure at all three post-natal time points.
The mixed model results shown in Table 3 indicate that high scores on mastery and social support were negatively associated with depression over time. Avoidant coping style and perceived stress were positively associated with depression over time. In addition, history of non-IPV-related trauma was not significantly associated with perinatal depression or the language in which the interview was conducted.
In this study, IPV-exposed women experienced elevated rates of depression throughout multiple time periods between pregnancy and 13 months after birth, which is consistent with and extends the findings of our previous report on the impact of IPV on depression during the prenatal period (Rodriguez et al., 2008). Rates of depression during their pregnancy and at two of the three time points over the first year after birth were in the 40-50% range among women with a history of exposure to IPV and were significantly higher than those in the non-IPV-exposed comparison group at each time period. Nearly 30% of IPV-exposed Latina mothers, compared to 5% of non-IPV exposed mothers, experienced depressive symptoms predictive of postpartum depression that, unlike “baby blues,” is not self-limiting and may continue throughout the first year after delivery. The continuation of depression among IPV-exposed women suggests that depression may be persistent in this group. The potential persistence of depression may be due to various factors including severity of depression, barriers to treatment, or lack of recognition or the problem. Our research did not address these factors, but future research should differentiate depressive symptoms among those in treatment versus those not in treatment.
Consistent with findings from cross-sectional studies of depression among Latinas (Heilemann et al., 2004), we found that the protective factors of social support and mastery were associated with decreased perinatal depression. Conversely, depression across all time points was higher for women with avoidant coping style and perceived stress, even after controlling for the effects of poverty and other demographic characteristics.
Unlike other studies on the role of acculturation in IPV and depression, however, we did not find acculturation to be predictive of depression over four time points, nor was it significantly associated with IPV exposure. These results may reflect study limitations in our use of preference for interview language (Spanish/English) as a proxy for acculturation, or our inability to distinguish different ethnic subgroups within our Latina sample. Studies using other measures of acculturation have found generational status, timing of immigration and time since immigrating to be highly associated with IPV and/or mental health status (Jasinski, 1998; Garcia, Hurwitz, & Graus, 2005). Furthermore, we did not distinguish type, severity, and recency of IPV exposure in this analysis. Others have found that mental health symptoms were more pronounced among women whose IPV exposures were more recent as opposed to remote (Bonomi et al., 2006). Related to IPV exposure, the age difference between the IPV exposed group and the non-exposed group may be due to the greater opportunity to experience violence related to older age. Future longitudinal studies should address these limitations.
Our findings have important implications for the care of both mothers and their children. The longitudinal design of this study allowed us to examine stability and change in depressive symptoms from pregnancy through the first year after delivery. While the pattern for depression scores between the two groups was similar, we found variability in the reports of depressive symptoms over this time period and noted a substantial increase in the proportion of women with depressive symptoms by seven months postpartum and persisting at 13 months postpartum. By examining depressive symptoms throughout the first year after delivery, this study contributes new findings to previous longitudinal studies of perinatal depression (Zelkowitz et al., 2008; American Medical Association, 1992).
In addition, this study examines the contribution of IPV exposure to perinatal depression patterns. Because perinatal depression was significantly higher for IPV-exposed women, compared to non-IPV-exposed women well after birth, our findings suggest that screening for perinatal depression should extend beyond the early post-partum period and continue periodically throughout the first year after delivery. The influence of IPV exposure on perinatal depression has important implications for prenatal and postpartum screening protocols. The occurrence of postpartum depression and its co-occurrence with IPV, can pose added burdens on individuals, society and other family members, especially children (Dubowitz et al., 2001; Morrel et al., 2003; Koverola et al., 2005; McFarlane et al., 2003; Whitaker, Orzol, & Kahn, 2006). Professional associations have issued practice guidelines that recommend screening for IPV and associated mental health disorders in various practice settings (American Medical Association, 1992; American Academy of Family Physicians, 1994; American Academy of Pediatrics, 1998). This is particularly important for Latinas of low income who rarely receive appropriate and adequate treatment for depression. The potential, adverse short-term and long-term consequences of untreated postpartum depression on maternal health and child development also speak to the need for an integrated and coordinated approach to outreach, screening, and intervention for mothers, infants and family members. The Institute of Medicine's recent report on child health in America noted that the complex relationships within family systems require a health care system that is integrated across delivery systems and developmental life stages of adults and children in order to promote child well-being across the life course (Institute of Medicine, 2004). More research is needed on ways to increase screening, promote access to interventions, provide more social support, increase protective strength factors, and integrate health services to address perinatal depression in the context of IPV. Quality improvement methodologies, such as benchmarking and pay for performance, may prove useful in changing current practices. More research on acculturation and ethnic differences can further contribute to designing culturally appropriate screening and intervention services that will have the greatest potential for success.
The researchers would like to thank Erin Richardson for her assistance with the manuscript and the women who graciously agreed to participate in this study.
Support: The National Institute Mental of Health
Michael A. Rodriguez, Department of Family Medicine, David Geffen School of Medicine at UCLA.
Jeannette Valentine, Center for Health Services Research, Semel Research Institute, David Geffen School of Medicine at UCLA; 3550 w. 6TH Street Suite 500, Los Angeles, CA, 90020; Phone: (213) 427-1651; Fax: (213) 427-2701.
Sawssan R. Ahmed, Department of Family Medicine, David Geffen School of Medicine at UCLA.
David P. Eisenman, Division of General Medicine and Health Services Research, Department of Medicine, David Geffen School of Medicine at UCLA; 911 Broxton Plaza, Room 225, Los Angeles, CA 90024; Phone: (310) 794-2452; Fax: (310) 794-3288.
Lekeisha A. Sumner, Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine at UCLA; BOX 951759, 300 Medical Center, Rm 1506, Los Angeles, CA 90095; Phone: (310) 267-0515; Fax: 310-206-9137.
MarySue V. Heilemann, School or Nursing at UCLA; 5-252 Factor Building, Los Angeles, CA 90024; Phone: (310) 206-4735; Fax: (310) 206-3241.
Honghu Liu, Division of General Medicine and Health Services Research, Department of Medicine, David Geffen School of Medicine at UCLA; BOX 951736, 911 Broxton Plz, Los Angeles, CA 90095 Phone: (310) 794-0700; Fax: (310) 794-0732.