Cheng, Diana MD; Horon, Isabelle L. DrPH
Studies within the past two decades have shown that homicide is a leading cause of pregnancy-associated death,1–7 defined by the American College of Obstetricians and Gynecologists (the College)/Centers for Disease Control and Prevention Maternal Mortality Study Group as a death from any cause occurring during pregnancy or within 1 year of pregnancy delivery or pregnancy termination regardless of the site or duration of the pregnancy.8 Unlike the medical causes of maternal mortality, such as hemorrhage and thromboembolism, homicides do not have a physiologic relationship to pregnancy. Nevertheless, some homicides, especially those due to intimate-partner violence, may not have occurred if the victim had not been pregnant. The prevention of intimate-partner homicide therefore could have a major effect on reducing the overall rate of pregnancy- associated mortality.
Unfortunately, population-based data on the proportion of pregnancy-associated homicides that are due to intimate-partner violence are nearly nonexistent. Studies done among pregnant and postpartum women in North Carolina9 and postpartum women in Georgia10 conclude that intimate-partner homicide was responsible for 36% to 38% of female homicides, respectively, although the total number of homicides studied in both states was small (n<30). Intimate-partner homicide made up at least 17% of 72 pregnancy-associated homicide cases in New York City, but perpetrator information was missing for the majority of cases.5 A common barrier inherent in the study of pregnancy-associated intimate-partner homicide is the need for supplemental data sources from multiple disciplines to establish both pregnancy status and perpetrator relationship to a homicide victim. Reliance on death certificate information alone to identify these deaths is problematic because these records are unlikely to indicate that a decedent was pregnant or had delivered recently and because domestic violence is not reported as a contributing cause of death.
Our objective in the current study was to use multiple surveillance methods to identify pregnancy-associated homicide cases among Maryland residents and to estimate the proportion that were perpetrated by a current or former intimate partner. Associated risk factors for these deaths also were examined.
MATERIALS AND METHODS
Data on pregnancy-associated deaths occurring among Maryland residents during the years 1993 through 2008 were collected from three main sources: 1) review of death certificates to identify those records in which a current or recent pregnancy could be identified through either cause-of-death information or a completed pregnancy checkbox, 2) linkage of death certificates of reproductive-aged women with corresponding live birth and fetal death records to identify a pregnancy within the year preceding death, and 3) review of medical examiner records for evidence of a current or recent pregnancy on the autopsy report or in the accompanying police record because all homicide victims in Maryland are examined routinely. Maternal race, ethnicity, age, level of education, marital status, and cause of death were obtained from death certificates. Birth and fetal death certificates for women who died during the postpartum period provided information on the date of the most recent delivery. Medical examiner autopsy records included information to estimate gestational age at the time of death. Information pertaining to the victim–perpetrator relationship was obtained primarily through local police records.
A death was categorized as a pregnancy-associated intimate-partner homicide if: 1) the perpetrator was a married person living with or estranged from his spouse or a current/former cohabiting intimate partner (including intimate partners of the same sex), in keeping with the definition of an intimate partner used by the Maryland Uniform Crime Reporting Program; and 2) the victim was a Maryland resident during pregnancy or within 1 year of delivery or pregnancy termination. This included pregnancies ending in live birth and fetal death, spontaneous or therapeutic abortion, ectopic pregnancy, and other nonviable outcomes. All homicide victims were Maryland residents at the time of death. However, the pregnancy within the year before death and the death itself need not have occurred in Maryland.
Although not collected specifically for this study, we accumulated articles and documents from newspapers, obituaries, circuit court proceedings, and local police and sheriff's offices that were of interest to our work on pregnancy-associated mortality. These documents were available to us for 38.2% of our homicide cases, and we used them to confirm information about the crime, perpetrator relationship, and pregnancy status. In three out-of-state deaths or deliveries, we relied on the newspaper or court documents as the sole source of pregnancy or perpetrator status.
The Maryland Department of Health and Mental Hygiene's institutional review board qualified this project as exempt research. Analyses were performed using SAS 9.2 (SAS Institute Inc., Cary, NC). The Fisher exact test was used to compare categorical variables. Statistical significance was set at P<.05.
Homicide (n=110) was the leading cause of pregnancy- associated death and accounted for 17% of deaths among Maryland residents occurring between 1993 and 2008, followed by cardiac disorders (16%) and accidents (11%). The perpetrator was identified in 95 of the 110 pregnancy-associated homicides and was found to be an intimate partner in 54.5% (95% confidence interval [CI] 42.5–63.9%, n=60) of total homicide deaths and a nonpartner in 31.8% (95% CI 23.1–40.5%, n=35). The remaining 15 cases (13.6%) were either unsolved or unavailable for review (Table 1). Therefore, for homicides in which the victim-to-offender relationship could be identified, 63.2% (95% CI 53.5–72.9%) were intimate-partner homicide cases. Firearms were the most common method of injury (61.8%), accounting for more deaths than all other methods combined in both intimate-partner and non–intimate-partner homicide cases. None of the fetuses survived the deaths of their mothers during pregnancy. Records from the local police were used to determine perpetrator status for all but two cases. An official court document was used to identify the perpetrator in one of these homicides, and a newspaper obituary was used to identify the perpetrator in the other.
The 66 cases of homicide in nonpregnant women (Table 2) included 64 women who had delivered liveborn neonates and two women whose fetuses were stillborn in the year before death. One woman was both 19 weeks pregnant and less than a year postpartum at the time of her death. She was categorized as pregnant at the time of death so as not to be counted twice. Medical examiner records were used to identify deaths occurring during pregnancy. A matched birth certificate was available for 63 of the 64 deaths that occurred after a live birth. The missing certificate was most likely from a birth that occurred outside of Maryland. A published newspaper article was the sole method used to identify the postpartum status of one woman who was a Maryland resident at the time of death but was not a resident of Maryland when she delivered. Although information on initiation of prenatal care was missing in 15.6% of cases of postpartum live-birth intimate-partner homicide, all remaining women had had prenatal care during their most recent pregnancies; 70.4% initiated care during the first trimester and 25.9% during the second trimester.
Victims of both intimate-partner homicide and nonpartner homicides were more likely to be African American, younger than 25 years, and unmarried. Women in the intimate-partner–homicide group were significantly more likely (P<.05) to have completed more than 12 years of education than were those in the non–intimate-partner–homicide group (23.3% compared with 5.7%), be married (28.3% compared with 8.6%), and be murdered in their homes (66.7% compared with 28.6%). Although the differences were not significant, women in the intimate-partner–homicide group were more likely to be 30 years of age or older at the time of death (23.3% compared with 11.4%) than were those in the non–intimate-partner–homicide group (Table 1) and more likely to have died during the first 3 months of pregnancy (25% compared with 11.4%) (Table 2). None of the intimate-partner homicides during early pregnancy occurred before the second month.
The pregnancy-associated homicide ratio, defined as the number of homicides during pregnancy or within the first postpartum year per 100,000 births, was 9.3 per 100,000 births for total homicides and 5.1 per 100,000 births for intimate-partner homicide cases. The ratio for intimate-partner homicide cases was approximately four times higher among African-American women than white women (10.4 compared with 2.7). African-American women younger than 25 years were at highest risk, with a ratio of 16.7 per 100,000 births (Table 3).
More than half of all Maryland pregnancy-associated homicides were committed by a current or former husband or intimate partner. This proportion is higher than the 16% to 38% reported in previously published studies that assessed pregnancy-associated intimate-partner homicides.5,9,10 National estimates of intimate-partner homicide in the general female population from the Federal Bureau of Investigation's “Supplementary Homicide Reports” indicate that current spouses, ex-spouses, or current intimate partners were responsible for approximately 30% of all homicides among American women (pregnant and nonpregnant) between 1993 and 2005.11 However, the exclusion of ex-intimate partners likely underestimates the number of intimate-partner homicide cases. Recent data from the Maryland State Police's Uniform Crime Report includes ex-partners but gave similar results, showing that current or former intimate partners were responsible for 27.7% of female homicides in Maryland from 2007 to 2008.12 Data for these government reports often are submitted by law-enforcement agencies before investigations are complete, leading to potential misclassifications and missing information about the perpetrator. Campbell's review of intimate-partner–homicide research shows that, when misclassifications are accounted for, most femicide studies estimate that 40% to 50% of homicides are due to intimate-partner violence.13 Moracco reports that 50% of homicides in North Carolina among women 15 years of age or older were due to intimate-partner violence.14 These higher proportions of intimate-partner homicide approximate the rate we found in our study of pregnancy-associated homicide cases.
Our current study shows a greater than fivefold higher pregnancy-associated homicide ratio than that reported by the only national study published1 of pregnancy-associated homicides (9.3 in Maryland compared with 1.7 in the national study) and a nearly twofold higher proportion of homicide deaths occurring during pregnancy (40% in Maryland compared with 21% in the national study). We attribute these differences to our use of multiple data systems, especially medical examiner records, which identified a large proportion of deaths during pregnancy. As Chang and colleagues acknowledge, the data set used for the national study is based on reports from states, few of which do enhanced surveillance of pregnancy-associated deaths. The higher Maryland figures also may be due in part to the high homicide rates in Baltimore City and the Washington, DC, area. In 2007, Maryland ranked 16th highest among states for homicide rates involving one female victim and one offender.15
Previous studies on pregnancy-associated homicide have not analyzed the occurrence of homicide by time interval during pregnancy or after delivery. If there were no association of homicide with pregnancy, we would expect approximately 14.3% of homicides to occur during each of the roughly seven equal time intervals we studied (three 3-month periods of pregnancy and four 3-month periods after delivery). The prevalence of intimate-partner homicide cases varied greatly with each 3-month time period, peaking at 25.0% during the first 3 months of pregnancy and decreasing to 5.0% during the first 3 months postpartum. When intimate-partner homicide during pregnancy is viewed as a whole, the high prevalence of intimate-partner homicide during the first trimester is offset by the low rates during the third trimester (8.3%). This may lead to an erroneous assumption that pregnancy offers protection against violence and obscures the possibility that the first trimester may be a high risk time for intimate-partner homicide, potentially triggering violence by the partner after the discovery of the pregnancy. Although the differences did not reach statistical significance, nonpartner homicides did not peak during the first trimester but were more evenly distributed (11.4–20.0%) over each of the seven 3-month intervals during pregnancy and the postpartum year, except the third trimester of pregnancy (2.9%).
Our finding that intimate-partner homicide most commonly occurs during the first 3 months of pregnancy makes screening for violence especially important before conception and early in pregnancy. Campbell and colleagues report that 79% of female intimate-partner homicide victims ages 18–50 had been physically abused by the same perpetrator on one or more occasions before death.16 Moreover, the prenatal period may be an especially vulnerable time for women. In a case-controlled study, McFarlane and colleagues report that women who were abused during pregnancy had three times the risk of homicide as did women who were not abused.17 All postpartum victims in our study with known prenatal care status had received prenatal care. This provides obstetricians an opportune time to assess routinely all prenatal patients for domestic violence during the initial visit, at each trimester of pregnancy, and at the postpartum visit as recommended by the College.18 Incorporation of the abuse assessment into the medical-history record helps to standardize the screen so that it has a better chance of being completed routinely.19 There are many screens for intimate-partner violence. One three-question assessment is presented below20 and also is available from the College's Web site21:
1. “Within the past year (or since you have been pregnant) have you been hit, slapped, kicked or otherwise physically hurt by someone?
2. Are you in a relationship with a person who threatens or physically hurts you?
3. Has anyone forced you to have sexual activities that made you feel uncomfortable?”
If a patient discloses violence, the trauma should be acknowledged, safety assessed, and assurance made that she is not responsible for the abuse. Educational materials and referral sources such as local shelters, hotline numbers, safety plans, and advocacy agencies, should be handed to the patient or made available in restrooms where they can be retrieved privately.
Women at risk of pregnancy-associated intimate-partner homicide may need safety planning. Firearms caused the deaths of nearly two thirds of intimate-partner homicide victims in our study, and gun ownership by an intimate partner is associated with a greater risk of danger or fatal outcome to women.22 Also, two out of every three intimate-partner homicide deaths in this study occurred at home. Safety planning that includes alternate living arrangements may be necessary for many women.
Limitations of this retrospective study include possible coding and reporting errors on birth and death certificates, which were the primary sources of our demographic and delivery data. In a study assessing concordance between data on HealthStart prenatal records and the birth certificate, Reichman reports accurate sociodemographic data.23 As with all maternal mortality surveillance, our numbers very likely still reflect an undercount of deaths during pregnancy and postpartum. Deaths that occurred among Maryland residents who were residents of other states when they delivered but moved to Maryland before they died would not be identified with our linked death and birth certificate surveillance because we would not routinely have access to a birth certificate from an out-of-state resident. Most early pregnancies also are undercounted because they are not commonly identified on routine autopsy records before 6 weeks of gestation. Deaths after a spontaneous or therapeutic abortion, ectopic pregnancy, or other nonviable pregnancy also are underestimated owing to the lack of accessible information about these events.
During the past century, the United States has had remarkable success in the dramatic reduction of maternal mortality from deaths due to hemorrhage, embolism, infection, and hypertension. As we continue our important work on reducing deaths from these causes, we must also recognize that pregnant and postpartum women are dying not only from physiologic causes but also from social causes such as intimate-partner violence. Informing women about the potentially fatal outcome of violence during pregnancy, especially during the first trimester, may motivate some women to obtain help before pregnancy. More importantly, women in an already tumultuous or abusive relationship should not become pregnant in the hopes that having a baby will stop the perpetuation of violence. For some women, this can be a fatal mistake. The obstetric field is uniquely positioned to play an important role in keeping women safe by screening for domestic violence before, during, and after pregnancy.
1. Chang J, Berg CJ, Saltzman LE, Herndon J. Homicide: a leading cause of injury deaths among pregnant and postpartum women in the United States, 1991–1999. Am J Public Health 2005;95:471–7.
2. Nannini A, Weiss J, Goldstein R, Fogerty S. Pregnancy-associated mortality at the end of the twentieth century: Massachusetts, 1990–1999. J Am Med Womens Assoc 2002;57:140–3.
3. Harper M, Parsons L. Maternal deaths due to homicide and other injuries in North Carolina: 1992–1994. Obstet Gynecol 1997;90:920–3.
4. Fildes J, Reed L, Jones N, Martin M, Barrett J. Trauma: the leading cause of maternal death. J Trauma 1992;32:643–5.
5. Dannenberg AL, Carter DM, Lawson HW, Ashton DM, Dorfman SF, Graham EH. Homicide and other injuries as causes of maternal death in New York City, 1987 through 1991. Am J Obstet Gynecol 1995;172:1557–64.
6. Horon I, Cheng D. Enhanced surveillance for pregnancy-associated mortality—Maryland, 1993–1998. JAMA 2001;285:1455–9.
7. Krulewitch CJ, Pierre-Louis ML, de Leon-Gomez R, Guy R, Green R. Violent deaths among pregnant women in the District of Columbia, 1988–1996. J Midwifery Womens Health 2001;46:4–10.
8. Atrash HK, Rowley D, Hogue CJ. Maternal and perinatal mortality. Curr Opin Obstet Gynecol 1992;4:61–71.
9. Parsons LH, Harper MA. Violent maternal deaths in North Carolina. Obstet Gynecol 1999;94:990–3.
10. Dietz PM, Rochat RW, Thompson BL, Berg CJ, Griffin GW. Differences in the risk of homicide and other fatal injuries between postpartum women and other women of childbearing age: implications for prevention. Am J Public Health 1998;88:641–3.
11. Fox JA, Swatt ML. Uniform crime reports [United States]: supplementary homicide reports with multiple imputation, cumulative files 1976–2007 [computer file]. ICPSR24801-v1. Ann Arbor (MI): Inter-university Consortium for Political and Social Research; 2009.
12. Crime in Maryland: 2007 uniform crime report. Pikesville (MD): Maryland State Police; 2008.
13. Campbell JC, Glass N, Laughon K, Bloom T. Intimate partner homicide: review and implications of research and policy. Trauma Violence Abuse 2007;8:246–69.
14. Moracco KE, Runyan CW, Butts JD. Female intimate partner homicide: a population-based study. J Am Med Womens Assoc 2003;58:20–5.
15. Langley M. When men murder women: an analysis of 2007 homicide data. Washington, DC: Violence Policy Center; 2009. Available at: http://www.vpc.org/studies/wmmw2009.pdf
. Retrieved November 2, 2009.
16. Campbell JC, Webster D, Koziol-McLain J, Block C, Campbell D, Curry MA, et al. Risk factors for femicide in abusive relationships: results from a multisite case control study. Am J Public Health 2003;93:1089–97.
17. McFarlane J, Campbell JC, Sharps P, Watson K. Abuse during pregnancy and femicide: urgent implications for women's health. Obstet Gynecol 2002;100:27–36.
18. Ettinger B. Intimate partner violence and domestic violence. In: Special issues in women's health. Washington, DC: American College of Obstetricians and Gynecologists; 2005. p. 169–82.
19. Wiist WH, McFarlane J. The effectiveness of an abuse assessment protocol in public health prenatal clinics. Am J Public Health 1999;89:1217–21.
20. American College of Obstetricians and Gynecologists. Assessments. In: Clinical updates in women's health care, primary and preventive care. Washington, DC: American College of Obstetricians and Gynecologists; 2007. p. 32.
22. Campbell JC, Webster DW, Glass N. The danger assessment: validation of a lethality risk assessment instrument for intimate partner femicide. J Interpers Violence 2009;24:653–74.
23. Reichman NE, Hade EM. Validation of birth certificate data. A study of women in New Jersey's HealthStart program. Ann Epidemiol 2001;11:186–93.
© 2010 by The American College of Obstetricians and Gynecologists.