Although deaths attributable to obstetrically related events, such as cardiac disease, infection, and hemorrhage, have improved, maternal mortality attributable to injury has remained constant.1 In fact, several studies have demonstrated that maternal injury is a leading cause of maternal mortality.2–9 Homicide and suicide are two important and potentially preventable causes of maternal injury. However, the study of homicide and suicide has been limited by: 1) a lack of information concerning victim-to-suspect relationships; 2) studies limited to localized samples; 3) little information regarding precipitating circumstances to these deaths; and 4) likely underreporting of maternal violent deaths, especially because of reliance on death certificates alone.10
We sought to address these gaps in the literature by conducting a secondary data analysis of maternal violent death from the Centers for Disease Control and Prevention's (CDC) National Violent Death Reporting System, a multistate database that collects data on violent deaths, including victim-to-suspect relationships, using multiple complementary data sources. This study aims to estimate the rates of pregnancy-associated (occurring during pregnancy or the first year postpartum) homicide and suicide in a multistate sample; compare these rates with other causes of mortality during the perinatal period; describe the demographic characteristics of victims of pregnancy-associated violent death; and estimate the prevalence of intimate partner conflict associated with these maternal deaths.
As previous data have suggested a strong association between intimate-partner violence and pregnancy-associated homicide, a significant association between intimate-partner violence and suicidal tendencies and suicides in women,12–16 and a potential relationship between intimate-partner violence and pregnancy-associated suicide exist,17 we hypothesized that intimate partner conflict may be related to pregnancy-associated violent death. This relationship requires the understanding of victim-to-suspect relationships and precipitating circumstances, variables that are available in the National Violent Death Reporting System.
MATERIALS AND METHODS
We conducted a secondary data analysis of the National Violent Death Reporting System, a multistate active surveillance system controlled by the CDC. The National Violent Death Reporting System began collecting violent death data in 2003. Participating states collect risk factor data on all violent deaths, including homicides, suicides, unintentional deaths by firearms, legal intervention deaths, and deaths of undetermined intent. Seven states participated in 2003 and 13 participated in 2004. In 2005, the number of participating states increased to 17, including South Carolina, Georgia, North Carolina, Virginia, New Jersey, Maryland, Alaska, Massachusetts, Oregon, Colorado, Oklahoma, Rhode Island, Wisconsin, California, Kentucky, New Mexico, and Utah. All states report statewide data except California, which gathers information only in a limited number of counties.18
The National Violent Death Reporting System is incident-based (not victim-based), meaning that it contains reports related to both victims and suspects associated with a given incident within one incident record. This links all victims and suspects with a given incident, permitting researchers to determine if the suspect was a partner of the victim. The system also uses multiple complementary data sources, including not only death certificates but also coroner or medical examiner records and police reports and contains approximately 250 unique data elements, including data on victim characteristics and precipitating circumstances of death.18 Each state health department's National Violent Death Reporting System office coordinates the abstraction process, which may include electronic transfer of data from primary sources or manual abstraction from the records maintained by primary sources at their offices.18 A coding manual is provided, and coding training is performed annually for all participating states.39 In addition, the CDC provides ongoing coding support through conference calls and an e-mail help desk.39 National Violent Death Reporting System states use multiple abstractors to perform blinded reabstraction of cases for reliability checks. The CDC also conducts a quality of control analysis, with a particular focus on abstractor-assigned variables. Further detail related to the National Violent Death Reporting System coding procedures is available at http://www.cdc.gov/ncipc/profiles/nvdrs.
The National Violent Death Reporting System collects information regarding pregnancy or postpartum status at the time of death. Information regarding pregnancy status is abstracted from the coroner or medical examiner records and the death certificate. The coding structure of the National Violent Death Reporting System pregnancy status variable was designed to match the U.S. standard death certificate, which was modified in 2003 to include checkboxes to differentiate whether a woman was pregnant, within 42 days postpartum, within 1 year postpartum, or not pregnant within the past year at the time of death.10 By 2005, only five of the participating National Violent Death Reporting System states were using this version of the U.S. standard death certificate.19 Therefore, two additional responses are available on the coding tree for states with death certificates that do not include a timeline that matches the U.S. standard death certificate: “not pregnant, not otherwise specified” and “pregnant, not otherwise specified.”18 For the purposes of our analysis, we treated values “pregnant at the time of death” and “pregnant, not otherwise specified” as pregnant at the time of death and values “not pregnant but pregnant within 42 days of death” and “not pregnant but pregnant 43 days to 1 year before death” as pregnant within 1 year of death.
The CDC/American College of Obstetricians and Gynecologists Maternal Mortality Study Group defined the term pregnancy-associated mortality to include “any death of a woman while pregnant or within 1 year of the termination of pregnancy, regardless of cause”; this includes deaths related and unrelated to pregnancy.4 In keeping with this terminology, we used the terms pregnancy-associated homicide and pregnancy-associated suicide to represent deaths during pregnancy or within 1 year postpartum attributable to homicide and suicide, respectively. In addition, we used the terms pregnancy-associated homicide rate and pregnancy-associated suicide rate to refer to the number of pregnancy-associated homicides and suicides per 100,000 live births, respectively. We used the term pregnancy-associated violent death to refer to deaths from homicide and suicide combined. The term “pregnancy-associated” refers to the timing of the death and does not mean that the homicide or suicide was caused by pregnancy or postpartum status.
The National Violent Death Reporting System contains six variables related to intimate-partner violence: victim-to-suspect relationship; intimate partner problem (suicides only); perpetrator of interpersonal violence (suicides only); victim of interpersonal violence (suicides only); intimate-partner violence–related (homicides only); or jealousy (“lover's triangle, ” homicides only).18
The victim-to-suspect relationship is abstracted based on the police report and the coroner or medical examiner records. Abstractors report the relationship as one of 29 relationship types or as unknown. For the purpose of our study, we defined a relationship as an intimate partner relationship if it was coded as any one of the following five relationship categories: Spouse; ex-spouse; girlfriend or boyfriend; ex-girlfriend or ex-boyfriend; or girlfriend or boyfriend, unspecified whether current or ex.18
For suicide victims, National Violent Death Reporting System abstractors record whether problems with a current or former intimate partner appear to have contributed to the suicide. The abstractors are instructed to code this variable as “yes” if, after reviewing the coroner or medical examiner records, police report, and child fatality report, if applicable, they found that at the time of the incident the victim was experiencing problems with a current or former intimate partner, such as a divorce, break-up, argument, jealousy, conflict, or discord. In the case of suicides, abstractors also code whether the victim was a known perpetrator or victim of interpersonal violence during the month before death. Interpersonal problems include, but are not limited to, intimate-partner violence.18
For homicide victims, National Violent Death Reporting System abstractors record cases in which a homicide is related to conflict between current or former intimate partners. Abstractors are instructed to code this variable as “yes” if, after reviewing the coroner or medical examiner records, police report, and child fatality report if applicable, they find that the death is perpetrated by an intimate partner or is associated with intimate-partner violence or conflict. The CDC defines an intimate partner as a current or former girlfriend or boyfriend, date, or spouse in relation to this variable. In addition, abstractors code whether jealousy or distress over an intimate partner's relationship or suspected relationship with another personal led to the homicide (coded as jealousy [lover's triangle]).18
For the purpose of this study, we limited our analysis to women of reproductive age (15–54 years) from the 16 states reporting complete data to the National Violent Death Reporting System. The CDC National Vital Statistics System reports birth and maternal mortality information for women in the category of “under 15. ” However, we excluded women younger than 15 years because of the extremely small sample size; the National Violent Death Reporting System prohibits reporting of cells showing or derived from fewer than five deaths (zero cells may be shown) and rates computed from cells containing less than 20 deaths or cases. Data were analyzed for the years 2003–2007 because this was the most current sample available for the National Violent Death Reporting System.
The primary focus of this analysis was to estimate the overall rate of pregnancy-associated homicide and suicide, describe the demographic characteristics of the victims associated with these deaths, and estimate the prevalence of intimate partner conflict related to these deaths. We calculated the rates of pregnancy-associated homicide and suicide by dividing the total number of pregnancy-associated homicides and suicides, respectively, by the number of live births in the sample population, using data from the national natality files of the CDC.20 We converted the results to maternal mortality rates by reporting the number of deaths per 100,000 live births. We used descriptive statistics to report demographic characteristics of victims and the prevalence of intimate partner conflict associated with the pregnancy-associated violent deaths in our sample. We were unable to analyze data for the perpetrator of interpersonal violence (suicides only), victim of interpersonal violence (suicides only), and jealousy (homicides only) because of high data coded as “no/not available/unknown” together for these variables, leaving us unable to differentiate missing data from data coded as “no.”
We used two-sample tests of proportions to compare sociodemographic variables among victims of pregnancy-associated homicide and pregnancy-associated suicide with women with live births within the National Violent Death Reporting System states during the same time period using data from the national natality files of the CDC.20 We also used a two-sample test of proportions to compare the pregnancy-associated homicide and suicide rates, respectively, in the National Violent Death Reporting System states using the U.S. standard death certificate compared with those that did not. P<.05 was considered statistically significant. This study was deemed Institutional Review Board-exempt by the Institutional Review Board of the University of Michigan Medical School.
In total, we identified 233 pregnancy-associated violent deaths, yielding an overall pregnancy-associated violent death mortality rate of 4.9 per 100,000 live births; 64.8% of the pregnancy-associated violent deaths in our sample (n=151) occurred during pregnancy (compared with the first year postpartum). The overall pregnancy-associated violent death rate was fairly stable over the study time period, ranging from 4.3 to 5.4 (Fig. 1). In addition, the rates of pregnancy-associated homicide and suicide were each higher than mortality rates attributable to common obstetric causes (Fig. 2).38
There were 94 counts of pregnancy-associated suicide in our sample, yielding a pregnancy-associated suicide rate of 2.0 per 100,000 live births (Fig. 2). Forty-three of the pregnancy-associated suicides occurred during pregnancy (45.7%), and 51 occurred postpartum. Older women were at greatest risk for pregnancy-associated suicide (Table 1). Women 40 and older represented 17.0% of pregnancy-associated suicides but account for just 2.8% of live births in the National Violent Death Reporting System states (P<.01). In addition, victims of suicide were significantly more likely to be white or Native American and more often were unmarried, as compared with women with all live births in the National Violent Death Reporting System states. The rate of pregnancy-associated suicide did not differ significantly between states that did and did not use the U.S. standard death certificate (2.0 compared with 1.9, P=.90). We were unable to compare the percentage of women who were Asian or Pacific Islander because of the low number of deaths in this group.
There were 139 counts of pregnancy-associated homicide in our sample, yielding a pregnancy-associated homicide rate of 2.90 per 100,000 live births (Fig. 2) and representing more than half (59.7%) of the total pregnancy-associated violent deaths; 77.7% of the pregnancy-associated homicides occurred in pregnant women (compared with during the first year postpartum). Women at the extremes of the age range were at the highest risk for pregnancy-associated homicide (Table 2). Women 24 years of age and younger accounted for more than half (53.9%) of the pregnancy-associated homicides in our sample but make up only one-third (33.6%) of all live births in the National Violent Death Reporting System states (P<.01). In addition, victims of homicide were significantly more likely to be aged 40 years or older as compared with women with all live births in the National Violent Death Reporting System states. African American women accounted for almost half (44.6%) of the pregnancy-associated homicides but only 17.7% of live births (P<.01). Victims of pregnancy-associated homicide were significantly more likely to be unmarried. The rate of pregnancy-associated homicide did not differ significantly between states that did and did not use the U.S. standard death certificate (3.0 compared with 3.7, P=.22). We were unable to compare the percentage of women who were Asian or Pacific Islander or Native American because of the low number of deaths in these groups.
In pregnancy-associated suicides, 54.3% of victims experienced problems with a current or former intimate partner that appeared to have contributed to the suicide, 45.3% of pregnancy-associated homicides were associated with violence from a current or former intimate partner, and 42.4% of suspects in pregnancy-associated homicides were a current or former intimate partner of the victim (n=59), and this is, by definition, intimate partner homicide. Similar to victims of all pregnancy-associated homicides, victims of pregnancy-associated intimate partner homicide were younger, with women aged 24 years and younger accounting for 44.1% of intimate-partner homicides but only 33.6% of all live births, although this did not reach statistical significance (Table 3). African American women accounted for 37.3% of pregnancy-associated intimate-partner homicides but only 17.7% of live births (P<.01). Victims of pregnancy-associated intimate partner homicide were significantly more likely to be unmarried as compared with all live births. We were unable to compare the percentage of women who were Asian or Pacific Islander or Native American because of the low number of deaths in these groups.
Our results indicate that pregnancy-associated homicide and suicide each account for more deaths than many other obstetric complications, which may be thought of as more “traditional” causes of maternal mortality, and confirm the need to focus on the relationships between sociodemographic disparities and intimate-partner violence with pregnancy-associated violent death. Our results demonstrate somewhat higher maternal mortality rates than those previously demonstrated in a national perinatal sample.23 Chang et al23 reported a pregnancy-associated homicide rate of 1.7 deaths per 100,000 live births between 1991 and 1999 in the Pregnancy Mortality Surveillance System. Suicides were even less common in their sample. These differences may reflect the data collection methods used by the two systems. The Pregnancy Mortality Surveillance System is designed to collect data on all pregnancy-associated deaths, regardless of cause. Reporting states use death certificate data or matched death-to-birth certificate data to identify deaths.23 The National Violent Death Reporting System is designed to collect data specifically associated with violent deaths. Abstractors code pregnancy and postpartum status based on findings from the victim's death certificate and the medical examiner's report.18 Mortality rates from other smaller studies of pregnancy-associated violent death have been closer to our results for suicide deaths and have demonstrated even higher mortality rates for pregnancy-associated homicide.4,24,25
Despite these differences in mortality rates between the National Violent Death Reporting System and Pregnancy Mortality Surveillance System, the characteristics of pregnancy-associated homicide victims between the National Violent Death Reporting System and others, including Pregnancy Mortality Surveillance System, are strikingly similar. Victims of pregnancy-associated homicide are more likely to be African American, younger, and unmarried.23,24 In addition, our finding of the association between intimate partner conflict and both pregnancy-associated homicide and suicide has been echoed by several studies in general and perinatal samples.1,2,17,22,24 In the National Violent Death Reporting System general population sample, more than one-half of female homicide deaths (59.1%) are associated with intimate-partner violence, and more than one-quarter of suicides in female victims (26.1%) were related to intimate partner problems.22 In a postpartum study, 38% of female homicide victims were killed by a boyfriend, husband, or ex-husband.26
Our study has several limitations. Although the National Violent Death Reporting System now collects data from many states, it is not fully nationally representative.22 Furthermore, because of the low rates of deaths in certain subpopulations (Asian or Pacific Islander and Native American), we were not able to compare data among all ethnic groups. National Violent Death Reporting System abstractors are limited by the completeness and quality of the reports they receive, and personnel, death certificates, and law enforcement protocols may vary from one jurisdiction to the next. For this reason, the National Violent Death Reporting System uses multiple complementary data sources and abstractors follow defined National Violent Death Reporting System primacy rules in coding data.22 Although the National Violent Death Reporting System codes pregnant and postpartum status from multiple data sources, pregnancy-associated deaths may still be underreported. Pregnancies, even if identified, may not be reported on death certificates, autopsies might not include examination for pregnancy,27 early gestation and late postpartum status may be missed on autopsy, and family members and friends may have been unaware of early or unwanted pregnancies. Also, because a majority of female deaths in the National Violent Death Reporting System are coded as unknown pregnancy or postpartum status (67.2%), our results may underestimate the number of pregnancy-associated violent deaths. In addition, we are unable to compare the rates of pregnancy-associated violent death in pregnant postpartum compared with nonpregnant nonpostpartum women because so many deaths were classified as unknown status. Finally, protective factor data would be helpful but are not collected by the National Violent Death Reporting System because reports associated with violent death often contain only circumstances associated with risk factors. Our data provide information regarding potential risk factors for maternal violent deaths but cannot prove causation. In addition, this analysis focuses on demographic data and prevalence of intimate partner conflict among victims of pregnancy-associated violent data but did not cover all of the potential precipitating circumstances that may be related to violent death. Our future research will involve analyses of other potential precipitating circumstances around maternal violent death, including substance abuse, life stress, and mental health diagnoses and treatment.
Despite these limitations, our study highlights the unfortunate but important role of homicide and suicide as contributors to pregnancy-associated mortality. These findings suggest that effective prevention methods aimed at perinatal psychosocial health are imperative. Unlike some obstetric complications, violence is potentially preventable. Research is moving forward in developing evidence-based guidelines for perinatal depression care,31 identifying successful strategies for engaging and training perinatal health care providers in delivering mental health care32,33 and engaging perinatal women in receiving mental health care services.34 Our findings also demonstrate the frequent association of intimate partner conflict with maternal violent death. Intimate-partner violence has been associated with adverse outcomes for both woman and neonate, including preterm labor and low birth weight, and may contribute to perinatal health disparities.35 Studies suggest that standardized screening for intimate-partner violence is associated with increased identification rates in pregnant women, and a recent intervention was shown to lower recurrence risk for intimate-partner violence victimization during pregnancy and the postpartum period.37 With continued focus on maternal violent death and a continued push toward the development of effective psychosocial interventions, particularly postscreening care, we may be able to reduce the effect of this unfortunate killer on American women, their children, and their families.
1. Shadigian E, Bauer ST. Pregnancy-associated death: a qualitative systematic review of homicide and suicide. Obstet Gynecol Surv 2005;60:183–90.
2. Krulewitch CJ, Pierre-Louis ML, de Leon-Gomez R, Guy R, Green R. Hidden from view: violent deaths among pregnant women in the District of Columbia, 1988–1996. J Midwifery Womens Health 2001;46:4–10.
3. Sachs BP, Brown DA, Driscoll SG, Schulman E, Acker D, Ransil BJ, et al.. Maternal mortality in Massachusetts. Trends and prevention. N Engl J Med 1987;316:667–72.
4. Jocums SB, Berg CJ, Entman SS, Mitchell EF Jr. Postdelivery mortality in Tennessee, 1989–1991. Obstet Gynecol 1998;91(5 Pt 1):766–70.
5. 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.
6. Harper M, Parsons L. Maternal deaths due to homicide and other injuries in North Carolina: 1992–1994. Obstet Gynecol 1997;90:920–3.
7. Baker N, Fogarty C, Stroud D, Rochat R. Enhanced pregnancy-associated mortality surveillance: Minnesota, 1990–1999. Minn Med 2004;87:45–7.
8. Ho EM, Brown J, Graves W, Lindsay MK. Maternal death at an inner-city hospital, 1949–2000. Am J Obstet Gynecol 2002;187:1213–6.
9. Horon IL, Cheng D. Enhanced surveillance for pregnancy-associated mortality–Maryland, 1993–1998. JAMA 2001;285:1455–9.
10. Hoyert DL. Maternal mortality and related concepts. Vital Health Stat 2007;33:1–13.
11. Chambliss LR. Intimate partner violence and its implication for pregnancy. Clin Obstet Gynecol 2008;51:385–97.
12. Leiner AS, Compton MT, Houry D, Kaslow NJ. Intimate partner violence, psychological distress, and suicidality: A path model using data from African American women seeking care in an urban emergency department. J Fam Violence 2008;23:473–81.
13. Abbott J, Johnson R, Koziolmclain J, Lowenstein SR. Domestic violence against women—incidence and prevalence in an emergency department population. JAMA 1995;273:1763–7.
14. Stark E, Flitcraft A. Killing the beast within: woman battering and female suicidality. Int J Health Serv 1995;25:43–64.
15. Seedat S, Stein MB, Forde DR. Association between physical partner violence, posttraumatic stress, childhood trauma, and suicide attempts in a community sample of women. Violence Vict 2005;20:87–98.
16. Pico-Alfonso MA, Garcia-Linares MI, Celda-Navarro N, Blasco-Ros C, Echeburua E, Martinez M. The impact of physical, psychological, and sexual intimate male partner violence on women's mental health: Depressive symptoms, posttraumatic stress disorder, state anxiety, and suicide. J Womens Health 2006;15:599–611.
17. Martin S, Macy R, Sullivan K, Magee M. Pregnancy-associated violent deaths: the role of intimate partner violence. Trauma Violence Abuse 2007;8:135–48.
18. Centers for Disease Control and Prevention. National Violent Death Reporting System (National Violent Death Reporting System) Coding Manual Revised [Online] 2008. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention (producer). Available at: http://www.cdc.gov/violenceprevention/pdf/NVDRS_Coding_Manual_Version_3-a.pdf
. Retrieved August 2, 2011.
19. Kung H, Hoyert DL, Xu J, SL M. Deaths: Preliminary data for 2005. Heath E-Stats 2007. Available at: http://www.cdc.gov/nchs/data/hestat/prelimdeaths05/prelimdeaths05.htm
. Retrieved August 2, 2011.
20. Centers for Disease Control and Prevention. National Vital Statistics System: birth data. Available at: http://www.cdc.gov/nchs/births.htm
. Retrieved February 18, 2010.
21. Xu J, Kochanek K, Murphy S. Deaths: Final data for 2007. Hyattsville (MD): National Center for Health Statistics; 2010.
22. Karch DL, Lubell KM, Friday J, Patel N, Williams DD. Surveillance for violent deaths–National Violent Death Reporting System, 16 states, 2005. MMWR Surveill Summ 2008;57:1–45.
23. 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.
24. Cheng D, Horon IL. Intimate-partner homicide among pregnant and postpartum women. Obstet Gynecol 2010;115:1181–6.
25. Parsons LH, Harper MA. Violent maternal deaths in North Carolina. Obstet Gynecol 1999;94:990–3.
26. 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.
27. Violence against women: data on pregnant victims and effectiveness of prevention strategies are limited. 2002. Available at: http://www.gao.gov/new.items/d02530.pdf
. Retrieved August 2, 2011.
28. Pallin DJ, Sundaram V, Laraque F, Berenson L, Schomberg DR. Active surveillance of maternal mortality in New York City. Am J Public Health 2002;92:1319–22.
31. Yonkers KA, Wisner KL, Stewart DE, Oberlander TF, Dell DL, Stotland N, et al.. The management of depression during pregnancy: a report from the American Psychiatric Association and the American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114:703–13.
32. Flynn HA, O'Mahen HA, Massey L, Marcus S. The impact of a brief obstetrics clinic-based intervention on treatment use for perinatal depression. J Womens Health (Larchmt) 2006;15:1195–204.
33. Miller L, Shade M, Vasireddy V. Beyond screening: assessment of perinatal depression in a perinatal care setting. Arch Womens Ment Health 2009;12:329–34.
34. Flynn HA, Henshaw E, O'Mahen H, Forman J. Patient perspectives on improving the depression referral processes in obstetrics settings: a qualitative study. General Hospital Psychiatry 2010;32:9–16.
35. Sharps PW, Laughon K, Giangrande SK. Intimate partner violence and the childbearing year: maternal and infant health consequences. Trauma Violence Abuse 2007;8:105–16.
37. Kiely M, El-Mohandes AAE, El-Khorazaty MN, Gantz MG. An Integrated Intervention to Reduce Intimate Partner Violence in Pregnancy A Randomized Controlled Trial. Obstet Gynecol 2010;115:273–83.
38. Berg CJ, Callaghan WM, Syverson C, Henderson Z. Pregnancy-related mortality in the United States, 1998 to 2005. Obstet Gynecol 2010;116:1302–9.
39. Centers for Disease Control and Prevention. Surveillance for violent deaths-National Violent Death Reporting System, 16 states, 2005. Surveillance summaries, April 11, 2008. MMWR Morb Mortal Wkly Rep 2008;57:SS–3.