Examining Homicide's Contribution to Pregnancy-Associated Deaths

Maternal mortality has been defined traditionally as deaths "related to or aggravated by" pregnancy complications, excluding "accidental or incidental causes," occurring during pregnancy or within 42 days of termination.¹ In other words, maternal deaths are those that would not have occurred if not for the pregnancy. This definition restricts the potential causal contribution of pregnancy to women's deaths, because only causes of death that are biologically related to the pregnancy are counted. However, some deaths are not biologically, but may be socially, related to the pregnancy and may not have occurred without pregnancy. To capture more completely all deaths occurring during or after pregnancy, the term "pregnancy-associated death," which includes "the death of any woman, from any cause, while she is pregnant—or within one year of termination of pregnancy," was developed.² Concurrently, enhanced maternal mortality surveillance techniques were being developed and tested, leading to previously underestimated maternal mortality rates being recalculated.^{3 - 6}

In this issue of THE JOURNAL, Horon and Cheng⁷ report the leading causes of pregnancy-associated death in Maryland between 1993 and 1998, using this expanded definition of maternal mortality and enhanced surveillance methods. After standardizing the data for the pregnant population to the age and race distribution of the nonpregnant population, the authors were surprised to find that homicide was the leading cause of death among pregnant or recently pregnant women in Maryland. In contrast, homicide was the fifth leading cause of death among nonpregnant women during the same time period. Homicides made up 20% of all pregnancy-associated deaths, more than twice as many deaths as embolisms, which comprised 9% of pregnancy-associated deaths.

The role of homicide in maternal mortality has been examined previously. Parsons and Harper⁸ linked reproductive-age womens' death certificate data to live birth and fetal death records in North Carolina between 1992 and 1994 and found that of 167 pregnancy-associated deaths, 22 (13%) were due to homicides. The authors did not use the term "pregnancy-associated deaths," per se, but defined maternal deaths as deaths of women from any cause while pregnant or within 1 year of pregnancy termination. Comparing injury deaths among postpartum women with those among nonpregnant, nonpostpartum women between 1990 and 1992, Dietz et al³ found that 29 (18%) of the 165 deaths among postpartum women were due to homicide. The authors defined "postpartum women" as those aged 15 to 44 years who had had a live birth in the year before their deaths. Risk ratios revealed that while postpartum women overall were not at increased risk of homicide, postpartum women aged 15 to 19 years were 2.63 times more likely to be killed than other women.

In New York City, Dannenberg et al⁹ examined maternal deaths due to injury between 1987 and 1991 and found that homicides made up 72 (25%) of the 293 deaths among pregnant or recently pregnant women. Maternal death was defined as deaths among pregnant women and women who had, according to medical examiner and autopsy records, been pregnant or delivered an infant in the past 6 months. The authors found that while homicides occurred among pregnant women at about the rate expected, they occurred significantly more often among pregnant black women. Fildes et al¹⁰ also reviewed all maternal deaths occurring in Cook County, Illinois, between 1986 and 1989 and found that homicides made up 25 (26%) of the 95 maternal deaths identified. These authors defined maternal deaths as deaths due to any cause among pregnant or 90-day postpartum women. In conjunction with the findings of Horon and Cheng,⁷ these findings reveal that across the United States, homicide is a considerable source of mortality among pregnant and postpartum women and that certain population subgroups may be at increased risk of homicide during pregnancy and postpartum.

When considering these findings, it is important to note that in the United States, homicide is a leading killer of young women, pregnant or not. In 1998, homicide was the second leading cause of death among women aged 15 to 24 years and the sixth among women aged 25 to 44 years.¹¹ While good evidence exists that nonlethal violence occurs in approximately 4% to 8% of all pregnancies and that violence during pregnancy may be more common than other conditions for which health care professionals routinely screen,¹² whether pregnancy is an independent risk factor for lethal violence is unclear. However, pregnancy as an independent risk factor for lethal intimate partner violence is currently under investigation.¹³ Thus far, research indicates that much of the violence that women experience during pregnancy is perpetrated by intimate partners¹⁴ and that, for some, intimate partner violence begins during pregnancy.¹⁵ The proportion of pregnancy-associated homicides committed by intimate partners is not reported in the study by Horon and Cheng,⁷ but evidence exists that a significant proportion of all female homicide victims are killed by their intimate partners. A recent report on female homicide in New York City between 1990 and 1997 revealed that among the 54% of cases that could be categorized according to intimate partner perpetrator status, approximately 40% of victims were killed by intimate partners.¹⁶

Homicide by intimate partners may offer a focal point for effective pregnancy-associated mortality prevention efforts because many of these women come into contact with the health care system before their deaths. For example, Campbell et al¹³ reported that 47% of completed or "attempted" intimate partner homicide victims were seen by health care professionals in the year before their deaths. In 1995, approximately 72% of reproductive-age women in the United States received reproductive health care,¹⁷ and many pregnant women receive some form of prenatal care during their pregnancies. The American College of Obstetricians and Gynecologists recommends that all female patients be screened routinely for domestic violence,¹⁸ yet, a recent survey of obstetricians and gynecologists found that just 17% of them screened at a first visit, and only 10% screened regularly at check-up visits.¹⁹ While some abused women delay entry into prenatal care,²⁰ when health care professionals screen for domestic violence and offer effective interventions, women report significantly more safety behaviors during and after pregnancy.²¹ To prevent intimate partner homicides of pregnant and postpartum women, health care professionals must be willing to undertake routine domestic violence screening and implement sensitive interventions where appropriate. If they are not, we will continue to read reports finding that only one fourth of clinicians whose patients were killed by intimate partners were aware of a history of or suspected current abuse by an intimate partner.⁸

Pregnancy-associated death represents a largely preventable source of premature mortality among young women in the United States and devastates the children, families, and communities left behind. The enhanced surveillance techniques used by Horon and Cheng⁷ and others reveal that when an expanded definition of maternal mortality is used to detect these deaths, threats from a woman's social environment, like homicide, may be found to be more deadly than those from her biological environment. Therefore, in addition to identifying traditional biological threats to pregnant and postpartum women's health, further research efforts should be directed toward evaluating social and biological risk factors for pregnancy-associated death. Further, strategies to promote safe motherhood must integrate these findings into their approach to maternal mortality reduction. Finally, despite the difficulties associated with screening for social conditions like violence, all conditions that are determined to contribute significantly to pregnancy-associated mortality should be assessed routinely and sensitively. A first step in this direction would be adherence among clinicians to the American College of Obstetricians and Gynecologists' recommendations for domestic violence screening.