January 12, 2005, 7:30 a.m.
The Mother of All Bad Stats?
A leading cause of public confusion.

Any claim, when oft repeated, may acquire the ring of truth. From tales of the deadly gut-busting power of mixing cola and candy pop rocks to the ghoulish origin of "Ring around the Rosie," people often accept myths as fact. These popular tall tales generally are harmless. Yet on some issues, such as violence against women, it's important that the public knows the facts.

The high-profile murders of pregnant women such as Laci Peterson in late 2002 and Bobbie Jo Stinnett — the 23-year-old whose baby was stolen from her womb in December — has prompted a widespread belief that pregnant women are more likely to die from homicide than from any other cause. An Associated Press story in April 2003, for example, was headlined "Murder: The Leading Cause of Death for Pregnant Women." In that article, Kim Gandy, president of the National Organization for Women, summed up the relationship between violence and pregnancy: "She's vulnerable. It's an easier time to threaten her."

But how do we know that homicide actually is the leading cause of death among pregnant women? The Washington Post recently ran a three-part series exploring the problem of the murder of pregnant women and cited a study conducted in Maryland that was published by The Journal of the American Medical Association (JAMA) in 2001 that appears to be the source of the information. Isabelle Horon and Diana Cheng, authors of the Maryland study, concluded that "A pregnant or recently pregnant woman is more likely to be the victim of homicide than to die of any other cause." Among 247 women in Maryland between 1993 and 1998 who died while pregnant or within a year of having been pregnant, 50 died of homicide. The next greatest killers were cardiovascular disorders (48), embolisms (21), accidents (18), and hemorrhage (17).

The data for the JAMA study deserves additional scrutiny. First, the authors defined a "pregnant woman" not as a woman carrying a fetus at the time of her death, but as all women pregnant or who had been pregnant within a year before their deaths. The expansive definition of "pregnancy-associated" deaths conveys a false sense that all of these deaths were directly related to the women having been pregnant. Almost three quarters of the deaths examined in the study occurred during the yearlong period after the woman was no longer pregnant. A year after a woman stops being pregnant (whether by giving birth, abortion, or miscarriage) seems a long time to assume that her death is related to pregnancy.

In addition, homicide wouldn't be the leading cause of death if the categories were defined differently. If different (but related) health disorders were combined, they would by far be the leading cause.

Dr. Harold Weiss, of the Center for Injury Research and Control, identified another way in which the study may be misleading. In a letter to JAMA following the publication of Horon and Cheng's study, Weiss highlighted how Maryland's policy regarding autopsies may contribute to an under-counting of death due to motor-vehicle accidents. The Maryland medical examiner's office requires an autopsy for all murder victims, but does not for all auto accidents. This means that while medical-examiner reports identified all murdered pregnant women — and, in fact, these reports were the source of 100 percent of Horon and Cheng's identification of pregnant homicide victims — the lack of data on women who die in motor-vehicle accidents means that Horon and Cheng likely missed some pregnant women who died in auto accidents.

Weiss also highlighted the pitfalls of assuming that the findings from a study conducted in Maryland are nationally representative. It turns out that Maryland has many more homicides per capita than the rest of the country. Between 1993 and 1998, among Maryland women of childbearing age (15 to 44 years), there were 499 homicides and 605 motor vehicle deaths — a ratio of .82 homicides for each motor vehicle death. Nationally, however, there were 19,306 homicides and 41,474 motor-vehicle deaths, or .47 homicides per motor-vehicle death. Thus even if homicides are the leading cause of death among pregnant women in Maryland, the same trend doesn't necessary follow nationally.

The General Accounting Office and Center for Disease Control have also researched the relationship between pregnancy and violence, and cast additional doubt on the alleged link between pregnancy and homicide. One report concluded that "current study findings suggest that for most abused women, physical violence does not seem to be initiated or to increase during pregnancy.... In one study we reviewed, only 2 percent of women who reported not being abused before pregnancy reported abuse during pregnancy. The same study also found that, for some women, the period of pregnancy may be less risky, with violence abating during pregnancy; 41 percent of the women who reported abuse in the year before pregnancy did not experience abuse during pregnancy."

Violence against women, including against pregnant women, is a serious problem. Each such death is a special tragedy — the loss of the woman compounded by the extinguished hope and promise of a new life. But overstating the frequency of these brutal acts minimizes the horror of each episode, confuses the public, and may lead to misallocated resources. We need a national discussion about how best to prevent violence against women — and an important element of that conversation must be separating fact from fiction.

— Carrie Lukas is the director of policy at the Independent Women's Forum.

*   *   *

YOU’RE NOT A SUBSCRIBER TO NATIONAL REVIEW? Sign up right now! It’s easy: Subscribe to National Review here, or to the digital version of the magazine here. You can even order a subscription as a gift: print or digital!