In the mid-1990s, Cara Krulewitch sat in a dark, cramped file room in the office of the D.C. medical examiner, poring over autopsies for days that became weeks, then months. She was an epidemiologist with the Centers for Disease Control and Prevention, assigned to the District.
Krulewitch wanted to see whether maternal deaths were being undercounted, as was common elsewhere across the country. Granted access to confidential death files, she assumed she would find more deaths from medical complications of pregnancy -- embolism, infection, hemorrhage -- than anyone knew.
What she stumbled upon instead was a surprising number of homicides: 13 of 30 maternal deaths, more than 40 percent. "I was just stunned," she recalled. "You assume it's a quirk in the numbers. A blip."
Krulewitch dug into medical archives and came across a 1992 journal article from Chicago and a 1995 study from New York City. In both, homicide had emerged as a significant cause of maternal death. It was difficult for the uninitiated to comprehend: Were pregnant women being killed in notable numbers?
"I didn't understand it at all," said Krulewitch, whose study was published in the Journal of Midwifery & Women's Health.
Her research came at a time when maternal mortality rates in the United States had fallen a full 99 percent from the last century, with fewer than 500 women a year dying of medical problems related to childbearing.
Health officials considered this a major achievement but also had set optimistic new goals to bring the death toll closer to what is called an irreducible minimum.
Still, there was a growing interest in doing a better job of capturing every possible case -- and taking note of homicides, suicides, car accidents and drug overdoses. In the larger public health world, the "social" causes of death were increasingly viewed as an important health issue.
"For a long time, violence was not defined as a public health problem," said Jacquelyn Campbell, who studies domestic homicides at Johns Hopkins University.
Even now, studies that analyze maternal homicide are relatively rare.
One of the most comprehensive studies came from Maryland, where researchers used an array of case-spotting methods, expecting to find more medical deaths than the state knew about. Instead they discovered that homicide was the leading cause of death, a finding published in 2001 in the Journal of the American Medical Association.
In 2002, Massachusetts weighed in with a study that also showed homicide as the top cause of maternal death, followed by cancer. Two of three homicides involved domestic violence. "This is clearly a major health problem for women," said Angela Nannini, who led the study.
Still, many questions remain unanswered.
"We don't even know what we don't know yet," said Elaine Alpert of Boston University. "We need to look not only at all the contextual factors that may have contributed to a mother's death, but also look to compare deaths and see commonalties between these cases."