And Then You're Dead

The Autopsy / Case File · Violence & Homicide

Killed By the Person Closest to Her

Nearly 4,000 women were killed by an intimate partner in four years of federal surveillance data. Two in three deaths involved a gun.

When a woman is murdered in the United States, the person most likely to have done it isn't a stranger. It's a current or former partner, most often at home, most often with a firearm. Here's a sourced look at what federal violent-death surveillance actually shows about the pattern, the specific window when the risk spikes, and the screening tool built to catch it before the fatal call.

And Then You're Dead · September 2026 · 8 min read

Join the discussion →
A military commander signs a domestic violence awareness proclamation before a 5K run/walk
An October awareness event, one of thousands held each year. The data behind it is federal, current, and specific.

The CDC's National Violent Death Reporting System pulls together coroner and medical examiner reports, police records, and death certificates from nearly every state, case by case, to answer a question ordinary crime statistics usually can't: not just how a woman died, but who did it, where, and what happened right before. Four years of that data, covering 2018 through 2021, produced a specific and consistent picture.

The Numbers

Among 3,991 women identified by NVDRS as victims of intimate partner homicide over that four-year span, the pattern held with unusual consistency year over year, rather than spiking around any single event or region.

0
women killed by a current or former intimate partner, identified in CDC's National Violent Death Reporting System, 2018–2021
0%
of those killings that involved a firearm

The surveillance data also captures the setting and circumstance with a level of detail an ordinary crime blotter doesn't. The overwhelming majority of these killings weren't chance encounters or crimes of opportunity committed by a stranger — they happened in the one place a victim might otherwise expect to be safest.

0%
of intimate partner homicides of women that occurred at the victim's own residence
0%
of these homicides committed by a male suspect

The rate held essentially flat between the first two years of the study period and the last two — not a crisis spiking suddenly into view, but a stable, ongoing mortality pattern that federal surveillance has now spent years documenting in the same detail as any infectious disease outbreak.

A Gun in the House Changes the Math

The firearm figure isn't incidental to the story — in the epidemiology of intimate partner violence, it's close to the whole story. A landmark case-control study led by Johns Hopkins nursing researcher Jacquelyn Campbell, drawing on interviews with the family members of homicide victims and with survivors of near-fatal violence, found that a gun's mere presence in a home where abuse is already occurring transforms the risk calculation entirely.

Access to a firearm increases the risk that an abusive partner will kill their victim by roughly 500%, independent of every other risk factor researchers controlled for.

That figure comes from formal risk-factor modeling, not intuition — it holds after accounting for prior threats, prior violence, substance use, and the relationship's history. A firearm doesn't just provide a method; it changes the odds an existing pattern of abuse ends in a killing rather than continuing as one more incident in a long history of them.

The Federal Gun Law That Doesn't Cover Everyone

Federal law has prohibited people convicted of a qualifying domestic violence misdemeanor, or subject to certain protective orders, from possessing a firearm since 1996's Lautenberg Amendment. For nearly three decades, though, that prohibition applied only to spouses, people who lived with the victim, or people who shared a child with them — not to a dating partner who had never lived with the victim, regardless of how serious or violent the relationship had been. Advocates and researchers came to call the gap the "boyfriend loophole," and it mattered because dating relationships, not marriages or cohabitations, make up a substantial share of intimate partner violence cases nationally.

0 yrs
duration of the federal firearm prohibition the 2022 Bipartisan Safer Communities Act extended to convicted misdemeanant dating partners, closing the boyfriend loophole at the federal level for the first time
0M
prohibited firearm purchases blocked by federal background checks since the system's 1994 launch — the enforcement backbone the 2022 law extended to a wider group of abusers

The 2022 fix closed the federal gap only for people actually convicted of a qualifying misdemeanor or under a qualifying protective order — it does nothing for the far larger number of abusive relationships that never produce a conviction or a court order at all, whether because a victim didn't report, a case wasn't prosecuted, or a judge declined to issue a restraining order. The law changed who counts as prohibited. It didn't change how many dangerous relationships actually reach that legal threshold in the first place.

The Most Dangerous Time Is Leaving

One of the most consistent, and most counterintuitive, findings in intimate partner homicide research is that the danger to a victim doesn't peak during the relationship's worst period of ongoing abuse — it peaks after the victim tries to leave. Separation represents a victim reclaiming control the abuser has spent the relationship establishing, and for a subset of abusers, that loss of control is itself the trigger for lethal violence.

Research funded by the National Institute of Justice, built on the same case-control methodology behind the firearm-risk finding, identified recent or attempted separation as one of the strongest predictors of intimate partner homicide across the cases studied — stronger than many of the risk factors that intuitively seem more alarming, like a documented history of the most severe prior violence. The take-home implication for anyone working with victims is uncomfortable: the advice to "just leave" is not straightforwardly safety advice. Leaving well, with a safety plan built around the specific danger period that follows it, is a different and more survivable thing than leaving abruptly.

A Screening Tool Built for the 911 Call

Some police departments have built that danger-period research directly into how officers respond to a domestic violence call. The Lethality Assessment Program, developed by the Maryland Network Against Domestic Violence, gives responding officers an 11-question screening instrument — has the abuser ever used a weapon, threatened to kill the victim or someone she loves, choked her, threatened suicide — that flags a subset of calls as carrying an elevated risk of homicide, in real time, before anyone leaves the scene.

0%
decline in intimate partner homicides in Maryland jurisdictions during the program's 2007–2012 implementation window
0
screening questions in the Lethality Assessment Program's checklist, administered by the responding officer at the scene

A high-risk score triggers an on-the-spot warm handoff: the officer calls a domestic violence hotline directly from the scene and puts the victim on the phone with an advocate before leaving, rather than handing over a pamphlet and a number to call later. Outcome research has found the program measurably increases victims' safety planning and help-seeking, on top of the homicide-rate decline observed during Maryland's rollout. It's now used by police departments in most states — but adoption remains a department-by-department choice, not a national requirement, which means whether a given 911 call gets this specific screening still depends heavily on which jurisdiction answers the phone.

The Disparity Inside the Pattern

The overall numbers above describe a national average. They obscure a racial gap that CDC researchers have documented as persistent across more than two decades of data, not a recent development or a single bad year.

0
homicides per 100,000 non-Hispanic Black women, the highest rate of any racial or ethnic group studied, versus roughly 1–2 per 100,000 for most other groups
0
homicides per 100,000 American Indian/Alaska Native women, the second-highest rate — both figures from CDC's own analysis of NVDRS cases, 2003–2014

The same CDC analysis found the gap "virtually unchanged" across the study period — not a disparity actively narrowing as public health attention and hotline funding have grown, but one that has simply persisted underneath the national totals reported each year. Intimate partner circumstances specifically, not homicide broadly, is where researchers found the disparity concentrated: Black and Native women weren't just more likely to be murdered overall, they were more likely to have been murdered by a partner when the case was resolved with a known relationship between victim and suspect. A national mortality pattern this well-documented, in other words, is really several overlapping patterns of unequal risk, not one evenly distributed one.

And Then You're Dead

The data here doesn't describe a random hazard, the kind no one could have seen coming. It describes a pattern precise enough that a police department can build an 11-question checklist around it and measurably cut the number of women who don't survive. The mechanism is documented, the highest-risk moment is documented, even the specific weapon is documented two-thirds of the time. What's still missing, case by case, isn't data. It's the intervention arriving before the moment the data already knows is coming.

That's it. That's the whole thing.

Comments

Loading…

Comments are anonymous and moderated.

Sources

Photo: U.S. Air Force photo by Airman 1st Class Gabrielle Spradling, via Wikimedia Commons / DVIDS (public domain).