The Autopsy / Case File · Violence & Homicide
The Eleven Questions
Most intimate-partner homicides are preceded by a police contact that never flags the danger. A short screening tool built in Maryland in 2005 gives officers eleven questions to ask instead — and a Department of Justice-funded study found it catches 92 percent of the people who go on to be nearly killed.
Domestic-violence calls are among the most common a patrol officer runs, and historically among the least likely to be treated as a preview of a killing. The Lethality Assessment Program changes that math with a research-based screen, administered at the scene, that sorts victims by measurable risk factors instead of an officer's instinct — and the outcome data behind it is now driving a push to make the eleven questions standard nationwide.
Join the discussion →
Before 2005, an officer who responded to a domestic-violence call in Maryland had a police report, a set of statutes, and whatever judgment the moment allowed. What the officer typically did not have was a structured way to tell a routine-looking argument apart from the specific pattern of behavior — strangulation, a threat with a weapon, an escalating pattern of control — that the research literature already associated with an intimate partner going on to kill. The Maryland Network Against Domestic Violence built the Lethality Assessment Program to close that gap: an eleven-question screen, administered by the first responder standing in the room, that turns decades of intimate-partner-homicide research into a checklist an officer can run in minutes.
A Screen Built From the Danger Assessment
The Lethality Screen didn't originate as a police tool. Its underlying risk factors trace to the Danger Assessment, a research instrument developed by Johns Hopkins University nursing researcher Dr. Jacquelyn Campbell from case-control studies of intimate-partner homicide and near-homicide. Maryland Network Against Domestic Violence adapted that research into a field-usable, eleven-question version for law enforcement in 2005, then built a two-part protocol around it: officers ask the questions at the scene, and any victim who screens as high-risk is connected, immediately and by phone, to a domestic-violence hotline advocate for safety planning — while the officer is still standing there.
The individual questions ask about factors the underlying research had already identified as carrying outsized weight in lethal and near-lethal cases: has the abuser ever used a weapon against the victim or threatened her with one, has he ever tried to choke her, is he unemployed, has he ever threatened to kill her or himself, does he own a gun, has the violence increased in frequency or severity over the past year, has she left him after living together in the past year, is he violently or constantly jealous, has he ever forced her to have sex when she didn't want to, does he follow or spy on her, and has he ever threatened to harm her children. None of the questions ask an officer to make a clinical judgment call; they ask for facts a victim can answer in a minute or two, standing in a doorway or a squad car, at the worst moment of her week.
What "It Works" Actually Means Here
Two different kinds of evidence back the program, and they answer two different questions. The first is predictive validity: does the screen correctly identify who is actually in danger? A Department of Justice-funded evaluation found the eleven-question screen had 92.86% sensitivity for predicting subsequent near-lethal violence — meaning the overwhelming majority of women who went on to be nearly killed had, in fact, screened as high-risk beforehand. Its positive predictive value was far lower, in the range of 13%, which is the flip side of a screen tuned to minimize false negatives: it flags many people who go on to experience less severe violence for every one who experiences a near-fatal assault, a deliberate design tradeoff because the cost of missing a genuinely lethal case is so much higher than the cost of over-flagging a case that turns out less severe.
The second kind of evidence is population-level outcome data: did the program actually reduce homicides where it was deployed? Maryland's own tracking found a 34% drop in intimate-partner homicides in the five-year period following the program's statewide police rollout, a figure the Maryland Network Against Domestic Violence and multiple state legislative summaries have cited since as the program's headline result. A separate academic evaluation, of a later Oklahoma implementation studied over roughly three and a half years across seven police jurisdictions, used a quasi-experimental design and found the program didn't significantly reduce the overall frequency of subsequent violence, but did measurably reduce its severity, increased victims' use of safety planning and hotline services, and left participating victims more satisfied with how police handled their case than victims in jurisdictions without the screen.
The screen isn't asking an officer to predict the future. It's asking a victim eleven questions the research already knows are the ones that separate a bad relationship from one where somebody is about to die.
Read together, the two findings aren't in tension so much as measuring different things: the Oklahoma study's null result on overall violence frequency describes what happens to the broad population of women who screen at any risk level, while Maryland's homicide-rate figure describes the outcome researchers designed the tool specifically to move — the rare, catastrophic end of the distribution. A tool built to flag the minority of cases heading toward a homicide can succeed at that narrow task while showing little detectable effect on the much larger number of cases that never would have ended in one.
From One State to Six Hundred Departments
Adoption has scaled well past Maryland's borders. As of the most recent nationwide tallies, more than 600 police agencies across 37 states have implemented some version of the Lethality Assessment Program, coordinated through a national training structure the Maryland Network Against Domestic Violence built to support departments adopting the model. That growth has happened without a federal mandate or a standardized national training requirement — individual states and departments have opted in one at a time, based largely on the same outcome data described above circulating through law-enforcement and domestic-violence-advocacy networks.
That patchwork is also the program's most obvious limitation. A screen that measurably reduces homicides only helps the victims whose department has actually adopted it, trained officers to administer it consistently, and built a live referral pipeline to an advocate who can pick up the phone in real time. In 2026, bipartisan federal legislation — introduced as the Preventing Domestic Violence Homicides Through Lethality Assessment Training and Technical Assistance Act — proposed authorizing the U.S. Attorney General to establish a national training center to standardize the program's rollout nationwide, an explicit acknowledgment that adoption has plateaued at "hundreds of departments," not "all of them," a full two decades after the tool was built.
The Point Where the System Already Has Contact
What makes the Lethality Assessment Program distinct from most domestic-violence interventions is where it intervenes: not at a shelter intake, not at a court hearing weeks later, but at the first police contact, before a case has been filed, before a protective order has been requested, sometimes before the victim herself has decided to leave. Advocates have long identified this contact point — the 911 call, the officer standing in the living room — as a uniquely high-leverage moment precisely because it's a point where the system already has a foot in the door regardless of whether the victim independently seeks help afterward. A woman who would never call a domestic-violence hotline on her own may still answer eleven questions from an officer already standing in front of her, and that single difference is the entire mechanism the tool is built around.
Critics of risk-assessment tools generally — including scholars who have raised concerns about lethality and danger-assessment instruments more broadly — have cautioned that any single screening tool risks over-relying on a checklist in place of trained judgment, or providing false reassurance to victims who screen low-risk but are still in real danger. The Lethality Assessment Program's own materials are explicit that a low-risk screen result doesn't mean no risk; the protocol still directs every victim, regardless of score, toward available resources. The tool's design intent has always been triage under real-world time and staffing constraints, not a diagnostic guarantee — a distinction that matters because the alternative it replaced wasn't a better assessment, it was frequently no structured assessment at all.
And Then You're Dead
Nobody's homicide file says the eleven questions were never asked. It says intimate-partner violence, or domestic dispute, or a case number from a call three months before the one that didn't end with anyone walking back out to the car. The gap between those two facts is a phone call to a hotline advocate that either happened at the scene or didn't — and whether it did came down, in six hundred departments and counting, to whether anyone had trained the officer standing in the doorway to ask.
That's it. That's the whole thing.
Read about intimate-partner homicide and the firearm method gap · or see every case file
Sources
- The Lethality Assessment Program's origin, the eleven-question Lethality Screen, its basis in Dr. Jacquelyn Campbell's Danger Assessment research, and background on the two-part protocol (screening plus immediate hotline referral) — Maryland Network Against Domestic Violence, "Position Paper: Effectiveness of the Lethality Assessment Program," June 2022, and the Office of Justice Programs abstract, "Lethality Assessment Program Maryland Model For First Responders: Learning to Read the Danger Signs"
- The 92.86% predictive sensitivity finding and the Oklahoma quasi-experimental evaluation (severity/frequency reduction, hotline engagement, victim satisfaction with police response) — National Institute of Justice, "How Effective Are Lethality Assessment Programs for Addressing Intimate Partner Violence?"
- The 34% drop in Maryland intimate-partner homicides, July 2007-June 2012, and the 2005 program launch date — cited via Rhode Island Current, "Police officers can help prevent domestic violence killings by asking the right questions"
- The 600-plus police agencies in 37 states adoption figure and the 2026 federal Preventing Domestic Violence Homicides Through Lethality Assessment Training and Technical Assistance Act — Capital News Service Maryland, "Elfreth legislation seeks to reduce domestic violence homicides nationwide"
- Background on scholarly critique of lethality and danger-assessment risk-screening tools generally — Evan Stark, Ph.D., MSW, "The Dangers of Dangerousness Assessment," Family & Intimate Partner Violence Quarterly
Photo: "Minneapolis Police Squad Car," by Tony Webster, via Wikimedia Commons (CC BY 2.0).
Comments
Loading…