Loading today’s stories
Science1 distinct publisher3 min readPublished
A six-year analysis of US violent death records finds single-, multi-victim and mass homicides differ in who dies and where, yet the same precipitant shows up in all three, which changes where prevention money reaches the most people.
The Scientist · Science desk
Compiled by The ScientistSomething wrong?How this is made
Read the intimate partner violence percentages back through the case counts and the allocation question gets concrete. Thirty-one percent of 892 mass-homicide deaths is about 277 [15]. Applied to the other two categories, 16.9% of 10,123 multi-victim deaths is about 1,711 [16], and 15.0% of 104,858 single-victim deaths is about 15,729 [17]. Together that is roughly 17,700 deaths in six years with IPV recorded as a precipitant, and mass homicides are about 1.6% of them [18]. Screening aimed at the single-victim population therefore sits upstream of roughly 57 times as many IPV-linked deaths as anything aimed at the mass-casualty tail [19].
The thing this doesn't tell you is whether those published shares count victims or incidents. The summary does not specify, and the ambiguity bites hardest where the numbers are smallest: 892 mass-homicide deaths came out of 178 incidents, about 5.0 victims apiece [20], against about 2.1 per multi-victim event [21]. "Precipitating circumstance" is also an abstraction from case records rather than a causal test. NVDRS assembles it by linking death certificates, medical examiner findings and law enforcement reports, as Annals of Epidemiology editor in chief Patrick Sullivan describes it [13]; a flag means investigators documented partner violence in the run-up, not that partner violence has been shown to have caused the death.
The composition also moves with the map. The 8.7% multi-victim share here is higher than in the earlier NVDRS analysis covering 2003 to 2017, and the authors offer two candidate explanations: more states joined the reporting system, or homicide patterns genuinely changed, with national rates rising over the same years [11]. Those two are not separable in this design, which is why the file supports claims about mix and not about risk.
One number resists a tidy gradient. Firearms were the leading method in every category, but the highest share falls in the middle, at 86.3% of multi-victim deaths against 76.4% of single-victim and 75.8% of mass [9][24]. Victim count is not what orders that ranking; setting is a better candidate, given that mass and multi-victim cases clustered in homes while single-victim deaths clustered on streets and in vehicles [7].
My reading, with its conditions attached: the widest-reach intervention implied by this table is IPV screening wherever the at-risk population already appears, because IPV is the one precipitant recurring across all three types [5] and because the volume sits in single-victim cases. The conditions are that this is a proportions argument rather than an effect estimate, that screening only helps if it connects to something that acts, and that mass-casualty-specific programs cover 0.77% of the deaths in this file [22], so they need a justification other than reach.
Ranked by verification strength, evidence, and original report placement.
A study published in Annals of Epidemiology analyzed 115,873 homicide deaths recorded in the CDC's National Violent Death Reporting System (NVDRS) from 2018 to 2023, categorizing them as single-victim, multi-victim (2-3 victims) and mass (4 or more victims, excluding the perpetrator).
Single-victim cases made up 90.5% of the deaths analyzed, or 104,858 deaths.
Multi-victim homicides claimed 10,123 lives across 4,789 events, 8.7% of the total, a higher percentage than in the prior study using NVDRS data from 2003 to 2017.
Mass homicides involving four or more victims accounted for less than 1.0% of the total: 892 deaths across 178 incidents, with individual cases spanning from 4 to 51 fatalities.
IPV was a precipitating circumstance in 31.0% of mass homicides, 16.9% of multi-victim homicides and 15.0% of single-victim homicides.
The highest proportion of mass homicide victims were non-Hispanic White (39.7%); the highest proportion of multi-victim and single-victim homicide victims were non-Hispanic Black (multi 47.8%, single 54.8%).
Distinct publishers with included, body-backed reporting in this cluster.
phys.org
1 article · September 2, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
science
Cyclosporiasis cases are easing, but one state's count is 95% of the national total1 distinct publisher
science
US life expectancy hit 79 in 2024, and the cushion over 2019 is about 73 days1 distinct publisher
science
Massachusetts pediatricians who keep vaccine-hesitant families are testing dismissal's premise1 distinct publisher
science
Two cases a year: a Louisiana amoeba death tells northern states little about their own lakes1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One paper, one retelling, numbers that reconcile
Every figure traces to a single peer-reviewed paper and a single account of it, down to the DOI phys.org prints at the bottom. In its favour: the parts add up exactly — 104,858 plus 10,123 plus 892 is the 115,873 deaths claimed — so the tallies are at least internally honest, and they come from linked death certificates, medical examiner findings and police reports rather than a survey. Against it: the circumstance coding that carries the whole IPV finding is never characterised for completeness, and the authors' own doubt about whether the multi-victim rise is real or an artifact of states joining NVDRS is raised and then left standing.
No uptake to measure
Nobody in this reporting has acted on the findings. The study gestures at 'tailored prevention strategies' and Sullivan calls them guideposts, but no health department, funder or program appears here changing a screening protocol or shifting a budget line. Uptake of a public-health finding is perfectly measurable — we simply have no measurement.
Emphasis lands on the rarest category
The number destined for quotation is IPV precipitating 31% of mass homicides — drawn from the smallest pool in the study, 892 deaths, under 1% of the total. Run the same percentages across the categories and IPV-associated single-victim deaths outnumber IPV-associated mass deaths by something like 57 to 1. phys.org does print all three shares, which is why this is a tilt rather than a distortion: the arithmetic is sound, the ordering is inverted.
The journal introducing its own study
One voice speaks in this story, and it belongs to the editor in chief of the journal that published the paper, commending both the paper and the federal data system behind it. Less a conflict than a closed loop: the journal announces, the outlet reproduces the announcement including the citation, and the question an outside epidemiologist would ask — is the multi-victim increase real? — goes unasked. The countervailing pressure is mild, since nobody here has a product to sell.
Solid counts, unverified reading
We would stand behind the tallies and hold the interpretation loosely. The counts are specific, self-consistent and traceable to a DOI. The conclusion drawn from them — three distinct problems needing three prevention playbooks — rests on circumstance codes of undisclosed completeness, reaches us from one outlet echoing one journal, and sits beside a firearm pattern that peaks in the middle category rather than rising with victim count, which is the sort of wrinkle a second read would have to explain.