#963 2002 · Cardiff University (Jonathan Shepherd) · Public health / policing
Cardiff cut street violence by piping hospital ER data to police instead of crime reports
the problem
Police violence maps were useless because most assault victims stitched up in the ER never filed a police report
background
Jonathan Shepherd, an oral and maxillofacial surgeon, spent his career reconstructing the faces of Cardiff's assault victims and noticed a pattern police never saw: most of the young men he treated for bottle and glassing injuries had never reported the attack. Police were mapping violence hotspots and allocating patrols from crime reports, but Shepherd's own audits found that as few as a quarter of the assaults treated in his emergency department had ever reached a police officer's notebook — fights outside unlicensed venues, brawls that ended before anyone dialed 999, victims who simply didn't trust the police to help.
The conventional fix was to get victims to report more, which decades of campaigns had failed to move. Shepherd's problem wasn't a shortage of violence data — it was that the data lived in the wrong building. Every stabbing, glassing and beating that mattered clinically already walked through a hospital's front door with a location, a time and a weapon attached to it, recorded for treatment purposes, in a building nobody had thought to ask.
what everyone would do
Every standard lever pointed at the victim or the offender: run more anti-violence campaigns urging people to report crime, put more officers on patrol based on existing crime-report maps, or lean harder on witnesses at the scene — all of which left the map built from the roughly one-quarter of assaults that ever reached a police officer.
what they saw
Shepherd saw the missing violence data already existed, itemized by location, time and weapon, inside hospital admission records — filed under a department that treats wounds, nobody had asked for a printout.
the move
Shepherd had emergency departments strip victims' A&E intake records of names and addresses but keep the location, time and weapon of each assault, then hand that anonymized data to police crime analysts every month to overlay on their own maps — turning hospitals, not victims, into the primary sensor for where violence was actually happening.
why it works
The mechanism swaps an unreliable, effort-dependent sensor (a frightened or distrustful victim deciding whether to call police) for a reliable, effort-independent one (a nurse filling in an intake form to bill for treatment). Because hospitals fill out that form regardless of whether the victim wants police involved, the resulting map captures violence that never generates a crime report — bar fights, gang disputes, domestic assaults — letting police and licensing authorities target patrols, bar hours and hotspots based on where people are actually getting hurt rather than where they're willing to complain.
the payoff
Violence-related hospital admissions roughly halved and violent crime fell about 42 percent in Cardiff between 2002 and 2013.
where it breaks
It needs an institution collecting the same underlying fact for a different, non-punitive reason, willing to strip identifying details before sharing, and a receiving agency able to act on aggregated location data without demanding names. It breaks if hospitals fear the data will be subpoenaed and used against patients, if consent and privacy law block anonymized sharing, or if the receiving side just adds more patrols instead of addressing the venues and hours the map reveals.
what came after
Renamed the Cardiff Model for Violence Prevention, it won the 2008 Stockholm Prize in Criminology and was formally adopted by police forces and health systems in Australia, the Netherlands, South Africa and several US cities and counties, with the CDC endorsing the approach nationally.
references
- [1]The Cardiff Model: Leveraging Data for Violence PreventionRobert Wood Johnson Foundation, 2018rwjf.org
- [2]The Cardiff Model for Violence PreventionCardiff University, 2019cardiff.ac.uk