Skip to main content

The Story of Police Shootings, as Told by Health-Care Data

Injury codes bolster the belief that police are injuring and killing African Americans at a higher rate mainly because they’re stopping and arresting them at a higher rate.

By Nathan Collins


(Photo: Scott Olson/Getty Images)

A new study of police stops and arrests in the United States confirms what some have argued for a while: Police are not that much more likely to injure or kill African Americans once they’ve made a stop or arrested someone, but they’re much more likely to stop and arrest African Americans in the first place.

If the result isn’t surprising—and at this point, it really shouldn’t be—the means by which Ted Miller and his colleagues at the Pacific Institute for Research and Evaluation reached their conclusion should be: Rather than rely on police-reported statistics, they relied on ICD-9 codes, the same system used to record colds, cancer, and broken bones.

“Deaths from legal police intervention are underidentified or undercounted in US Vital Statistics, FBI Supplemental Homicide Reports, and Bureau of Justice Statistics Arrest-Related Deaths programme data,” the research team writes in Injury Prevention.

Indeed, the government’s own statistics don’t even agree, and certain statistics from the Federal Bureau of Investigation and Centers for Disease Control and Prevention likely report only about 50 to 60 percent of police homicides, sometimes because reports simply don’t mention police involvement, according to research published in March. Meanwhile, researchers know very little about non-fatal injuries inflicted by police during traffic stops, arrests, and other activities.

For every 10,000 African Americans, 1,404 were stopped and arrested.

Fortunately, there’s a standard system for recording all kinds of injuries and fatalities at the hands of police (though it wasn’t created for that purpose): the Healthcare Cost and Utilization Project, a Department of Health and Human Services-run initiative that exists mainly to aid policy researchers interested in improving the American health-care system.

In service of that goal, however, HCUP tracks individual case reports using the ICD-9 system of diagnostic codes, along with its successor, the ICD-10. Those systems include police homicides and injuries, so Miller and his team had a less biased way to assess the physical damage done by police. (Still, the team relied on the Guardian’s count for 2015 fatalities, since prior studies suggest journalistic accounts are as accurate as anything else researchers have tried.)

“In 2012, an estimated 55,400 people were killed or hospital-treated in legal police intervention incidents,” Miller and his team write. About 2 percent—1,063 people in all—were killed, and another 5 percent had to stay in a hospital; the rest were treated in an emergency department and released, the team estimates. About 95 percent of the deaths involved gunshots, while most of the rest were the result of tasers.

More important than the raw numbers, however, is who gets stopped, arrested, and ultimately injured and killed. For stops and arrests, the team turned to FBI statistics (which they adjusted for underreporting) as well as the 2011 Police Public Contact Survey, which keeps track of factors like age, gender, and race.

The results: Police are not more likely to kill African Americans, Hispanics, or Native Americans once they’ve stopped them, but they are more likely to stop them in the first place—a narrative consistent with several otherrecent findings. Roughly a third of all people police stopped or arrested were treated in an emergency room, landed in the hospital, or died, and statistics suggest those numbers are largely independent of race. However, the team estimates that, in 2011, for every 10,000 African Americans, 1,404 were stopped and arrested, compared with only 503 non-Hispanic whites.