The Wall Street Journal has an interesting piece on the role of trauma care, declining homicide rates, and a possible increase in assaults with guns or knives over the past decade, which may come as a surprise. It raises more questions than it answers, but it's a worthwhile addition to the discussion.

First: we know that people are getting killed less often in America. There are various theories for this: the baby boom aged (an older population will generally mean less violence); the cocaine boom passed; there's less lead in the environment; and so forth, and it doesn't have to be one or the other. All these suggest that there's just less violence in general. Another is that trauma care has improved—which doesn't mean fewer people are getting attacked, just that we're better at keeping them from dying. And it's definitely true on the field of battle:

Combat survivability is at an all-time high in Operations Enduring Freedom and Iraqi Freedom. Ten percent of all injuries resulted in death, as opposed to Vietnam, where the fatality rate was 16.1 percent, or World War II, with a 19.1 percent fatality rate.

It would seem that would translate domestically, and the Journal piece describes how military and emergency medicine complement each other, passing innovations back and forth. What's harder is figuring out what the effect is.

After a steady decline through the 1990s, the annual number of homicides zigzagged before resuming a decline in 2007, falling from 16,929 that year to an estimated 14,722 in 2010, according to FBI crime data.

That everyone's pretty confident in. Then:

At the same time, medical data and other surveys in the U.S. show a rising number of serious injuries from assaults with guns and knives. The estimated number of people wounded seriously enough by gunshots to require a hospital stay, rather than treatment and release, rose 47% to 30,759 in 2011 from 20,844 in 2001, according to data from the Centers for Disease Control and Prevention's National Electronic Injury Surveillance System-All Injury Program. The CDC estimates showed the number of people injured in serious stabbings rose to 23,550 from 22,047 over the same period.

CDC data says that non-fatal serious injuries are way up. I ran the data on "Violence-Related Firearm Gunshot Nonfatal Injuries" for all dispositions (hospitalization, treat and release, etc), and there's a similar pattern: 45,316 in 2001 and 59,208 in 2011, though that's down from 61,406 in 2008. FBI stats for nationwide aggravated assaults (assaults with deadly weapons) have gone in the opposite direction, quite dramatically.

Delving further: a 2003-2011 comparison of FBI stats for aggravated assault broken down by weapon doesn't offer much of a drop in assaults by firearm (136,672 to 136,371) or "knife or cutting instrument" (129,536 to 121,786). There's a bigger drop in hands/fists/feet/etc. (190,913 to 172,931) and other (254,680 to 222,154). 

So it's confusing. And that confusion is confusing:

Criminologists say they are cautious about using such medical statistics to draw conclusions because of year-to-year inconsistencies in the number of medical institutions reporting data. The FBI collects annual homicide and aggravated assault statistics but doesn't have reliable numbers for gun and knife attacks.

Jens Ludwig, a law professor and the director of the University of Chicago Crime Lab, said he was leery of any number beyond reported homicides.

"Homicide is the one thing we're measuring well," he said. "Everything else is subject to much more uncertainty," including varying numbers of emergency departments contributing data, as well as differences in how injuries are classified.

Basically, we know a lot about when, where, and how people die. What do we know about when they don't die? Not much:

Mortality rates of gunshot victims, meanwhile, have fallen, according to research performed for The Wall Street Journal by the Howard-Hopkins Surgical Outcomes Research Center, a joint venture between Howard University and Johns Hopkins University. In 2010, 13.96% of U.S. shooting victims died, almost two percentage points lower than in 2007. (Earlier data used different standards, making comparisons useless.)

Death is the ultimate data point. Before that there's a lot of variance.


Photograph: oskay (CC by 2.0)