What murder statistics don’t say

The city of Toronto has broken its record for homicides in a single year after a Sunday afternoon shooting in an apartment building in Scarborough. Ninety people were by then killed in the city (and another since), a scary number that will naturally create unease in Toronto and attract national attention.

So I suppose it falls to me, an Edmontonian who is a student of morbid statistics, to point out that this recordshattering number is still not so awful.

Toronto has 2.7 million people: it’s almost three times larger than Edmonton. Out here we are enjoying a quiet year, by our standards, and have had 26 homicides to date, so Toronto’s rate of violent death is slightly higher.

But this is very unusual. Our homicide total last year was 45, or 47 if you count police killings (which we should); the local record (set in 2011) is 48. When Toronto gets closer to three times those totals, we’ll start getting nervous on your behalf.

What, if anything, should we really make of homicide counts like these? Homicides are in theory useful as a benchmark for overall criminality: they represent the hard, inescapable edge of a wedge of social violence, or even of general disorder. It is enormously difficult, and presumably very rare, to kill a person without having anybody at all notice. Virtually all homicides are thus recorded in official statistics, and most of them make the newspaper as a matter of course.

But the numbers of homicides for any Canadian city are so paltry from year to year that they are easily influenced by chance. There is an obvious example, a reason Toronto has smashed through its previous record for deadly violence: we count deaths rather than acts of lethal violence, and so the terrible van carnage that occurred in North York City Centre in April counts as 10, rather than one or even zero.

The perpetrator of the van attack was a lone nut who chose a risky method of inflicting mass death. When a terrorist used the same modus operandi in downtown Edmonton last fall, crashing a rented vehicle into a crowded sidewalk, he managed to not even hurt anyone especially badly. The difference is just dumb luck.

But there is another influence that is hardly ever discussed in Canada. Toronto’s old record year for homicide was 1991. That is almost a generation ago. Surgical trauma care is continuously evolving — or one assumes, and hopes, that it is.

Military statistics, which are even more careful and complete than the ones police agencies keep in cities, suggest that doctors have learned a lot about keeping victims of violence alive.

As the medical writer Atul Gawande observed in a 2004 journal article, American physicians made little progress in keeping wounded soldiers alive between the Revolutionary War and the end of Vietnam. Deaths from “combat injury” as a fraction of sufferers went only from perhaps one in three to maybe one in four over all that history. But in the U.S.’s post-2001 wars, the figure was more like one in 10.

This should probably affect how we regard homicide as an indicator of serious violence. In this year’s annals of Toronto bloodshed, the flip side of the dreadful North York van attack was the Danforth shooting in July. A disturbed man walked down a crowded Greektown street randomly shooting passersby and restaurant patrons: he wounded 15 people in all. Because of outstanding surgical care given at three different Toronto hospitals, 13 of these individuals were saved.

What would that figure have been in 1991? Indeed, if Toronto had the same quality of trauma care in 2018 that it did in “record-setting” 1991, and emergency-room doctors had learned nothing useful in the meantime, how high would the city’s year-todate homicide total now be? Over 100? 120?

There is no telling — but there are reasons to be skeptical about the assumption of constant progress in trauma care. The question of improved medicine “masking” a more violent urban environment is an active issue in the United States, with its high-stakes debate over firearms in the culture and the constitution.

Moreover, the U.S. has cities like Chicago, where the all-time annual homicide record is a sickening 943 (1992). In a place like that, the police are mighty quick to be judged on a bad year for murders, and they are equally quick to take credit for a good one, even if some should be going to doctors and paramedics.

It seems clear that there was some “masking” before the year 2000. But if it has existed since — if there has been further uniform progress in the performance of the trauma-care system — it is harder to find statistically.

There was a bit of a clamour in 2012 when the U.S.’s National Electronic Injury Surveillance System seemingly revealed a reduction in the case-fatality rate for intentional shootings; the brute figures showed that over a decade or so (20032012) it had been cut from 25 per cent to 18 per cent.

Unfortunately, statisticians were (in 2017) able to show that these figures were affected by poor sampling, and that the rate had actually held steady, at 22 per cent.

HOMICIDES ARE IN THEORY USEFUL AS A BENCHMARK FOR OVERALL CRIMINALITY.