This week Zachary Crockett of the Priceonomics blog highlighted some eye-popping statistics on high-speed police pursuits -- you know, the kind that you see on COPS, or that local TV crews chase using helicopters. Here are the numbers that really stand out to me:

Crockett points to a 2007 study in the journal Prehospital Emergency Care, which found that these crashes take about 323 lives each year. To put it in perspective, that's more than the number of people killed by floods, tornadoes, lightning and hurricanes -- combined. These numbers come from the National Highway Traffic Safety Administration's fatal accident database, so they only count deaths directly related to vehicle accidents involved in these chases. If a person is chased down by cops and eventually shot, for instance, that death wouldn't show up here.

But the most shocking thing is that innocent bystanders -- meaning people not at all involved with the chase -- account for 27 percent of all police chase deaths, or 87 deaths per year. If that number seems high to you, just start Googling. Some recent headlines:

This underscores a key fact that may seem obvious: high speed police chases are incredibly dangerous not just to the people involved in them, but to everyone who crosses their path. And given that many chases happen in urban areas, on densely populated city streets, the hazard to residents is high.

Given the high risk, you might assume that cops only give chase to the most violent criminals, in circumstances in which the hazards of a high-speed chase are outweighed by the risk posed by the criminals, right? But you'd be wrong.

Ninety one percent of high-speed chases are initiated in response to a non-violent crime, according to a fascinating report from the International Association of Chiefs of Police and the National Institute of Justice. They analyzed nearly 8,000 high-speed chases in the IACP's database. What they found was that the overwhelming majority of pursuits -- 91 percent of them -- were not initiated in response to a violent crime. Most -- 42 percent -- involved a simple traffic infraction. Another 18 percent involved a stolen vehicle. 15 percent involved a suspected drunk driver.

So you can start to see the problem here -- is it worth risking life and limb, barreling through town at high speed to catch somebody who ran a red light? Or who failed to signal a turn? If a driver is drunk, does it make sense to engage him in a high-speed pursuit?

Questions like these are making some localities revisit their high speed pursuit policies. There's even a non-profit dedicated to the issue. Some companies are busy devising technical solutions to it as well.

But meanwhile, the drumbeat of fatal police accidents goes on.