Traffic Injury Prevention Journal releases study of NYC Bicyclists

The journal Traffic Injury Prevention recently published a Department of Health and Mental Hygiene study that examined fatal bicycling accidents in New York City from 1996 to 2005. The Times Cityroom blog focused on the alcohol findings

Some 21 percent of autopsies for New York City bicyclists who died within three hours of their accidents detected alcohol in the body

Ms. Stayton cautioned that the data was somewhat limited because of nature of toxicology tests. Of the 225 bicyclists who died in fatal accidents during the 10-year time period, only 176 were tested for alcohol. But because alcohol is metabolized with time, only 84 of those cases the bicyclist had died within three hours of the crash were considered to have valid tests. Of those, 18 — or 21 percent — showed signs of alcohol.

The study is gated so this brings up a lot of questions that I can't easily get answers to. How many pedestrians and drivers who died within three hours of their accidents had alcohol in their body? Why were only 176 tested - is their a selection bias? Are we talking alcohol-impaired or just "some alcohol"? Was alcohol a contributing factor? What percentage of cyclists at any given time have some alcohol in their system? (This isn't so crazy high, 32 percent of total traffic fatalities in 2006 involved an alcohol-impaired driver). Of course, it certainly seems reasonable that biking while alcohol-impaired increases your risk and is a very bad idea (at the very least, it is illegal).

Of course, walking may not be any safer

From a 1993 “End Paper” on a pedestrian death report in the Times: “If you must drink, don’t walk: 60 percent of pedestrians 16 and over who were killed at night in 1991 had blood alcohol concentrations of 0.10 or more, enough to get you arrested for drunk driving. In New York City, 37 percent of the pedestrians killed last year showed traces of alcohol and/or other drugs.”

Other statistics

Only 3 percent of the bicyclists who died were wearing helmets, according to available data. Head injuries contributed to three-quarters of bicycle deaths.

The overwhelming share of the deaths, 92 percent, involved a moving motor vehicle

Fourteen deaths were bicycle-only, and four deaths involved striking a parked vehicle, including one which had an open door.

Adam Voiland tries to garner lessons from the NYC report. I'll try to pull some new facts out.

Cyclists and pedestrians make up 11% of commuters but 54% of fatalities.

8% of deaths did not involve a motor vehicle - this is lower than the number I've seen reported elsewhere.

"Only one fatal crash with a motor vehicle occurred in a marked bicycle lane. This fatality occurred in Prospect Park, as a result of a motor vehicle colliding with a bicyclist. A total of 10 other fatal crashes occurred in or near a marked bicycle lane. Six fatal bicycle crashes with motor vehicles occurred in close proximity to, but not inside of, a marked bicycle lane. One fatality occurred on a city street near a bicycle lane, but did not involve a motor vehicle. The last three fatalities occurred inside a marked bicycle lane located within a park, but did not involve a motor vehicle."

There were 7 fatal crashes that involved open car doors. One hit the door, four were hit swerving to avoid the door and three hit the door, fell and were then hit by a motor vehicle.

Even using the more recent data (2004-2005), when helmet use became more prevalent and record-keeping better, 97% of cycling fatalities involved cyclists without helmets (sample size of 38). Between 14-20% of high school students report wearing a helmet. Observational studies from 2004 show 22% of on-road cyclists wear helmets. 13% of injured cyclists were wearing a helmet. This is some very strong evidence supporting a relationship between helmet use and survivability.

In a crash scene investigation, investigators determine the apparent contributing factors for the fatal crash. Bicyclist-only factors were assigned 42% of the time. Driver only factors 20%. "In interpreting these data, it is important to remember that in some fatal crashes the motor vehicle driver’s recollection of the crash is the only one available, which may bias these findings to show more bicyclist errors compared to motor vehicle driver errors." This is backed up when one looks at the factors assigned to non-fatal crashes: Bicyclist-only factors were assigned 13% of the time. Driver only factors 35%. That's a big difference.

For some reason, the number of fatalities has remained flat, but injuries have come down quite a bit. Number of cyclists has not increased much.

Comments

[Contrarian]

The study came out in 2006. The conclusions about helmet use and bike lane study both suffer from severely dubious methodology.

Here are some stats from the body of the report:

89% of fatalities at intersections
11% midblock

It also says:

Marked bicycle lanes are areas on roadways that are clearly delineated for use by bicyclists. Bicycle lanes are located on city streets, and in parks where the lanes are shared with pedestrians. Using the NYC DOT’s database of bicycle lanes, fatal crashes were examined for their proximity to marked bicycle lanes.
Only one fatal crash with a motor vehicle occurred in a marked bicycle lane. This fatality occurred in Prospect Park, as a result of a motor vehicle colliding with a bicyclist. A total of 10 other fatal crashes occurred in or near a marked bicycle lane. Six fatal bicycle crashes with motor vehicles occurred in close proximity to, but not inside of, a marked bicycle lane. One fatality occurred on a city street near a bicycle lane, but did not involve a motor vehicle. The last three fatalities occurred inside a marked bicycle lane located within a park, but did not involve a motor vehicle.

They're being a little disingenuous there, conflating bike paths with bike lanes, and then saying there was only one bike lane death because there was only one bike path death. The other bike lane deaths are described as "in or near" a bike lane. What does that mean?

But here's the real whopper: they threw out 90% of the data before they started. Of the 225 deaths, only 11%, or 25, did not occur at intersections. Those are the only ones they counted. Of those 25, 11 occurred "in or near" bike lanes.

If you wanted to spin the data differently, your conclusion could have been: "Bike Lanes Death Traps: 44% of deaths occur in or near bike lanes, which make up 1% of roads!"

(New York has 19,000 lane-miles of roads.)

I would also dispute that this study is strong evidence of a relationship between helmet use and survivability. Rather, I would say it is strong evidence that accident investigators have difficulty determining whether accident victims were wearing helmets.

You did good though to pick up the relationship between a cyclist surviving an accident and the driver being blamed.

[Washcycle]

On the helmet use, IF (and that's a big if) you think the numbers are correct, it seems to connect helmet use to survivability. Every study falls apart if the data is incorrect. I couldn't easily find a reason to not trust the data, but then I also couldn't figure out how they gathered it.

[guez]

I'm going to agree with Washcycle here. The prima facie evidence, in this study, is that helmet-wearing prevents injuries. This effectiveness of helmets should of course be separated from the question of helmet laws. I will grant that it does not settle the question once and for all, but it is as strong, if not stronger, as much of evidence marshaled on this blog in defense of other aspects of cyclist advocacy.

As Sheldon Brown once said, "helmets good, helmet laws bad."

[Jonathan Krall]

"There were 7 fatal crashes that involved open car doors. One hit the door, four were hit swerving to avoid the door and three hit the door, fell and were then hit by a motor vehicle."

So the way to survive being doored is to turn *away* from traffic. That is, to aim straight at the jackass who opened the f---ing door!

[Contrarian]

guez, I think you're giving this study way too much credit.

If you look at the helmet issue, it quickly becomes clear that the benefits of helmets are controversial -- if a benefit does exist, it is so small that it is difficult to establish with any certainty. Then this study comes along claiming that the benefit is huge, with survival rates for helmet wearers seven times that of non-wearers. Which is more likely: that all the previous studies were wrong, or that this study is flawed?

To give an analogy, global warming is a controversial subject. When scientists calculate the change in global temperature over the past century they get conflicting numbers, some positive some negative. The results seem to be centered around about 1.2F, but the magnitude is small enough and the variance is large enough that it's clear that the effect is subtle, and there is disagreement about whether it actually exists.

This study is the equivalent of a climatologist claiming that the global temperature has increased by ten degrees in the last century. It's just so out of bounds it defies credibility.

The same holds true for the study's conclusions on bike lane safety. If you spend any time looking at bike lanes you know that any safety is so small and so subtle as to be essentially undetectable. Most bike lane advocates have abandoned the safety argument in favor of second-order arguments about convenience and perception.

What's disheartening is that a purportedly scientific journal would accept this study.

[Contrarian]

"There were 7 fatal crashes that involved open car doors. One hit the door, four were hit swerving to avoid the door and three hit the door, fell and were then hit by a motor vehicle."

Interesting use of the passive voice here. The accidents "involved open car doors." Instead of, say, people who opened doors into the paths of cyclists.

The biggest shortcoming of this study is that it focuses solely on the behavior of cyclists. For instance, it mentions that about 90% of the victims of fatal crashes are male, and suggests that aggression may be a factor. What they don't mention, but I read in another study of similar data, is that 90% of the drivers involved in fatal bike crashes are male as well.* Since this is so far from what you would expect if crashes occurred at random, the conclusion is that driver behavior is overwhelmingly likely to be the primary factor in cyclist safety -- far greater than facilities, helmet use, or cyclist training.

Yet changing driver behavior is something that is essentially off the table, both here and in New York.

*That study pointed out that roughly one in three drivers flees the scene and is never identified; the 90% number is of drivers who are identified.

[Contrarian]

A detailed critique of NYPD's methodology in reporting bicycle crashes is here:

http://www.rightofway.org/research/cyclists.pdf