Bike Crash update
Since my bike accident, I’ve had what appears to be a full recovery. To recap, I fell off my bike behind the Frist Museum (probably due to slippery stones and too strong brake application), hitting my head, resulting in a brief period of unconsciousness, followed by a longer period of dreamworld. My visit to the Vanderbilt ER lasted about five hours, and involved CT scans and stitches. For about a week I had what felt like a weight on my head, which then lifted. My knees remained sore for about a month (bruised somehow), my head cut gradually healed, and finally a new bike helmet arrived. The Vanderbilt Emergency services bill arrived in the mail ($7,500.00), but that amount was discounted down to negotiated rates by the insurer, and then covered except for the deductible and copay, so that my part (I think) was only about $250. By early January, I returned to riding my bike for short distance transportation, though with a promise to Kathleen to avoid “dark and stormy” and “slippery.”
The risk question
Ever since I began biking for most of my short trips around town (about 3-4 years ago), I’ve wanted to know the answer to this question: what are my risks on a bike? Is my idea of the positive health effects of biking overridden by the increased risks of injury or even death from bike accidents? Following my bike crash in late November, and feeling more nervous on a bike for a while, I returned to the question in earnest. After all, I have been a bike advocate, urging more people to ride. Am I doing the right thing? The following review of the evidence focuses on fatalities rather than injuries. (I’m saving that for another day.)
Safety in Numbers
One of the more interesting findings from studies of biking casualties is that there is safety in numbers. The more cyclists there are, the fewer injuries and deaths per cyclist. One study reported that in recent years there has never been an increase in bike riding that had been followed by an increase in bike deaths. Deaths always remain stable or even declined.
There are several possible reasons for this finding:
- increased riding follows on improved, safer infrastructure
- more bikes on the road make drivers more aware of bikes and more careful, and/or
- more bikes on the road mean drivers are more likely to also be bike riders, and therefore more careful.
Infrastructure Helps: Portland, Oregon bridge crossing gives cyclists and walkers a safe place to cross into the city
Biking in Portland, Oregon, demonstrates safety in numbers. During my visit there last summer, I counted 65 bikes in a 5 minute period coming off this bridge (around 8 am on a weekday morning). A bit later, 8:30, I rode across the same bridge inbound and saw the bike counter tick 1,435. This is one of four bridges going into downtown Portland, and all have walk/bike pathways. During 2013 Portland had no bike deaths, and in the prior 10 years, there were three other years with no bike deaths.
As an advocate for biking, the safety in numbers hypothesis gives me some comfort that getting more people to ride bikes will not lead to more biking deaths, but there is still the question of how safe are we now, in Nashville or some other place in the US, where riding is far less common than in places like Portland, Oregon, or overseas in Amsterdam or Copenhagen.
In the US commuter cycling is rare overall. About 0.6% of workers use the bike as their main commuting vehicle and the share is somewhere between 0.3% and 0.5% in Nashville (about 800 cyclists). The share rises to about 7 percent in Portland, and is much higher in some college towns. In the Vanderbilt and Belmont areas, the share is probably close to 5 percent. But, for the most part US motorists are not used to cyclists, and are not cyclists themselves, making biking more risky here than, say, in Europe.
The real danger is, of course, the motor vehicle. From the bike’s perspective, the motor vehicle is a weapon. Bike/motor vehicle crashes are a small part of overall bike accidents, but represent about 80% of bike fatalities.
I didn’t find a study which told me what I wanted to know about risk in the US (there were good European studies), so I had to construct the relevant information from several sources. My effort is summarized in the table below.
Risk of transportation fatalities
Annual Risk per Risk
Deaths risk per 100 ml per ml
2012 ml persons miles hours
Motor vehicle (all) 33,561 106.9 0.8 .32
Automobile occupant 12,271 39.1 0.3 .12
Pedestrian 4,743 15.1
Motorcycle 4,957 742.0
Bicycle (adults) 884 10.8 11.7 1.17
Commuter I** 335 151.7 4.6 .46
Commuter II** 67 30.9 1.9 .19
Commuter III** 708 187.7 11.7 1.17
British cyclists (all) 145 2.0 0.46
British cyclists (active) 0.25
Dutch cyclist (all) 0.7 0.12
Dutch car rider 0.1 0.09
Note: The information in this table was developed from the following sources: (1) Malcolm Wardlaw “Assessing the actual risks actually faced by cyclists,” TEC, Dec. 2002, (2) The National Household Travel Survey, 2009, (3) A survey of North American Bicycle Commuters, William E. Moritz, Transportation Research Record, 1997, (4) the National Highway Traffic Safety Administration, Motor Vehicle Crashes, 2012, and (5) Center for Disease Control, CDC “Wonder” interactive data base, 2010.
** These represent three different approaches to calculating risks for commuter cyclists. Explanations follow in the text below.
Let’s sort this out. In 2012 about 884 adults were killed on a bicycle, while about 33,000 were killed in or by motor vehicles(Column 1). Large trucks were the most deadly kind of vehicle, with a 5.5 kill ratio: 5.5 third parties killed by the truck in crashes for every truck occupant killed and representing about 10% of the deaths. (Beware of those tractor-trailers.) It turns out that pickup trucks killed more cyclists than passenger cars. (Another vehicle to be wary of.)
The 884 bike deaths represented a risk of 10.8 chances in one million (column 2) of being killed among those who rode a bike at least once last year (about 35% of the adult population or 81 million adults). The risk of riding in or driving a passenger car was about 4 times the bike risk (39.1 per million). For comparison purposes, the risk of dying of anything prematurely (before age 75) was about 3,712 in one million, and half of that risk involves cancer and heart (diseases in part linked to smoking, poor eating, and lack of exercise). The risk from firearms is about 100.9, the non-traffic risk from drugs about 172.2, and the non-traffic risk from alcohol about 105.5. The smoker risk trumps all. Among the 20 percent of the population who smoke, the annual risk of death was 9,161 in one million, or about 1 percent. The overall bicycle risk is tiny by comparison, but not zero.
Exposure is important
The average cyclist went about 92 miles in 2009 (9 hours at 10 mph), whereas the average car rider went about 13,000 miles (325 hours at 40 mph). This leads to a risk per-mile (column 3) and per-hour (column 4) of exposure that is the reverse of the annual risk (column 2).
The risk of bike riding is about 40 times the risk of car riding on a per-mile basis (11.7 versus 0.3 per 100 million miles). The risk per hour, a fairer comparison, still indicates a risk ratio of about 10 to 1 (1.17 per million hours to 0.12). Bike riding, then, is about 10 times more hazardous than car riding. However, since few people spend much time on a bike, the “annual risk” (column 2) favors cyclists.
But what about those urban bike commuters, who spend a lot of time on a bike? I’ve taken three different approaches to that risk (Commuters I-III in the table). I explain these in a note at the end of this column. Based on my assessment of the studies I reviewed, I think the average risk for the commuter cyclist today is somewhere between commuter I and commuter II, and may tilt towards commuter II where there is good infrastructure and more cyclists.
On a per-hour basis, commuter II is about 1.6 times as risky as riding in or driving a car, and less risky than the overall risk of motorized vehicles. The motorized vehicle risk includes motorcycles (almost 5,000 of the 33,000 deaths) but also it includes the car or truck as a weapon: pedestrians and cyclists struck by cars and trucks.
Now take a look at the British Cyclists and the Dutch cyclists. British active cyclists are in the same risk realm as the Commuter II estimate. The Dutch, who are far more likely to ride bikes than Americans, have very low risks. In Amsterdam (data not shown) bikes are used for more trips than cars, and are safer (fewer deaths) than cars, though the figure for nationwide deaths (in the table above) is slightly above Dutch motor vehicle deaths.
The estimates above are based on averages. My particular risks, whether in a car or on a bike, can be managed to improve substantially on the averages. I can lower my bike commuting risk below the average by careful route choice (lightly traveled residential streets or where there are bike lanes), by avoiding rainy days, by not biking at night, by avoiding drinking and biking, by wearing a helmet, and by avoiding sinking into deep thought while biking. (The last caution is one of the drawbacks of cycling versus walking.)
For example, about half of all bike fatalities occur between 6pm and 6am (after dark). My perception is that most commuter-riding, by far, occurs during the day, suggesting a much lower risk from daytime riding on a per-hour and per-mile basis than the table implies. (However, during the winter the daytime commuter can be caught in the early morning or early evening shadows.) Also, in about 20-25 percent of fatalities (Insurance Institute for Highway Safety, 2010), a possible causal factor was that the bicyclist was determined have a BAC of .08 or higher. Individuals who lose their license due to a DUI may be forced onto bikes (no license required), but without having solved their drinking problem. There is also the drunk driver, and he (sometimes she) tends to be out late at night. 65% of traffic fatalities between 12 midnight and 3am involve impaired drivers (BAC>.08).
The Nashville Risk and an Example
In Tennessee, between 2005 and 2013 (inclusive) there were an average of about 6 to 8 bicyclists killed each year on the road. In Nashville (Davidson County) there were 10 cyclists killed over the past 11 years. However, during the past 5 years, there has been only one cycling traffic death (in 2012).
That one death was a young man biking home from work around 9:30 pm. The following details come from the accident report. He was hit from behind by a car, with crash details suggesting the cyclist was towards the left side of the lane. There is some evidence the driver may have been texting or receiving texts at the time or during some time within a few minutes of the 911 call, though the text sent may have occurred before the accident and the texts received could have been ignored. The driver claims he didn’t see the cyclist before hitting him and was not speeding at the time of impact. The investigating officers didn’t see a need to assess the driver (age 24) for alcohol or drugs. The driver did not leave the scene. People riding in the vehicle behind the driver made the 911 call.
The cyclist had reflectors on his bike, but no flashing rear lights, and the investigating officer inferred that the driver should have seen the reflectors from a distance of more than 400 feet, giving him about 6 seconds to apply the brakes.
The cyclist had a revoked drivers license (DUI) and a blood alcohol test revealed a BAC of .172 (above the legal limit of .08) and marijuana was also detected. The cyclist was seen 30 minutes prior to the accident at McDonald’s by two juvenile observers, who thought him behaving strangely, talking to himself and rocking back and forth, described either as under the influence or suffering a mental illness episode. One of the observers took a video, using a phone camera, which was also observed by the investigating officer. Another observer who lived near the accident heard a driver yell angrily (cursing) at the cyclist near where the accident occurred about five minutes before the observer heard sirens. That same observer saw the cyclist riding on the right side of the road next to the curb at that time (not in the middle of the lane). There was no indication from the report that the cursing driver was the driver who hit the cyclist.
The investigation conclusion was that the cyclist should have been seen by the driver, and that the driver was probably inattentive, whether involved with texting or some other distraction. The investigator also suggested that the cyclists being in the left side of the lane rather than to the right was also a factor.
However, the attorney general decided against filing charges against the driver after a review of the case. The AG mentioned the BAC level of the cyclist, but said it was unknown whether that was a factor. (In other words, the cyclist was hit from behind, while pedaling in the traffic lane, and whether the cyclist was impaired didn’t seem to be a factor.) The AG didn’t address the distraction issue, but did discuss where the cyclist was in the lane, the poor visibility afforded by reflectors, and the inapplicability of the 3-foot law in this case. In another development, I was told by someone who had followed the case that the driver’s insurance company made a settlement with the cyclist’s family.
I’ve provided some details about this case because I think there are some important lessons implied in the description. Driver distractions, which were a likely factor, have become a major issue in accidents (especially the texting distraction), as is impairment from alcohol or drugs. Bike lighting is important for night-time biking, though there was a difference between the investigator’s view (driver should have seen the bike) and the attorney general’s view (reflectors aren’t enough, though legally reflectors meet the minimum standard). Finally, major arterial roads (this was a four lane with a middle suicide lane with a 45 mph speed limit) are not well suited for cyclists, especially at night. As I mentioned in previous paragraphs, risks are higher at night, and using strategies to mitigate night-time risks (bike lighting, choosing roads with slower moving, more limited traffic) are important.
The risk of doing nothing
My own bike accident reminded me that biking has risks, but my general experience with biking and the calculations above suggest that the risks are within the bounds of the normal everyday risks of living in an urban environment. The exercise benefits of commuter biking on health haven’t been included in the calculations, but when included in European studies, the cyclist risk is well below the risk of car riders.
Nashville is improving its bike commuting infrastructure (with bike lanes, and more recently with protected bike lanes, and with greenways that can serve commuting needs). Evidence has show elsewhere that as biking increases, and infrastructure improves, biking becomes as safe as or safer than driving a car. We have a ways to go in the US and in Nashville, but are heading in the right direction in Nashville.♦
Risk calculation details
The bike commuter is the individual who uses his (or her) bike to get places—to work, to a friend’s house, to a meeting, to the grocery store, or to a restaurant. Studies have found that such cyclists travel around 1,600 to 3,300 miles each year on a bike (160-330 hours), resulting in far greater exposure to accidents than the average bike rider (92 miles or 9 hours). There are about 2 million adults nationwide who bike for at least 30 minutes a day (and presumably may commute on a bike) according to the most recent National Household Travel Survey. What is their risk?
To come up with reasonable per mile or per hour estimates, the 884 annual bicycle deaths must be allocated between two groups: the commuters and everyone else? If you make a linear projection (Commuter III in the table), based on annual miles ridden by bike commuters (1600 miles) compared to the overall average for cyclist (92 miles), then bike commuters would make up about 80% of the bike mileage, and also 80% of the deaths (708), despite being only about 3% of those who ride bikes.
There are two alternative approaches that I think are more likely to match reality. One involves inferences from a study (Moritz, 1997) of US bike commuters conducted in the mid-90s. That study documented Emergency Room use among bike commuters, finding that 10% of bike commuters (annually) had crashes, and about half of those (57%) went to the ER. Taking the current (2009) estimate of the number of commuter cyclists, estimating their ER use (based upon the Moritz findings), and comparing it to total ER use (by adult cyclists), suggests that about 37% of cyclist ER visits are from bike commuters versus the far more numerous occasional bike riders. Applying that percentage to total bike deaths, a much lower estimate of the contribution of bike commuters to fatalities results (335 fatalities rather than 708). Note that the Moritz study (Commuter I) found commuting cyclist to average about 3,300 miles per year rather than 1,600 miles used with Commuter II.
A second alternative approach (Commuter II) follows on an English study (Wardlaw, 2002) which used a fatality pattern found with automobile crashes to develop estimates for commuter cyclists. The auto study found that as miles ridden doubled, the number of fatalities didn’t double, but instead went up by 30%. The rationale for this for bikes is that biking experience (gained from daily commuting) leads to lower risk. Using that method, about 67 of the 884 cycling deaths would be linked to bike commuters (assuming 1600 miles ridden).