Risk Assessment Follies
July 30, 2015
I heard recently, “X was telling me how you were saying that people shouldn’t wear bicycle helmets”. This is not correct; I know better than to violate tribal norms like that, and bicycle helmets have been measured to reduce head injuries by 64%, offset by a pile of sweat and a 36% increase in neck injuries (i.e., a net reduction in horrible injuries). However, the focus on and attention devoted to helmets-helmets-helmets for bicycling-bicycling-bicycling is not at all rational. If helmets make sense for bicycles, they make sense for other activities with a similar risk of head injury, there are other bicycle risk mitigations that work about as well that we scarcely mention, there are other similarly-sized risks with mitigations that are seemingly not to be discussed at all, and there are much larger risks that most people willingly expose themselves to every day that bicycling (with or without a helmet) mitigates quite nicely.
(Most risk stats cited here are from OECD, “Cycling, Health and Safety” which contains a compendium of studies on bicyling health and safety. You can read it online.)
For comparison, consider that use of daytime running lights has been measured (just one study, but a pretty good one) to reduce risk of a serious crash (one resulting in an insurance claim or hospital visit) by about 50%. Not 50% of some injuries – 50% of serious crashes. No additional sweat, no additional offsetting increases in other injuries, just a flat reduction in crashes. If we took a rational approach to risk reduction, we’d hear about as much about this as we do about wearing helmets.
Or, compare the risks of riding a bicycle to the risks of riding in a car. You might think that riding a bike is vastly less safe, but per-trip (in the US), it is only about twice as dangerous and 25 times safer than riding a motorcycle. Head injuries are also plenty dangerous in automobiles, which in the US cause 14% of all traumatic brain injuries and 26% of all TBI-related deaths. Researchers in Australia have proposed a cheap and lightweight “headband” for use in cars; they estimate it would save A$380 million (US$276 million) in “reduced societal harm” in Australia alone. If we took a rational approach to risk reduction, and that rational approach had us recommending helmets for bicycles, it would almost certainly also have us seriously discussing the use of helmets in automobiles. It would be surprising if the line between “not that risky” and “too risky” fell so cleanly and definitely between the two activities.
There are other cycling risks that are scarcely discussed at all. It turns out that particulates – especially from some older diesel engines, especially from small engines with uncontrolled emissions – are very bad for us. How bad? Studies cited in the OECD survey suggest that the pollution risk for cyclists is about half as large as the crash risk for cyclists – however, it is not just cyclists that bear this risk. Pedestrians walking near streets are disproportionately affected, and drivers are affected as well. Drivers experience the greatest concentration, but pedestrians are exposed for the greatest duration and inhale the most air per mile. Summed over all the exposed population this is a good-sized risk, because almost everyone is exposed. Given a rational approach to risk, we’d expect to hear a whole lot about the need to get stinky small engines off the roads (replace them with small electric scooters) and the need to ban or upgrade diesel engines in urban areas (newer ones are cleaner).
And finally, there are the other huge risks that cyclists avoid that non-cyclists endure in apparent blissful ignorance. Again in the OECD survey (page 44, table 1.2) we can see a summary of studies on the health benefits of cycling. Five studies in three countries, all adjusted for confounding factors, giving the relative risk ratio (for annual mortality) for cyclists (in at least one study, bicycle commuters) compared to non-cyclists. These range from 0.66 to 0.79 – that is, if we instead regard cycling as the norm, non-cyclists increase their annual risk of death somewhere between 25% and 50% (these numbers are all-cause mortality, so they include crash risk – which in the US appears to be in the ballpark of a 5% increase in annual risk; more than I like, but not nearly large as the no-exercise risk). The majority of the US population is exposed to this much-larger risk. Someone with a rational approach to risk would not waste much time at all worrying about the relatively small risks to bicycle riders (all of them) as long as so much of the population was exposed to this much larger risk.
However, we’re not rational, so let me again endorse the wearing of helmets in the United States. Studies show it will make you slightly safer. There’s all sorts of risk reductions we should be looking into, but we don’t talk about those because we’re not rational.