How dangerous is it to ride a bicycle without a helmet?

Epistemic Sta­tus: An hour of googling com­bined with some highly du­bi­ous statis­tics and very rough notes. Take with at least 500 grams of salt.

I don’t like wear­ing bike helmets, and have a bunch of friends who were hor­rified at me not wear­ing helmets when I oc­ca­sion­ally ride a bi­cy­cle. On their re­quest, here is a rough cost-benefit anal­y­sis of rid­ing a bike with­out a helmet to my office and back (a to­tal of ~20 min­utes):

I de­cided to break down the ques­tion into the fol­low­ing three sub­ques­tions:

  • How great are the benefits from ex­er­cise when rid­ing a bike?

  • What is the baseline risk of rid­ing a bi­cy­cle?

  • How ele­vated is the risk by not wear­ing a bi­cy­cle helmet?

How great are the benefits from ex­er­cise when rid­ing a bike?

https://​​www.red­dit.com/​​r/​​bi­cy­cling/​​com­ments/​​2zx61c/​​on_micro­morts_cy­cling_and_risk/​​

Ac­cord­ing to the first epi­sode of More of Less (haven’t got­ten to the sec­ond one yet), the benefits of cy­cling out­weigh the risks by about 5-10 times.

https://​​www.bike­fo­rums.net/​​com­mut­ing/​​896396-risk-anal­y­sis-micro­mort-micro­life.html

Micro­morts are a unit of risk mea­sur­ing a one-in-a-mil­lion prob­a­bil­ity of death. You can say that a mode of trans­port re­sults in so many miles per micro­mort and the re­sult is a num­ber you can re­ally use. For cy­cling, this is 10-20 miles, for driv­ing it is about 230 miles.
Micro­lifes are a mil­lionth of an (av­er­age re­main­ing) lifes­pan = 30mins of life, so ac­tivi­ties which ex­tend your life or de­crease it in a chronic rather than acute way can be mea­sured.
In look­ing at the risk as­so­ci­ated with cy­cling to work, peo­ple of­ten quote the mort fac­tor You also have to in­clude the life fac­tor. First 20mins of ex­cer­cise = +2 micro­lifes . Se­den­tary be­havi­our = −1 microlife

https://​​en.wikipe­dia.org/​​wiki/​​Microlife

First 20 min­utes of mod­er­ate ex­er­cise: 2 microlifes
Sub­se­quent 40 min­utes of mod­er­ate ex­er­cise: 1 microlife

How­ever, this is more com­pli­cated by the other costs of wear­ing a helmet, as well as a prefer­ence of mine to not die at a par­tic­u­larly young age. All-cause mor­tal­ity is re­ally low in my age-range, so bi­cy­cling might be a sig­nifi­cantly higher pro­por­tional in­crease for that age range, and with shorter GCR timelines I might not care su­per much about my long-term health, which also re­duces the value of ex­er­cise (though then there are also im­me­di­ate cog­ni­tive benefits of ex­er­cise be­yond all-cause mor­tal­ity that start tak­ing effect im­me­di­ately, which seems maybe like a suffi­cient coun­ter­bal­anc­ing con­sid­er­a­tion).

Con­cretely, if I cy­cle for 20 min­utes ev­ery day and that is my only ex­er­cise, I gain about 2 micro­lives. If I have some other source of ex­er­cise, I gain more around 0.5 micro­lives. On av­er­age I think it’s more some­thing around 1 micro­live for the 20 min­utes of cy­cling. So let’s go with that num­ber for now.

What is the baseline risk of rid­ing a bi­cy­cle?

Wikipe­dia for one micro­mort: Trav­el­ling 10 miles (16 km) (or 20 miles (32 km)) by bi­cy­cle (ac­ci­dent)

Aver­age bi­cy­cle speed (Wikipe­dia): 15.5 km/​h

Micro­morts per hour of cy­cling (baseline): ~1

That re­sults in about a gain of a third of a micro­mort over 20 min­utes of cy­cling.

How ele­vated is the risk by not wear­ing a bi­cy­cle helmet?

https://​​www.tand­fon­line.com/​​doi/​​abs/​​10.1080/​​15389588.2017.1298748?jour­nalCode=gcpi20

Re­sults: Es­ti­mates of helmet effec­tive­ness were similar from odds ra­tios (ORs) us­ing hos­pi­tal con­trols or from rel­a­tive risks (RRs) us­ing helmet use es­ti­mates (Seat­tle: OR = 0.339, RR = 0.444; Vic­to­ria: OR = 0.500, RR = 0.353). Ad­di­tion­ally, the odds ra­tios us­ing hos­pi­tal con­trols were similar when con­trols were taken from a larger co­hort for head in­jury of any sever­ity (Seat­tle: OR = 0.250, alt OR = 0.257; NSW: OR = 0.446, alt OR = 0.411) and for se­ri­ous head in­jury (Seat­tle: OR = 0.135, alt OR = 0.139; NSW: OR = 0.335, alt OR = 0.308). Although rele­vant ex­po­sure data were un­available for The Nether­lands, the odds ra­tio for helmet effec­tive­ness of those us­ing rac­ing, moun­tain, or hy­brid bikes was similar to other es­ti­mates (OR = 0.371).

Ok, so if we as­sume that all micro­morts come from head in­juries, then this sug­gest a re­duc­tion of around 23 in risk.

https://​​aca­demic.oup.com/​​ije/​​ar­ti­cle/​​46/​​1/​​278/​​2617198

Re­sults: A to­tal of 43 stud­ies met in­clu­sion crite­ria and 40 stud­ies were in­cluded in the meta-anal­y­sis with data from over 64 000 in­jured cy­clists. For cy­clists in­volved in a crash or fall, helmet use was as­so­ci­ated with odds re­duc­tions for head (OR = 0.49, 95% con­fi­dence in­ter­val (CI): 0.42–0.57), se­ri­ous head (OR = 0.31, 95% CI: 0.25–0.37), face (OR = 0.67, 95% CI: 0.56–0.81) and fatal head in­jury (OR = 0.35, 95% CI: 0.14–0.88). No clear ev­i­dence of an as­so­ci­a­tion be­tween helmet use and neck in­jury was found (OR = 0.96, 95% CI: 0.74–1.25). There was no ev­i­dence of time trends or pub­li­ca­tion bias.

Ok, so this also sug­gest some­thing around the 23 num­ber, so I think that’s a pretty de­cent bet.

For ar­gu­ments sake, let’s just as­sume that all of the bad con­se­quences of ac­ci­dents are cov­ered by mea­sur­ing head-in­juries, re­sult­ing in a to­tal risk in­crease by a fac­tor of 3.

Con­clu­sion (very sketchy)

Ok, so if you add this up su­per naively in terms of micro­morts, then we get a (1/​3) * 3 = 1 micro­mort loss at a 1 micro­life gain for my av­er­age office com­mute. If we com­mit a statis­ti­cal atroc­ity and just add those up naively, we end up with a net gain of 0, sug­gest­ing that cy­cling with­out a helmet is roughly equally risky as do­ing some ran­dom seden­tary ac­tivity.

There are ob­vi­ously a lot of com­pli­cat­ing fac­tors to this, but I am not sure in which di­rec­tion they point. I care more about short-term in­jury risk than I care about long-term gain from ex­er­cise, but I also care about the cog­ni­tive benefits of ex­er­cise and want to max­i­mize my peak-po­ten­tial more than my av­er­age po­ten­tial.

I also know that you can’t just add micro­morts to micro­lifes to­gether, though it seems hard to figure out what the cor­rect thing to do is, and I only set aside around an hour of time for this.

Over­all, my con­clu­sion is that if I have the choice be­tween rid­ing my bike with­out a helmet, and stay­ing at home, I should prob­a­bly be mostly in­differ­ent be­tween the two. Since the al­ter­na­tive is usu­ally pay­ing for an Uber to my office, or walk­ing, or rid­ing with a helmet which I find quite ac­tively an­noy­ing, I think I will con­tinue rid­ing with­out a helmet for now, in the ab­sence of me notic­ing some new con­sid­er­a­tions or ev­i­dence (or dis­cov­er­ing some way to be less an­noyed by helmets).

Or maybe some­one feels mo­ti­vated and does a bet­ter anal­y­sis than I did here and cor­rects me. My over­all es­ti­mate on the im­por­tance of this isn’t su­per high, so I prob­a­bly won’t do much more anal­y­sis of it.