Thursday, September 8, 2011

Risk Analysis - ROM # Expected Rando Deaths

Reading the statements from the candidates for the RUSA board elections this autumn, I noticed that one candidate wants to do a study regarding safety and enhancing same for randonneurs.  The catalyst behind his thinking is that one randonneur has died while doing a brevet in each of the last four years (accepting his "fact" without checking, but I do wonder why he apparently excluded permanents).

Ignoring the curious combination of sentences "... skiers can reduce their risk if they choose.  But brevet riders do not have this option", the thing I immediately thought was:

I wonder how many deaths can be expected among randonneurs during brevets, and permanents?

Being an actuary by training, I decided to do a "quick back-of-the-envelope" estimate of those questions -- it will take me longer to write this up than it took me to do the calculations.

I needed to make reasonable estimates for a few key variables:

  • number of randonneurs, 
  • amount of time spent doing randonneuring rides, and 
  • expected death rates from all causes. 
One can consult the membership databases on the RUSA website, under RBA resources to get a list of all active members.  I did that a few months ago, and found the number was around 3000 active members.  (If I was going to do more than a "quick back-of-the-envelope" estimation, I would, of course, confirm that number.)

I went on-line and quickly found three sources of publicly available death rates by age, or by age and sex.  These included all causes of death, and while randonneurs might be healthier than the average, I'm only intending a quick rough-order-of-magnitude (ROM) calculation; these seemed acceptable for the purpose.

I chose to use the third of the three shown above.  IMO it has a couple advantages:  (1) death rates are available by age and sex should I decide to try to do a more refined calculation, (2) I inherently have a better idea of what processes may have been used to develop the death rate estimates.  (Since this is just a ROM back-of-the-envelope estimate, I did not attempt to find and study the methodology of the three above possible sources.)

I still needed to determine which death rate(s) to use from the SSA Table.  I wanted to develop a range for the answer to my posed question(s), but I didn't want to make a huge project of this (typing this far has taken longer than finding the necessary assumption estimates AND doing the calculations).  I decided to use the death rate for Males-age-50 as the low end of the spectrum and Males-age-60 for the high end of the spectrum.

That left me needing only an estimate for the time people spend doing brevets (and permanents).  If I had access to the entire RUSA database in a data-downloadable fashion, I could crunch the number for everyone, and come up with ... the exact amount of elapsed time for RUSA members on brevets and permanents that they completed ... that would still leave out time information for DNFs, and this is supposed to be a quick ROM estimate.  I needed something I could get my hands on quickly.  The answer:  my time spent so far this year doing brevets and permanents.  Am I representative of the entire membership?

Well ... I am 55 and male.
I did one SR this year (with no extras, so far).
More brevet kms than quite a few; much less than many.

I am a mid-to-back-of-the-packer.
So it probably takes me a bit longer than average to complete a brevet.

I am no Lone Star rando, but I have ridden more permanent kilometers this year than most.
As I write this, I have completed 18 permanents so far this year (4 not in the RUSA results, yet).
I have also completed 12 permanent populaires so far this year.

One could argue that I have more RUSA kms and time doing rando this year than "average".
But, I am only trying to get a ROM estimate based on readily available data.

What did I find?

Based  on:  3000 members, my time in-motion and overall elapsed time, death rates as indicated:

Expected Deaths While Engaging in Randonneuring
Time in Hrs Time in Days
Kms In-Motion Elapsed In-Motion Elapsed
Breverts 1,500 64.013 85.033 2.667 3.543
Perms 4,926 215.351 261.933 8.973 10.914
Total 6,426 279.364 346.967 11.640 14.457

Annual Death Rates Expected Deaths, In-Motion Expected Deaths, Elapsed
Male 50 Male 60 Male 50 Male 60 Male 50 Male 60
0.00551 0.01141 0.1 0.3 0.1 0.3
0.00551 0.01141 0.4 1.0 0.4 1.0
0.00551 0.01141 0.5 1.4 0.5 1.4
  • Expected deaths while randonneuring at my rate:  0.5 to 1.4 per year.
  • Expected deaths while doing brevets at my rate:  0.1 to 0.3 per year.  

Based on this Rough Order of Magnitude estimate, and leaning more on the "Total" numbers which include the permanents, I think I agree with Lois Springsteen's response to Mr. Berk's previous suggestion to RUSA regarding the safety research:  thank-you, no.

(I suspect that Jonathan Berk is a Ph.D. based on the content in his "saferrando" blog; please accept my apology if you would prefer to be "Dr. Berk", but on the other hand, we are all just randonneurs in fancy underwear here -- no need for fancy titles.)

Martin Shipp


  1. maybe you'd like to post this on Google Randon, to counter that bad "back of the envelope" number that is running around.

  2. re: "bad 'back of the envelope' number" ... that is your characterization, not mine.

    I didn't notice any expected estimate in Mr. Beck's blog ... but I could have missed it. (I stayed with "Mr." instead of "Dr." to stay consistent with my fancy underwear comment.)

    I've never joined the "Google Randon", I know it relatively painless, but, "lynnef", maybe you'd like to do the honors of making a link or whatever. (My thought on that front might change ... I am tired right now as I spent all day test riding a friend's prospective perm populaire, and other bike riding stuff.)