View Single Post
Old 08-16-14, 03:41 AM
  #8527  
meanwhile
Senior Member
 
meanwhile's Avatar
 
Join Date: Jun 2008
Posts: 4,033
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Likes: 0
Liked 0 Times in 0 Posts
This analysis provides an up to date summary of research on helmet effectiveness (and should be read by silly people who quote discredited papers from the 80s without checking their validity):

http://www.google.co.uk/url?sa=t&rct...,d.d2k&cad=rja

Opposite there is a long tradition of casualty surveys and laboratory experiments (virtual or not)
which in large majority find an overwhelmingly positive effect of bicycle helmets. The extent of the
reported effect does however vary strongly from zero to + 85%. That last figure comes from an older
but very influential study, of which it has since been established that it is untenable (also according
to the authors!). This number however is impossible to retract from the Internet 5 . In a recent fact
sheet SWOV puts it at +45% 6 . In June, in a previous fact sheet, it was still +15%. No explanation was
given for this significant adjustment. The figure of +45% is based on an influential meta‐study of
traffic fatalities 7 .

Recently, this meta‐study was subjected to scrutiny again by the Norwegian Elvik 8 . Elvik found two
methodological errors in the previous meta‐study. Correcting these and adding new studies to the

Paper for the “Nationaal verkeerskundecongres” Netherlands, 2 November 2011
3 meta‐study, Elvik establishes an effectiveness of 15%. He also notes that the effectiveness in the
reported studies decreases over time. When limited to the studies of the last decade, Elvik finds
there is no effect (0% effectiveness).

The author of this contribution can add the following, there is an additional methodological error
that Elvik overlooked 9 . Unjustly all studies use wrong odds ratios to estimate the risk ratios. The
effects are very difficult to map due to lack of relevant data, but in all cases it leads to an
overestimation of the effectiveness of bicycle helmets. A first indicative estimate is that the
effectiveness of bicycle helmets is overestimated by at least 11 percent points, possibly considerably
more. With this correction the alleged difference between theory and practice has in fact practically
disappeared.
To summarize, the current best estimate based on meta‐studies for the effectiveness of bicycle
helmets has an upper limit of 6% 10 and no effect (0%) lies within the unreliability interval.


..That's the bottom line: the biggest number you can pencil in for 100% helmet use would be a 6% reduction in cyclist deaths.
meanwhile is offline