All the evidence says…

A recurrent theme in helmet promotion is that “all the evidence says that helmets reduce head injuries”

[we found it] impossible to definitively quantify the effectiveness or otherwise of cycle helmets based on the literature reviewed. – The potential for cycle helmets to prevent injury – A review of the evidence, Transport Research Laboratory report PPR446

This is wrong on two levels.

No evidence says that anyway!

First and foremost the research does not say that helmets prevent injury, it says that those cyclists who choose to wear helmets were, in the small groups studied, less likely to suffer head injury – in fact less likely to suffer injury of any kind, or to have damage to their bikes in crashes.  This does not scale to the population level.  In fact, the countries where head injury rates are lowest are the ones with the lowest helmet wearing rates.

Pro-helmet evidence comes from small-scale case-control studies.  These all have a problem called confounding – to put it simply, they can’t easily distinguish between the chooser and the choice.  If the most cautious cyclists are the ones most likely to wear helmets, which seems reasonable, is this likely to affect the results?  The evidence which supports helmet use is the same kind of evidence which showed that combined hormone replacement therapy reduces risk of coronary heart disease – except that it doesn’t.  Subsequent clinical trials found that it actually caused a small increase in risk, but the kinds of women who asked for HRT were more likely to be health conscious on other ways.  For such an obvious finding this caused a quite disproportionate amount of consternation in the epidemiological community!

More importantly, none of the other types of evidence support the pro-helmet results, either in magnitude or in sign.

Some evidence says the opposite

The existence of evidence which conflicts with the pro-helmet line comes as a surprise to many.  The fact that the pro-helmet evidence is of a type which is notoriously weak is rarely mentioned by helmet proponents, and the existence of conflicting evidence is rarely, if ever, mentioned at all.

Most of the conflicting evidence comes from whole-population studies.  For example, America’s Consumer Products Safety Commission produced some figures which showed that cyclist head injury rates have increased as helmet use has risen.  The same happened in Australia after a helmet law was passed.  The largest ever study, by Rodgers of CPSC, which surveyed eight million cyclist injuries, found that there was a small but not significant increase in risk of injury and a small and significant increase in risk of fatality for helmeted riders.

The most widely-studied population may be that of New Zealand, where a mandatory all-ages helmet law was passed in 1994.  This resulted in a doubling of helmet use to over 95% in a single year, and no measurable change in cyclist head injury rates.  There was a numerical drop in head injuries, and an equivalent numerical drop in non-head injuries, both of which are consistent with the large concurrent observed drop in cycling, but the proportion of cyclists’ injuries which are head injuries trended no better than for pedestrians.

True Believer syndrome

TRL were commissioned by the Department for Transport to review the evidence on cycle safety in general, including helmets. In their report PPR446 they note that “[it was] impossible to definitively quantify the effectiveness or otherwise of cycle helmets based on the literature reviewed” – but then went on, controversially, to propose a small figure for potential lives saved based on their own novel and purely speculative model – something the review was supposed to avoid.

This illustrates a fundamental and recurrent issue with helmet proponents: if the figures don’t give the answer they want, they either change the question or invent new figures. And when they get the answer they want, well, that’s good enough, no need to check further. Which is why a paper by Cook & Sheikh in Injury Prevention went to press with a glaring schoolboy error in the maths, as pointed out in a response by James Annan: Cook and Sheikh calculate helmet effectiveness to be given by the ratio 3.6/5.8 = 60%. However the correct expression to use is 13/7 = 186%. In other words, “helmet effectiveness” is so high that each helmet does not just save its wearer, but a non-wearer too. At this rate, head injuries would be eliminated completely if just a little over half of all cyclists wore them! This is clearly ludicrous. As with the “crisis in epidemiology” resulting from the false link drawn between coronary heart disease and hormone replacement therapy, authors must suspend belief, maintain scepticism and not be seduced by mechanism.