Barry Pless claimed that critics of helmet laws rely on fatality data. Yet my review considered all injuries serious enough to require hospital admission in all jurisdictions where helmet wearing increased substantially (more than 40 percentage points). There were no obvious responses in percent head injury.[1]

In contrast, the large, obvious falls in both non-head and head injuries with Victoria's helmet law – see graph (right) – refute Pless' claim that times series studies are "weak". Helmets cannot prevent non-head injuries, so the only plausible explanation for the large drop in non-head injuries is a decrease in cycle-use. Coupled with pre- and post-law observational surveys (640 person-hours per year at the same sites, observation times and the same time of year) showing a 36% reduction in cyclists counted, this represents overwhelming evidence that the main effect of the law was to discourage cycling.

Confusion arises when people hear only half the story. A busy researcher, reading in the Cochrane review of Thompson et al.[2] that Victoria's helmet law reduced head injuries by an estimated 40% would have no idea that non-head injuries also fell substantially. It would be natural to jump to the conclusion that the effect was due to increased helmet wearing. Even though Pless finds the arguments "tiresome, almost boilerplate", people need to see the data on both head and non-head injuries so that constructive debate can take place.

Despite the irritation, real progress has been made, as demonstrated by Pless' comment that "no sensible helmet advocate has argued that a typical bike helmet provides adequate protection against several tons of moving metal". Only about 47% of non-helmeted cyclists die of head injury (a proportion that, after adjusting for age, sex and other confounders appears no different to the 33% of helmet wearers dying of head injury), yet two recent papers in Injury Prevention claimed or assumed that helmets prevent an astonishing 73% and 65% fatalities.

The moderation from claims that helmets prevent 65-73% of fatalities (i.e. all head injury deaths and a large proportion of other deaths) to "inadequate protection" is a welcome improvement. Debate can now focus on the evidence (summarised in my review and elsewhere) that risk compensation, losses in health and environmental benefits from reduced cycling, and reduced safety in numbers probably outweigh any benefits of helmet laws.

Health benefits vs cost of reduced cyling
The health benefits of cycling are large. In Denmark (where only about 3% of cyclists wear helmets), the modest amount of daily cycling needed to ride to work reduces mortality by 40%.[3] A UK study of mainly non-helmeted cyclists found that the health benefits of cycling, measured in years of life gained, outweigh the injury risks, measured in years of life lost, by about 20 to 1.[4]

In contrast, the benefits of helmet laws are small. Estimated head injury reductions from New Zealand's helmet law ranged from zero (if trends were fitted in the model) to about 19% (ignoring trends). A peer-reviewed paper calculated the saving in hospital costs. The most optimistic estimate for a helmet bought to satisfy the law was a saving of NZ$0.65 over its 5-year lifespan, i.e. 13 cents per helmet per year![5]

Estimates are slightly higher (about NZ$2.85/helmet/year) if no consideration is given to the cost of reduced cycling nor willingness to pay to avoid inconvenience and discomfort of helmet-wearing, but willingness to pay to avoid pain and inconvenience of injury is included. Cyclists who are free to choose can weigh up the risks and the inconvenience, perhaps wearing a helmet for a rapid mountain descent but not for a short trip to the shops on a hot day. One possible explanation for the lack of benefit from helmet laws is that cyclists are able to judge the risks and know what gear is appropriate.

Simple criteria vs "solid" research designs
Simple criteria, such as the absence or presence of a response (and the size of that response) are an effective way of judging the value of an intervention. For example, fatality data provide very convincing evidence that speed cameras and random breath testing are highly effective road safety measures. But if no response had been evident (either for fatalities or serious injuries), would we argue that the measures were effective, the lack of response being due to the "weakness" of time series data? Or simply concentrate on measures that produced large, obvious responses?

Case-control studies that Pless describes as "solid research designs" led to the conclusion that hormone replacement therapy (HRT) reduces the risk of heart disease by 50%. This was later dismissed when randomised control trials showed that HRT can actually increase the risk of heart disease.[6]

A case-control study of children injured in an activity that could entail use of protective equipment (PE) led Barry Pless to conclude that risk compensation by children was "highly doubtful".[7] Yet when completing an obstacle course in a gym, wearing a helmet and wrist guards increased risk taking (measured by tripping, falling and bumping into things) by 60% and 49% for 10-12 and 7-9 year old girls and 48% for 10-12 and 7-9 year old boys. Just as the best way to measure risk compensation is to observe the behaviour of the same children with and without PE, the most reliable way to determine the outcomes of helmet laws is to evaluate what happens when such laws are passed. Consequences such as risk compensation or reduced safety in numbers do not have to be quantified separately. They are instead (quite correctly) treated as consequences of legislation.

All costs and benefits should be considered. To assess potential returns from use of scarce resources, cost-benefits should be compared those for other road safety measures. For example, one team of researchers concluded from real-life crash data that helmets for motor vehicle occupants might prevent 28%, 40% and 26% of minor, moderate and severe brain injuries. Thus a helmet law for motorists (in addition to seatbelts) could save $1.9 billion (over 5 years, all vehicles equipped with airbags) to $2.2 billion (50% with airbags).[8] This works out at more than $100 per helmet, an order of magnitude greater than the most optimistic estimate of the benefit of NZ's helmet law.

Although benefits of helmet laws for vehicle occupants might be offset by risk compensation, there would be no problem of discouraging healthy exercise and environmentally friendly transport. The higher estimated benefits and fewer drawbacks imply that any future calls for bicycle helmet laws should be postponed until helmet laws for motorists have been implemented and shown to be beneficial.

References
1 Robinson DL. No clear evidence from countries that have enforced the wearing of helmets. BMJ 2006;332:722-725.
2 Thompson D, Rivara F, Thompson R. Helmets for preventing head and facial injuries in bicyclists (Cochrane Review). In: The Cochrane Library, Issue 3. Oxford: Update Software, 2003.
3 Andersen LB, Schnohr P, Schroll M, Hein HO. All-cause mortality associated with physical activity during leisure time, work,sports, and cycling to work. Arch Intern Med 2000;160(11):1621-8.
4 Adams J, Hillman M. The risk compensation theory and bicycle helmets. Inj Prevent 2001;7(2):89-91.
5 Taylor M, Scuffham P. New Zealand bicycle helmet law-do the costs outweigh the benefits? Inj. Prevent. 2002;8:317-320.
6 Lawlor DA, Davey Smith G, Ebrahim S. Commentary: The hormone replacement-coronary heart disease conundrum: is this thedeath of observational epidemiology? Int. J. Epidemiol. 2004;33(3):464-467.
7 Pless IB, Magdalinos H, Hagel B. Risk-compensation behavior in children - myth or reality? Arch Pediatr Adolesc Med 2006;160:610-614.
8 McLean A, Fildes B, Kloeden C, Digges K, Anderson R, Moore V, et al. Prevention of Head Injuries to Car Occupants An Investigation of Interior Padding Options: Federal Office of Road Safety: Rpt CR160. Available at http://www.monash.edu.au/muarc/reports/atsb160.pdf 1997.