Cycle helmets - public policy aspects

This document predates the creation of http://www.nohelmetlaw.org.uk

Background
On the face of it the idea that cyclists should wear helmets would appear to be common sense. So why is it so controversial, and why is legislation so vigorously resisted by cyclists' groups? The purpose of this briefing is to present to you the background to this surprisingly contentious issue, with a particular focus on the public policy issues.

Briefly, we believe that cycle helmet laws represent the wrong solution to the wrong problem. They have been tried, and have failed, and it is time to move on to a more effective and less victim-focused approach to safety issues. As HL Mencken said, to every complex problem there is a solution which is simple, neat and wrong...

Basis For Objections
Opposition to helmet legislation is often perceived as being on libertarian grounds, much as the 1970s campaigns against compulsory motorcycle helmet use (another intervention where it is notoriously hard to find any tangible benefit in the road traffic casualty statistics). This is not the case. Although some oppose helmet laws on that basis, the major opposition from cyclists is based on a variety of compelling evidence gathered from around the world.

Questions Of Policy
In order to gauge whether helmet legislation it is merited, we must first establish what criteria might justify an intervention of this kind. We believe that the following criteria should apply to any proposed legislation:


 * Is the activity in question unusually dangerous?
 * If so, is the danger inherent in the activity, or consequent on some other cause?
 * Is the activity unusually likely to result in the kinds of injury causing concern?
 * Is the proposed intervention designed to prevent injuries of a severity which might justly be of public concern?
 * If the intervention has been tried elsewhere, did it work?
 * If other interventions are available, is the proposed one the most effective?
 * Is the proposed intervention free of potential unintended adverse consequences?

In the case of cycle helmets it is the firm belief of cycle safety experts and cycling bodies alike than the answer in each case is: no.

Is cycling unusually dangerous?
There have been a number of studies which quantified the effect on overall life expectancy of regular cycling. Without exception they show increased lifespan and improved health. These studies are based on large populations and are repeatable across different study groups. This is not in the least controversial - the idea that cycling is beneficial to health has been the basis of public policy for some years. One might dispute the precise scale of the benefit - whether it is an average of two years' greater life expectancy, as per the BMA study, or Mayer Hillman's calculation of the benefits outweighing the risks by a factor of 20:1. Whichever you believe, these studies, being based on real populations, account in full for any risk potentially posed by cycling.

It makes little sense, therefore, to suggest that cycling is an unusually dangerous activity, although to do so is in a sense self-fulfilling as it deters cycling, and evidence form around the world shows that the best way to make cycling safer is to get more poeple cycling.

Is the risk inherent to cycling?
Consider the risk inherent in riding along a deserted road in broad daylight. It would be fair to characterise this risk as negligible. What turns this benign mode of transport into a supposedly dangerous pursuit? Clearly it is the presence of motor traffic.

Several studies have assessed the responsibility of the various parties in car v. bike collisions. Each has concluded that the cyclist is less likely to be to blame than the driver. To a first approximation, the following is true for the UK:


 * Number of cyclists injured or killed in car crashes where cyclist is to blame: 700
 * Number of drivers injured or killed in these crashes: 0
 * Number of cyclists injured or killed in car crashes where driver is to blame: 1,700
 * Number of drivers injured or killed in these crashes: 0

The problem of injury is one of motor traffic, not cycling. This is made abundantly clear from the annual road traffic casualty statistics: cyclists account for only one in seven of the vulnerable road users killed on the roads each year. In the case of children the contrast is stark: road traffic crashes account for one in ten child injury hospitalisations but half of all fatalities. Again, only a minority are cyclists.

It seems clear that the dangerous activity is not cycling. Moreover, even if all cyclist head injuries were prevented altogether (a most unlikely scenario) this would still barely dent the injury and death toll.

Is cycling unusually likely to result in head injuries?
Hard evidence is hard to come by. For some reason those with access to Government funding and data have not seen fit to quantify the proportion of cyclists' injuries which are head injuries, relative to other comparable activities. Recently, data for child hospital admissions for England over a seven-year period were provided to CTC, the national cyclists' organisation, and analysed. It was found that the proportion of cyclists head injured was similar to that for all transport injuries, significantly lower than that for pedestrians. Since pedestrian injuries generally include only road traffic collisions, the balance being recorded as trips and falls, and cyclist injuries may be from poor surfaces, ice or whatever, the indices for those injuries with and without recorded motor vehicle involvement were calculated. The values were, not surprisingly, comparable, and child pedestrians accounted for around four times as many head injuries as child cyclists.

Once again the conclusion is clear: what is unusually likely to produce head injury, especially sever head injury, is not cycling but motor traffic. And once again cyclist head injuries form a small minority of all fatalities and serious injuries from this cause.

Are helmets designed, in any case, to protect against these injuries?
The fact that the severity of injuries in road traffic crashes is substantially elevated, is not, we believe, seriously disputed. Helmet legislation is almost invariably positioned as a "road safety" measure, not least because of the obvious difficulties in regulating the behaviour of private individuals on private land. Most serious cyclist injuries, and almost all fatalities, come from road traffic crashes.

This, then, is the "problem" to be solved.

Are helmets designed for that kind of crash? No. The dominant standard in Europe, EN1078, specifies a drop test which is equivalent to an impact speed of roughly 12mph - this is the equivalent of a fall to the ground from a stationary or slow-moving bike. The forces developed in road traffic crashes typically exceed the rated capacity of the best motor racing helmets. This is inherent in the design compromises which must be made for a helmet which is, in the end, designed to be used while doing physical work.

To suggest that helmets are a road safety intervention at all, flies in the face of reality. Indeed, it might even be considered dangerously misleading, as the well-known mechanisms of risk compensation mean that, if anythign, the benefits of an intervention should always be understated and never under any circumstances exaggerated.

Finally we would submit that even this misses the point: the source of danger is neither cycling nor cyclists. Even where cyclists' behaviour might be viewed as irresponsible, the consequences in the absence of motor traffic are generally minor.

Have helmet laws been tried? Did they work?
Some countries already have helmet laws. These may be for children only, or for all ages. In no case has there been any measurable reduction in the levels of cyclist injuries. Broadly speaking, helemt laws have one of two outcomes:


 * The law is not enforced: there may be some reduction in cycling due to the false portrayal of cycling as dangerous, but thelmet usage rates do not change much and neither do injury rates. Early data from Alberta showed a significant and worrying rise in injury rates, but numbers of serious injuries are in any case so small as to make accurate year-on-year comparisons difficult.
 * The law is enforced: there are substantial and sustained reductions in cycling, especially utility cycling, and particularly among teenagers. Injury numbers drop broadly in line with the drop in cycling, head injuries as a proportion of all injuries do not change due to the law. Other measures may reduce head injury rates for all road users, but cyclists experience no improvement which is not equally apparent in other road user groups. The two most closely studied examples here are Australia and New Zealand.

If helmets are genuinely effective against head injuries then it should be obvious from the road accident statistics for countries which have achieved substantial step-change increases in helemt wearing. This turns out not to be the case. This chart shows the percentage of hospital admissions which were head injuries, for a seven year period, for motor vehicle, non motor vehicle and all injuries. Knowing that helmets are designed primarily to be effective in low speed falls, the figure for non motor vehicle injuries in particular should reveal the helmet affect very clearly. Over this period the adult helmet wearing rate rose from 30% to 43%, then in a single year climbed to 92%. It then declined back to 87%. Looking at the chart below, can you tell in which year the helmet use rate more than doubled to over 90%?



The answer is: 1994. In the same year new rules on drink driving were introduced, and enforcement was greatly increased. This may go some way to explaining the relatively good trend for motor vehicle injuries. The trend for pedestrian injuries is similar.

A cost-benefit analysis was performed which started from the premise that the helmet law had reduced serious injuries by 19%. This figure was arrived at by ignoring the evident pre-existing trends. Even with this obvious distortion, the calculation still showed that the helmet law represented a net cost to the country, even before any allowance was made for the health costs of reduced levels of cycling.

Are other solutions available; would they be more effective?
Health and Safety best practice dictates that the approach to risk management should be: first, to reduce risk at source; second, to reduce exposure to risk; third and last to apply personal protective equipment. Cycle helmet laws reverse this order. Worse: they subvert it entirely, because it turns out that the majority of those injuries causing concern are beyond the limited protective scope of cycle helmets.

There are very few published evaluations of the relative efficacy of different cycle safety interventions. Usually any study is conducted in order to support a pre-existing agenda, be it conspicuity aids or helmets. There are two well-known cases, one from TRL and one being a survey of BMA members. Both put helmets last on the list, and both put reducing motor traffic speeds first.

As has been noted above, many pedestrians and cyclists die or are seriously injured in road traffic collisions every year. Few, if any, drivers are injured in these collisions. Cyclists are usually not to blame in injury accidents involving them, and where blame is apportioned to the pedestrian or cyclists the consequences would clearly have been greatly reduced if it were not for the presence of a motor vehicle. Reducing danger at source by controlling motor danger is clearly going to be more effective than attempting to mitigate a subset of injuries in a minority of the injured, using a device which was designed for a less demanding job altogether. But more than that: it is fundamentally more equitable, since the danger, such as it is, exists almost exclusively because of the car.

Recent experiments such as the "naked streets" initiatives have shown that where genuine efforts are made to reduce danger at source, not only does this result in greater safety for all - including the far more numerous pedestrians - but it also stimulates active travel, walking and cycling, and offers common space back to local residents, giving social benefits.

Are there any possible unintended adverse consequences?
There are two well documented adverse consequences of helmet promotion: deterrence of cycling and risk compensation.

Deterrence of cycling is a double whammy because not only does the helemt law fail to improve the head injury rate, it actually undermines one of the key factors which is known to make cycling safer. There is robust evidence to suggest that the more people cycle, the safer it gets. Logically one might expect the numbers of injuries to rise as more people ride, but the risk per mile travelled goes down - and in some cases total injuries go down as well.

A decade after the Australian laws were passed cycling is creeping back to pre-law levels, but utility cycling, which is especially beneficial, is still depressed, and the recovery does not match the increase in population over that time. Other countries have seen significant rises in levels of cycling over the same period, and much of the recovery in Australia was sport cycling following the Olympic games.

Risk compensation is a widely documented theory first proposed in the 1980s. The theory states that we respond to changes in perceived risk, by changes in our risk-taking behaviour. It has been shown to happen in the case of almost every road safety innovation: primary safety devices such as ABS brakes; secondary safety, including seat belts and airbags; and engineering changes, with accident migration around treated blackspots. It has been specifically documented in the case of child cyclists and helmets.

In the case of helmets it is possible that the gross exaggeration of the potential benefits which has been a feature of pro-compulsion campaigns may have had the effect of nullifying much of the limited benefit which helmets can confer. Certainly there is no sign that the modest changes in helmet use in the UK over the last few decades are reflected in changes in head injury rates: they seem to rise and fall quite independently of each other, much as in New Zealand.

It is undoubtedly especially irresponsible to give the impression that helmets can save your life in a road traffic crash. No manufacturer could or does make that claim, the law would not permit it.

All Cyclists Are Not Equal
Another fundamental problem with one-size-fits-all legislation is that it assumes that the risk profile is similar for all cyclists. In fact, cycling is a very broad church indeed, encompassing John Major's vision of elderly spinsters riding to Church, children on tricycles, long-distance tourists, commuters, people riding a hundred yards to the shops, riders taking their folding bikes on trains, kids doing stunts on BMXs, serious downhill mountain bikers and weekend racers - there is no such thing as a typical cyclist, no such thing as a typical cycle journey. While it may seem that the inconvenience of wearing a helmet for a Sunday afternoon club ride is trivial, a few seconds at each end, the need to store a helmet when off the bike can make it quite impractical for a short trip to town visiting severla shops, or a mixed-mode journey where the bike must be folded rapidly and carried on a crowded train, along with the day's work in a briefcase or bag.

To discount any potential effect on some riders based on a perception of "a typical rider" risks missing this fundamental point. It may be that a mature adult is capable of assessing the likely protective effect of a helmet, and ensuring that the additional risks taken do not outweigh them. It is highly unlikely that a child will make such a sophisticated judgment anything like as accurately. Bell advertise their helmets to the youth market as "courage for your head", and mountian bike magazines routinely describe helmets and even full body armour as allowing the rider to push that bit closer to the edge.

It is also apparent that different cycling behaviours have different risk profiles. For example:


 * A particular problem for urban cyclists is that of left-turning heavy goods vehicles. Conditioned to keep left at all times, cyclists will begin to pass a goods vehicle on the left in traffic, only to have it turn left across them - and sometimes crush the cyclist against railings, ironically placed there to improve "road safety" by preventing pedestrians from crossing at the junction. Training and education, for both cyclists and goods vehicle drivers, obviously has the potential to be highly effective in reducing the incidence of such cases.
 * Fear of traffic drives many cyclists to use the pavement, which brings them into conflict with motor traffic at junctions. Collisions resulting from crossing at junctions rather than negotiating them in the traffic stream will account for a number of the cases where an injured cyclist is said to be at fault. Evidence indicates that the risk at junctions is around six times greater for cyclists out of the main traffic stream, whether on the pavement, a pavement cycle path, or a cycle lane.
 * Some youths engage in stunts. The involvement of the bicycle is incidental here - roller shates and skateboards are also used. Interestnigly, this is one group where helemt use is widespread, and often hard--shell helmets (almost unknown among road cyclists) are used. This does not result in greater safety, instead the stunts become more ambitious.
 * New riders especially may simply fall off. Teaching children to ride by using training wheels (stabilisers) inhibits the learning of balance and may make this worse, but falling off is a brief transient phase in any new rider's skill development. Adult learners and those who take up a radically different form of cycle, such as a recumbent or unicycle, experience the same learning curve. This phase of learning, which results in exactly the type of crahs for which helmets are designed, lasts perhaps a few months at most, in some cases only a few days.

Little has been done to analyse the fundamental causes of cyclist injury, especially when compared with the incredible volume of data published on the injuries themselves. Where analysis has been done it very often points up interventions or adaptations which can reduce the incidence of crashes, often with the potential to benefit others as well as cyclists. We suggest that more work along these lines is a substantially higher priority than compelling the use of helmets, which are not proven to address the problem - if indeed it is the right problem in the first place.

Questions of evidence
In every case the push for legislation has come in the first instance from "injury prevention" activists, usually with a background in emergency medicine. These groups typically have a narrow focus: they see injuries; they perceive that, all other things being equal, these injuries might have been less severe had protective equipment been in use; they call for the use of protective equipment to be compulsory. The problem ,of course, is that all other things are rarely equal.

The "injury prevention" approach to interventions is two-pronged:


 * define a "problem"
 * offer a "solution"

There are at least three fundamental flaws with this approach:


 * it is founded on investigating only the consequences of crashes, so fails to account for well documented differences in behaviour between helmeted and unhelmeted cyclists
 * it proposes cycling as the cause of the injuries, when very often the cause is actually a road traffic collision; the same false conclusion is not applied to pedestrian injuries
 * in failing to account for the differences between those injuries resulting from simple crashes and those from motor vehicle collisions, it loses sight of the fact that no helmet in the world is designed, tested or warranted for the kinds of energies involved in motor vehicle collisions: the problem, in as much as it exists, is a problem not of cycling, but of motor traffic.

Define a "problem"
The problem is generally defined in terms of the solution: this happens because not enough cyclists wear helmets. In the case of cycle helmets in particular the studies which underpin promotion all start from the basis that helmets are necessary or desirable, and express differing degrees of bewilderment at cyclists' inability to see this "obvious" fact. The possibility that cyclists may have reviewed the available evidence and decided that helmets are unnecessary or irrelevant is not even alluded to. In fact, most pro-helmet studies don't even acknowledge the existence of conflicting evidence. They can get away with this because the evidence against masive efficacy tends to be published in transportation safety journals or other sources, where the author can find a publisher at all for such a heretical view, whereas the bar to publication of het another pro-helmet observational study in the medical press is set fairly low. This is changing over time with Malcolm Wardlaw and others at least being allowed to raise the possibility that all is not what it seems, but the majority of reviews of evidence still include exclusively Medline sources and so will not pick up, whether by accident or by design, the data which contradicts the pro-helmet case.

Cyclists have always suffered a certain level of injuries, some of which are to the head (the tendency of high-wheelers - penny-farthings or "ordinaries" - to pitch the rider face down onto the ground was a key driver for the adoption of the safety bicycle, which we recognise today as the norm for bike design). Despite this, the level of cyclists' head injuries does not seem to be greater than for comparable activities, in as much as data is available. It is a stable feature of "injury prevention" research that little effort is devoted to proviing contextual information which might reveal the predefined solution to be irrelevant.

Some figures are quoted for the scale of the "problem". These are often speculative, sometimes simply false. For example, a figure of 50 child cycliust head injury deaths per year, quoted in a debate in the British Parliament in 2004, is five times the actual number as given in a Parliamentary written answer only a few months earlier.

Other figures are accurate but irrelevant. While it may be true to say that around 2/3 of cyclists' fatalities are due to head injury, this figure is typical for all trauma deaths; many figures are quoted in isolation, without context or comparison with other activities.

Offer a "solution"
As evidence to support their calls for compulsion helmet promoters will generally cite studies which make substantial claims for injury reduction form helmet use. Of these the most widely quoted is a 1989 study from Seattle by authors Thompson, Rivara and Thompson.

This study gives by far the highest figures of any study of its type, 85% of head injuries and 88% of brain injuries, but arrives at those figures by a process whose flaws are by now well known. For example, it compares mainly young black male riders cycling along on urban streets, with predominantly white middle class families riding on bike trails, and attributes the different injury profiles to differences in helmet use. It also considers the helmet usage rate of the latter group to be typical despite co-author Rivara's own contemproaneous street counts showing a much lower rate. Simply substituting these lower figures reduces the calculated benefit to zero, within the limits of statsistical error. Alternatively you can take the authors' figures and demonstrate that helmets prevent 75% of broken legs.

Notwithstanding the faults in the study, it is misleading in two other ways.

First, it is presented as concluding that helmets prevent 85% of head injuries and 88% of brain injuries. This is not, in fact, what the study says. The authors acknowledge that the different injury profiles may indeed be a result of different levels of risk-taking and different exposure to risk, even though decide that the effect is minor. The study is not predictive ion the way it is represented, it is quantifying the differences in rates of a certain type of injury between two different populations of cyclists, the differences between which include, but are not limited to, differences in helmet wearing rates. The extent to which the difference is attributed to that one factor is largely subjective - and these authors were already staunch helmet advocates before the study was commenced.

Second, the terms "head injury" and "brain injury" are rarely defined when the study is cited. In the study a head injury is any injury to the head or scalp, including cuts and grazes, and a brain injury is almost always a simple concussion. While these injuries can be distressing or concerning, particularly to the parent sand family of those injured, they are very rarely life threatening or even particularly serious.

In fact, no study can truly demonstrate that helmets prevent X% of injuries, because the only type of study which could be considered to do so would be a randomised controlled trial. These are generally considered unethical in the case of safety equipment. Interestingly, in recent times, a fundamental weakness in hospital-based observational studies has been exposed. A number of studies and meta-analyses showed a link between hormone replacement therapy (HRT) in women and reductions in risk of coronary heart disease (CHD). Just as with helmet studies, there was a repeatable effect and a plausible mechanism for its delivery. But when randomised controlled studies were carried out, it was found that the HRT group actually had a slightly higher incidence of CHD. The reason for the error was self-selection bias. The women volunteering for HRT were predominantly from the higher socio-economic groups and had, on average, better diet and were more health-conscious.

Exactly the same self-selection bias has been documented in cycle helmet studies, most vividly by Spaite et. al., who commented: "A striking finding was noted when the group of patients without major head injuries was analyzed separately. Helmet users in this group still had a much lower mean ISS [injury severity score] and were much less likely to have an ISS greater than 15 than were nonusers. [...] We conclude that helmet nonuse is strongly associated with severe injuries in this study population. This is true even when the patients without major head injuries are analyzed as a group".

Conclusion
Cycle helmets have been around for approximately a quarter of a century, during which time their adoption has varied between near-zero (e.g. India, China, the Netherlands) and near-total (e.g. New Zealand). There is no association between helmet use, or increases in helmet use, and reductions in head injury.

On helmets:
 * Cycle helmets are not a road safety product, and are not designed or sold as such.
 * Cycling is, in general, a safe and healthy activity.
 * What risk exists is not inherent in cycling, but scomes either from particular risk-taking behaviours or, more particularly, from motor traffic.

On helmet laws:
 * Not surprisingly, given that helemts are not a road safety product, where helmet laws have been tried as a road safety measure, they have failed.
 * Helmet laws raise a false perception of cycling as dangerous, and deter cycling.
 * Helmet laws cost more in helmet purchases alone than they save in injuries.

Questions to ask:
 * Is there genuine evidence that cycling is unusually dangerous?
 * If so, is the danger inherent in the activity, or consequent on some other cause?
 * Is cycling more likely to cause head injuries than other activities we consider safe, like walking?
 * Are helmets designed to prevent serious injuries, rather than mainly trivial ones which are of no real public concern?
 * Helmet laws have been tried in other countries: what were the results?
 * Is helmet use the first, best way to improve cyclist safety? Or would training, traffic calming and other measures have a greater effect?
 * Is a helmet law free of unintended adverse consequences?
 * Is the population of cyclists in any case sufficiently homogeneous that any "one size fits all" solution is likely to be appropriate?

Some false or misleading claims are usually advanced by helmet advocates:
 * The figure 85% of head injuries and/or 88% of brain injuries is disgnostic of ill-informed or deliberately misleading information. They come from a single study whose flaws are well documented; no other study has come close to these figures and no real world population has demonstrated even a tiny fraction of this kind of reduction.
 * Conflating numbers of trivial injuries with consequences of serious ones. The vast majority of head injuries suffered by cyclists are cuts and bruises, and the vast majority of what are described as brain injuries are simple conussions.
 * Arm-waving and lack of context. It is easy to say that hundreds of cyclists are killed every year, but when this is taken in the context of an estimated six million children and up to thirty million adults owning bicycles, the risk looks much smaller. It is easy to say that "many" injuries would be prevented, since the scientific definition of "many" is any number greater than five, but without hard figures this, too, is meaningless.
 * Implying that efficacy against trivial injuries implies equal efficacy against more serious ones. In fact some studies go further and claim that helmets, alone of all protective equipment, are more efffective against serious injuries than against trivial ones.
 * Use of numbers of injuries where the correct measure is risk per unit exposure. For example, some sources claim the Australian laws as a resounding success due to a reduction in the number of head injuries. The fact that the reduction in cyclists on the roads was at least as great, and trends for head injury rates therefore remained unchanged, is not acknowledged.
 * Shroud-waving. Every individual cycle crash is different, and it is generally impossible to establish if a helmet would have made any dfference at all in any given crash, let alone a crucial difference; very often it transpires that some other intervention may have prevented the crash altogether, which woudl clearly have a much higher chance of success. In any cas, public policy based on rare cases, however tragic, is rarely sound, as the Dangerous Dogs Act and the Community Charge (predicated on the mythical poor old lady living alone in a large house) have shown.