BMA

As you may know the BMA recently adopted a policy of advocating helmet legislation, following publication of a report by its Science Committee. This report was an execrable piece of work, including such easily verifiable falsehoods as BeHIT's "50 fatalities" claim. It has now been toned down somewhat but remains poorly researched and highly subjective. It is clear that as long as they keep giving the job of writing this report to a helmet zealot whose primary resource is BeHIT press releases, the BMA are going to continue to look foolish. As HL Mencken said, "for every complex problem there is a solution which is simple, neat and wrong."

The paper is here: http://www.bma.org.uk/ap.nsf/Content/Cyclhelmet;

Cyclehelmets.org have a critique.

Here are my comments:

I am inclined to the view that there is a fundamental weakness in the process by which this report has been prepared and amended: the author(s) appear so intent on proving a point that they are accepting at face value claims which anybody active in the field of cycle safety will readily identify as speculative, contentious, or even plain false.

I note that some of the more easily disproved statements have been removed, but there are still false or distorted claims, some of which are readily identifiable as emanating from the same source as the bizarre "50 fatalities" figure which has been removed. No attempt has apparently been made to verify these claims, despite this previous and plainly embarrassing experience. Moreover, it is apparent that the authors of the paper not only have not read the wider literature, it appears they have not even read the sources they cite - claims are attributed to one source when that source is in turn citing another paper already referenced by the BMA's authors. This is appallingly sloppy!

A paper on which a body as prestigious as the BMA intends to base lobbying for public policy change should at the very least meet the standards required for publication in a journal such as the BMJ. I strongly suspect that if this paper were submitted to the Editors it would be rejected out of hand, and certainly following peer review. The scope of research is narrow, the sources highly selective and taken in the main from helmet compulsion activists. Context is almost entirely absent. What is the evidential basis for singling out cycling as a candidate for special treatment in this way? By what measures is cycling either uniquely dangerous or uniquely productive of head injuries? The authors do not say.

In short, the paper reads like a press release from the national association of hand-wringers - it is lousy science and an appalling basis for policy in any body with claims to scientific credibility. It falls a very long way short of the breadth and depth of the previous review which it apparently replaces. In short, it does the BMA no credit to replace a policy based on a review by the internationally renowned Meyer Hillman, with one based on a document whose authors are not even identified by name, and who display in several areas a profound ignorance of the current state of knowledge on cycle safety.

The paper says: "The evidence from those countries where compulsory cycle helmet use has already been introduced is that such legislation has a beneficial effect on cycle-related deaths and head injuries." This is simply not true: the facts from Australia and New Zealand, for example, show that the laws were followed by a steep decline in the numbers of cyclists, and consequently the *numbers* of head injuries dropped, but the *rates* (per mile travelled or as a proportion of all reported cyclist injuries) remained the same or worsened. In the six months following the passage of a law in Alberta, Canada, the cyclist head injury rate approximately doubled compared with the trend for the previous two years. While this may well be an artifact (cycling injuries are, despite the shrill cries of the helemt lobby, relatively rare), it is certainly not possible to justify the claim as made. Indeed, if the claim were true, Road Safety Minister David Jamieson could not have stated in a letter to Michael Jack, MP, that the British Government knows of no data linking increases in helmet use with improvements in cycling safety.

The paper says: "Recent evidence has indicated that the introduction of compulsory legislation does not have a significant negative effect on cycling levels." This refers solely to the situation in a single province of Canada, where the law was not enforced. Wearing rate trends did not change as a result of legislation - and neither did head injury rate trends. This has been pointed out before, I believe, but remains in the report, although discussion of this "recent data" has been placed separately at the end of the paper, which has the effect of obscuring the source somewhat.

The paper says: "Such legislation in the UK should not discourage cyclists and lead to a more sedentary lifestyle with consequent health risks." This is outright speculation, especially since the BMA appears to want an enforced law, and the evidence from jurisdictions where a law has been enforced strongly supports the idea that cycling is discouraged. It is perverse to claim in one breath that cycling is a safe and healthy activity, and in the next that it cannot be undertaken withut special protective equipment.

The paper says: "Cycle helmets are now compulsory in Australia, New Zealand, Spain, (etc.)" While true, this does not constitute evidence. Some 75% of countries have laws requiring one to drive on the right, not the left as we do. Does that mean we must follow suit? To constitute evidence in favour of helmets the list of "law" countries must be measured against the following criteria:

(1) Do these countries have, as a result of their laws or otherwise, a safer cycling record than the UK? (2) Did these countries demonstrate an improvement in cyclist safety as a result of these laws?

The answer to 1 is: no. One example of a country with a markedly better cycling safety record than the UK, is the Netherlands. They have a legislative framework for cyclist safety, but in their case it is a law of "strict liability" where any motorist injuring a vulnerable road user (pedestrian or cyclist) is presumed to be at fault. This, unlike helmet laws, goes to the heart of the source of danger, rather than attempting to mitigate the effects of it, and it benefits pedestrians as much as it does cyclists. The BMA should not need to be reminded that the major cause of injury death in children is road traffic crashes, and only a small minority of these involve bicycles.

The answer to 2, according to the British Government (and most impartial observers) is also no, but to an extent it depends on your definition of improvement. As stated above, while both Australia and New Zealand saw numerical declines in cyclist head injuries following the passage of laws, these were more than matched by declines in numbers cycling (whether or not one accepts that these were due to the laws) and in neither case did the injury rate drop. Graphs of percentage head injury in New Zealand show that the decline in cyclist head injury percentage tracks that of all road users for the five years before and after the law, years during which helmet wearing rates rose from around 40% to over 95%, and then fell again to under 90% - anecdotal evidence suggests that rates may now be as low as 50% in some areas. There is, in short, no credible evidence that the New Zealand law (to name but one well-documented example) has delivered any benefit from its helmet law. One of Scuffham's papers explicitly counts reduced cycling as a benefit, in defiance of the well-supported view that the health benefits of cycling outweigh the costs. As Bob Davis puts it in Death on the Streets, the transitive and intransitive meanings of danger have been confused.

Even if pre-existing trends and trends for non-cyclists are ignored, so the results of all injury reductions for cyclists are attributed solely to helmets (which is clearly not the case since similar improvements were observable in the non-cycling population), the laws in New Zealand still resulted in less than a quarter of the injury reduction claimed in advance - and cost-benefit analysis indicates that even with this most favourable distortion the law still represented a net cost to the economy.

The paper says "Studies in a number of these countries have shown that high usage rates of helmets as a result of legislation is associated with a reduction in cycle related deaths and head injuries." Really? According to the US Consumer Products Safety Commission, the risk of head injury has approximately doubled during the period when helmet use rose from under 20% to around 50%. Perhaps the author means "some studies" - in which case these should be named, and, given that BMA is supposedly an impartial and professional body, balanced by those which tell a conflicting story. Such as Rodgers' 1987 study of eight million cycle injuries, which showed that helmet wearers were significantly more likely to suffer a fatal accident than non-wearers.

The paper says "Evidence supporting the wearing of cycle helmets continues to mount.". Really? The majority of the published original research into helmets predates the BMA's last review, in 1997, and what has been published since is predominantly meta-analysis. There are a few genuinely new pieces of evidence, such as the 2004 paper by Mok et. al in Injury Prevention, which showed that risk compensation could be detected in the use of protective equipment (specifically including cycle helmets) by children - a point which the paper ignores entirely. Problems with observational studies and self-selection bias are, of course, topical in the medical community.

The paper says "It is estimated that 90,000 road-related and 100,000 off-road related cycling accidents occur every year in the UK, of which 53% (100,000) involve children under sixteen". The cited reference is a secondary source, the original being the paper's own reference 5, and here the figure is revealed as speculative, with no source cited. There is also no measure of the severity of these injuries, no reference to the causes (whether single-vehicle or involving motor traffic, and no indication of the rate of injuries as a proportion of all cyclist crashes. My own analysis of hospital admissions data for children in England over a seven year period leads me to conclude that the proportion of admissions in child cyclists which are due to head injury is typical for all injury admissions, somewhat lower than for child pedestrians.  The paper makes no attempt to place the figures in context, or to provide any measure of how these figures might compare with, say, injuries due to trips and falls, which account for half of all child injury admissions.  Cycling accounts for around 6.5% of serious child head injuries, and many of these will involve motor traffic.  75% of children have bicycles. One solid fact is that the total number of admissions to hospital for cycling head injuries is around 2% of the figure stated, so the vast majority of these injuries, if the figure can be justified at all, are the kind of trivial cuts and scrapes which, in a more innocent age, were accepted as a normal part of growing up. It is interesting to speculate what proportion of these minor injuries would have been prevented by any kind of headgear; no thought is given to this, the assumption is that only a helmet can prevent "head injuries".

The paper says: "In 2002, 594 children and 1,801 adults were killed or seriously injured as a result of road-related cycling accidents." While this may be true, the paper is surely not seeking to claim that cycle helmets are effective in crashes involving motor traffic? This is a wholly unsupportable idea, as even the most cursory reference to the standards will reveal, yet what other purpose can there be in mentioning the figure? I also wonder: why were the 2003 figures not used? And what about the 2,828 child and 5,644 adult pedestrians killed or seriously injured the same year? In this context surely the law to be pressed for is again that of presumed fault in crashes involving vulnerable road users, which would yield benefits in a much larger target group. The KSI figure for child cyclists in 2002 is about the same as that for bus users - does the BMA think that bus travel is especially dangerous? The statement also places the emphasis of danger in the wrong place. Bicycles, on their own, present little or no danger. Most serious and fatal juvenile head injuries occur in collision with a motor vehicle.

The paper states: "Significantly, with child cyclists, 85 per cent of accidents occur off road where primary prevention measures such as cycle lanes, vehicle speed reduction and driver education are ineffective" and cites as evidence a paper written by Lee and Mann of the Bicycle Helmet Initiative Trust, a single-issue pressure group (the group responsible for the 50 fatalities claim and other speculations). Published criticisms of this paper have also apparently not been followed up. Aside from the speculative nature of this figure, there is the question of how many of these might be serious. Since child cyclist hospital admissions divide approximately 50:50 between motor traffic injuries and injuries where no motor traffic involvement is recorded (i.e. not more than half of all admissions are due to off-road riding, and probably considerably fewer), the clear implication is that off-road crashes are on average substantially less serious than on-road crashes. This is entirely consistent with another finding form the DoH data, that the primary indicator for the presence of serious head injury is the involvement not of bicycles, but of motor vehicles. Cyclists form a small minority of this group.

The paper then goes on to cite some of the research which is routinely quoted by helmet advocates, without noting (a) how the claimed figures compare with the average for all claims made in such studies (i.e. have the largest claims been cherry-picked?); (b) what, if any, published criticisms exist of these studies; (c) how these studies relate to real-world comparisons of helmet usage and head injury trends, such as those available from the countries with helmet laws, and what evidence might there be to explain the obvious disparity. The absence of (3) is partly explained by the failure of this paper even to mention the existence of this disparity, despite its having been noted in the academic press at least as far back as 1987, and widely discussed as a result of the 2004 Private Member's Bill introduced by Eric Martlew, MP.

The paper cites reference 5 as the source for "a 30-month study of 3,854 cyclists showed that helmet usage decreased the overall risk of brain injury by 65 per cent and severe brain injury by 74 per cent in all age groups". This is a secondary source, the original is cited - it comes from the Snell Memorial Foundation, which has a vested interest in helmet promotion. Was the reference not checked? The term brain injury is also used without explanation. Few non-expert readers will realise that concussion is counted as brain injury - the public picture of "brain injury" is chronic intellectual impairment or persistent vegetative state, whereas much of the population will have experienced brain injury (concussion) at some stage and had no lasting ill-effects.

The idea that helmets are more effective against more severe injuries than against less severe ones is also hard to credit - it is saying, in effect, that a helmet tested for the equivalent of a fall from a stationary riding position will offer more protection against a serious brain injury than against a mild concussion. Not only is this somewhat implausible, it ignores the fact that the most serious brain injuries are considered to be those diffuse injuries caused by rapid twisting forces. I know of no evidence of helmets protecting against such forces, and there is some speculation that they may make them worse.

The paper's reference 7 states "An Australian study showed that wearing cycle helmets reduces both the incidence of facial injuries by 28 per cent and their severity" - this is another second-hand reporting, the original study is cited in the source. No mechanism is proposed by which helmets might prevent facial injuries; indeed this claim might well be taken by a more sceptical reviewer as evidence of confounding factors or self-selection bias in the sample (a factor common to many widely-reported observational studies, not by any means restricted to cycle helmets). Instead it is repeated without comment, further evidence of a lack of critical judgment on the part of the report's author(s).

The paper states: "A Cochrane review considering five case-control studies from the UK, Australia and the USA illustrates a large and consistent protective effect from cycle helmets, reducing the risk of head and brain injury by 65 to 88 per cent and injury to the upper and mid face by 65 per cent". This is unforgivably lax. The review in question includes, by the nature of the Cochrane review, the comments and criticisms which have been appended over time. None of these are noted. Specifically, the 88% figure is quoted despite its having been arrived at by attributing to helmets the differences in injury rates between predominantly white middle-class families riding on leisure trails, with mainly young, black, urban road cyclists. I believe the authors have since acknowledged that the figure cannot, in fact, be justified by the figures they present. It has also been pointed out that substituting the helmet wearing rate from co-author Rivara's own contemporaneous street counts for that from this homogeneous and atypical control group, brings the claimed benefit down to zero within the limits of statistical accuracy. The Cochrane review has also been criticised because the majority of cases covered come from the authors' own studies, the studies selected all lie at the upper end of the range of estimates of efficacy, and counter-evidence such as that available from Australia and New Zealand by that time was deliberately excluded.

The paper states: "A study of primary school, secondary school and adult cyclists in New Zealand demonstrated a 19 per cent reduction in head injuries to cyclists in the three years after the introduction of legislation". This cites the work of Scuffham, directly contradicting the same author's own previous analysis which showed no benefit. The change from no benefit to 19% benefit (well below the forecasts and the estimates in the Cochrane review, a fact for which the paper makes no attempt to account) was arrived at by the simple expedient of careful selection of the timeframe, ignoring an unexplained rise in fatalities the year before, and explicit exclusion of pre-existing trends. This same author has shown that even with this claimed 19% benefit, the costs of the law still outweigh the benefits. If the analysis is instead carried out over five years before and after the law, and comparison made to the head injury rate for all road users, no benefit is detectable. This was Scuffham's original thesis and it stood up to robust criticism from helmet advocates at the time.

The paper states: "In Victoria, Australia, an increase in helmet use from 31 per cent prior to legislation to 75 per cent one year after was accompanied by a decrease in head injuries by 40 per cent in the following four years". This is meaningless without the accompanying data for pedestrians and other road users, and data for exposure. In fact there was no evidence of any reduction in the proportion of head injuries, the number of head injuries per cyclist, or the number of head injuries per mile travelled, as a result of the helmet law (although the simulataneous application of more stringent policing for impaired driving, among other things, does appear to have had an impact for all road users). Levels of cycling dropped substantially in the years following the law, by an amount at least as great as the decline in head injuries. The authors cite the Cochrane review as the source for this claim, but this, too, is a secondary source: the claim originates in a 995 paper by Carr et. al. In this original source it is noted that the authors cannot tell from their analysis whether the reduction was due to reductions in cycling or increases in helmet wearing. The fact of the substantial reduction in cycling following the law has been sufficiently widely discussed that to claim a 40% reduction in head injuries without acknowledging the parallel reduction in cycling levels amounts either to gross ignorance (in which case what are these authors doing writing a policy paper in the first place) or deliberate intent to mislead.

The paper states: "As with any other legislation enforcement is as important as the law itself. Without compliance the law is at best ineffective. To achieve maximum compliance, the legislation should be complemented by mass educational and promotional campaigns. Evidence from Australia and New Zealand showed that educational campaigns prior to the introduction of legislation resulted in an increase in helmet wearing from two per cent to up to 95 per cent". So it may have done, but as previously noted it made no measurable difference to head injury rates. If the goal of legislation is to promote helmet wearing then the laws succeeded; if it is to reduce head injuries, then they appear to have failed. If the aim is to deliver, overall, better health for the nation, then it is clear to me that the helmet laws are a public health own-goal, a fact now admitted by Australia's Land Transportation Safety Board, the body responsible for the laws.

There is a thorough discussion of the effect of these laws here: http://www.cyclehelmets.org/papers/c2022.pdf. Look carefully at the charts, and remember that the steepest rise in helmet use (to 95%) happened in the year before the law became effective.

The paper cites: "An education programme in Reading that promoted cycle helmet use among children and teenagers resulted in a local increase in usage from 18 per cent to 60 per cent and a concurrent decrease in cycle-related injuries". This is the Lee and Mann paper in Arch. Dis. Child again, which appears to be the source of around half the claims made in the BMA paper. The authors' vested interest is not noted, and neither are the criticisms of this paper - for example, there were no measures of levels of cycling to determine potential deterrent effect, despite evidence from TRL that this was a possible consequence which should therefore be controlled for; and the fact that much of the claimed reduction in head injuries turns out to have happened before the programme started. In fact this report provides so little hard data that it is impossible to test the authors' conclusions.

The paper cites again the Cochrane review, criticisms of which are previously noted. Again no account is taken of the criticisms and comments which form part of that review.

The paper states: "At a practical level, enforcing the legislation can be achieved through on-the-spot fines or tickets issued by police and traffic wardens, while schools can ensure all children wear helmets on journeys to and from school. Cycle helmet legislation and other safe cycling promotions are not mutually exclusive, and there is a clear role for the simultaneous introduction of more primary prevention measures including cycle lanes, driver education and vehicle speed reduction initiatives." No evidence is presented regarding the relative merit of various "safe cycling" promotions, but international evidence for the efficacy of cycle facilities (to choose but one) in reducing the danger posed to cyclists by motor traffic is hard to come by. John Franklin, widely acknowledged as Britain's foremost authority on safe cycling, has spent some time analysing the injury and fatality record of segregated facilities and concluded that, in the main, they are ineffective in their stated aim of reducing risk, since they generally trade a reduction in a small risk (being hit from behind) for a substantial increase in a much larger risk (conflicts at junctions). To simply assume that cycle facilities are a "safe cycling" initiative is to display an ignorance of the wider context which is, again, unforgivable in a policy paper for a body as influential as the BMA.

The paper states: "The BMA believes that cycling has many advantages to the individual in terms of improved health and mobility, as well as to society via, for example, reduced air pollution and traffic congestion. Even in the current hostile traffic environment, the benefits gained from regular cycling are likely to outweigh the loss of life through accidents for regular cyclists." This is uncontroversial, but hardly compatible with the vastly inflated claims of danger, and of helmet efficacy, made elsewhere. If the figures as presented in the paper are taken at face value, helmets might be expected to prevent of the order of 100,000 or more head injuries annually, and by implication several thousands of serious, debilitating or fatal injuries. This suggests that either the wording "many advantages" is an understatement of the highest order, or the figures are simply wrong.

The paper suggests a number of possible cycle safety initiatives. No attempt is made to prioritise these, or to provide any measure of the likely benefit. In context, a poll of the BMA's own members placed helmets last in the hierarchy of possible effectiveness of various measures. Indeed, I am not aware of any objective assessment of the relative merits of various cycle safety interventions which puts helmets anywhere other than last. And rightly so: prevention of injury certainly does not begin with attempts to mitigate the last link in the chain, when the cyclist hits the ground.

And that is my biggest problem with the tone of the entire paper. Helmet laws seek to place the focus entirely on the cyclist, and on passive actions taken by the cyclist to mitigate the consequences of crashes. The causes of crashes are ignored, as are the potential effects of addressing those causes on other road users. The paper mentions, in passing, reduction of vehicle speeds. It has been estimated that a reduction of mean vehicle speeds by 10mph could save over 1,000 lives annually. By comparison, the most optimistic credible estimate I have seen for the number of children's lives which could be saved by a rigidly enforced helmet law is: one.

When we consider public policy in respect of cycling, we should be thinking: do we emulate the USA and Australia, whose cycle safety record is worse than our own (and who are, possibly not coincidentally, the two most obese nations on Earth), or the Netherlands? The paper repeats a claim which comes from Angela Lee of the Bicycle Helmet Initiative Trust that: "[...] evidence from Australia indicating that cycling levels decreased following the introduction of legislation [...] was found to be outdated and contained distortions from variables including a reduction in the legal age of driving that meant more teenagers travelled in motor vehicles." This is simply wrong. The legal age for a learner driver's permit (not a licence) was changed by one year in one state. This could not possibly account for more than a small fraction of the observed drop in cycling levels nationally among this age group, let alone made any significant impact on the overall numbers cycling, and in fact there is no reliable evidence that it had any effect at all.

The paper goes on to say: "A study from Ontario, Canada, demonstrated that the introduction of helmet legislation did not reduce numbers of children cycling" - this, as has been pointed out, is because the law was not enforced - and cycling levels were, in any case, poorly and inconsistently measured. Lack of enforcement meant that wearing rates did not show the same step change after passage of the law as they did in New Zealand and Australia, so this evidence, limited as it is, does not, as claimed, contradict that of the far more closely studied Australasian laws - and certainly not in the context of the BMA's apparent preference for an enforced law.

The paper states: "Research published in June 2003 for the Department for Transport revealed a growing trend for wearing cycle helmets in the UK." Up to a point. Wearing rates were declining for children, the group said to be most deserving of compulsion.

The paper does not note the apparent substantial reduction in cyclist injuries in London following the introduction of the congestion charge, against a background of increasing cycle use.

Finally, the references are appallingly narrow. I do not have access to a tiny fraction of BMA's resources but my personal database of helmet and cycle safety research publications amounts to several hundred papers and abstracts, with comments and critiques almost a thousand separate documents.

We are told that 3/4 of cyclists do not wear helmets: this is presented as a problem, and legislation as a solution. We are also told that 3/4 of motorists will not obey speed limits, and this is advanced as a reason why speed limits should not be enforced. Once again the transitive and intransitive meanings of danger are being confused - and I suspect it is deliberate.