No Mystery Here: Cars Piling Up

A Review of J Barnes' NYT Article on Bike Helmet Efficacy (August 2001)

D. A. Clarke

In August of 2001, the New York Times ran an article by Julian Barnes, entitled "A Bicycling Mystery: Head Injuries Piling Up." This article discusses the conflict between claims made for the efficacy of bicycle helmets in preventing injury and fatality, and actual cyclist injury and fatality statistics in the US. Cyclist head injuries, according to Mr Barnes, have actually increased during the period during which helmets became increasingly popular among American cyclists. He discusses several possible explanations for this phenomenon, including what is called "adaptive behaviour," "risk homeostasis," or "risk compensation".

It was a great relief to me, and to many other cycling advocates and road safety or pedestrian/bike rights advocates, to see -- at last -- something like the beginning of real public debate on the effectiveness of the last two decades of "bike helmet" campaigns and laws. Mr Barnes' NYT article was the first appearance I can recall (in the last decade or two and in the mainstream press), of any rational discussion of risk compensation -- even though it is both common sense, and common knowledge among insurance adjusters and the like. In my experience this is the first time that any "major daily" has admitted that bike helmets are anything other than a miraculous cure-all. I congratulate the author on breaking the mysterious voluntary silence of the US media on this issue.

All essays on issues of road safety should perhaps start with the disclaimer that "Traffic, like God, Football, and Politics, belongs to that select group of subjects on which everyone, when the spirit seizes him, instinctively feels that he can speak with overriding authority and conviction." (Cohen and Preston 1968, quoted by J.S. Wilde in Target Risk II) In my own defence I can only say that I have been a cyclist for over 30 years, on American roads -- without, I should perhaps add, ever incurring any injury more serious than a minor bruise or scrape. For the last couple of years I have made a particular study of bike safety and traffic fatality, with an emphasis on pedestrian and cyclist road rights. I would like to discuss what my own experience leads me to perceive as some minor weaknesses, problems, or shortcomings in Mr Barnes' article. I sincerely hope that Mr Barnes (and other journalists) will be writing a great deal more in future on this subject, continuing a productive discussion.

I was disappointed that Mr Barnes' discussion opened with

...researchers have found [that helmets] can reduce the severity of brain injuries by as much as 88 percent.
This (in)famous 88 percent figure is almost certainly from Thompson RS, Rivara, and Thompson DC (The New England Journal of Medicine Vol 320:21 pp1361-7, 1989): "A case-control study of the effectiveness of bicycle safety helmets". Mr Barnes may not be aware that this paper has been much criticized ( summarizes some of this critique). The same team, in 1996, revised their claims: they concluded in this later study that helmets decrease risk of head injury by 69%, brain injury by 65% and severe brain injury by 74%.

Even with this revision, TR&T remain probably the most optimistic of helmet advocates. Many other studies have arrived at significantly lower mitigation estimates. Furthermore, there is a startling disparity in the literature between the conclusions presented in very small case studies (usually based on admissions or treatment records from individual medical facilities) and those presented in larger studies based on regional or national statistics.

Here is just one example of a study using a larger sample than either of the well-known TR&T papers: Langley and Scuffham in 1997 performed an analysis of serious head and non-head injuries to New Zealand cyclists, correlated with helmet wearing survey data, 1989 to 1992 (prior to introduction of NZ mandatory helmet law). Helmet wearing rates increased over the period from 46% to 84% for primary schoolchildren, 23% to 62% for secondary schoolchildren, and 21% to 39% for adults. All serious injuries to cyclists decreased substantially for children, but not for adults. Serious head injuries as a percentage of all serious injuries remained constant for all groups, with no apparent difference between bicycle-only and motor vehicle related crashes. However, percentages of mild concussions and lacerations to the scalp decreased more than other cyclist head injuries. The failure to achieve the expected decline in serious head injury could be attributable to a variety of factors, including the incorrect fitting and wearing of the helmet. (Note that cycle use declined over the study period by 19%.)

This is not the only case in which an analysis of large-scale patterns of injury and fatality leads to far more cautious and modest conclusions than one will find in TR&T's published papers and letters. This is an area of significant controversy, and I fear Mr Barnes is misleading his readers by suggesting implicitly that there is any kind of consensus among researchers on the 88 percent mitigation figure. It represents one extreme boundary of expert opinion, not the middle ground.

I recommend John Franklin's page


    which summarizes a reasonable cross section of the literature (and from which I cribbed the preceding cogent summary of Scuffham/Langley 97). One possible starting point for a journey through the variety of online references by "helmet skeptics" and cycling advocates is my own


    Mr Barnes wrote "Children's head injuries declined until the mid-1990's..."

    Many claims have been made that a steady decline in child cyclist fatalities was due to the implementation of child helmet laws in the US. See my paper (with Riley Geary, currently under review at Injury Prevention, preprint at for a brief discussion of such claims, which are often used to "prove" that adults will "also benefit" from mandatory helmet laws.

    In summary, the decline in both walking and cycling among children in automobile-focussed societies (US, Australia, UK, NZ) is sufficient to account for declining child pedestrian and bike fatalities; also, it is notable that pedestrian and bike fatalities tend to decline together, even though helmet use has not been promoted or enforced for child pedestrians. This tends to strengthen the case for diminished exposure as the cause of declining fatalities, rather than mechanical intervention.

    Mr Barnes wrote

    Still, with fewer people riding bicycles, experts are mystified as to why injuries are on the rise. "It's puzzling to me that we can't find the benefit of bike helmets here," said Ronald L. Medford, the assistant executive director of the safety commission's hazard identification office.
    As mentioned above, there are widely varying findings in the literature, and heated controversy on the ground. Mr Medford's remark seems somewhat naive in the context of this long-standing and vehement (if not widely publicized) debate. The US figures described in Mr Barnes' article are consistent with whole-population studies from other countries. This "experts are mystified" phrase in the text has evoked sarcastic commentary on some cyclists' newsgroups and mailing lists. If adult cycling injuries and fatalities have not declined in parallel with the statistics for child cyclists -- and if they are now in fact rising -- this does not strike most contemporary cyclist advocates as "mystifying".

    The disparity in child and adult statistics in the US may be attributable simply to the fact that adults did not giving up cycling as quickly as children did; children have mostly been discouraged or forcibly dissuaded by parental and school authority (for example, there are schools whose administrations actually forbid children under a certain age to ride a bicycle to school). Adults are not subjected to such restrictions.

    Most people know intuitively, without any expert "studies", that children simply do not ride bikes as much as they used to. For those who require expert testimony, CDC has published several press releases on the national "crisis" in diminished child physical activity, restriction of children's freedom of play and movement, etc. I direct the reader's attention also to this excellent presentation (1999) by Dr Mayer Hillman:


    on the negative impacts of automobile dependency on child social and emotional development.

    What seems very obvious to most cyclists who spend many hours each week on America's roads is what we also know from auto industry statistics: the ratio of the US car population to the US human population has been rising steadily over the last several decades. The number of cars on the road has increased significantly; the number of miles driven per year is alarming (Americans have collectively driven over 2 trillion vehicle miles every year since 1988).

    The average American already spends 434 hours (18 full days a year), in a car. Soon the figure will become unsustainable as a vicious circle of population growth, more car registrations, fewer occupants per vehicle, increasing adoption of car travel over other modes of transport, and more car journeys takes over. The report cites the city of Indianapolis as an example. There population increased by only 17 per cent (150,000 persons), between 1982 and 1997, but the number of vehicle miles driven went up by 103 per cent -- creating as much congestion as if the population had increased by a million persons.

    Multiply this scenario by the 32 per cent increase in the United States population that took place between 1970 and 1997; the doubling of car registrations and the 65 per cent increase in licenced drivers, and all this amounts to a predictable deadline for national gridlock. Everybody knows this, just as we all know that we will die -- but we don't believe it. It is the best kept secret in the world.

    -- Martin Pawley, 'The Age of Immobility' (Aug 2000)

    Or in the vernacular: "Traffic is so much worse these days." In addition to the risk per mile for drivers and passengers, there is a risk per mile and a risk per hour driven which is imposed on persons outside the confines of the vehicle. This exposure obviously rises with VMT per annum and with the number of cars (hence, drivers) on the road during any given hour.

    But speed also plays a part: collisions at 20 mph and below are unlikely to result in disabling injury or fatality to more vulnerable road users; collisions at 30 mph and up are very likely to inflict major injury, and the risk of fatality rises steeply with the vehicle's speed. Therefore, risk exposure for pedestrians and cyclists tends to decrease with advanced congestion (gridlock) as the overall speed of motor traffic slows to a crawl; people outside cars then experience less risk. In our cyclist injury and fatality statistics, we may be seeing the initial rise of a bell-type curve whose decay slope is only a few years away -- along with terminal gridlock.

    On faster roads such as rural highways, freeways, interstates, etc. over the last two decades, speed limits have been raised. Speeding has become epidemic even in residential and commercial districts; police departments report themselves "unable" to enforce existing speed laws, mostly because the slope of the increase in the number of cars on the road is far steeper than the slope of expansion in police budgets and staffing. Adding to the problem are general policies such as the infamous highway engineers' "85 percent rule" (successfully challenged this summer in California) which have set speed limits according to driver preferences, not according to general public safety requirements.

    Perhaps the most significant trend is that on every type of road, larger and heavier vehicles such as SUVs and minivans have become overwhelmingly popular, changing the nature of traffic all over the country. Many people, myself included, feel that safety for both cyclists and pedestrians has been reduced as a result. The weight of an SUV such as the notorious Ford Explorer increases the probability of its causing fatality in a collision with a lighter car; why would we not expect it to be also more lethal or injurious in any collision with a cyclist, moped or motorcycle rider, or pedestrian? The size (especially the height) of these popular vehicles obstructs the view of both cyclists and drivers of conventional cars; cyclists are "hidden" behind these hulking vans, SUVs and super-trucks, and are less likely to be seen by cross traffic and turning traffic.

    Mr Barnes wrote: "Even so, bicyclists suffered 73,750 head injuries last year, compared with 66,820 in 1991..."

    I hope he will forgive me for saying bluntly that I find this is a rather misleading statement. Why do I think so? Because of the absence of any context by which the reader could evaluate or understand this number.

    According to statistics published by Johns Hopkins and on the "Caregiver" web site, the number of cyclist head injuries per annum which required hospitalization was about 7,700 in a typical recent year. The total number of US head injuries which required hospitalization in the same year was between 500,000 and 750,000. So cyclists presented, on average, less than 2 percent of head injuries serious enough to require any hospital attention. Furthermore, this JHI figure for "serious" cyclist head injuries is only about 10.5% of the 73.5 thousand injuries to which Mr Barnes refers above. The rest are presumably so trivial as to require no treatment: the equivalent of skinned knees and minor bruises.

    I should note that the same sources indicate that there are from 75,000 to 100,000 head injury deaths in the US per annum. There are only about 500 to 650 cyclist head injury deaths per annum: cyclists on average contribute less than 1 percent of all deaths from head injury. [Here I would like to thank Ken Kifer whose research supplied the figures in the last 2 paragraphs.]

    It seems reasonable to conclude that if we compared cyclist head injuries (all severities) which Mr Barnes states at 73.5K/annum above, with national head injuries (all severities, all causes), cycling would still contribute a small drop in a large bucket.

    By citing the largest possible number (all head injuries no matter how trivial) to describe cycling risk, and then associating this figure with specific, frightening anecdotes of maximum risk (fatality and disabling injury), the article gives an exaggerated impression of the danger of cycling. Without any wider context the reader -- who probably has no idea of the annual rate of minor accidental injury in a population approaching 300M -- is left with a "large 5 digit number" which seems very impressive and makes cycling look quite scary.

    It is also unclear whether the increase in head injuries is general, or correlated with severity. Are trivial head injuries increasing faster than serious ones? If so, the increasing popularity of off-road riding (which is generously productive of scrapes, minor gashes, bruises, etc) may account for an upward trend at the more trivial end of the injury scale. If, on the other hand, it is major injury which we see increasing, then road hazard might be a more likely source to consider.

    Two out of three of Mr Barnes' anecdotes focus, as I mentioned, on maximum risk: fatality and permanent disability, often associated (in all fields of human endeavour, sport, transport, etc) with head injury. But the total number of cycling fatalities per annum is extremely small, and the average cycling injury is very trivial. Furthermore, only a very small percentage of cycling injuries are head injuries.

    In other words, an ambiguity is left unresolved: is it the case that head injuries are now a higher percentage of all cyclist injuries than they were before? Or are cyclists generally involved in more collisions and experiencing more injuries of all kinds than before? Most readers, already predisposed to believe that cycling particularly endangers the brain, will conclude that this special risk has intensified.

    I think we should also bear in mind that very few cyclists manage to die on the road, or even incur life-threatening injury, without the "assistance" of a motor vehicle operator. Rather than focussing obsessively on the small number of cyclists who perish on our roads and describing them as victims of that dangerous vehicle, the bicycle, we might (I think more productively) take a different approach.

    We might remember that about 35,000 people annually are killed while riding in or driving automobiles; about 5,000 people are killed as pedestrians, struck by automobiles; and fewer than 800 people are killed each year as cyclists -- most of them struck by automobiles. (Here we should understand "automobiles" to include trucks, taxis, etc). We might consider whether the automobile -- since it is responsible, as a consumer product or as a technology, for over 40,000 deaths per annum -- should be analyzed and discussed as a public health risk... rather than the relatively benign bicycle.

    The leading cause of head injury in the US, by the way, is being in a motor vehicle during a crash. Runner-up causes are falls (especially on stairs), contact sports, and personal violence; bicycles appear very far down the list. Yet for a quarter century, Americans have been told repeatedly by trusted authorities that cycling presents an overwhelming risk of fatal or disabling head injury against which mechanical precautions must be taken. For perspective:


    When an article such as Mr Barnes' quotes cyclist head injury statistics without placing them in the context of relative severity, in the context of cyclist injuries of all types, or in the context of national injury statistics for all causes -- a misleading impression is given. The apparent risks of cycling are greatly exaggerated, and the real risks of automobile use are minimized or concealed.

    The uninformed reader may well be left, after reading the NYT article, with an amplified feeling that cycling is terribly dangerous, and that it is a particular contributor to head injury. Previously, this same uninformed reader might have believed that cycling was safe only if a helmet was worn; Mr Barnes' article undermines this naive faith in an inadequate technology, but does nothing to counteract the exaggerated fear -- in fact, may serve to stimulate it further. This seems to me very unfortunate. Many people may conclude that cycling is even more dangerous than they had thought, if they are bereft of the miraculous saving power of helmets.

    I think we need to look at cycling in the context of, and on the scale of, other ordinary risk-taking behaviours, rather than continuing to exaggerate its risks by examining them in isolation and, as it were, under a magnifying glass. When discussing risk and public health, we need to look at "risk inflicted" instead of focussing exclusively on "risk run." In this respect I found the article unsatisfying. It adhered to traditionalist definitions of risk and to a traditionally disproportionate focus on the low-risk activity of cycling.

    Mr Barnes wrote

    During the same period, overall bicycle use has declined about 21 percent as participation in in-line skating, skateboarding and other sports has increased...

    This implicitly perpetuates yet another unfortunate misperception of cycling. Most Americans (including police and other civil authorities) think of bicycles as toys, and cycling as strictly a sporting or entertainment activity. However, for a few million US cyclists, the bicycle is a practical form of daily transportation.

    Dismissal of cycling as strictly for entertainment or recreation leads to the neglect of bread-and-butter improvements in road safety, in favour of elaborate recreational pathway systems (for weekend or pleasure riding) which do not serve popular destinations like restaurants, shops, schools, jobs, etc. Thus, a one-sided view of cycling tends to neglect improvements in access for cyclists to essential goods and services, and thus in turn discourages "average people" from using their bikes for everyday, practical transportation. There is a direct connection with road safety here. If people used bikes more for routine errands, they would replace automobile trips which inflict risk on other citizens, with bicycle trips which (by comparison) inflict virtually no risk on others.

    The NYT article does not mention that American children in particular have not merely abandoned cycling as a play activity in favour of other sports like skating and skateboarding; they have, by and large, stopped riding their bikes to school or to after-school activities -- i.e., they have stopped using bikes for basic transportation. Most kids of middle income and upper income families are now chauffeured everywhere by "soccer mom" or a paid babysitter; this, of course, increases the motor vehicle traffic which makes people more afraid of cycling.

    Thus the exaggeration of cycling danger, and the simultaneous redefinition of cycling as a trivial entertainment activity, form a dismal feeback loop in public policy. Overestimation of road danger leads to a demand for expensive segregated "bike facilities" rather than a commitment to making the public roads safer for pedestrians and bikes. Dismissal of cycling as a trivial "play" activity means that authorities, and indeed most of the (driving) public, do not consider it important that such segregated pathway systems should serve any practical destinations. After such facilities are constructed, cyclists who persist in using the public streets are sometimes seen as "wilfully reckless" or "ungrateful;" some motorists even harass cyclists on the road and ask them why they do not "stick to the bike paths," etc.

    The habitual categorization of cycling with "in-line skating and skateboarding" as "sports" rather than with public transit, walking, and the use of mopeds and motorcycles (all alternatives to the private automobile for quotidian transportation needs), does a disservice to American cyclists.

    By expressing (more than once) my disappointment at the article's failure to challenge received opinion about road safety, cycling risk, etc., I obviously imply that that the received opinion -- the traditional approach to discussing and understanding road safety -- is less than optimal. In fact I feel strongly that a new approach is needed. Significant social costs have been incurred by our adherence to what we might call the "car-centric" theory of road safety, of which an overemphasis on bike helmets is just one facet. I divide this problem into two (interrelated) issues: the dangerizing of cycling, and the exaggeration of the protective power of cycle helmets. These issues are discussed in a separate essay.

    In conclusion I would like to say again how encouraged many cycling advocates, experienced cyclists, and road safety activists were by the appearance of the NYT article! Though my reservations and criticisms above may seem extensive, they are minor compared to the great positive potential if this smouldering 20-year debate is finally brought before the general public. Exciting opportunities may arise for a whole new discussion, in which we might finally be able to address the real complexities (psychological, mechanical, and political) of road safety.

    I would like to thank Mr. Barnes very much for writing an article which will no doubt draw fire from all sides. Convinced "helmeteers" will, I expect, be flaming him for daring to cast doubt on the rigid doctrine of helmet efficacy. Indeed, Randy Swart (the force behind the "Bicycle Helmet Safety Institute") has already done so:

    (Swart's response is itself weak on some signal points; see my critique of his rather sweeping dismissal of risk compensation).

    But only by having the courage to challenge received opinion in the pages of a free press, can we maintain a healthy atmosphere for public intellectual life -- or get anywhere near the truth in contentious public issues. I offer these comments in a spirit of good fellowship and free inquiry; and I hope that Mr Barnes and others will find at least some of them interesting, and conducive to further discussion.
    De Clarke
    Santa Cruz, CA