Studies of cyclist fatalities contradict the claims of Cummings et al.
Cummings et al. assume that bike helmets prevent 65% of deaths. Yet a study of cyclist crashes in Brisbane concluded that helmets would prevent very few fatalities. All deaths were caused by bike/motor vehicle collisions. For 13 of the 14 non-helmeted cyclists who died, there was no indication that a helmet would have made any difference. The authors were very concerned about brain damage from rotational injuries and recommended developing a test to measure sliding impact friction of helmets.
Cyclist deaths were also investigated in Auckland. 16 of 19 non-helmeted cyclists died from multiple injuries, so helmets would not have changed the outcome. Only one cyclist died of head injuries in a bike-only crash, the most likely situation for a helmet to help. That cyclist died despite wearing a helmet. The authors concluded: "This study indicates that the compulsory wearing of suitable safety helmets by cyclists is unlikely to lead to a great reduction in fatal injuries, despite their enthusiastic advocacy."
In the three years after helmets were made compulsory in New South Wales, Australia, 80% of fatally injured cyclists wore helmets, an almost identical proportion to population wearing rates (75% of children, 84% of adults), again suggesting that helmets are ineffective at preventing fatalities.
Comparing Australia-wide fatalities in 1988 (before any helmet law) with 1994 (when all states had enforced laws and about 80% helmet wearing); cyclist, pedestrian and all road user deaths fell by 35%, 36% and 38% respectively; head-injury deaths fell by 30%, 38% and 42%. Thus the reductions for cyclists were less than for other road users. Factoring in the reduction in cycling, cyclists were probably at greater risk with compulsory helmet laws than without them.
In inner London, 58% of cyclist fatalities were caused by collisions with heavy goods vehicles, as were 30% of those in outer London. The idea that a polystyrene helmet could be of significant benefit in such circumstances borders on the absurd. The well-known tragic case of 4 helmeted cyclists killed by a car travelling at 50 miles/hr demonstrates that cyclists often die in impacts too severe for a helmet to help.
Riley Geary explained that helmet-wearing status in the FARS database grossly underestimates the true value – many state agencies do not have a check-box for helmet use on their forms and unknowns seem to have been incorrectly recorded as non-wearers. The claims of Cummings et al., based on incorrect helmet wearing rates, and an assumption of the ability of helmets to prevent mortality that bears no relationship whatsoever to information from fatality data, might be dismissed as "enthusiastic advocacy", a classic case of GIGO (garbage in, garbage out).
But in a world of limited resources, there is a sinister side to unrealistic and exaggerated claims – they divert funding away from measures that really could save lives. The only cycling fatality of which I have personal knowledge happened where an off-road cycleway intersects a minor road. It was a difficult crossing; cyclists had to ride carefully through a line of parked cars. By the time cross-traffic was visible, cyclists were almost in front of it, a problem that had been drawn to the attention of the local council. Despite his helmet, a teenager died of head injury after an emergency operation failed to stop the swelling in his skull.
The evidence cited above indicates that forcing cyclists to wear helmets saves very few lives. Other measures, such as guidelines to prevent car parking in places where it obscures sightlines, exploring ways of reducing the disproportionate numbers of circulating cyclists hit by motorists entering roundabouts, random breath testing, speed cameras, and fixing up accident blackspots could save many more.
Perhaps the authors of this article would like to estimate how many more lives might be saved if the considerable efforts currently spent exhorting cyclists to wear helmets were instead spent on making the roads safer for cyclists?
1. Cummings P, Rivara FP, Olson CM, Smith KM. Changes in traffic crash mortality rates attributed to use of alcohol, or lack of a seat belt, air bag, motorcycle helmet, or bicycle helmet, United States, 1982-2001. Inj Prev 2006;12(3):148-154.
2. Corner JP, Whitney CW, O'Rourke N, Morgan DE. Motorcycle and bicycle protective helmets: requirements resulting from a post crash study and experimental research. Federal Office of Road Safety, Report CR 55., 1987.
3. Sage M, Cairns F, Koelmeyer T, Smeeton W. Fatal injuries to bicycle riders in Auckland. N Z Med J. 1985;98:1073-4.
4. Robinson DL. Head injuries and bicycle helmet laws. Accid Anal Prevent 1996;28:463-475.
5. Curnow WJ. The Cochrane collaboration and bicycle helmets. Acc Anal Prevent 2005;37(3):569-73.
6. Gilbert K, McCarthy M. Deaths of cyclists in London 1985-92: the hazards of road traffic. BMJ 1994;308:1534-1537.
7. BBC News. Four cyclists killed in car crash, http://news.bbc.co.uk/1/hi/wales/north_west/4592412.stm 8 January 2006. (accessed February 2006).
8. Geary R. Faulty FARS Bicycle Helmet Use Data & Implications for Effectiveness. Injury Prevention 2006:Electronic letter, 29 June, http://ip.bmjjournals.com/cgi/eletters/12/3/148.
9. Robinson DL. Accidents at roundabouts in NSW. Road and Transport Research 1998;7:3-12.
10. Robinson DL. No clear evidence from countries that have enforced the wearing of helmets. BMJ 2006;332:722-725.
Faulty FARS bicycle helmet use data & implications for effectiveness
Dear Editor and Authors,
Dr. Geary makes a very important point regarding the validity of the data on the use or non-use of bicycle helmet use abstracted from FARS and recently published by Cummings, et al, in their June, 2006 paper. This issue is one of the most important limitations and challenges in the use of narrative analysis. We previously struggled with a similar issue in a study published in Injury Prevention with regards to PPE use among welders who had experienced a work-related eye injury [Lombardi et al., 2005]. It is important to reiterate one of our stated limitations since it appears relevant to the current discussion.
"The narrative analysis method is also limited by the completeness and consistency of the available text data [Lincoln et al., 2004]. Additionally, sensitivity is likely to be better than specificity—that is, when keywords are found in the narrative they probably indicate real contributions to the incident/injury. It is unknown whether there are words that were truncated, forgotten, lost in conversation, or abbreviated by those reporting or recording the claim. Thus, narrative analysis likely underestimates the magnitude of these contributing factors/circumstances to eye injuries. (insert helmets)"
As always, however, I find the innovative use of the FARS data by Cummings, Rivara et al. to provide many excellent examples on the utility of surveillance and administrative data systems for risk factor identification and injury prevention.
P Cummings, F P Rivara, C M Olson, and K M Smith Changes in traffic crash mortality rates attributed to use of alcohol, or lack of a seat belt, air bag, motorcycle helmet, or bicycle helmet, United States, 1982–2001 Inj Prev 2006; 12: 148-154
Lombardi DA, Pannala R, Sorock GS, et al. Welding related occupational eye injuries: a narrative analysis. Inj Prev 2005 Jun;11(3):174-9.
Lincoln AE, Sorock GS, Courtney TK, et al. Using narrative text and coded data to develop hazard scenarios for occupational injury interventions. Inj Prev 2004;10:249–54.
Faulty FARS bicycle helmet use data & implications for effectiveness
Cummings, et al, in their June, 2006 paper, "Changes in traffic crash mortality rates attributed to use of alcohol, or lack of a seat belt, air bag, motorcycle helmet, or bicycle helmet, United States, 1982-2001"  apparently assume that the data on bicycle helmet use among fatally injured bicyclists contained within the Fatality Analysis Reporting System (FARS) database is at least as valid as that for motorcycle helmet use and seat belt use. However, even a cursory examination of the data indicates FARS was underestimating actual helmet use among fatally injured bicyclists by up to an order of magnitude or more during the period 1994- 98, when FARS first began recording such data; and though the situation has improved considerably since then, FARS continues to underestimate overall bicycle helmet use in the US by a factor of two or more as of 2004 (the most recent data available) .
Like most other data elements reported under FARS, data on helmet use is derived from individual Police Accident Reports (PARs) collected by the various state agencies that deal with traffic crashes. Since data on bicycle helmet use is not considered a particularly high priority in most jurisdictions, relatively few state PAR forms have the type of simple check-off box commonly associated with seat belts or motorcycle helmets (e.g. "used", "not used", or "unknown") that allow for easy transcription into the FARS database. In these cases (i.e. the vast majority involving bicyclists), any information on bicycle helmet use must be obtained from the narrative of the crash prepared by the attending police officer; and if no definite mention is made as to whether a bicycle helmet was used or not (which is still all too common given the relatively low priority in determining actual helmet use among involved bicyclists), such cases should be recorded as "unknown" according to the FARS coding protocols. Unfortunately, it appears that nearly all of these cases that should have been coded as "unknown" (including a considerable number where the bicyclist actually was using a helmet, but such usage was either never noted or overlooked in the narrative) were instead coded as "not used"—particularly in the initial period of 1994-98.
One strong indicator that the FARS bicycle helmet use data should not be fully trusted is the fact that the "unknowns" are so few in number in the first place. It is simply not credible that a low priority data element such as bicycle helmet use would have a precision associated with it that is a factor of 20 better than that seen for much higher priority data elements such as seat belt or motorcycle helmet use (0.5% "unknowns" vs. 11% or 10%). Even more persuasive is a direct comparison of FARS data with equivalent state data. Though very few states make any real effort to determine bicycle helmet use in their annual traffic crash summary reports, two that have done so for at least a decade, California and Florida, together account for ~30% of all US bicycle fatalities.
California data from the StateWide Integrated Traffic Records System (SWITRS) indicates that 13.2% of fatally injured bicyclists were using a helmet during the period 1994-98 , but only 3.4% supposedly were doing so according to FARS . Likewise, Florida data from the Department of Highway Safety and Motor Vehicles (DHSMV) indicates that 6.5% of their fatally injured bicyclists were using a helmet during the same period , but only 0.2% (i.e. just 1 out of nearly 600) according to FARS . And while the reliability of bicycle helmet use data in FARS has clearly improved in recent years (15.0% vs. 17.6% CA SWITRS data for the period 2001-03, and 5.5% vs. 6.8% FL DHSMV data for the period 1999-2004), it is clear that FARS continues to undercount such use in too many cases. Indeed, while overall bicycle helmet use in FARS has reached 11% for the period 2001-04, a number of states within FARS now routinely record bicycle helmet use rates in excess of 20% (CO, GA, HI, ID, NV, OK, TN, WA), and a few actually record use rates in excess of 40% (MA, NE, VT, WY) .
It should be obvious by this point that the overall 2% bicycle helmet use figure FARS indicated in 1994 (and as recently as 1998) has no basis in reality, and the assumption by the authors that helmet use among fatally injured bicyclists was essentially nil before 1994 is fundamentally flawed. Since SWITRS bicycle helmet use data extends back fairly reliably to 1990, it is noteworthy that even in the earlier 1990-93 period, helmet use had already reached 8.4% among fatally injured CA bicyclists, and that perhaps a dozen other states may have had helmet use rates at least similar to or greater than that of CA.
It is also worth pointing out that during this earlier period, SWITRS data indicates that helmet use among non-fatally injured CA bicyclists grew steadily from 6.3% in 1990 to 9.0% in 1993, before jumping to 13.5% in 1994 and 16.7% in 1995 (at least partially in response to the passage of a mandatory helmet law in1994 covering all CA bicyclists under the age of 18), and eventually reached a plateau level of 20-22% from 1999 on. Since the overall helmet use rate averaged just 7.5% among non-fatally injured CA bicyclists during 1990-93, it could actually be argued that bicycle helmets had no beneficial effect at all in preventing fatalities, though later data suggests this more likely was just an artifact of non- fatal helmet use being less reliably recorded during the earlier period.
Over the next 10 years (1994-2003), non-fatal helmet use averaged 18.76%, compared to 15.52% among fatally injured CA bicyclists, which suggests that bicycle helmets have at best only been ~17% effective in preventing fatalities statewide (selective recruitment effects have almost certainly resulted in a positive bias, so it remains quite possible there is no net safety benefit associated with bicycle helmets at the whole population level). Since this result is far lower than the ~65% effectiveness the authors assumed for bicycle helmets based on a single case control study , it seems clear that either the assumed effectiveness of bicycle helmets has been wildly inflated relative to real world data, and/or risk compensation effects have essentially negated any safety benefits bicycle helmets may have to offer in the event of a crash- -by apparently "encouraging" helmeted cyclists to crash more often and/or get into more serious crashes.
 Thompson DC, Rivara FP, Thompson RS. Effectiveness of bicycle helmets in preventing head injuries: a case-control study. JAMA 1996;276:1968–73.[Abstract]
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of Injury Prevention.
View free sample issue >>