The federal government is withdrawing its long-standing claim that bicycle helmets prevent 85% of head injuries, in response to a petition filed by WABA under the federal Data Quality Act.
In 1989, a study in Seattle estimated that helmets prevent 85% of head injuries. Efforts to replicate those results during the 1990s confirmed that helmets reduce injuries, but not nearly as much as the Seattle study suggested. Yet public health advocates, government web sites, and the news media have continued to repeat the 85% factoid to the point that it has become a mantra.
Bad information can cause problems, even when it is promoted with the best intentions. If people think that helmets stop almost all head injuries, consumers will not demand better helmets, and legislators may think it makes sense to require everyone to wear one. So we asked two federal agencies to correct the misinformation, and they recently agreed to do so.
How Effective are Bicycle Helmets?
Helmets absorb the shock from a crash. If your head strikes the ground or a vehicle, your brain could be seriously shaken by the sudden deceleration. Helmets should decrease that shaking. The deceleration will be more gradual as your head depresses the foam in the helmet, rather than striking a hard surface. Helmets can also prevent head fractures by spreading the force of the impact, like the difference between being hit on the head by a rock or a beach ball with the same weight.
That’s the theory. But how often do helmets actually prevent head injuries? It’s hard to tell. Experiments on people are unethical. So researchers instead collect hospital data on people involved in bicycle crashes.
In 1989, a team of researchers from Seattle collected data about cyclists who went to area hospitals after a crash. The team was led by Robert S. Thompson, MD, who directed preventive care for the Group Health Cooperative of Puget Sound. Only 7% of the cyclists with head injuries wore helmets, but 24% of those without head injuries did wear helmets. Based on a statistical analysis they estimated that helmets had reduced the risk of a head injury by 85%. The study was published in the New England Journal of Medicine.
Doctor Thompson’s study was a “case-control study.” This type of study originally showed the link between smoking and cancer. “Case-control” is a misnomer because there is no true control group. Epidemiologists often say that case-control studies are a good way to show whether something has a good (or bad) effect on health, but not to accurately quantify that effect.
So the fundamental contribution of the Thompson study was to demonstrate that helmets do reduce the risk of head injuries. But public health advocates recognized that the 85% estimate was a good factoid for risk communication: it means that failing to wear a helmet makes you more than 6 times as likely to experience a head injury. Government web sites and newspapers repeated this factoid, to the point where it has become ubiquitous in discussions about bicycle helmets.
Meanwhile, dozens of researchers sought to replicate the Thompson findings in their own communities. They also found that helmets reduce the risk of head injuries. But they found less of a beneficial effects than Dr. Thompson found in Seattle. Some of the studies also found that helmets increase the risk of neck injuries.
In 2001, a review of all published studies found that helmets reduce the risk of head injuries by 45–71%, and increase the risk of neck injuries by 0–86%. That “meta-analysis” was updated in 2011: Helmets reduce head injuries by 25–55%, but because of the increased risk in neck injuries, the combined reduction in head and neck injuries is only 2–26%.
Yet government web sites, public health advocates, and the news media continue to repeat the 85% estimate.
Misinformation encourages helmet laws, discourages better helmets
Bicycle safety is one of WABA’s central missions, and we have strongly supported bicycle helmets for the last few decades. We require helmets on all rides that we organize. One of our sponsored projects is the Bicycle Helmet Safety Institute (BHSI), which reviews bicycle helmets and encourages improvements in their design. (BHSI raises its own funds, and is not supported by WABA membership dues.)
In the 1990s, we supported proposals to require children under the age of 16 to wear bicycle helmets, which eventually became law.
Yet we draw the line when it comes to laws that require adults to wear helmets. Several researchers have demonstrated that such laws do little to promote safety; but they discourage bike share and other uses of bicycles for short trips. So this year we fought hard against a bill in the Maryland General Assembly that would have required all adults to wear bicycle helmets on any trip, no matter how short. Fortunately, objections from cyclists persuaded the sponsor of the bill not to push it forward–at least this year.
Thanks to occasional articles in the Washcycle, local cycling advocates have known for years that public health advocates overstate the effectiveness of helmets. But with all the ways by which drivers and cyclists misunderstand each other while navigating the roads, helmet effectiveness has not ranked high in our list of misconceptions to fix.
That changed this year. The Maryland Department of Transportation supported the mandatory helmet bill, based on the web site of the National Highway Traffic Safety Administration (NHTSA), which says that helmets prevent 85% of head injuries. An article in the Washington Post questioned why cyclists opposed the mandatory helmet bill, and stated that helmets prevent 80% of head injuries, according to the federal Centers for Disease Control (CDC). The mandatory helmet law was promoted by people who were relying on incorrect information on federal agency web sites.
As we prepared our testimony on the bill, we realized that most helmet research has been focused on making helmets cool, rather than more protective. Better ventilation and more fashionable designs might encourage more people to buy and wear helmets, but it does not make someone safer. Could that be because everyone is assuming that helmets are already 85% effective?
If people thought that helmets are less than 50% effective, might there be a greater focus on what really matters—a better helmet?
WABA pushed agencies to correct the misinformation
Last February, I sent emails to both CDC and NHTSA, pointing out that the 85% estimate is incorrect, and providing citations to newer research. A few weeks later, CDC thanked me for pointing out the new research. I spoke with an epidemiologist over the phone, who told me that CDC would remove the error. She confirmed the conversation in a letter.
NHTSA staff told me that they were too busy to discuss the matter. That led us to conclude that a more formal request would be necessary: The Data Quality Act requires information on federal web sites to be accurate and supported by appropriate research. So I asked NHTSA to provide the underlying documentation. NHTSA confirmed that the 85% figure was based on the Thompson study.
On March 15, we sent our formal “request for correction” asking NHTSA to either remove the statement that helmets are 85% effective, or revise the quantitative estimate so that it accurately reflects the published literature. 
(Jim Titus is on WABA’s Board of Directors and a resident of Prince George’s County)
 We also asked NHTSA to support its claim that helmets are “the single most effective to prevent head injury resulting from a bicycle crash”, but it was unable to do so.
 Our petition also asked NHTSA to “delete all statements … asserting that wearing a helmet is the single most effective way (or device) to prevent a head injury, unless this claim has been substantiated by a peer-reviewed study showing that helmets are more effective than other ways or devices for preventing head injuries.”
 NHTSA did not, however, agree to our request that the agency either substantiate or remove the claim that “wearing a helmet is the single most effective way (or device) to prevent a head injury.” NHTSA said that WABA had not met its burden of proof. Evidently, WABA and NHTSA disagree on whether NHTSA is required to provide at least one study showing its statement to be correct, before WABA would be required to show the statement to be wrong. We are thinking about whether to appeal.
One thing this reiterates is that we see in studies what we want to see, rather than the information that's really there.
As one who (there was no option, really) cycled from childhood without a helmet, I was always skeptical of the published results. For one thing - the general structure of helmets suggested, to me, that while they would be useful in reducing or eliminating concussion, in the case of minor crashes, they would be of no use whatsoever, in the case of a major collision (cyclist into side or rear of van, or thrown by high-speed impact).
However, in order to encourage my daughter, at age 8, to use a helmet without having a major family fight every time, I took to using a helmet myself almost thirty years ago, just as the current designs were becoming common and inexpensive.
What I found, in my own habits were two things: a) that a helmet (despite the voiced "knowledge" of the nay-sayers) did not change my riding habits or style; and b) that within a matter of a few months, I felt uncomfortably naked without that helmet - even though I was aware that its protection was only partial.
The mandatory helmet lobby has - at least IMO - done cycling no benefit, as a means to promoting a healthy life - which means we use the bike as simple transport, rather than as a dedicated "recreational vehicle". What would be far more effective - either than mandatory helmet use (and the demonization of those who don't), or any other measure designed to make cycling "safe", would be what I got, sixty years ago - education, lessons in DRIVING the bike, as a real part of traffic.
The biggest problem is that far too few riders know how to work in traffic, with it, and persuade the motorists around them to let them co-exist. Simply put - any child being introduced to the bike needs lessons at least as comprehensive as those she will get, ten to fifteen years later, when she goes to get a Motor Vehicle Operator's Permit. The cyclist who learns how to live with, and co-operate with motor traffic, when he is the prime target of the risk will, as a motorist, be more aware of the rest of traffic - including other cyclists - and will find (as I did) that learning to drive a car is easy - if you already know how to drive, in the first place.
Thanks for your comments. The request for correction to NHTSA addressed this issue a bit on page 3, where we said
<blockquote>The appearance of bias need not imply that NHTSA was intentionally biased. NHTSA may have simply repeated claims made by seemingly authoritative public health8 and safety advocates.9 The public health community adopted a similarly selective reading of the scientific literature, in an effort to promote wider use of helmets and enactment of mandatory helmet laws.10 Nevertheless, the imperatives for government agencies and public health advocates may be different: Public health advocates often need a simple, clear statement to get people to pay attention. Caveats can leave people with a mixed message, while creative exaggeration is often inherent to good marketing. Government information, by contrast, must be accurate and unbiased, even if the true state of knowledge is a mixed message or difficult to explain. (see original for footnotes). </blockquote>
Aside from that paragraph which was part of a WABA statement, I do not speak for WABA. Randy speaks for BHSI and Shane speaks for WABA on these types of issues, so think of me as just another blogger willing to opine.The text and comment thread touched on some of those issues on the version of this article appearing at http://greatergreaterwashington.org/post/19036/feds-will-stop-hyping-effectiveness-of-bike-helmets/
Our story this week is about federal agencies violating the Data Quality Act, which requires them to disseminate accurate and unbiased information that can be documented. Private citizens, by contrast, can opine all they want. We would not expect Doctors Thompson and Rivara to necessarily decide that Elvik's estimate is better than their own, and each may have close friends and colleagues who will remain loyal to their results regardless of decisions made by impartial tribunals--at least until a form of clearly superior research is undertaken.
BHSI is essentially an independent project which is under WABA's umbrella, but WABA does not provide it with funding. WABA provides an umbrella for good projects doing important things to promote cycling safety, but there is no rigid requirement requiring everybody to agree on everything. Imagine, for example, a professor at a University who is doing great things but happens to have a different point of view on one very important and polarizing issue. If that professor happens to be "wrong", then perhaps her colleagues will eventually persuade her that she is wrong. But we certainly would not want University administrators telling her to change her conclusions. There would, of course, be disclaimers to make sure that no one thinks that the professor's view represents the University.
Thanks very much for the reply, Randy. no idea why the font is so big...!
The quote is from this page:
However, I'm still curious as to you the BHSI persists with the 66-88 figure, when the article above clearly points out that, at best, the range should be much lower. Indeed some more recent research (eg Elvik) puts the head injury reduction closer to 15-25%. It looks a lot like cherry-picking a result to suit an agenda.
@invisiblehand -the reason for bringing it up is that BHSI receives funding from WADA, and is affiliated with it. Seemed strange that WADA is asking the government to correct their statements regarding the efficacy of bicycle helmets, but not asking the BHSI to do the same.
Thanks for finding that page. I fixed it. It was overdue for revision anyway.
WABA does not fund BHSI, consumers do. BHSI contributes 5% of its income to WABA.
The definitive stats on effectiveness are here:
Thanks for the response, and for fixing up that page. I guess it slipped through.
I'm interested in the cochraine papers though - they do get cited a lot in this context as if they are some kind of bible. Why do you feel that a paper published 7 years ago, which itself it a meta-study of papers published up to 24 years go is 'definitive'? There have been many other, newer studies since then, published in reputable and prestigious journals by distinguished academics. This is how science works - it progresses, and our understanding of issues tends to improve. There's nothing special about the cochraine review that makes its findings somehow impervious to the march of science - indeed i could point to many cochraine papers that have since been superseded.
If you are going to state a range for effectiveness, shouldn't you publish the actual range in the literature (which would look something like 15% - 88%)? If you are going to ignore some of the science then you are misleading people. If you have good reason to doubt that all of the myriad papers that have suggested figures lower than 63% are wrong, then it would be more honest to state that, along with your objections to the methodology / findings in those cases?
Just a thought.
Just a small point, and please correct me if I am wrong, but Thompson et al. published their Cochrane review in 1999. It's unclear to me whether these later dates that one sometimes sees are more than the online equivalent to a textbook publishes like Wiley putting out a second edition, where differences may be little more than correcting minor errors in the first edition.
The <a href="http://bike.risingsea.net/docs/Legislation/helmet/NHTSA-request-for-correction-from-Titus-WABA.pdf#page=4">petition at Table 1 </a>provides some elaboration on the question, with Attewell (2001) and Thompson (1999) covering the same time span. The Cochrane review had a smaller number of studies due to more restrictive criteria, so it got different results. But note that one study that met the criteria fir the Cochrane review was left out of the calculations just because it only reported a standard odds ratio (the McDermott study). It appears that including the McDermott results in the Thompson Cochrane Review would have provided results fairly close to the Attewell study done at about the same time.
So why does the BHSI still feature the 'prevent up to 88 per cent of cyclists' brain injuries' line?
@chillikebabNot found on any current BHSI page. We have been using a range: 63 to 88 percent, from the 1999 study, and will be revising that to this statement provided by Diane Thompson, one of the authors:
Helmets provide a 66 to 88% reduction in the risk of head, brain and severe brain injury for all ages of bicyclists. Helmets provide equal levels of protection for crashes involving motor vehicles (69%) and crashes from all other causes (68%). Injuries to the upper and mid facial areas are reduced 65%.
Isn't that a question for BHSI?
Well done! This article from the June issue of Bicycling also has some great information on the safety issues surrounding helmets http://www.bicycling.com/sites/default/files/uploads/BI-June-13-Helmet.pdf