Comments on an Article in the
July 29, 2001
Summary: This 2001 article in the New York Times shows how bad data result in bad conclusions when poorly researched by the media and exploited for shock value.
The July 29, 2001 Sunday New York Times had an article on its front page by Julian E. Barnes on helmets. CNN also ran the item, and it was syndicated to other newspapers. We have been dealing with its myths for years.
The article is headlined "A Bicycling Mystery: Head Injuries Piling Up." Its theme is that head injuries per active cyclist have increased 51 per cent in the US despite increased helmet use and less bicycle use. The tone is that helmets should have prevented this if they were as effective as advocates claim they are. If you are willing to sign up for their website access you can read it
on the Times web page. They will not permit us to post it here.
The article may be designed as what Internet flamers would call "a troll" - - something to stir up reaction and provoke readers to reply. Juicy for a newspaper, but a headache for helmet promoters if readers don't take the time to analyze the content.
For starters, we as helmet advocates would indeed like to see bicycling head injuries go away with helmet use. We are concerned that they have not been lowered more. But the article's tone is that helmets are widely used and have failed to reduce the rate of head injuries, dismissing the other factors involved with a light touch. This is not a balanced article. But then it is not unusual to find that in a newspaper article when you are familiar with the subject matter.
The reporter accepted without question the CPSC contention that half of US bicycle riders wear helmets. He also focused on injuries rather than deaths. Anyone familiar with both sets of statistics should realize that CPSC's numbers are wishful thinking, and that injury stats have always been much less reliable than death counts. A lot less than 50 per cent of US cyclists are using helmets. (North Carolina's actual head count showed 17 per cent state wide.) The Insurance Institute for Highway Safety reports that 98 per cent of the cyclists killed in 1999 were not wearing helmets. (That number is suspect, to say the least, since an analysis of the data by Riley Geary indicates it actually might show that only 2 per cent definitely had helmets on and for the rest we don't really know.) There were 746 bicyclists killed in crashes with motor vehicles in 1999. This is 1 percent fewer than in 1998 and down 26 percent since 1975. (CPSC still cites 900 as the annual toll.) But to judge the effectiveness of helmets based on numbers like these -- negative or positive -- is to ignore a hundred other factors that affect the safety of cyclists on the roads. We have no idea where the 51 per cent increase in head injuries per active cyclist comes from, but ask any long-term cyclist and they will tell you it is not correct from their simple observation. It would be massively evident if it were correct.
Any trend in the rate of head injuries reported should be matched with accurate exposure data on how many miles are being cycled, a key statistic we have never had. The author concludes somehow that the increase in off-road riding is not likely to be adding to the head injury numbers, even though most cyclists who do both know that per mile there are more crashes and more injuries of all kinds in off-road cycling than road cycling. Again, we lack the exposure data we need. The article brushes lightly by the fact that traffic has made US roads more dangerous for cyclists and pedestrians. That is a fact of life for every cyclist who uses roads. Did anyone expect helmets to compensate for that? Statistics are dependent on those who gather them, and the recognition of mild traumatic brain injury has advanced considerably in the last ten years, probably affecting the number of head injuries reported. Other factors that have changed on the roads include the popularity of SUV's, whose height and bulk probably would contribute to more cyclist and pedestrian head injuries.
The article also attempts with examples and quotes from experts to convince the reader that helmeted cyclists take more risks because they think they are protected. While there may be some riders like that, we believe that most riders understand that their arms, legs, faces, torsos and other body parts not covered by a helmet are not protected from anything by the helmet on their heads. They do not compensate for the risk because they do not believe that a helmet removes much of the risk of pain and suffering from a crash. If we were talking about full body armor there might be some additional risk taking, but helmets obviously do not keep the rider's most tender parts intact. That raises a question about the article. The riders quoted had injured necks, legs or knees, and were quoted as blaming their helmets for making them feel safe and by implication causing the injury. We do not know what kind of people were quoted, or in what context they made their remarks, but they just do not ring true to us as a representation of general attitudes among cyclists. We do not believe that bicycle helmets give a false sense of security, any more than seat belts, air bags, motorcycle helmets, smoke detectors, steel-toed boots or child safety seats do. Anti lock brakes on cars are cited as a parallel by one of the experts, even though crashes related to them are generally due to drivers not realizing that they require a very different braking action in a panic stop. From the quote it appears that this expert thinks people drive faster because of their anti lock brakes, and would be driving slower and having fewer crashes if they did not have them. In another example the article describes a rider hit by a pickup truck on a highway. His helmet was knocked off his head by the force of the impact and did not prevent his death. There is a debate going on in the injury prevention community about risk compensation and bicycle helmets, with each side unable to prove its case conclusively with the available data, but insisting that the burden of proof falls on the other side.
Helmet promotion at the current level has not reached the majority of US riders. For that reason alone it will not cause head injuries to go away. And it should be evident that bicycle helmets will not prevent injuries to other body parts, or prevent death when hit by a truck at highway speeds. Those are straw men. But helmets do protect their wearers in an amazing number of crashes. (We have posted some crash stories illustrating that.) They reduce deaths and injuries case by case, no matter how a given set of poorly-defined statistics may appear. We have always supported the full-environment approach to bicycle safety, with helmets as one useful tool, but this article attempts to turn that approach into evidence that helmets are not working. The lack of balance and perspective should be evident here.
In short, this article may be a troll, it plays on the fact that cyclists' head injuries have not disappeared following our helmet promotion efforts so far, and it will be quoted in internet helmet wars for the next decade. It was not an admirable piece of journalism, and the Times should know better. Maybe they do.
To understand how we fit helmets as a tool into the broader context of promoting bicycle safety, please see this page on the National Strategies for Advancing Bicycle Safety. We are members of the Steering Committee that produced the final strategies document.
Check out this page for a look at the CPSC data the NY Times used. The numbers for kids under 15 are not statistically significant, and all head injuries are lumped together, so a cut chin needing stitches counted as a "head injury." There is no way to separate out the injuries that helmets should mitigate.
Here is another comment by John Sabelli, a veteran in the hockey helmet field.
John Allen has published an extensive analysis of the data problems in the CPSC study on his website.
Postscript: In the years following publication of this article, editorial problems at the New York Times became well known. The article should be viewed in that context.