Citation

Hagel BE, Macpherson AK, Rivara FP, Pless IB. Br. Med. J. BMJ 2006; 332(7543): 725-726.

Affiliation

Montreal Children's Hospital Research Institute, Montreal, Canada.

Copyright

(Copyright © 2006, BMJ Publishing Group)

DOI

10.1136/bmj.332.7543.725

PMID

16565133

PMCID

PMC1410864

Abstract

Robinson's opposition to helmet laws is contrary to published evidence on the effectiveness of bicycle helmets. At least six independent studies have reported a protective association between wearing bicycle helmets and head injuries. Furthermore, systematic reviews of the relation have all noted a protective effect of helmets. Similarly, six studies have examined the relation between helmet laws and head injuries, and all found a reduction in head injuries after legislation was enacted.

What do the data show?

Robinson suggests that the percentage of bicycle related injuries that are head injuries seems to be declining and that this decline started before the enactment of the law. However, her own figures also show that helmet laws are successful in increasing helmet use and seem to be associated with a decrease in the percentage of head injuries. The effect of helmet use is most evident in her fig 2, where the increase in the percentage of cyclists wearing helmets corresponds with a decrease in the percentage of head injuries. The correlation coefficient for the percentage helmet use and percentage head injury is -0.8 for children and -0.9 for adults. The corresponding r2 of 0.64 for children and 0.81 for adults suggests that much of the variation in the percentage of head injuries is explained by helmet use. Thus, as the proportion of helmeted cyclists increases, the proportion of bicycle related head injuries decreases.

All of her data are based on time series or ecological designs, without any concurrent comparison groups. Such studies are considered to provide weak evidence. With ecological studies, investigators cannot determine whether all cyclists sustaining head injuries were wearing helmets. Confounding variables may also influence both the exposure and outcome variables in the context of a time series or ecological study. For example, a fall in the number of bicyclists in the 1990s may simply reflect an increase in in-line skating or other recreational activities.

Robinson dismisses the evidence from all the case-control studies because of problems adjusting for confounders in observational studies leading to what she believes to be biased, misleading results. The studies in the Cochrane review that Robinson criticises deal with the issue of helmet effectiveness, not evaluation of helmet legislation.

Finally, Robinson invokes the idea of risk compensation, suggesting that wearing helmets may encourage cyclists to take more risks or motorists to take less care when they encounter cyclists. However, she also cites studies showing that, if anything, helmet users take fewer risks than non-helmeted cyclists. However, if risk compensation were operating, the bike helmet studies in the Cochrane review would have greatly underestimated the protective effect of helmets. Furthermore, adjustment for crash circumstances (such as motor vehicle involvement) in Thompson and colleagues's study would remove any risk compensation effect. The notion that driver's take less care when encountering helmeted cyclists is pure speculation.

Health arguments

Robinson's position rests on the assumption that the evidence points to a reduction in cycling after legislation and the adverse fitness consequences that she assumes follow. Although we disagree with the science, let's assume legislation does discourage a large proportion of cyclists. The crucial question is whether the decision not to ride, for however long, truly diminishes physical fitness. In other words, is the trade-off between head protection and fitness worth it?

The answer depends on knowing, in each age group affected, the average distance traveled and the speed at which most trips are made because both time and energy influence aerobic fitness. Morris's classic study of over 9000 British government employees showed that people between the ages of 45 and 64 had to cycle for at least an hour a day or for at least 40 km a week to decrease their risk of coronary heart disease compared with those who were sedentary. It seems unlikely that most leisure cyclists, adults or children, are cycling that far or that frequently.  Without evidence that those who allegedly stopped cycling rode enough to confer a heart health benefit or that they did not take up another healthy activity in its place, Robinson cannot conclude that decreases in cycling are harmful to health.