Author's replyBMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6937.1164b (Published 30 April 1994) Cite this as: BMJ 1994;308:1164
- Caroline H C Acton
- Director, Brisbane bicycle study, University of Queensland, Royal Children's Hospital, Herston, Queensland 4029, Australia.
EDITOR,--I and my colleagues generally agree with Patricia Priest's criticism and comments. Our discussion particularly aimed to emphasise the potential inadequacies of the case-control design, and we were careful to suggest that results from our study cannot stand alone and that a cause and effect relation might be suggested only by large population based cohort studies. By high compliance with helmet wearing we meant levels similar to those quoted in our results, with about half the population wearing helmets, to ensure adequate person time of experience in which to consider events of head injury. We agree that in the case of extremely high compliance such a study would not be ideal for the reason cited.
Priest rightly corrects our interpretation of the direction of the change in risk. Given the magnitude of the protective effect that we estimated, however, such a bias alone is unlikely to be large enough to nullify the noted association.
There was much internal debate about our choice of a suitable control group for this study. Certainly, a further restriction to include only non-head injuries to cyclists of a severity requiring admission to hospital would control for any bias in risk taking behaviour, although perhaps at the expense of statistical power, generalisability of the study, and other logistic considerations. There was little evidence in our study that helmets cause injuries to other parts of the body. Though we did not specifically address questions about vision and headturning, we asked for details of the cause of the accident and injury, which elicited no reports of poorly designed or poorly fitting helmets as a cause. Consequently, as Priest concurs, even though we chose disease controls, it is most plausible that the noted effect is a protective effect of helmet wearing against upper head injury. Thompson et al used similar controls and showed an even stronger effect.1
We do not agree that the trends of injury suggest that helmets may increase the rate of bicycle related injury overall, but we allow that no effect on the overall rate is possible. As Priest suggests, it is more likely that, while the overall rates do not change, severe injuries are reduced by helmet wearing.
Our study has produced results that are similar to those of studies in other populations. While we acknowledge that there are limitations to the case-control design, the weight of evidence consistently suggests that helmet wearing substantially reduces the risk of head injury in children.
Log in using your username and password
Log in through your institution
Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
Sign up for a free trial