Should snowboarders wear helmets?

Helmets protect skiers and snowboarders against head injuries, but do they also protect the neck? Kristina Fister explains how a matched case-control and case crossover study was used to investigate the effect of wearing helmets on head and neck injuries

This month's paper is Hagel BE, Pless IB, Goulet C, Platt RW, Robitaille Y. Effectiveness of helmets in skiers and snowboarders: case-control and case crossover study. BMJ 2005;330:281-3. You can read it by going to studentbmj.com and clicking on the link.

Why do the study?

Wearing a helmet while skiing or snowboarding seems obviously a good thing if you happened to fall. Common sense tells us that. But the authors of this paper thought that it may not be so simple. Previous research has showed that helmets effectively prevent injuries to the head, brain, and face in cyclists, and one small study with serious limitations showed that the same could be true for skiers and snowboarders.

Other research has found that helmets may worsen the injury, especially in children, due to the biomechanical association between the head, neck, and helmet. In the present study, the authors decided to tackle this issue and state this aim in the first sentence of the abstract and the last sentence of the introduction.

How to tackle the question?

After researchers form a specific question, they need to decide which study design is the most appropriate to answer it. Among other things, they need to choose whether to intervene (experiment) or just observe what happens naturally.

In theory, it may be possible to randomly assign some skiers and snowboarders to wearing a helmet and some to skiing without a helmet. Or it could be possible to do a cohort study, in which a large group of skiers would have to be defined at the beginning and monitored for a long time. Researchers could then compare injuries according to categories of as many known, thought of, and measured risk factors, including helmet wearing. But the number of participants, length of the study, and the cost would be immense for those studies to gather enough data to make valid conclusions.

Our researchers chose a case-control design, which is lower in the hierarchy of evidence than both randomised trials and cohort studies, but is appropriate for rare harmful events. This design enables conclusions to be drawn with far fewer participants—the 4377 who skied in the 19 largest ski areas in Quebec, Canada, in one season.

Skiers or snowboarders who fall, injure themselves, and receive medical attention are routinely recorded by the ski patrol. Directors and chief ski patrol members of all participating ski areas agreed to send their report forms to our researchers every two or three weeks in the 2001-2 ski season. The researchers took the basic characteristics of the injured people (age, sex, and type and date of injury) and contact details from the hospital where they were treated. Researchers then posted questionnaires or telephoned potential cases (those whose injuries were to the head, face, or neck) and controls (who were also injured while skiing or snowboarding on the same hills, but did not hurt their head, face, or neck), and asked them about the relevant circumstances of their fall. This order of investigation—that is, choosing participants on the basis of outcome and going back in time to collect possibly relevant data, makes this a retrospective study.

For participants whose injuries were only to the head, the researchers decided to also use a case crossover design. This is similar to the case-control, but the chosen cases serve as their own controls. This is done by comparing the circumstances of their skiing on the day of the injury to those of the previous day on the slope.

Can we trust the results?

When assessing the credibility of epidemiological reports, three sources of error should be kept in mind: bias, chance, and confounding. Interpretation of findings in view of these errors is largely subjective. For example, you may not agree with your peer to what extent a method of data collection (for example, postal compared with telephone questionnaire) matters for the validity of the data. You could, for example, argue that people will remember more details about the circumstances of their injury (that maybe happened more than a month ago) if they have the time to think and are not rushed by the investigator on the telephone. On the other hand, your colleague may believe that people are generally not going to take much time to fill out the questionnaire anyway, and in fact it may be that an investigator on the phone prompts patients to think about the circumstances of their fall more carefully. Neither of these scenarios is wrong and they probably both affect the results. More information about bias, confounding, and chance is on studentbmj.com.

CHRISTOF SONDEREGGER/SWISS TOURIST BOARD
“Hey dude, where's my gravity?”

What are the results?

The simplest way of summarising the results of this study is that wearing a helmet may reduce risk of head injury by 29%. The true proportion may be greater, as in cycling, if helmets are worn incorrectly, are in poor condition, or are not designed for skiing or snowboarding. Wearing a helmet may increase the risk of neck injuries, however. How did the authors come to these simple but powerful conclusions?

The adjusted odds ratio for helmet use in participants with any head injury was 0.71 (95% confidence interval 0.55 to 0.92). This means that people with head injury (a subgroup of cases) wore helmets less often than those without (controls). The authors also say that this indicates a 29% reduction in the risk of head injury.

Similarly, wearing a helmet reduced the risk of severe head injury (participants who required evacuation by ambulance for head injuries) by 56% (95% confidence interval 0.24 to 0.81). The case crossover design gave a similar result, an odds ratio of 0.43, but with a wider confidence interval (0.09 to 1.83). Here, the confidence interval including 1 means that from the case crossover design alone, we would not have been able to say whether cases with head injuries wore the helmets more or less often on the day of the injury than on the previous day, when they were not injured. This is not a welcome finding for authors who are looking to answer a research question, but our researchers would have been happy to see the practically identical adjusted odds ratio in both designs.

Results for neck injuries are much harder to interpret. The adjusted odds ratio for helmet use for participants with any neck injury was 0.62 (0.33 to 1.19), and for participants who required evacuation by ambulance for neck injuries it was 1.29 (0.41 to 4.04). This means that we cannot be sure whether wearing a helmet was protective or harmful for people who acquired injuries to the neck. A sensitivity analysis, however, indicated that wearing a helmet increases the risk of neck

What are the implications?

Although the researchers' initial rationale for doing the study was to shed light on the protective or possibly harmful effects of helmets to neck injuries in skiers and snowboarders, they did not get conclusive results to answer this question. They did contribute to the existing evidence, however, and their results strengthen two arguments: helmets are effective in protecting from injuries to the head and face, but they might be associated with a higher risk of neck injuries. You can read more about the implications of these findings in the Rapid Responses (http://bmj.bmjjournals.com/cgi/eletters/330/7486/281). But for now skiers and snowboarders are left with the question, “Do you value your head more than your neck?”¹

More information about bias, confounding, and chance

Kristina Fister
Email: kfister@bmj.com

Roger Robinson, editorial registrar, BMJ


studentBMJ 2005;13:89-132 March ISSN 0966-6494

1. Abbasi K. Questions and answers [Editor's choice];. BMJ 2005;330. (5 January.)