Cook A, Sheikh A. Injury Prevention 2003;9:266-267
This paper uses statistics from national surveys of cycle helmet use and head injury data from English hospital admissions to formulate a relationship between the two datasets from 1st April 1995 to 31st March 2001. Pedestrian head injury admissions are used as a control to monitor trends in hospital admissions. The authors found that cyclist head injury admissions declined over the period more than those for pedestrians, and they consider the most plausible explanation for this is the increase in helmet wearing by cyclists. They go on to estimate that helmets were successful in preventing 60% of serious head injuries, a figure consistent with that obtained from case control studies.
The following peer feedback was published only on-line:
The Transport Research Laboratory reported that helmet wearing amongst cyclists rose from 16.0% in 1994 (Taylor and Halliday, 1996) to 17.6% in 1996 (Bryan-Brown and Taylor, 1997) to 21.8% in 1999 (Bryan-Brown and Christie, 2001), and the authors used these statistics against which to compare trends in head injury admissions. However, these figures for helmet use were averages based on observations on main roads.
For children, helmet use declined from 17.6% in 1994 to 15.0% in 1999 (Bryan-Brown and Christie, 2001), yet Cook & Sheikh mistakenly attribute the reduction of 8.32% in head injuries among children during this period to an increase in helmet wearing. In reality, head injuries fell as helmet use also fell. Moreover, in the period of declining helmet use, child cyclists saw more rapidly declining head injuries than child pedestrians.
The paper has been much criticised for not considering other changes in the cycling environment that might have contributed to a fall in head injuries. Although Cook & Sheikh give some acknowledgement to the possibility that traffic calming and engineering measures targeted at cyclists may have benefited cyclists, they make no reference to other trends that might do likewise. Examples include the proportion of school-age children cycling (a particularly vulnerable group and one which declined disproportionately during the 1990s), the volume and speed of motor traffic, and general road safety campaigns (which especially targeted vulnerable road users during the study period). (Robinson, 2004)
Caution always needs to be exercised in extrapolating from small, gradual changes in helmet wearing as the effect cannot be distinguished from the other gradual changes in behaviour that occur naturally over time. Much more weight should be given to what happens when helmet wearing changes dramatically, as is the case upon the imposition of mandatory helmet laws. Yet in these instances the proportion of head injuries has not improved. (Wardlaw, 2004)
The authors' calculation that helmets prevent 60% of head injuries is completely wrong - the direct ratio of 3.6/5.8 (reduction in head injuries relative to pedestrians / helmet wearing increase) is not the appropriate calculation. If the relative cyclist head injury rate went down by 3.6 percentage points, this represents a fall in head injuries of 13% of the original head injury rate of 27.9%. However helmet wearing only increased from 16% to 21.8%, meaning a drop in the proportion of unhelmeted riders of 5.8/(100-16) = 7%. Thus the proportional fall in head injuries was approximately twice the increase in helmet wearing. If the increase in helmet wearing was a major factor in the reduction in head injuries, they saved as many injuries in people who did not wear them as in people who did! Clearly this is not tenable and there are other confounding influences which swamp the effect (if any) of helmet wearing. (Annan, 2004)
The report concludes with the sentence: "As evidence continues to mount, and consistency emerges between different study designs, the position of the sceptics becomes increasingly untenable". Alas, the emotive language here is characteristic of the bias evident throughout the paper.
Annan JD, 2004. Fundamental error in "Trends in serious head injuries". Injury Prevention on-line .
Bryan-Brown K, Christie N, 2001. Cycle helmet wearing in 1999. Transport Research Laboratory Report 487.
Bryan-Brown K, Taylor S, 1997. Cycle helmet wearing in 1996. Transport Research Laboratory Report 286.
Robinson DL, 2004. Reasons for trends in cyclist injury data. Injury Prevention 2004;10:126-127.
Taylor S, Halliday M, 1996. Cycle helmet wearing in Great Britain. Transport Research Laboratory R156.
Wardlaw MJ, 2004. Policy must be evidence based to succeed. Injury Prevention on-line .