METHODS AND RESULTS

The Hospital Episode Statistics database records all admissions to NHS hospitals in England. It records planned “elective” admissions and unplanned “emergency” admissions. Presentation at accident and emergency departments is not included, although for more serious injuries presentation may proceed to admission. An injured cyclist would usually present first at the accident and emergency department of a hospital and then be admitted if necessary. Deaths occurring before arrival at hospital would not be recorded. All cyclists and pedestrians injured in transport accidents are included, whether or not a motor vehicle was involved.

We extracted records of cyclists (cause code: V10–V14 or V17–V19) and pedestrians (cause code: V0) admitted as emergencies between 1 April 1995 and 31 March 2001. Only first episodes with completed discharge information were retained. Head and brain injury, hereafter referred to as head injury, was defined as fracture of vault or base of skull (S020 or S021), intracranial injury (S06), or other and unspecified injuries of head (S09). This grouping includes some minor conditions such as concussion, but these would need to be sufficiently severe to warrant hospital admission. International Classification of Diseases, 10th revision (ICD-10) codes were selected to match as closely as possible with the earlier study of cyclists using ICD-9 (injury codes 800, 801, 850–854 and cause codes E8261, E810–E825 fourth digit = 6). All other injuries were classified as “other”.

Data were grouped by month and head injuries were expressed as the percentage of admissions. Linear regression modelling was used to assess head injury trends over time. Overall changes among cyclists and pedestrians were modelled; cyclist data were also analysed by age (adult ≥16 years).

During the six year study period there were a total of 69 591 emergency cyclist admissions (0.2% of total emergency admissions). Of these, 16 436 (23.6%) were admitted for head injury. The percentage of cyclist head injuries decreased from 27.9% (n = 3070) in 1995/96 to 20.4% (n = 2154) in 2000/01, an estimated change of –8.49% (95% confidence interval (CI) –6.75 to –10.2) (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Linear regression of head injury as a percentage of total admissions, cyclists and pedestrians, England 1995–2000.

Significant decreases were seen among adult and child cyclists. The percentage of head injuries among adults in 1995/96 was 24.7% (n = 1129), and estimated to change by –8.09% (95% CI –6.18 to –10.01). Among children, the percentage of head injuries was 29.6% (n = 1625) in 1995/96, and estimated to change by –8.32% (95% CI –5.35 to –11.28).

A total of 53 207 emergency pedestrian admissions occurred in the six years, of which 13 193 (24.8%) were due to head injury. Pedestrian head injuries declined significantly from 26.9% (n = 2256) in 1995/96 to 22.8% (n = 1792) in 2000/01, an estimated change of –4.94% (95% CI –3.79 to –6.10) (fig 1). The decline was similar among both adults and children, from 24.7% to 21% among adults and 33.2% to 29.2% among children. Modelling pedestrians and cyclists together, there was a greater reduction in cyclist head injuries (p=0.03).

DISCUSSION

This analysis, which was confined to the period in which ICD-10 was operational, shows that cycling related head injuries fell in children and adults during a period of increased helmet usage. During the same time period pedestrian head injury admissions also fell, suggesting a changing threshold for head injury admissions that might explain the reduction in cyclist serious head injury admissions. However, the reduction among cyclists was significantly greater than that among pedestrians.

We consider helmets to be the most plausible explanation for the extra reduction in cyclists, but other changes in the cycling environment may also have had an effect. Over recent years more cycle lanes and traffic calming measures have been introduced but, while these may reduce total cycling injuries, it is less likely that they would disproportionately reduce head injury. Changes might also have occurred in hospital admission policies but these would be expected to affect both cyclist and pedestrian groups equally. Pedestrians were chosen as a comparator group because changing patterns of transport use may have affected other groups, for example the average speed of motorcycle collisions has probably reduced as a result of rapid growth in the use of small motorcycles.

Assuming that the difference between cyclists and pedestrians (3.6%) is attributable to helmet wearing, and with helmet wearing having increased by 5.8%, we estimate that helmets prevent 60% (3.6/5.8) of serious head injuries. An estimated efficacy of 60% is very consistent with that of 63%–88% obtained from case-control studies of cycle helmet efficacy. As evidence continues to mount, and consistency emerges between different study designs, the position of the sceptics becomes increasingly untenable.