Public Health (1991), 105, 327 334
© The Society of Public Health, 1991
Pedal Cyclists, Crash H e l m e t s and Risk Mark McCarthy, FFCM Department of Community Medicine, University College London, London WC 1E 6EA
As a rate per million kilometres travelled, the 'risk' of cycling appears to be high in relation to other forms of transport. Yet, in absolute numbers, there are far fewer cyclist deaths than pedestrian or motor vehicle occupant deaths, and most deaths and serious injuries to pedal cyclists are caused by other road users principally motor vehicles. The large majority of pedal cyclist deaths are due to head injuries after collision with a motor vehicle. It is therefore commonly proposed that cyclists should wear crash helmets for their own 'safety'. Helmets may protect against fall injuries, but current models are not designed to withstand the impact of collisions with motor vehicles. Evidence for the benefit of pedal cyclists wearing helmets is limited: the existing studies cannot exclude the possibility of different risk-taking behaviour, either by cyclists or by motor vehicle drivers, for helmet wearers compared with non-wearers. A public health policy towards reducing pedal cyclist deaths should seek prevention of accidents, rather than protection from their consequences. Cycling in greater safety would reduce the 'risk' per kilometre travelled, but more cycling might not reduce total cyclist deaths or injuries--because of greater exposure. The 'risk' of cycling--the risk of injury or death is a complex mix of exposure, 'danger' of the environment, and the perceived risk affecting our precautionary preventive behaviour.
Introduction The pedal cycle is an accepted and effective means of transport, used worldwide. It is energy efficient, low in pollution and enjoyable for business transportation and leisure. But, as with any form of transport, cycling has risks. This paper examines cycling risk, and the proposal that cyclists should wear helmets.
Describing cycling risk Cycling risk is usually described as the n u m b e r of individual events (the numerator) divided by the population exposed (the d e n o m i n a t o r ) - - f o r example, cyclist injury rates per 100 million kilometres. Both n u m e r a t o r and d e n o m i n a t o r need to be chosen carefully. Injury occurs in some, but not all, accidents involving a cyclist. R o a d injuries should be reported by law, but studies questioning cyclists retrospectively ~ or using casualty attendances (presumably more serious injuries) 2,3 suggest that there is substantial under-reporting to the police. Reporting will vary according to self-injury or collision, questions o f insurance, and the availability of police to report to. A policeman's decision on the severity of an injury is usually m a d e without medical examination. 4 The 'severe' category ranges from a fractured finger or substantial lacerations to multiple fractures requiring intensive care. The variation in recording, between policemen and for different county police forces, has never been assessed. Thus police injury statistics are epidemiologically unreliable.
Mark McCarthy
328
Road deaths in Britain are reported both by the Ministry of Transport, from statistics provided by the police of deaths within 30 days o f an accident, and by the Office of Population Censuses and Surveys usually following a coroner's inquest and including all deaths associated with an accident up to one year after the accident. The variation between these two sources is not marked and it is logical to use transport deaths data as numerator when using transport data also as denominator. Data including 'killed and seriously injured' together are likely to be much less accurate than death data alone. Data published by the Department o f Transport present the denominator for accidents, injuries or deaths as distance travelled. 4 This implies that distance covered is the prime objective of travel (to arrive rather than to travel). The approach favours long distance journeys, for which clearer data are available--short journeys are usually underestimated in travel surveys. Denominators of time taken in travelling, or per journey, would give different pictures, but such data are not provided. Also, using travel as denominator implies that risk relates only to the traveller, but externalities of travel--for example, lost productive time; energy consumed; pollution--affect the whole society. F r o m a public health viewpoint, however, it is the absolute population number, or rate per million population, that is more important than the rate per traveller.
Transport mortality Comparisons of risk between different modes o f transport need reliable numerators, so that death data are preferable. When transport 'safety' is considered as death rates per distance travelled, motor cycling appears the most 'dangerous', followed by pedal cycles, moving down a hierarchy of 'safety' to air travel, the 'safest' form of transport. These different forms o f transport are not strictly comparable, since it would be difficult to travel the distance of a plane journey by pedal cycle, and vice versa. The 'safety' o f some forms of travel is achieved through rigorous preventive p r o c e d u r e s - - f o r example, in air transport, by pilot medical examinations; agreed aircraft inspection procedures; and pre-takeoff countdown checks. Who is at risk in road transport crashes? Table I presents national data reported by the police on single vehicle crashes involving a pedestrian, and two-vehicle crashes, in which one or more people were killed. While more than 200 cyclists were killed in 1978, cyclists were reported as the 'victim' in almost all accidents. These data can also be considered as victim ratios, the number o f cyclists killed in crashes compared with the death o f one other road user. The cyclist victim ratios in Table I are 0.3 per pedestrian, 1.0 per m o t o r cyclist
Table I
Road deaths by user category, Great Britain 1988 (excluding accidents with three or more, and type not reported, accidents) (Department of Transport 1989, Table 23) (4).
Subject(*) Pedestrian Pedal cycle Motor cycle Motor vehicle
Pedestrian
Object* Pedal cycle
Motor cycle
-6 91 1591
2 1 3 178
5 3 13 359
Motor vehicle 4 1 6 1146
(*) for example, 1591 pedestrians and 4 motor vehicleusers were killed in pedestrian/motor vehicleaccidents.
Pedal Cyclists, Crash Helmets and Risk
329
and 178 per m o t o r vehicle occupant. Cyclists rarely kill others on the road, but are themselves overwhelmingly killed by m o t o r vehicles.
Epidemiology of cycling accidents Although data on pedal cyclist deaths are more reliable than injury data, there have been more epidemiological studies o f cycling accident injuries than deaths--perhaps because they are much more frequent. To describe the epidemiology o f cycling injuries, it is necessary to use data from hospital accident departments as well as police records. As noted above, British studies show hospital accident departments to be more complete sources o f data on cyclist injuries than police records. In Sweden, a country usually noted for the high quality of its public services, it was estimated that official information based on police reports identified 1% of accidents causing slight injuries, l f f 20% o f moderate injuries but 100% o f fatalities. 3 Child and adult cycle accidents should be considered separately, since cycling behaviour, experience, exposure and injuries are different in the two groups. In the United States, reports indicate that cycling accidents occur mainly in young people: for example, in a study based on police reports in Maryland, 5 75 % o f the injured cyclists were under 15, and 91% were under 20. While many of the reported accident department studies have focused on childhood cycling injuries, 6-t° half o f the 25,000 cyclist casualties in Britain recorded annually by the police, and two-thirds o f deaths, are to people of 20 years or more (4: table 29). While children injured in accidents are mainly cycling for pleasure, many adult cyclists are injured travelling to and from w o r k ? There are numerically more accidents in good weather during daylight than at other times, related to greater use of cycles in the afternoon and summer. 3'7'9''° Several series indicate that both junctions and clear straight roads are places for accidents. Most cycling accident injuries are minor; but accidents between cyclists and cars more often give severe injuries. 9 A study o f coroner's reports confirmed that a cyclist's death is almost always the result o f a collision with a m o t o r vehicle. H In a study o f police records in the USA, Williams 5 sought to identify the 'responsibility' for cycle - m o t o r vehicle collisions. Responsibility was allocated according to details o f the collision course. Cyclists were 'responsible', for example, when emerging from a drive or side road, and crossing a junction with a yield or stop sign. Motorists were 'responsible', for example, when making a left turn or striking a cyclist from behind. However, there were substantial age differences: almost all children under 15 had cyclist 'responsible' accidents, whereas almost two-thirds of adult cyclists (age 20 and over) were in accidents where the motorist was 'responsible'. Another study showed that fatal collisions o f child pedal cyclists are much more likely to occur with an overtaking motorist than non-fatal collisions, indicating greater motorist 'responsibility' in fatal accidents. ~2
Protection of pedal cyclists There have been frequent recommendations, in medical journals and national reports, for cyclists to protect themselves from head injuries by wearing crash helmets. 'When are cyclists going to wear helmets? '~3 'The medical evidence for cyclists to wear head protection is strong'. 14'Death from head injury occurs in about 70% o f cases, and cycle helmets offer a chance o f reducing this number'. ~5 'If bicyclists used helmets, many fatalities and serious head injuries would not OCCUr'. t6 Up to one-third o f cyclist injuries are head injuries. Serious head injuries, in which the
330
Mark McCarthy
cyclist is unconscious for more than 15 minutes, are much rarer--about 2-5% of all injuries. ~7,~s Death results from less than 1% of all injuries, but head injury is the commonest cause of death? 9'2° And, as has been shown, almost all deaths are due to collision with motor vehicles. However, pedal cyclist helmets meeting British Standard 6863 provide protection 'only when the rider falls onto the road without other vehicles being involved'. 2~ A study of American helmets showed that many would not protect the head even in a fall of 3 feet.~2A design standard that would protect the head fully in a motor collision would be too heavy and limiting ever to be worn. There have been no epidemiological studies of cycle helmet protection against death. The best evidence of a protective effect against injury was a case-control study using the accident departments of 5 hospitals in Seattle. 22 235 head injured cyclist cases were compared with 433 control cyclists attending for other injuries and with 558 cyclists, not attending hospital but identified through a health maintenance organisation as injured within the previous year. 20% of cyclists in both control groups said they wore helmets compared with 7% of the head injured cyclists--and, within this latter group, only 4% of 99 cyclists who received brain injuries. These differences remained significant after controlling for age, sex and education in a multiple regression analysis. But the study design cannot exclude the possibility of uncontrolled covariance--that helmet wearers were lower risk-takers, or had less exposure, and were therefore less likely to have serious accidents. Two studies have compared the head injuries of pedal cyclists attending an accident department with injuries of motor cyclists. Waters 23 comparing people age 13 years and over in Nottingham, found similar proportions of injured pedal and motor cyclists dying from head injuries, while motor cyclists also died from other injuries, to the body and legs. In Oxford, a study comparing people 16 and over gave similar findings.~8.~9The inference of these studies is not clear, since motor cyclists in Britain all wear helmets. There are also denominator problems when comparing proportions of all injuries because of the lower accident department attendance rates of pedal cyclists. In a roadside survey of cyclists in Vermont, USA, ~4% of 516 cyclists recalled striking their head in the previous 18 months. Seven of 13 unhelmeted cyclists received head injuries, whereas none of eight helmeted cyclists were injured. No information was available on the severity of injury. Dorsch 24mailed a questionnaire to Australian 'bicycling enthusiasts', and received replies on 197 head injuries. Two-thirds of this group had been wearing helmets, of various types. The authors considered that injury severity was reduced for wearers, and extrapolated the results to infer that helmets would also reduce fatalities. Only a minority of head injured people attending hospital accident departments are pedal cyclists. Brookes et al. 25 analysed head injured patients attending accident departments according to age and road user category. About one-third of injured child pedal cyclists attending accident departments have head injuries, but only one in ten adult cyclists. By contrast, 42% of child, and 23% of adult pedestrians had head injuries; and 20% of children and 58% of adult vehicle occupants had head injuries. The population risk of head injury death, and risk of death per head injury, for pedal cyclists is lower than for other road users. In Oxford, a University town with a high proportion of cyclists, motor vehicle deaths were seven times greater than cyclist deaths in absolute frequency (64 compared with 9), while pedestrian deaths were ten times more common than cyclist deaths as a proportion of the injured (27 deaths, 5.4% of all injured compared with 9, 0.5% of all injured). ~9In California also, motor vehicle occupants have far higher brain injury rates than pedal cyclists]6 Why have recommendations for, and
Pedal Cyclists, Crash Helmets and Risk
331
studies of, helmet use been addressed at cyclists, rather than higher risk road users--pedestrians and motor vehicle drivers?
Cycling risk: primary prevention A public health programme towards cyclist deaths and injuries should seek to prevent accidents through education, legislation and environmental change. However, there has been little epidemiological analysis of these approaches. Education of child cyclists is often considered a task of parents. The view, put in one medical editorial, that 'probably most parents are unaware of the high accident rates associated with cycling'27 is unlikely: ownership of child pedal cycles in households is high, but parents are increasingly reluctant to let their children use roads for travel. Training is given to some children in Britain by the police, but there has been no study of the benefit of these educational programmes. One retrospective study of injured children 9 showed that serious injury was no less frequent among children given bicycle safety instruction than those not given it. There has been no study reported of education to motorists about their risk for cyclists. Legislation affects pedal cyclists mainly through laws for other road users, for example, one way streets, pedestrians' rights of way on crossings, and exclusion from using motorways. Cyclists could benefit substantially from obedience to the law by motor vehicle users, especially limitation of alcohol consumption and speed limits, and possibly by more severe penalties for drivers who injure or kill other road users. Attitudes of the police may also need to change. A high school cyclist, writing an editorial on cycling safety in an American paediatric journal, described reporting a fractured ankle from a hit-and-run motorist, and receiving the reply from a police officer 'I don't want bicycles on my roads'. 28 Perhaps the most interesting legislative proposal has been by Howarth, 29 who recommended (in the context of child road injury reduction) that vulnerable road users--pedestrians and cyclists--should have the right of way in suburban residential roads that pedestrians have on zebra crossings. This could be implemented by defining the reversed right of way to exist wherever there is no kerbside yellow line. Cycling deaths often occur in ordinary settings--open roads, good visibility, speed restricted areas. Improvement of the environment for cyclists depends on limitation of motor vehicle parking and movement. Cyclists can be separated from other road users in bicycle lanes. Two studies 19,3° have attributed lower injury rates to cycle lanes, but small numbers and confounding factors make these claims difficult to evaluate. A comparison in Sweden 3 of 334 bicycle accident injuries on ordinary roads with 124 accidents on cycle paths indicated that path accidents more often involved adult cyclists, were more frequently at junctions and were more often between cyclists and pedestrians. The types of injuries did not differ between the two settings.
Health promotion through cycling: changes and effects The public health reason for supporting cycling is the likely improvement in health resulting from exercise. Morris has shown diminished cardiovascular risks in men taking moderate exercise 31 and there may also be mental health benefits. The Health Education Authority promotes cycling as a positive benefit to health. Transport planning needs to take account of these benefits to promote cycling as an alternative to motor vehicle travel. What is a possible agenda for health promotion through cycling? The first need is to
332
Mark McCarthy
control motor vehicles in suburban settings by environmental changes creating 'traffic calming', including widespread vertical deflections ('sleeping policemen') on road surfaces; many more pedestrianised and semi-pedestrianised roads; 32 and changing the law on right of way on the road. Second, cyclist needs should be respected in all road layouts--not only cycle tracks, and cycle junctions across main roads, but also adequate cycle space on the inside lane of all roads, especially at traffic lights. Third, broad transport policies should promote alternatives to motor vehicle travel--stopping business allowances for cars, investing in better public transport (especially for commuters), enhancing inner city environments. 33 Yet, paradoxically, these policies may not reduce cycle deaths and injuries, for two reasons. First, lowering the apparent risks of cycling will bring more cyclists onto the roads--adults and children who are currently too frightened to cycle, and greater exposure will bring more accidents. Transport statistics might show a fall in deaths or injuries per kilometre travelled--each individual journey could still be 'safer'--but the population risk, deaths or injuries per million population, could rise. In the Netherlands, one in three people cycle to work, compared with one in twenty in Britain; so the pedal cyclist death rates per head of population in the Netherlands are also higher than in the United Kingdom. The 'safest' behaviour for any individual cyclist, and for the population as a whole, is to stay at home. The second issue is risk compensation. Wilde 34 has suggested that human behaviour in complex social settings cannot be expected to have a simple measurable action-response effect. The complementary responses of others around the original factor may cancel out the expected change. This idea of 'risk compensation' has been shown by Adams 35 to be important in assessing the impact of transport policies. For example, it may be preferable for a pedal cyclist to appear a 'risky' road user (flying a flag at the back of a bicycle, or wearing bright clothing), so that drivers keep a good distance from the cyclist. In contrast, appearing as a 'low risl~' road user, for example, being less visible by wearing sombre clothing, or by keeping close to the kerb, could have the opposite effect of making drivers less worried about hitting the cyclist--and thus more likely to do it. It has proved difficult to measure risk compensation happening in practice because of its 'homeostatic' feedback. 35 In the case of seat belt wearing for front seat car drivers and passengers, however, the lack of population effectiveness of the British seat belt law was predicted in advance by proponents of risk compensation. 36 Yet it is relevant to cycle helmet use. Cyclists wearing helmets are probably by nature more cautious people than other cyclists, perhaps accounting for the observed differential head injury rates between wearers and non-wearers. But they are already risk takers in daring to venture out on a bike at all. If these cyclists begin to feel safer wearing helmets, some may change their behaviour in compensation. Equally, if motorists come to expect cyclists to wear helmets, and regard them as safer, the motorists' respect for cyclists may also change. The sum of all these individual risk behaviour changes may not be the benefit predicted by a simple action-response model. Conclusion
There are more than 200 pedal cyclist deaths and 25,000 injuries each year in Britain. Cyclists, like pedestrians, rarely cause deaths directly (because they contribute little impact energy), but are killed, usually in towns, by motor vehicles. Medical recommendations have focused on victim--education (especially for children), separation (cycle lanes) and, especially, cyclist protection (helmets). There are fewer cyclist head injury deaths than
Pedal Cyclists, Crash Helmets and Risk
333
pedestrian or m o t o r vehicle head injury deaths, but helmets have not been recommended for other road users as well. Perhaps this is because helmets that would protect against collisions with other vehicles would be too heavy to be acceptable to users. If, by separating or reducing m o t o r vehicles, cycling appears 'safer', then more cyclists, especially children, will return to the roads. The actual injury rate will only be reduced if m o t o r vehicle drivers change their driving behaviour. The benefits will be not only ecological; m o r e exercise, through cycling, will also give more health.
References 1. Wasserman, R. C., Monty, M. J., Emery, A. B. & Robinson, D. R. (1988). Bicyclists, helmets and head injuries: a rider-based study of helmet use and effectiveness. American Journal of Public Health; 78, 122(~1221. 2. Hobbs, C. A., Grattan, E. & Hobbs J. A. (1979). Classification of injury severity by length of stay in hospital. Crowthorne, Transport and Road Research Laboratory. (TRRL Report 871). 3. Lind, M. G. & Wollin, S. (1986). Bicycle accidents. Acta Chirugica Scandinavica, supplementum 531, 147. 4. Department of Transport (1989). Road Accident Great Britain 1988. London, HMSO. 5. Williams, A. F. (1976). Factors in the initiation of bicycle-motor vehicle collisions. American Journal of Diseases in Children, 130, 370. 6. Bjornstig, U. (1984). Pedal cycling accidents--mechanisms and consequences. Acta Chirugica Scandinavica, 150(4), 353-359. 7. Nixon, J., Clacher, R., Pearn, J. & Coreoran, A. (1987). Bicycle accidents in childhood. British Medical Journal, 294, 1267 1269. 8. Illingworth, C. M., Noble, D., Bell, D., Kemn, I., Roche, C. & Paseoe, J. (1981). 150 bicycle injuries in children: a comparison with accidents due to other causes. Injury, 13, 7-9. 9. Selbst, S. M., Alexander, D. & Ruddy, R. (1987). Bicycle-related injuries. American Journal of Disease in Childhood, 141(2), 140-144. 10. Joly, M. F., Foggin, P. M., Zuagulis, I. & Pless, I. B. (1989). Bicycle accidents among children in the urban environment. Canadian Journal of Pubfic Health, 80(5), 351-354. 11. Fife, D., Davis, J., Tate, L., Wells, J. K., Mohan, D. & Williams, A. (1983). Fatal injuries to bicyclists: the experience of Dade County, Florida. Journal of Trauma, 23(8), 745 755. 12. Friede, A. M., Azzara, C. V., Gallagher, S. S. et al. (1985). Epidemiology of injuries to bicycle riders. Paediatric clinics of North America, 32(1), 141-151. 13. Anonymous (1988). When are cyclists going to wear helmets? [Editorial]. Lancet, 1, 159-160. 14. Bull, J. P. (1988). Cyclists need helmets. British Medical Journal, 296, 1144. 15. Gloag, D. (1988). Strategies for accident prevention: a review of the present position. In Department of Health and Social Security. Strategies for accident prevention: a colloquium. London, HMSO. 94 16. National Committee for Injury Prevention & Control (1989). Injury prevention: meeting the challenge. American Journal of Preventive Medicine, 5(3) (suppl.), 118. 17. McDermott, F.T. (1985). Head injury predominance: pedal cyclists versus motor cyclists. Medical Journal of Australia, 143, 232-234. 18. Simpson, A. H., Unwin, P.S. & Nelson, I. W. (1988). Head injuries, helmets, cycle lanes and cyclists. British Medical Journal, 296, 1161-1162. 19. McCoy, G. F., Johnstone, R. F., Nelson, I. W. & Duthie, R. B. (1989). A review of fatal road accidents in Oxfordshire over a 2-year period. Injury, 20, 65 68. 20. Worrell, J. (1987). Head injuries in pedal cyclists: how much will protection help? Injury, 18, 5-6. 21. British Standards Institute (1990). Catalogue. London, British Standards Institute, 328. 22. Thompson, R.S., Rivara, F.P. & Thompson, D.C. (1989). A case-control study of the effectiveness of bicycle safety helmets. New England Journal of Medicine, 320(21), 1361 1367. 23. Waters, E. A. (1986). Should pedal cyclists wear helmets? Injury, 17, 372-373.
334
Mark McCarthy
24. Dorsch, M.M., Woodward, A.J. & Somers, R.L. (1987). Do bicycle safety helmets reduce severity of head injuries in real crashes? Accident Analysis & Prevention, 19(3), 183-190. 25. Brookes, M., MacMillan, R., Cully, S. et al. (1990). Head injuries in accident and emergency departments. How different are children from adults? Journal of Epidemiology and Community Health, 44, 147-151. 26. Kraus, J. F., Black, M.A., Hessol, N. et al. (1984). The incidence of acute brain injury and serious impairment in a defined population. American Journal of Epidemiology, 119, 186-201. 27. Smith, T. (1987). Children, bikes and money. British Medical Journal, 294, 1244-1245. 28. Allen, C. E. & Boorland S. W. (1987). Bicycle safety [Editorial]. American Journal of Diseases of Children, 141(2), 136 137. 29. Howarth, C. I. & Gunn, M. J. (1980). Pedestrian safety and the law. In: Chapman, A. J., Wade, F. M., Foot, H. C. (Eds.) Pedestrian Accidents. Chichester, Wiley, 265-290. 30. Armson, C. J. & Pollard, C. W. (1986). Child cyclist injuries: a prospective study. Medical Journal of Australia, 144(3), 144-146. 31. Morris, J. N., Clayton, D. G., Everitt, M. G., Semmence, A. M. & Burgess, E. H. (1990). Exercise in leisure time: coronary attack and death rates. British Heart Journal, 63, 325-334. 32. Gunnarsson, S.O. (1987). Strategies for reduction of traffic risks through urban and traffic planning. In: Kohler, L., Jackson, H. (Eds.) Traffic and children's health. Stockholm, Nordic School of Public Health, 187-211. 33. Morton, S. & Hannah, J. (1988). Why transport is a health issue. HFA 2000 News (Faculty of Public Health Medicine of the Royal Colleges of Physicians), 7, 1 2. 34. Wilde, G. J. S. (1982). The theory of risk homeostasis: implications for safety and health. Risk Analysis, 2, 209-225. 35. Adams, J. G. U. (1985). Risk and Freedom. London, Transport Publishing Projects. 36. McCarthy, M. (1989). The benefit of seat belt legislation in the United Kingdom. Journal of Epidemiology and Community Health, 43, 218-222.
© The Society of Public Health, 1991
Pedal Cyclists, Crash H e l m e t s and Risk Mark McCarthy, FFCM Department of Community Medicine, University College London, London WC 1E 6EA
As a rate per million kilometres travelled, the 'risk' of cycling appears to be high in relation to other forms of transport. Yet, in absolute numbers, there are far fewer cyclist deaths than pedestrian or motor vehicle occupant deaths, and most deaths and serious injuries to pedal cyclists are caused by other road users principally motor vehicles. The large majority of pedal cyclist deaths are due to head injuries after collision with a motor vehicle. It is therefore commonly proposed that cyclists should wear crash helmets for their own 'safety'. Helmets may protect against fall injuries, but current models are not designed to withstand the impact of collisions with motor vehicles. Evidence for the benefit of pedal cyclists wearing helmets is limited: the existing studies cannot exclude the possibility of different risk-taking behaviour, either by cyclists or by motor vehicle drivers, for helmet wearers compared with non-wearers. A public health policy towards reducing pedal cyclist deaths should seek prevention of accidents, rather than protection from their consequences. Cycling in greater safety would reduce the 'risk' per kilometre travelled, but more cycling might not reduce total cyclist deaths or injuries--because of greater exposure. The 'risk' of cycling--the risk of injury or death is a complex mix of exposure, 'danger' of the environment, and the perceived risk affecting our precautionary preventive behaviour.
Introduction The pedal cycle is an accepted and effective means of transport, used worldwide. It is energy efficient, low in pollution and enjoyable for business transportation and leisure. But, as with any form of transport, cycling has risks. This paper examines cycling risk, and the proposal that cyclists should wear helmets.
Describing cycling risk Cycling risk is usually described as the n u m b e r of individual events (the numerator) divided by the population exposed (the d e n o m i n a t o r ) - - f o r example, cyclist injury rates per 100 million kilometres. Both n u m e r a t o r and d e n o m i n a t o r need to be chosen carefully. Injury occurs in some, but not all, accidents involving a cyclist. R o a d injuries should be reported by law, but studies questioning cyclists retrospectively ~ or using casualty attendances (presumably more serious injuries) 2,3 suggest that there is substantial under-reporting to the police. Reporting will vary according to self-injury or collision, questions o f insurance, and the availability of police to report to. A policeman's decision on the severity of an injury is usually m a d e without medical examination. 4 The 'severe' category ranges from a fractured finger or substantial lacerations to multiple fractures requiring intensive care. The variation in recording, between policemen and for different county police forces, has never been assessed. Thus police injury statistics are epidemiologically unreliable.
Mark McCarthy
328
Road deaths in Britain are reported both by the Ministry of Transport, from statistics provided by the police of deaths within 30 days o f an accident, and by the Office of Population Censuses and Surveys usually following a coroner's inquest and including all deaths associated with an accident up to one year after the accident. The variation between these two sources is not marked and it is logical to use transport deaths data as numerator when using transport data also as denominator. Data including 'killed and seriously injured' together are likely to be much less accurate than death data alone. Data published by the Department o f Transport present the denominator for accidents, injuries or deaths as distance travelled. 4 This implies that distance covered is the prime objective of travel (to arrive rather than to travel). The approach favours long distance journeys, for which clearer data are available--short journeys are usually underestimated in travel surveys. Denominators of time taken in travelling, or per journey, would give different pictures, but such data are not provided. Also, using travel as denominator implies that risk relates only to the traveller, but externalities of travel--for example, lost productive time; energy consumed; pollution--affect the whole society. F r o m a public health viewpoint, however, it is the absolute population number, or rate per million population, that is more important than the rate per traveller.
Transport mortality Comparisons of risk between different modes o f transport need reliable numerators, so that death data are preferable. When transport 'safety' is considered as death rates per distance travelled, motor cycling appears the most 'dangerous', followed by pedal cycles, moving down a hierarchy of 'safety' to air travel, the 'safest' form of transport. These different forms o f transport are not strictly comparable, since it would be difficult to travel the distance of a plane journey by pedal cycle, and vice versa. The 'safety' o f some forms of travel is achieved through rigorous preventive p r o c e d u r e s - - f o r example, in air transport, by pilot medical examinations; agreed aircraft inspection procedures; and pre-takeoff countdown checks. Who is at risk in road transport crashes? Table I presents national data reported by the police on single vehicle crashes involving a pedestrian, and two-vehicle crashes, in which one or more people were killed. While more than 200 cyclists were killed in 1978, cyclists were reported as the 'victim' in almost all accidents. These data can also be considered as victim ratios, the number o f cyclists killed in crashes compared with the death o f one other road user. The cyclist victim ratios in Table I are 0.3 per pedestrian, 1.0 per m o t o r cyclist
Table I
Road deaths by user category, Great Britain 1988 (excluding accidents with three or more, and type not reported, accidents) (Department of Transport 1989, Table 23) (4).
Subject(*) Pedestrian Pedal cycle Motor cycle Motor vehicle
Pedestrian
Object* Pedal cycle
Motor cycle
-6 91 1591
2 1 3 178
5 3 13 359
Motor vehicle 4 1 6 1146
(*) for example, 1591 pedestrians and 4 motor vehicleusers were killed in pedestrian/motor vehicleaccidents.
Pedal Cyclists, Crash Helmets and Risk
329
and 178 per m o t o r vehicle occupant. Cyclists rarely kill others on the road, but are themselves overwhelmingly killed by m o t o r vehicles.
Epidemiology of cycling accidents Although data on pedal cyclist deaths are more reliable than injury data, there have been more epidemiological studies o f cycling accident injuries than deaths--perhaps because they are much more frequent. To describe the epidemiology o f cycling injuries, it is necessary to use data from hospital accident departments as well as police records. As noted above, British studies show hospital accident departments to be more complete sources o f data on cyclist injuries than police records. In Sweden, a country usually noted for the high quality of its public services, it was estimated that official information based on police reports identified 1% of accidents causing slight injuries, l f f 20% o f moderate injuries but 100% o f fatalities. 3 Child and adult cycle accidents should be considered separately, since cycling behaviour, experience, exposure and injuries are different in the two groups. In the United States, reports indicate that cycling accidents occur mainly in young people: for example, in a study based on police reports in Maryland, 5 75 % o f the injured cyclists were under 15, and 91% were under 20. While many of the reported accident department studies have focused on childhood cycling injuries, 6-t° half o f the 25,000 cyclist casualties in Britain recorded annually by the police, and two-thirds o f deaths, are to people of 20 years or more (4: table 29). While children injured in accidents are mainly cycling for pleasure, many adult cyclists are injured travelling to and from w o r k ? There are numerically more accidents in good weather during daylight than at other times, related to greater use of cycles in the afternoon and summer. 3'7'9''° Several series indicate that both junctions and clear straight roads are places for accidents. Most cycling accident injuries are minor; but accidents between cyclists and cars more often give severe injuries. 9 A study o f coroner's reports confirmed that a cyclist's death is almost always the result o f a collision with a m o t o r vehicle. H In a study o f police records in the USA, Williams 5 sought to identify the 'responsibility' for cycle - m o t o r vehicle collisions. Responsibility was allocated according to details o f the collision course. Cyclists were 'responsible', for example, when emerging from a drive or side road, and crossing a junction with a yield or stop sign. Motorists were 'responsible', for example, when making a left turn or striking a cyclist from behind. However, there were substantial age differences: almost all children under 15 had cyclist 'responsible' accidents, whereas almost two-thirds of adult cyclists (age 20 and over) were in accidents where the motorist was 'responsible'. Another study showed that fatal collisions o f child pedal cyclists are much more likely to occur with an overtaking motorist than non-fatal collisions, indicating greater motorist 'responsibility' in fatal accidents. ~2
Protection of pedal cyclists There have been frequent recommendations, in medical journals and national reports, for cyclists to protect themselves from head injuries by wearing crash helmets. 'When are cyclists going to wear helmets? '~3 'The medical evidence for cyclists to wear head protection is strong'. 14'Death from head injury occurs in about 70% o f cases, and cycle helmets offer a chance o f reducing this number'. ~5 'If bicyclists used helmets, many fatalities and serious head injuries would not OCCUr'. t6 Up to one-third o f cyclist injuries are head injuries. Serious head injuries, in which the
330
Mark McCarthy
cyclist is unconscious for more than 15 minutes, are much rarer--about 2-5% of all injuries. ~7,~s Death results from less than 1% of all injuries, but head injury is the commonest cause of death? 9'2° And, as has been shown, almost all deaths are due to collision with motor vehicles. However, pedal cyclist helmets meeting British Standard 6863 provide protection 'only when the rider falls onto the road without other vehicles being involved'. 2~ A study of American helmets showed that many would not protect the head even in a fall of 3 feet.~2A design standard that would protect the head fully in a motor collision would be too heavy and limiting ever to be worn. There have been no epidemiological studies of cycle helmet protection against death. The best evidence of a protective effect against injury was a case-control study using the accident departments of 5 hospitals in Seattle. 22 235 head injured cyclist cases were compared with 433 control cyclists attending for other injuries and with 558 cyclists, not attending hospital but identified through a health maintenance organisation as injured within the previous year. 20% of cyclists in both control groups said they wore helmets compared with 7% of the head injured cyclists--and, within this latter group, only 4% of 99 cyclists who received brain injuries. These differences remained significant after controlling for age, sex and education in a multiple regression analysis. But the study design cannot exclude the possibility of uncontrolled covariance--that helmet wearers were lower risk-takers, or had less exposure, and were therefore less likely to have serious accidents. Two studies have compared the head injuries of pedal cyclists attending an accident department with injuries of motor cyclists. Waters 23 comparing people age 13 years and over in Nottingham, found similar proportions of injured pedal and motor cyclists dying from head injuries, while motor cyclists also died from other injuries, to the body and legs. In Oxford, a study comparing people 16 and over gave similar findings.~8.~9The inference of these studies is not clear, since motor cyclists in Britain all wear helmets. There are also denominator problems when comparing proportions of all injuries because of the lower accident department attendance rates of pedal cyclists. In a roadside survey of cyclists in Vermont, USA, ~4% of 516 cyclists recalled striking their head in the previous 18 months. Seven of 13 unhelmeted cyclists received head injuries, whereas none of eight helmeted cyclists were injured. No information was available on the severity of injury. Dorsch 24mailed a questionnaire to Australian 'bicycling enthusiasts', and received replies on 197 head injuries. Two-thirds of this group had been wearing helmets, of various types. The authors considered that injury severity was reduced for wearers, and extrapolated the results to infer that helmets would also reduce fatalities. Only a minority of head injured people attending hospital accident departments are pedal cyclists. Brookes et al. 25 analysed head injured patients attending accident departments according to age and road user category. About one-third of injured child pedal cyclists attending accident departments have head injuries, but only one in ten adult cyclists. By contrast, 42% of child, and 23% of adult pedestrians had head injuries; and 20% of children and 58% of adult vehicle occupants had head injuries. The population risk of head injury death, and risk of death per head injury, for pedal cyclists is lower than for other road users. In Oxford, a University town with a high proportion of cyclists, motor vehicle deaths were seven times greater than cyclist deaths in absolute frequency (64 compared with 9), while pedestrian deaths were ten times more common than cyclist deaths as a proportion of the injured (27 deaths, 5.4% of all injured compared with 9, 0.5% of all injured). ~9In California also, motor vehicle occupants have far higher brain injury rates than pedal cyclists]6 Why have recommendations for, and
Pedal Cyclists, Crash Helmets and Risk
331
studies of, helmet use been addressed at cyclists, rather than higher risk road users--pedestrians and motor vehicle drivers?
Cycling risk: primary prevention A public health programme towards cyclist deaths and injuries should seek to prevent accidents through education, legislation and environmental change. However, there has been little epidemiological analysis of these approaches. Education of child cyclists is often considered a task of parents. The view, put in one medical editorial, that 'probably most parents are unaware of the high accident rates associated with cycling'27 is unlikely: ownership of child pedal cycles in households is high, but parents are increasingly reluctant to let their children use roads for travel. Training is given to some children in Britain by the police, but there has been no study of the benefit of these educational programmes. One retrospective study of injured children 9 showed that serious injury was no less frequent among children given bicycle safety instruction than those not given it. There has been no study reported of education to motorists about their risk for cyclists. Legislation affects pedal cyclists mainly through laws for other road users, for example, one way streets, pedestrians' rights of way on crossings, and exclusion from using motorways. Cyclists could benefit substantially from obedience to the law by motor vehicle users, especially limitation of alcohol consumption and speed limits, and possibly by more severe penalties for drivers who injure or kill other road users. Attitudes of the police may also need to change. A high school cyclist, writing an editorial on cycling safety in an American paediatric journal, described reporting a fractured ankle from a hit-and-run motorist, and receiving the reply from a police officer 'I don't want bicycles on my roads'. 28 Perhaps the most interesting legislative proposal has been by Howarth, 29 who recommended (in the context of child road injury reduction) that vulnerable road users--pedestrians and cyclists--should have the right of way in suburban residential roads that pedestrians have on zebra crossings. This could be implemented by defining the reversed right of way to exist wherever there is no kerbside yellow line. Cycling deaths often occur in ordinary settings--open roads, good visibility, speed restricted areas. Improvement of the environment for cyclists depends on limitation of motor vehicle parking and movement. Cyclists can be separated from other road users in bicycle lanes. Two studies 19,3° have attributed lower injury rates to cycle lanes, but small numbers and confounding factors make these claims difficult to evaluate. A comparison in Sweden 3 of 334 bicycle accident injuries on ordinary roads with 124 accidents on cycle paths indicated that path accidents more often involved adult cyclists, were more frequently at junctions and were more often between cyclists and pedestrians. The types of injuries did not differ between the two settings.
Health promotion through cycling: changes and effects The public health reason for supporting cycling is the likely improvement in health resulting from exercise. Morris has shown diminished cardiovascular risks in men taking moderate exercise 31 and there may also be mental health benefits. The Health Education Authority promotes cycling as a positive benefit to health. Transport planning needs to take account of these benefits to promote cycling as an alternative to motor vehicle travel. What is a possible agenda for health promotion through cycling? The first need is to
332
Mark McCarthy
control motor vehicles in suburban settings by environmental changes creating 'traffic calming', including widespread vertical deflections ('sleeping policemen') on road surfaces; many more pedestrianised and semi-pedestrianised roads; 32 and changing the law on right of way on the road. Second, cyclist needs should be respected in all road layouts--not only cycle tracks, and cycle junctions across main roads, but also adequate cycle space on the inside lane of all roads, especially at traffic lights. Third, broad transport policies should promote alternatives to motor vehicle travel--stopping business allowances for cars, investing in better public transport (especially for commuters), enhancing inner city environments. 33 Yet, paradoxically, these policies may not reduce cycle deaths and injuries, for two reasons. First, lowering the apparent risks of cycling will bring more cyclists onto the roads--adults and children who are currently too frightened to cycle, and greater exposure will bring more accidents. Transport statistics might show a fall in deaths or injuries per kilometre travelled--each individual journey could still be 'safer'--but the population risk, deaths or injuries per million population, could rise. In the Netherlands, one in three people cycle to work, compared with one in twenty in Britain; so the pedal cyclist death rates per head of population in the Netherlands are also higher than in the United Kingdom. The 'safest' behaviour for any individual cyclist, and for the population as a whole, is to stay at home. The second issue is risk compensation. Wilde 34 has suggested that human behaviour in complex social settings cannot be expected to have a simple measurable action-response effect. The complementary responses of others around the original factor may cancel out the expected change. This idea of 'risk compensation' has been shown by Adams 35 to be important in assessing the impact of transport policies. For example, it may be preferable for a pedal cyclist to appear a 'risky' road user (flying a flag at the back of a bicycle, or wearing bright clothing), so that drivers keep a good distance from the cyclist. In contrast, appearing as a 'low risl~' road user, for example, being less visible by wearing sombre clothing, or by keeping close to the kerb, could have the opposite effect of making drivers less worried about hitting the cyclist--and thus more likely to do it. It has proved difficult to measure risk compensation happening in practice because of its 'homeostatic' feedback. 35 In the case of seat belt wearing for front seat car drivers and passengers, however, the lack of population effectiveness of the British seat belt law was predicted in advance by proponents of risk compensation. 36 Yet it is relevant to cycle helmet use. Cyclists wearing helmets are probably by nature more cautious people than other cyclists, perhaps accounting for the observed differential head injury rates between wearers and non-wearers. But they are already risk takers in daring to venture out on a bike at all. If these cyclists begin to feel safer wearing helmets, some may change their behaviour in compensation. Equally, if motorists come to expect cyclists to wear helmets, and regard them as safer, the motorists' respect for cyclists may also change. The sum of all these individual risk behaviour changes may not be the benefit predicted by a simple action-response model. Conclusion
There are more than 200 pedal cyclist deaths and 25,000 injuries each year in Britain. Cyclists, like pedestrians, rarely cause deaths directly (because they contribute little impact energy), but are killed, usually in towns, by motor vehicles. Medical recommendations have focused on victim--education (especially for children), separation (cycle lanes) and, especially, cyclist protection (helmets). There are fewer cyclist head injury deaths than
Pedal Cyclists, Crash Helmets and Risk
333
pedestrian or m o t o r vehicle head injury deaths, but helmets have not been recommended for other road users as well. Perhaps this is because helmets that would protect against collisions with other vehicles would be too heavy to be acceptable to users. If, by separating or reducing m o t o r vehicles, cycling appears 'safer', then more cyclists, especially children, will return to the roads. The actual injury rate will only be reduced if m o t o r vehicle drivers change their driving behaviour. The benefits will be not only ecological; m o r e exercise, through cycling, will also give more health.
References 1. Wasserman, R. C., Monty, M. J., Emery, A. B. & Robinson, D. R. (1988). Bicyclists, helmets and head injuries: a rider-based study of helmet use and effectiveness. American Journal of Public Health; 78, 122(~1221. 2. Hobbs, C. A., Grattan, E. & Hobbs J. A. (1979). Classification of injury severity by length of stay in hospital. Crowthorne, Transport and Road Research Laboratory. (TRRL Report 871). 3. Lind, M. G. & Wollin, S. (1986). Bicycle accidents. Acta Chirugica Scandinavica, supplementum 531, 147. 4. Department of Transport (1989). Road Accident Great Britain 1988. London, HMSO. 5. Williams, A. F. (1976). Factors in the initiation of bicycle-motor vehicle collisions. American Journal of Diseases in Children, 130, 370. 6. Bjornstig, U. (1984). Pedal cycling accidents--mechanisms and consequences. Acta Chirugica Scandinavica, 150(4), 353-359. 7. Nixon, J., Clacher, R., Pearn, J. & Coreoran, A. (1987). Bicycle accidents in childhood. British Medical Journal, 294, 1267 1269. 8. Illingworth, C. M., Noble, D., Bell, D., Kemn, I., Roche, C. & Paseoe, J. (1981). 150 bicycle injuries in children: a comparison with accidents due to other causes. Injury, 13, 7-9. 9. Selbst, S. M., Alexander, D. & Ruddy, R. (1987). Bicycle-related injuries. American Journal of Disease in Childhood, 141(2), 140-144. 10. Joly, M. F., Foggin, P. M., Zuagulis, I. & Pless, I. B. (1989). Bicycle accidents among children in the urban environment. Canadian Journal of Pubfic Health, 80(5), 351-354. 11. Fife, D., Davis, J., Tate, L., Wells, J. K., Mohan, D. & Williams, A. (1983). Fatal injuries to bicyclists: the experience of Dade County, Florida. Journal of Trauma, 23(8), 745 755. 12. Friede, A. M., Azzara, C. V., Gallagher, S. S. et al. (1985). Epidemiology of injuries to bicycle riders. Paediatric clinics of North America, 32(1), 141-151. 13. Anonymous (1988). When are cyclists going to wear helmets? [Editorial]. Lancet, 1, 159-160. 14. Bull, J. P. (1988). Cyclists need helmets. British Medical Journal, 296, 1144. 15. Gloag, D. (1988). Strategies for accident prevention: a review of the present position. In Department of Health and Social Security. Strategies for accident prevention: a colloquium. London, HMSO. 94 16. National Committee for Injury Prevention & Control (1989). Injury prevention: meeting the challenge. American Journal of Preventive Medicine, 5(3) (suppl.), 118. 17. McDermott, F.T. (1985). Head injury predominance: pedal cyclists versus motor cyclists. Medical Journal of Australia, 143, 232-234. 18. Simpson, A. H., Unwin, P.S. & Nelson, I. W. (1988). Head injuries, helmets, cycle lanes and cyclists. British Medical Journal, 296, 1161-1162. 19. McCoy, G. F., Johnstone, R. F., Nelson, I. W. & Duthie, R. B. (1989). A review of fatal road accidents in Oxfordshire over a 2-year period. Injury, 20, 65 68. 20. Worrell, J. (1987). Head injuries in pedal cyclists: how much will protection help? Injury, 18, 5-6. 21. British Standards Institute (1990). Catalogue. London, British Standards Institute, 328. 22. Thompson, R.S., Rivara, F.P. & Thompson, D.C. (1989). A case-control study of the effectiveness of bicycle safety helmets. New England Journal of Medicine, 320(21), 1361 1367. 23. Waters, E. A. (1986). Should pedal cyclists wear helmets? Injury, 17, 372-373.
334
Mark McCarthy
24. Dorsch, M.M., Woodward, A.J. & Somers, R.L. (1987). Do bicycle safety helmets reduce severity of head injuries in real crashes? Accident Analysis & Prevention, 19(3), 183-190. 25. Brookes, M., MacMillan, R., Cully, S. et al. (1990). Head injuries in accident and emergency departments. How different are children from adults? Journal of Epidemiology and Community Health, 44, 147-151. 26. Kraus, J. F., Black, M.A., Hessol, N. et al. (1984). The incidence of acute brain injury and serious impairment in a defined population. American Journal of Epidemiology, 119, 186-201. 27. Smith, T. (1987). Children, bikes and money. British Medical Journal, 294, 1244-1245. 28. Allen, C. E. & Boorland S. W. (1987). Bicycle safety [Editorial]. American Journal of Diseases of Children, 141(2), 136 137. 29. Howarth, C. I. & Gunn, M. J. (1980). Pedestrian safety and the law. In: Chapman, A. J., Wade, F. M., Foot, H. C. (Eds.) Pedestrian Accidents. Chichester, Wiley, 265-290. 30. Armson, C. J. & Pollard, C. W. (1986). Child cyclist injuries: a prospective study. Medical Journal of Australia, 144(3), 144-146. 31. Morris, J. N., Clayton, D. G., Everitt, M. G., Semmence, A. M. & Burgess, E. H. (1990). Exercise in leisure time: coronary attack and death rates. British Heart Journal, 63, 325-334. 32. Gunnarsson, S.O. (1987). Strategies for reduction of traffic risks through urban and traffic planning. In: Kohler, L., Jackson, H. (Eds.) Traffic and children's health. Stockholm, Nordic School of Public Health, 187-211. 33. Morton, S. & Hannah, J. (1988). Why transport is a health issue. HFA 2000 News (Faculty of Public Health Medicine of the Royal Colleges of Physicians), 7, 1 2. 34. Wilde, G. J. S. (1982). The theory of risk homeostasis: implications for safety and health. Risk Analysis, 2, 209-225. 35. Adams, J. G. U. (1985). Risk and Freedom. London, Transport Publishing Projects. 36. McCarthy, M. (1989). The benefit of seat belt legislation in the United Kingdom. Journal of Epidemiology and Community Health, 43, 218-222.
