Bicycle Helmet Safety Institute

The Effectiveness of Bicycle Helmets:A Review

Summary: Back in 1995 this was a top resource for starting helmet research. It is still useful.

Revised Edition Prepared by Dr. Michael Henderson
for the Motor Accidents Authority of New South Wales, Australia.
1995

[Reorder Number MAARE-010995] [ISBN 0 T310 6435 6]

CONTENTS

EXECUTIVE SUMMARY

1 INTRODUCTION

1.1 Bicycling and the need for head protection
1.2 The emergence of an Australian Standard

1.3 Legislation for helmet use

1.4 The present review

2 INJURIES TO PEDAL CYCLISTS: AN OVERVIEW

2.1 Bicycle casualties in Australia 2.2 Data from other countries 2.3 The importance of head injury

3 CHARACTERISTICS OF PEDAL CYCLE CRASHES

3.1 Crash patterns 3.1.1 Australia 3.1.2 United States of America 3.1.3 Europe 3.2 The kinematics of a bicycle collision 3.3 The influence of alcohol

4 THE BIOMECHANICS OF HEAD INJURY

4.1 The anatomy of the head 4.2 Types of injury 4.3 Brain injury experimentation 4.4 Impact injury to the brain 4.5 The application of biomechanics to head protection 4.6 Helmet construction in practice 4.7 Standards

5 THE EFFECTIVENESS OF BICYCLE HELMETS

5.1 Measurement of effectiveness 5.2 Motorcycle helmet effectiveness 5.3 The first studies of bicycle helmet effectiveness 5.4 Crash reconstruction and helmet studies 5.5 Case comparison studies 5.6 Time series analyses 5.7 The effect of legislation 5.7.1 Early promotion of helmet use 5.7.2 The first legislation 5.7.3 Early results in Victoria 5.7.4 Results in New South Wales 5.7.5 Overseas studies 5.8 The effect of other strategies for increasing helmet use

6 SUMMARY AND CONCLUSIONS

6.1 The importance of cycling 6.2 Injuries to bicyclists 6.3 Characteristics of bicycle crashes 6.4 The development of head protection 6.5 The introduction of legislation 6.6 Studies of effectiveness 6.7 The maintenance of effectiveness

7 REFERENCES

EXECUTIVE SUMMARY

The importance of bicycling injury

Bicycle helmets substantially reduce the risk of head injury in a crash. This is shown by the biomechanical and epidemiological evidence reviewed in this paper. Scientific research has uncovered hard evidence on the benefits of bicycle helmet wearing, quite independent of issues related to the acceptability and effects of legislation.

Bicycling is a worldwide activity and an important means of transport for millions of people. Worldwide bicycle sales have grown far more rapidly than car sales over the last 20 years, so that the number of new bicycles produced is now three times the number of new cars.

Head injuries have emerged as a serious problem for bicyclists involved in accidents, and for the community as a whole because to a large part the cost of an individual's injury is a cost to the community.

Over the 20 years 1970 to 1990, bicyclist fatality rates per l00,000 people have fallen by an average of 1.0 per cent each year, but this is a rate of fall less than one-third of that shown by other road-user groups.

Nonfatal injuries resulting from bicycle accidents are grossly under-reported in official road accident statistics. Six times as many cyclists are admitted to NSW hospitals as are recorded as hospitalised in police/RTA road accident statistics. In NSW in 1990, hospital data show that pedal cyclists (2,108) were numerically the road users third most likely to be admitted to hospital as the result of a road crash compared to other road users, after vehicle drivers (3,954) and passengers (2,972) and before pedestrians (1,958) and motorcycle riders (1,792).

Injury rates are especially high in children and in males. In NSW in 1993, RTA data show that 102 cyclists aged five to 16 years were killed or seriously injured. This is 36 per cent of all cyclists recorded as killed or seriously injured. Of these 102, 85 were male.

In Australia, recent mass data indicates that 25 per cent of bicyclists admitted to hospital, and 44 per cent of those killed, had head injury as their single most important injury. These figures do not include multiple injuries, among many of which are unrecorded head injuries. Head injury is a cause of death in 80 per cent of cyclists' deaths and 33 per cent of reported injuries in Victoria, and several other studies have shown that, depending how the statistics are collected and analysed, bicycle crashes result in serious head injuries in one-quarter to two-thirds of bicyclists admitted to hospital, and up to 80 per cent if the collisions involved a motor vehicle. Up to 80 per cent of deaths among bicyclists are due to severe head injury.

Bicyclists admitted to hospital with head injuries are 20 times as likely to die as those without.

The characteristics of a bicycle accident

Bicycle crashes occur mainly during times of heavy traffic, and during daylight. Three-quarters of crash victims are male, with a high proportion being teenagers on school trips and young adults on work trips. Most collisions between bicycles and cars occur at intersections or where cyclists or drivers enter a roadway. The commonest injuries are to the limbs, followed by injuries to the head.

Collisions between bicycles and motor vehicles result in worse injuries than when the cyclist has fallen off without a collision. The primary impact is with the bicycle and the lower limbs of the cyclist. The body of the cyclist is then thrown up over the front of the car. Impact with the windscreen of the car is common at impact speeds as low as 25 km/h. The cyclist's head almost always hits the bonnet, the lower centre part of the windscreen or the A pillars that support the ends of the windscreen. The body of the cyclist is further injured by contact with roof structures, and at impact speeds of 55 km/h and over the cyclist is likely to be thrown completely over the car. There is then a new risk of head injury as the cyclist hits the ground.

Little is known in Australia about the involvement of alcohol in bicycle accidents. In the United States, 23 per cent of fatally-injured adult bicyclists have been found to be legally intoxicated, and most of them were males.

Head protection: principles and practice

Because of the obvious importance of head injury, from the earliest stages of bicycle accident analysis attention was concentrated on head protection. Early standards for bicycle safety helmets complied with the requirements of safety advocates, but failed the test of consumer acceptance. There then started a long process of education and persuasion, together with detailed modifications to the original Australian standard, aimed at wider acceptance and acceptability of pedal cycle helmets.

When a head is impacted, violent forces of acceleration are applied to the brain. These may be both linear and rotational. Resulting forces in the brain result in deformations throughout the brain tissue. Acceleration alone, in the absence of fracture, can result in functional injury to the brain.

Injuries to the scalp, skull and brain may be inflicted by a variety of mechanisms, and to protect against these different mechanisms requires a variety of approaches. The two fundamental principles for helmet design are centred on the use of padding to absorb energy and on the distribution of impact loadings.

The primary objective of head protection is the minimisation of brain tissue distortion on impact. A protective helmet usually consists of two parts: the outer shell and a padded layer for energy absorption. Most bicycle helmets are now made without a hard outer shell.

It is the plastic foam liner of the helmet that is responsible for absorbing the energy of impact through its own destruction. It should have a well defined relatively constant crushing strength, and be essentially plastic in the nature of its crushing.

The effectiveness of head protection

In July 1990 Victoria made the wearing of pedal cycle helmets compulsory, and through 1991 and 1992 NSW and the other states and territories followed suit. Nationwide, official figures show that deaths among pedal bicyclists have fallen from around 100 each year some 10 years ago to about half that number currently. Most of that fall has occurred in the years since 1989.

Several scientific studies have now been conducted into the effectiveness of bicycle helmets. Helmet design and construction is based on known mechanisms of head and brain injury. Evidence that helmets are effective depends on laboratory work, field in-depth investigations, and statistical analysis. Among the findings of the better studies are the following:

• The effectiveness of crash helmets for motorcyclists has been studied for decades, and they are known to reduce the risk of severe head injury by about one-third.
• The most careful, conservative estimates from good studies show that the reduction in risk of head injury to a bicyclist as a result of wearing a helmet is in the order of 45 per cent. In other words, at the very minimum a helmet halves the risk of head injury.
• Other estimates from controlled studies give even higher risk reduction figures. Depending on the type of impact and the severity of injury, the reduction in the risk of head injury as a result of wearing a helmet has been shown in several studies from all over the world to be in the range of 45 per cent to 85 per cent.
• Those who do not wear helmets are several times more likely to sustain injury to the brain tissue than riders who do.
• For children, an Australian study has shown that the risk of injury is reduced 63 per cent for head injury and 86 per cent for loss of consciousness, when a helmet is worn. For loss of consciousness, the risk is over seven times higher among non-helmet wearers than among helmet wearers.
• In the two years after the compulsory helmet legislation was introduced in Victoria, the number of bicyclists with head injuries decreased by 48 per cent and 70 per cent in each of the two years, relative to the last year before the law.
• In Queensland, the rate of head injury from bicycle crashes fell by more than half following the introduction of a helmet-wearing law; admissions to hospitals with bicycle-related injuries other than to the head remained unchanged over the same period.
• Helmets designed to the Australian and Snell standards provide a margin of protection in the real world greater than the respective standards require.
• Old-style helmets that do not comply with the Australian Standard reduce the risk of head injury by little or nothing.
• The vast majority of head impacts occurring in the real world of traffic are easily survivable if a Standards-approved helmet is worn.
• No studies have come to conclusions contrary to the above.

The maintenance of effectiveness

It seems clear from the Australian experience, and from American legislation affecting helmet use by children in very many states since 1990, that legislation is the only effective way rapidly to increase wearing rates to 80 per cent or so.

Even where legislation is in force, however, it remains the case that there is a recalcitrant group of cyclists who will not wear safety helmets for one reason or another. There is a wide disparity in wearing rates area by area and group by group in NSW, and this shows that the pressures against wearing helmets are also very different. It is probably the case that these differences are not directly related to legislation as such, but to personal reactions to helmet use and beliefs about helmet effectiveness. Where there are doubts about helmet effectiveness, such beliefs should be corrected as a matter of urgency.

There are other factors that affect the effectiveness of helmets.

Serious head injuries have been found by research to occur when the helmet comes off a rider's head, or the head is struck predominantly below the rim of the helmet.

These injuries are often the result of misuse. In New South Wales and other administrations it has been shown that a high proportion of helmets--especially those being used by young riders--are fitted loosely or otherwise poorly, are placed wrongly on the back of the head, or are worn without the chin straps being fastened. Unless such deficiencies are corrected, neither the helmets nor the laws requiring their use can reach anything like their full effectiveness.

1
INTRODUCTION

1.1 Bicycling and the need for head protection

Bicycling is a world-wide activity and an important means of transport for millions of people. Worldwide bicycle sales have grown far more rapidly than car sales over the last 20 years, so that the number of new bicycles produced is now three times the number of new cars.

But to ride a bicycle is not free of risk, any more than other modes of transport are free of risk. Excellent evidence from all over the world consistently shows that bicycle riders who go without head protection are roughly three times more likely to suffer head injuries in a crash than those who wear a helmet. Further, a bicyclist who sustains a head injury is some 20 times more likely to die than a rider who suffers other kinds of injury.

But the notion that pedal cyclists should wear protective helmets was once seen as ridiculous. Helmet use for motorcycle riders was "different". Motorcycles were perceived as fast and dangerous machines, and crashing the bike carried a clear and unarguable risk of death or injury. Therefore, opposition to helmet use for motorcyclists has always been (in Australia) muted, and based on arguments for civil liberty rather than on the effectiveness of helmets

On the other hand, pedal cycles have long been perceived as relatively slow, and falls and collisions merely inconveniences mostly suffered by children. The freedom to have one's hair flying in the wind was seen as much more important than the small risk of head impact if a rider was so unfortunate as to fall, or be knocked off the bicycle. But some twenty years ago, these perceptions started to change.

1.2 The emergence of an Australian Standard

An important step was the establishment of a committee by the then Standards Association of Australia (SAA). This was charged with the task of writing a standard for pedal cycle helmets. The original request for the formation of the committee came from competition riders and other serious cyclists who were in various ways dissatisfied with the head protection then offered, such as it was. The head "protection" of the time was provided by no more than a skull cap made of foam rubber or plastic strips covered with leather. This became known as the "hairnet" style of head protector.

During the early meetings of the new SAA committee, cyclists' representatives were horrified to find from head protection experts that if the brain was to be protected from injury at impact speeds as low as 20 km/h, then the helmets would have to be up to 50 mm thick, encasing padding material in a hard shell and with very few if any holes for ventilation. The only model available in the world at the time (1973) had just been put into production by the Bell Corporation in the United States, and although cyclists might reluctantly have conceded that it was probably effective, it was seen by most of them as a cumbersome and uncomfortable impediment to enjoyable cycling.

It was realised by the SAA committee that for head protection, the requirements of pedal cyclists were very similar to those of horse riders and many others engaged in other activities carrying a risk of head injury at comparable speeds. Therefore, the first edition of the Australian Standard, which was published in 1977, covered "lightweight protective helmets for use in pedal cycling, horse riding and other activities requiring similar protection". In terms of its general requirements and test procedures, the standard was based on the standard for motorcycle helmets. It included specifications for shock absorption, penetration resistance, strength of the retention system, peripheral vision and labelling.

It was soon found that while bicycle helmets complying with the new standard satisfied the requirements of safety advocates, they failed the test of consumer acceptance. There then started a long process of education and persuasion, together with detailed modifications to the original standard, aimed at wider acceptance and acceptability of pedal cycle helmets. The first helmets to be made in Australia were produced by Guardian Safety Products in 1979 and by Rosebank (the "Stackhat") in 1983. Particularly in the latter case the helmet was aimed squarely at making the product attractive, and in 1985 many other Australian-made helmets started appearing on the market with consumer acceptability being seen as important as safety performance. Amendments to the standard made it possible for the manufacturers to omit the hard and rather heavy outer shell, and a new wave of colourful, aerodynamic helmets was hitting the market by the late eighties.

1.3 Legislation for helmet use

In July 1990 Victoria made the wearing of pedal cycle helmets compulsory, and through 1991 and 1992 New South Wales and the other states and territories followed suit. At that time no other country or administration in the world had made the wearing of helmets for pedal cyclists compulsory.

When the wearing of car seat belts was made compulsory in Australia the proportion of vehicle occupants wearing restraints rose very rapidly to 90 per cent and more, and has remained at high levels thereafter. However, in the case of bicycle helmets compliance with the various state regulations has never been as high as for seat belts, and use rates have been patchy. This is especially the case among people in the younger age groups, who are arguably are the ones who need the most protection.

The reasons for this non-acceptance of helmet wearing by a substantial proportion of pedal cyclists are not well understood. They probably differ by age group. Among teenage adolescents, especially males, there is undoubtedly a perception that helmets are "daggy" and lack macho. Peer pressures will act against the ready acceptance of safety equipment by all in such groups. When helmets are worn, these pressures may result in a slapdash approach, including failure to fasten the chin strap, or rakishly perching the helmet at the back of the head. Both these practices (not surprisingly) have been shown to be related to an increased risk of head injury.

Youngsters and their parents may also be dismayed by what they see as the high cost of bicycle helmets. This is, however, a highly competitive market, and bargains can be found. Because even the least expensive helmet on sale in Australia must comply with the Australian standard, there need be no worry that to buy a new helmet at a discount rate is a threat to safety.

Among older riders, failure to use head protection may at least in part be due to a degree of fear and misunderstanding about the benefits of head protection. For example, many riders mistakenly believe that because they ride "safely", the only real risk they face is being hit by a motorised vehicle, and that in such a case the helmet cannot provide protection. There is strong scientific evidence however, documented in this report, to show that helmets do provide such protection.

Riders doubtful about the protective effect of helmets cannot be aware of the solidity of the scientific evidence showing that helmets work very well. Several excellent studies will be cited in this report to show that wearing a helmet provides at least as much protection for the head as a seat belt does for the body of a restrained car occupant, and there can be very few adults these days who doubt the value of seat belts.

There remains a proportion of the riding population who are opposed to legislation requiring the use of helmets on grounds of principle. They simply cannot accept that society has the right to make them do something that protects only them. But as will be shown in this report, head injuries impose a high cost to society. The worst injuries may require lifetime care, at a cost that is carried by all taxpayers, including those who have no interest whatsoever in cycling or head protection. Nobody in today's society can maintain that an injury to their own person imposes no cost--financial or in mental anguish--to anyone else. When protection can be proven by good science--as is the case for helmets and seat belts--then even those who hold most firmly to civil libertarian principles must concede that to compel protection for a few does bring benefit to many.

1.4 The present review

Concerned at the heavy drain on medical and rehabilitation resources resulting from head injury related in part to non-use of helmets by pedal cyclists, the Motor Accidents Authority commissioned the present review in order more clearly to establish and clarify the various issues relating to head injury, head protection and helmet effectiveness.

The report is structured as follows. Following this introduction, in Section 2 statistics are reviewed on injuries suffered by pedal cyclists, both in Australia and in other countries. The emphasis is placed on head injuries. The typical characteristics of a bicycle accident are described in Section 3.

In Section 4, the mechanism of head injury is described, as are features of helmet design and construction aimed at reducing the likelihood of head injury on impact.

Section 5 reviews the several good published studies of helmet effectiveness, undertaken both in Australia and overseas. These include both quasi-experimental case comparison studies and studies which have followed the effects of legislation on helmet use. Also reviewed are data on the present use of bicycle helmets, and some information on increasing such use through education and legislation.

2
INJURIES TO PEDAL CYCLISTS: AN OVERVIEW

2.1 Bicycle casualties in Australia

Injuries to bicyclists are an important cause of death and serious injury in the community. These injuries can result in substantial disability, both in the short and long term. The statistical statement for road traffic accidents in NSW (Road Safety Bureau, 1993) records that in 1993 eight pedal cycle riders were killed, 272 seriously injured (which means in practice admitted to hospital) and 1156 were less seriously injured. Nationwide, official figures show that deaths among pedal bicyclists have fallen from around 100 each year some 10 years ago to about half that number currently. Most of that fall has occurred in the years since 1989.

Over the 20 years 1970 to 1990, bicyclist fatality rates per 100,000 people have fallen by an average of 1.0 per cent each year (Anderson et al, 1993). This is a lesser improvement than shown for all other road-user groups. During the same period the average annual decrease in fatality rates for all road users was 3.3 per cent, with the rate for car drivers and passengers dropping by 3.9 per cent and 4.0 per cent respectively during the period.

Official statistics are generated by police accident reports, and while fatal injuries are quite accurately recorded there is no doubt that official road accident figures hide a serious level of under-reporting of non-fatal injuries among pedal cyclists. The same will be true for Australia as a whole. There is an enormous difference between, on the one hand, the number of pedal cyclists recorded in police accident statistics (used by the Roads and Traffic Authority in New South Wales) as having been admitted to hospital, and official hospital records data on the other. In 1990, the latest year for which direct comparisons are possible, 349 pedal cyclists were recorded in road accident data in NSW as having been hospitalised after a crash (Road Safety Bureau, 1993). For the same year, hospital admission figures show that 2,108 pedal cyclists were admitted to hospital in NSW as a result of a road crash (Federal Office of Road Safety, 1993). Thus, six times as many cyclists are admitted to NSW hospitals as are recorded as hospitalised in police/RTA road accident statistics.

In NSW in 1990, hospital data show that pedal cyclists (2,108) were numerically the road users third most likely to be admitted to hospital as the result of a road crash compared to other road users, after vehicle drivers (3,954) and passengers (2,972) and before pedestrians (1,958) and motorcycle riders (1,792).

For the nation as a whole, police accident records indicate that in Australia in 1990 1,571 pedal cyclists were admitted to hospital; however, hospital data for the same year show that there were 6,412 cyclists who appeared in hospital admission records, or more than four times as many.

In Western Australia the two data sources--hospital admissions and police accident reports--are linked. In that state there has been a systematic attempt to compare and contrast the characteristics and rates of bicycle injuries, using data from police reports and hospital admissions over a 15 month period (Piggott et al, 1994). Authors of this recent study estimated that there were at least 3,700 bicyclists injured each year in Western Australia, with the majority of the injured being aged less than 20 years. Police reports of casualties among bicyclists arose primarily from collisions between bicycles and motor vehicles. In contrast, the majority of hospital admissions were not due to such collisions but to bicycle-only crashes. Further, severe injuries were more common in collisions between bicycles and cars than in bicycle-only crashes. Bicyclist admissions to hospital as a result of collisions with motor vehicles were associated with a three-fold higher incidence of moderate or more severe injury that admissions among bicyclists who had not collided with another vehicle. After bicycle-motor vehicle collisions, over 25 per cent of patients stayed in hospital for more than a week. However, after bicycle-only crashes, only 10 per cent of admitted patients stayed longer than one week. In the United States, collisions with motor vehicles result in an estimated 90 per cent of bicycle-related deaths (Baker et a1, 1992; Sacks et al, 1991) and 33 per cent of bicyclists' brain injuries (Kraus et al. 1987).

Nearly half all the police reported casualties in Western Australia arose from right angle collisions between bicycles and motor vehicles, and a further one-quarter or so arose from sideswipe collisions. About half all collisions occurred at intersections.

Injury rates are especially high in children and in males. In NSW in 1993, 102 cyclists aged five to 16 years were killed or seriously injured. This is 36 per cent of all cyclists recorded as killed or seriously injured (police data: Road Safety Bureau, 1994). Of these 102, 85 were male.

All injury carries a substantial cost for society. The NSW Staysafe Standing Committee on Road Safety attempted in 1988 to estimate the cost of bicycle accidents in NSW, but was presented by very conflicting evidence. The committee was able to come only to the tentative conclusion that an estimate of $100 million a year could be the annual cost of bicycle crashes in NSW. This would equate to about 6 per cent of the cost of all road collisions (Staysafe 12, 1988).

2.2 Data from other countries

Overseas data show that patterns of injury in other countries are similar to those recorded in Australia, including the predominant importance of head injury in causing death and incapacitation. Long-term sequelae have been found to include behavioural disturbance: an American study of children who had been admitted to hospital, conducted more than four months after their discharge, revealed that 32 per cent had persistent behaviour changes and 83 per cent were to some extent limited in their physical activity for a median period of six weeks (Nakayama et al, 1990). Even minor head injury can cause emotional, psychological and intellectual problems (Rimel et al, 1981).

In December 1993 the US Consumer Product Safety Commission (CPSC) released a study that extensively reviewed bicycle use and bicycle injury (Rodgers, 1993). The study included a survey of hospital emergency department visits, and a survey of exposure to risk involving 1,254 telephone interviews (which relied on cyclists to estimate their riding exposure). In the United States there are about 580,000 bicycle-related injuries seen in hospital emergency departments each year. About 90 per cent of these injuries involved bicyclists, with the rest being passengers or pedestrians. Of the injured bicyclists, 62 per cent were male and 37 per cent were under age 10. One-third of all injuries involved the head or face, and 27 per cent of these injuries were regarded as potentially serious, being fractures, internal injuries or concussion. Eightyseven percent of head injuries involved collisions with a motor vehicle. Young children sustained more head injuries, with about 50 per cent of their injuries involving the head or face.

This is one of the few studies that have taken exposure into account. Riders in the 5-14 year age group had the highest injury rate at 17 per 1,000 riders. Injury rates for riders over aged 15 were considerably lower than the child injury rate. However, examining the injury rate per hours of bicycle use, the rate for those over 64 was found to be similar to the rate for the 5-14 year olds. Overall, children under age 15 were at more than five times greater risk for a bicycle injury than older riders.

The CPSC estimated that social costs in America for bicycle injuries and deaths totalled about $8 billion, or about $120 per year for each of the nation's 67 million bicycle riders. The risk of injuries on neighbourhood streets was found to be seven to eight times greater than on bicycle paths. About 90 per cent of the bicycle deaths involved crashes with motor vehicles on public roads. Analysis of the CPSC data indicates that a substantially higher bicycle fatality risks exists for males, for older bicyclists and for bicyclists who ride after dark when risks are adjusted for either the number of bicyclists or the time spent bicycling.

Differences also exist between fatal and non-fatal risk patterns, taking the age of the cyclist into account. These differences can be explained by the interaction of the types of accidents that result in death (as opposed to non-fatal injury) with the riding patterns and behaviour of bicyclists. Deaths among bicyclists are associated with collisions with motor vehicles and riding after dark to a much greater extent than non-fatal injuries. Since the more serious and life-threatening accidents involving collisions with motor vehicles occur on public roads, bicyclists' deaths are much more likely to involve cyclists who ride on roadways and major thoroughfares and bicyclists who ride after dark. These groups are more likely to include men and older riders. In addition, of course, drinking patterns may also contribute to the higher fatality risks for males and older riders, when exposure is taken into account. Further, older people are more likely to die as a result of a given severity of injury.

Most cyclist fatality rates depend on population figures rather than the population of bicyclists or some other measure of riding exposure. Such rates indicate a high risk for children. The CPSC study, if it is accepted that the methodology is sound, indicates that these higher rates for children do not hold when the fatality risk is adjusted by riding exposure. However, from a public health point of view it is the population-based figures that set priorities for countermeasures, and because there are many more children riding bikes than old people they inevitably figure more prominently in the statistics.

The analysis by Baker et al (1993) has been the only other study in the United States that attempted to adjust fatality rates by estimating riding exposure, and the CPSC and Baker studies are in broad accord. Baker estimated that the exposure-adjusted fatality risk for males was about 2.4 times the risk for females, and the fatality risk for 50-64 year old bicyclists was 3.5 times that of bicyclists 5-15 years of age.

2.3 The Importance of Head Injury

McDermott and Klug (1982) in a ground-breaking study over a decade ago in Victoria examined the different patterns of injury between motor cyclists, all of whom wore helmets, and pedal cyclists among whom fewer than 5 per cent wore helmets in the study period (1975-1980). Significant differences were shown. The incidence of concussion, fractured vault and base of skull and of intracranial trauma was significantly higher in pedal cyclists than motor cyclists. The frequency of fractured vault of skull was significantly higher in pedal cyclists than in motor cyclists sustaining only head injury. Although motor cyclist casualties overall then outnumbered pedal cyclist casualties by two or three times, there were about twice as many pedal cyclists as motorcycle riders suffering only head injury.

In Australia, recent mass data indicates that 25 per cent of bicyclists admitted to hospital, and 44 per cent of those killed, had head injury as their single most important injury (O'Connor, 1993). These figures do not include multiple injuries, among many of which are unrecorded head injuries. Wood and Milne (1988), using insurance-based figures, estimated that head injury is a cause of death in 80 per cent of cyclists' deaths and 33 per cent of reported injuries among cyclists in Victoria.

The recent Western Australia study (Piggott et al, 1994) is another of the very many that clearly identify the importance of head injury as a cause of death and morbidity among pedal cyclists. The prevalence of head injuries among hospital admissions was 32 per cent, the same as Woods and Milne documented in Victoria, although for only 5 per cent was the head injury categorised as being more than minor (Abbreviated Injury Scale 2 or more). Another recent Australian study pointing to the importance of head injury is one by the National Injury Surveillance Unit (NISU). They showed that in the year 1990 the body region sustaining the most severe injuries in pedal cyclists who were killed was the head in nearly half the cases, and this did not include those cases in which multiple injuries occurred with unspecified head injuries (O'Connor, 1993). Even for nonfatal injuries, the head is the body region sustaining the most severe injury for about one-quarter of all pedal cyclists admitted to hospital.

Several other studies have shown that, depending how the statistics are collected and analysed, bicycle crashes result in serious head injuries in one-quarter to two-thirds of bicyclists admitted to hospital, and up to 80 per cent if the collisions involved a motor vehicle. Up to 80 per cent of deaths among bicyclists are due to severe head injury (Kruse et al, 1980; Rivara, 1985; Guichon and Myles, 1975; Illingsworth et al, 1981; Thorson, 1974; Fife et al, 1983; McDermott, 1984; McKenna et al, 1984). The disability caused by head injuries may be of long standing (Guichon and Myles, 1975; Buntain, 1985).

The comprehensive review by Baker et al (1993) showed that in the state of Maryland (considered to be reasonably representative of the US population), the prevalence of bicyclists with head injury was 40 per cent for all ages combined. It was, however, highest in the youngest age groups, declining dramatically from 56 per cent for ages 1-4 to 19 per cent for ages 55 and older. The percentage with head injury was higher for females than males, 46 per cent versus 38 per cent . In more than one-third of these head injury cases in Maryland, the primary diagnosis was concussion. The most common associated injury was facial injury. Of all bicyclists with head injuries, only 8.5 per cent had skull fractures without evidence of injury within the skull. This percentage was somewhat higher in head-injured children less than 15 years of age, 11 per cent of whom suffered only skull fractures.

Baker's Maryland data showed that bicyclists admitted to hospital with head injuries are 20 times as likely to die as those without.

Emergency room data for the whole of the United States (Baker et al 1993) show that admission is most common among bicyclists with injuries to the head. Head injuries comprised 38 per cent of all reasons for hospital admission, totalling almost 7,700 per year in the United States. Injuries to the neck were found to be rare.

3
CHARACTERISTICS OF PEDAL CYCLE CRASHES

3.1 Crash patterns

3.1.1 Australia

Most information on the circumstances of bicycle crashes is derived from police accident reports, which are generally limited to crashes involving motor vehicles. In Western Australia, the number of police reported casualties per day from bicycle crashes was found to be approximately double for weekdays compared to weekends. On weekdays, crashes were more prevalent in the late afternoon, with 40 per cent of crashes occurring between 3 pm and 6 pm. On weekends, 50 per cent of crashes occurred between 9 am and 3 pm (Piggott et al, 1994).

An earlier study of bicycle crashes in Western Australia (Travers Morgan, 1987) showed that bicycle crashes occurred mainly during times of heavy traffic, and during daylight. Three-quarters of crash victims were male, with a high proportions being teenagers on school trips and young adults on work trips. Most collisions between bicycles and cars occurred at intersections or where cyclists or drivers enter a roadway. Obstructions to vision appeared to be a factor in these collisions. The commonest injuries were to the limbs, followed by injuries to the head.

A Monash University study in 1988 (Drummond and Jee, 1988) was the first in Australia to allow for exposure to risk in analysing crash patterns for bicyclists. The work was primarily aimed at investigating whether it would be safer for young children to ride on footpaths than the roadway (it was found to be so). The data showed that 11 to 17-year-old cyclists had a much higher risk of accident than two other age groups, up to 10 years and over 18. This was so for all kinds of accidents except for the kind of collision that results from a cyclist suddenly emerging on to a road in mid-block: this was found to carry the highest risk for the youngest riders. The data in this study did not allow analysis for smaller age groupings, so there is no confirmation of the American indication that exposure-related risk rises for the oldest groups.

Another study in Victoria specifically examined fatal bicycle accidents occurring during the night (Hoque, 1990). Although the author estimated that less than 10 per cent of total bicycle travel takes place at night, some 25 per cent of fatal bicycle accidents in Victoria occur during night time. The predominant problem appeared to be that bicyclists were being killed in accidents involving motorists coming from behind them. Most night time bicycle fatalities occurred on arterial road links and in high speed limit zones (75 km/h and over). In 90 per cent of night-time accidents the cyclists were hit by an overtaking motorist; the day-time equivalent figure was 40 per cent.

3.1.2 United States of America

American data on fatal injuries to bicyclists show that more than half of fatal collisions occur in urban areas and about one-third at intersections. The circumstances of crashes were found to vary with age. Children less than 15 years old are more likely than older cyclists to be killed in urban areas, on local roads, and at intersections. Older cyclists are more likely than children to be killed on divided highways after dark and by hit and run drivers (Baker et al, 1993).

3.1.3 Europe

In London, analysis of the characteristics of the accidents resulting in the deaths of nearly 200 cyclists showed that other motor vehicles were involved in almost all of them (Gilbert and McCarthy, 1994). Among these vehicles, heavy goods vehicles were involved in 30 per cent in outer London and nearly 60 per cent in inner London. This is a much higher proportion than has been recorded in other countries, including Australia. Passenger cars were involved in 54 per cent in outer London and 26 per cent in inner London. Some 35 per cent of those who died were children aged 16 years or less. These authors point to earlier studies which have tended to focus on children rather than adult cyclists, especially in the United States. As a pointer towards the political difficulties in implementing protective legislation in some countries, including the UK, these authors appear to hold the belief that limitation of heavy goods movements in the cities has a higher priority for cyclists' injury reduction than the use of safety helmets.

An emergency room study in Denmark over a two-year period compared over 1,000 bicyclists injured in collisions with more than 1,500 who were injured in cyclist-only accidents (Larsen, 1994). It was found that young bicyclists aged 10-19 years of age had the highest incidence of injuries caused by accidents involving collisions. Among the collisions, the crashes have different characteristics according to the colliding vehicle. In one group of crashes were the collisions with road users the author terms "soft" (other bicyclists, mopeds and pedestrians), as opposed to collisions with "hard" road users (motor vehicles and motorcycles). Bicyclists were most commonly injured in collisions during weekdays in the day time, and mainly in the summer. Some three-quarters of the collisions with cars occurred at intersections, mainly when the bicycle or car entered the intersection from a side road. The median stay in hospital was less than four days. Collisions with pedestrians and other bicyclists mostly occurred on bicycle tracks, and most of the collisions resulted only in minor injuries, although these often involved a stay in hospital and some absence from work or school.

3.2 The kinematics of a bicycle collision

A careful field study of real-world cycle accidents was conducted in Germany (Otte, 1989). The predominant collision was between a cyclist and the front of a car. The subsequent kinematics were shown to be similar to car-to-pedestrian or motorcycle collisions. The primary impact is with the bicycle and the lower limbs of the cyclist. The body of the cyclist is then thrown up over the front of the car. Impact with the windscreen of the car is common at impact speeds as low as 25 km/h. In this study the mean collision speed was 36 km/h. The cyclist's head almost always hits the hood, the lower centre part of the windscreen or the A pillars that support the ends of the windscreen. The body of the cyclist is further injured by contact with roof structures, and at impact speeds of 55 km/h and over the cyclist is likely to be thrown completely over the car. This, of course, presents a renewed risk of head injury as the rider hits the ground.

3.3 The influence of alcohol

Alcohol involvement in bicycling injury has not been well documented in the world literature. Li and Baker (1994) used data from the US Fatal Accident Reporting System to examine blood alcohol concentrations among fatally injured bicyclists aged 15 years or older during the years 1987-91. Of 1,711 bicyclists who were killed and tested (63 per cent of the total) 32 per cent were positive and 23 per cent legally intoxicated. Adjusted for age, time of crash and other variables, males were 3.3 times as likely as females to be positive for blood alcohol and 3.9 times as likely to be legally intoxicated. Those who died aged 25-34, and those who died from crashes at night, also had a significantly increased likelihood of being positive for blood alcohol and legally intoxicated. Even among those aged 15-19 years who were legally prohibited from drinking in the United States, 14 per cent had positive blood alcohol concentrations.

The proportion of bicyclists tested for alcohol appeared to be independent of year, day of week, time of day and other circumstances surrounding the crash. Even allowing for the fact that bicyclists suspected of being under the influence are more likely to be tested, these figures indicate that between one-quarter and one-third of adult bicyclists in the United States may have been affected by alcohol when they crashed. The authors of this study concluded that the role of alcohol should be taken more seriously into consideration in developing strategies for bicycle injury control and prevention.

4
THE BIOMECHANICS OF HEAD INJURY

The above review of the epidemiology of injuries to bicyclists clearly shows the predominant importance of head injury as a potential cause for death and permanent disability. Indeed, for physical trauma in general, the brain is the human organ that it is most important to protect. This is because injuries to the structures of the brain cannot be corrected through present medical technology, and the consequences of brain injury are often disastrous. Injury to the brain and other parts of the central nervous system affects control and function of very many other parts of the body, and it is by affecting the way the body functions that brain injury can lead to secondary injury to other parts.

4.1 The anatomy of the head

The anatomy of the head is important to an understanding of the mechanism of brain injury.

The outer layer of the head is composed of the scalp, which is some 5-7 mm thick and consists of three layers: the skin (including hair), a layer of connective tissue below the skin, and a layer of muscle and fibre. Beneath the scalp there is a loose layer of connective tissue and a fibrous membrane that covers the bone of the skull. The scalp provides some protective function to the skull through its firmness and some degree of mobility when sideways forces are applied to it.

The skull is a highly complex set of bones that enclose the brain, eyes, ears, nose and teeth. It varies in thickness from about 4 to 7 mm. Eight bones combine together to form the case that encloses the brain. While the inner side of the upper part of the brain case (the cranial vault) is relatively smooth, the base of this enclosed space is very irregular. It has several depressions and ridges, and many small holes through which arteries, veins and nerves pass. In addition, there is a larger hole (the foramen magnum) through which the spinal cord leaves the underside of the brain at the brain stem.

Within the skull there are three layers of membrane, the meninges, which separate the brain from the bone and support the blood vessels and nerves as they enter and leave the brain. Folds of these membranes project into fissures between the right and left sides of the brain and between the cerebrum (the main part) of the brain and an associated structure, the cerebellum. The outer layer is known as the dura mater. The meninges are lubricated by the cerebrospinal fluid (CSF) which also fills spaces within the brain (the ventricles), all helping to protect the brain tissue from mechanical shock as well as providing nutrient for the brain tissue.

The central nervous system (CNS) consists of the brain and the spinal cord. The CNS consists of a tightly packed network of nerves and supporting tissue.

4.2 Types of injury

Brain injuries fall into two categories: diffuse injuries and focal (localised) injuries. Diffuse injuries consist of swelling of the brain, concussion and what is now termed diffuse axonal injury (DAI). Focal injuries consist of haematomas (bruising and localised collections of blood) at various layers within the meninges, haematomas within the brain, and contusions (bruises) of the brain itself.

Brain injuries sustained by road accident victims fall into the diffuse type in three out of four cases, with one out of four cases of brain injury being of the focal type. The two most important causes of death are acute subdural haematoma (bleeding under the dura mater) and diffuse axonal injury. The injuries most often associated with a good or moderate recovery are cerebral concussion and contusion of the cortex, the outer part of the brain tissue.

Diffuse injuries can range from mild concussion to severe injury of the nervous tissue. Mild concussion may not result in loss of consciousness, but merely some confusion and disorientation. This is common and reversible, and often not brought to medical attention. Classic concussion involves immediate loss of consciousness following the head impact. Consciousness is lost for less than 24 hours and is reversible. Loss of memory (amnesia) for events before and after the head impact is present, and the duration of this amnesia is a good indication of the severity of the concussion. In about one-third of the cases of concussion there are no other lesions of the brain to be demonstrated. In more severe cases, concussion is associated with bruising of the brain and fracture of the skull (Gennarelli and Thibault, 1982). The eventual outcome depends on the severity of these associated brain injuries, but the vast majority can look forward to a good recovery within one month. However, about 2 per cent of these patients might in the end have severe deficit in brain function and another 2 per cent might have moderate deficit.

Injury resulting in immediate loss of consciousness which lasts for more than 24 hours has a much poorer outlook. After one month, only one in five of these cases will be showing a good recovery, with 50 per cent ending up with moderate to severe deficit and 20 per cent surviving in only in a vegetative state. About 7 per cent will have fatal outcomes.

Diffuse axonal injury involves immediate loss of consciousness which lasts for days or weeks. At the end of one month 55 per cent of patients are likely to have died, and others will have severe deficit. Diffuse axonal injury is characterised by the presence and persistence of signs related to disruption of the nerves within the brain and brain stem. Microscopic examination of the brain shows tearing of the nerves throughout the brain.

Diffuse brain injuries may all be complicated by swelling of the brain within the skull, and this is associated with a higher mortality rate. Because the internal volume of the skull cannot be increased, the brain and brain stem (which controls several vital physiological functions) is pushed downwards through the foramen magnum, and this causes further disruption of nervous tissue.

Subdural and extradural haematomas result from lacerations and tearing of the blood vessels in the space between the brain and the interior of the skull. These haematomas may be associated with skull fracture, but this is not necessarily the case.

Cerebral contusion is found in some 90 per cent of brains where head injury has resulted in death, with CT scans among nonfatally injured patients showing an incidence of contusion varying between about 15 per cent and 40 per cent (Melvin et al, 1993). Contusions may occur at the site of impact (coup contusions) and at sites remote from the impact (contrecoup concussions). The contrecoup lesions have more important effects than coup lesions. They occur predominantly over the fronts and sides of the brain as a result of internal impact against the irregular bony floors and sides of the skull cavity. Contusions are frequently although not necessarily associated with skull fracture, and are generally more severe when a fracture is present. Mortality from contusions ranges from about 25 per cent to 60 per cent, with older adults being more likely to die than children.

There is no consistent correlation between simple linear fractures of the skull and injury to the brain tissue. More complicated skull fractures result from more severe impact and are more likely to be associated with damage to the nervous tissue.

4.3 Brain injury experimentation

Experimental study of brain injury has fallen over the years into three categories: head impact, head acceleration, and deformation of the brain.

Head impact studies were first conducted in primates as early as 1941 (Denny-Brown and Russell, 1941). Most of the early experimentation involved impacting the primates' skulls in such a way that head movement after impact was unconstrained. This resulted in complex three-dimensional movement of the head and thus in a high degree of variability in measured responses. Later experimentation did attempt to constrain the head, but reproducibility of results was not greatly improved (Melvin et al 1993). Attempts at similar experimentation with non-primate models led to little more success.

The third group of experiments has involved direct deformation of brain tissue in the laboratory. The rationale for this work is that brain tissue is injured by stresses and pressures within the cerebrum and brain stem resulting from impact. The main advantage of such work is the comparative lack of variation in results for impact to impact, and it has been possible for researchers to derive a rather closer relationship between impact severity and brain injury than has previously been possible in the experimental situation (Lighthall et al, 1990).

4.4 Impact injury to the brain

Application of external impact forces to the head can result in local injury to the scalp, bones of the skull and to the brain tissue resulting from concentrated loadings. In the case of the brain, local injury at the site of impact can occur as a result of penetration of the skull by the impacting surface or local deflection of the skull without skull fracture. External forces can also deform the skull case as a whole, which can cause pressure disruptions throughout the brain.

When a head is impacted, violent forces of acceleration are applied to the brain. These may be both linear and rotational in nature. Resulting forces in the brain result in deformations throughout the brain tissue. An analogy can be seen in a jelly set on a plate. If the plate is sharply rotated, rotation of the top of the jelly will follow only after a delay; meanwhile, there will be tensions and distortions set up within the jelly mass. At an early stage it was shown that acceleration alone, in the absence of fracture, could result in functional injury to the brain (Gennarelli et al, 1982).

In addition to generating these internal stresses within the brain tissue, head impact can result in a relative motion between the different regions of the brain, and between parts of the brain with respect to the interior of the skull. The brain tissue is deformed as it bears upon irregular internal skull surfaces, and veins that bridge these surfaces can be disrupted or broken. The extent to which the brain tissue is internally deformed depends on the location of the impact point, the nature of the distribution of the force and the nature of the resulting motion of the head. In addition, as a result either of direct head impact or head motion secondary to impact with other parts of the body, the brain stem and spinal cord within the neck can be stretched as a result of secondary motion of the head.

Clearly, the complex nature of all the above relationships makes it impossible to define one single mechanism for brain injury as a result of head impact, let alone unarguably to associate a given degree of injury with a given type of impact force.

It has been shown that loss of consciousness is more readily produced by high levels of angular (rotational) acceleration than high levels of translational (straight line) acceleration. Ommaya and Gennarelli (1974) subjected two groups of animals to similar levels of acceleration loading. Pure translation (linear acceleration) did not produce diffuse injury, although focal lesions were produced. It was only when rotation was added to the translation that diffuse injury types were seen. Further work along these lines has clarified some relationships between different kinds of rotational acceleration and brain injury, but because the models used are small animals scaling the results to apply to humans is extremely difficult.

The first brain injury studies centred on direct blows to the head and measurement of the resulting linear accelerations associated with such blows. Using human cadavers, Lissner et al (1960) at Wayne State University summarised the relationship they found between acceleration levels, impulse duration and the onset of linear skull fractures. Their results indicated a decreasing tolerable level of acceleration as the duration of acceleration increased. This relationship became known as the Wayne State Tolerance Curve, and became the foundation of most current indices for head injury tolerance. The findings were extended by work with human volunteers in sled testing and the next step in the derivation of a head injury criterion became the Gadd Severity Index (GSI; Gadd, 1966). In 1972 the US National Highway Traffic Safety Administration proposed a modification of the GSI that has become known as the Head Injury Criterion (HIC), and this is currently used to assess the potential for head injury in car crash test dummies. As can be seen, however, the complex patterns of impact, stress and tissue deformation encompass a much higher degree of complexity than this relatively simple criterion would indicate.

There is also the issue that many if not most bicycle helmets are worn by children. Corner et al (1987) showed that there is considerable flexibility in the child's skull, which will deform readily on impact. This is why a child who has suffered only a mild head impact is usually admitted to hospital for observation. The elastic deformation of the child's skull can result in quite extensive diffuse brain damage. This would indicate that children's helmets should be constructed differently from adults', but there is only limited progress in that direction among manufacturers and standards-setting organisations. Lane (1986) came to similar conclusions, particularly in the case of the smallest children, in his review of the biomedical considerations for child bicycle helmets

4.5 The application of biomechanics to head protection

Injuries to the scalp, skull and brain may be inflicted by a variety of mechanisms, and to protect against these different mechanisms requires a variety of approaches. The two fundamental principles for helmet design are centred on the use of padding to absorb energy and on the distribution of impact loadings.

The above discussion will have demonstrated that the primary objective of head protection is the minimisation of brain tissue distortion on impact. Any mass will accelerate when a force is applied to it. During impact, as noted, accelerations may be linear or rotational. During the impact process energy is transferred, and because the head is not rigid, deformation and injury may be the result. Because energy cannot be created or destroyed, it must be transferred or absorbed. Therefore, the basic aim of head protection is to reduce the forces that could injure the head by absorbing some of the kinetic energy through the deformation or destruction of something else. That is the function of the protective helmet.

The extent to which the forces generated at impact can be reduced is a function of how much deformation of the helmet's structure may be achieved and the force required for that deformation. This in turn will depend on the strength, the amount and the shape of the padding material and on its relationship to the head. Padding materials may be categorised either as plastic or elastic. If padding is plastic, it will not recover from any deformation that occurs during impact loading. If on the other hand the padding material is elastic, it will recover its original shape. As it does so, the head will resume its initial velocity but in the opposite direction (in other words it will bounce). The maximum force developed will be the same, but the time during which the head is loaded will be doubled.

Most padding materials are neither perfectly elastic nor perfectly plastic, and the selection of material will depend to some extent on the activity which threatens the head. For example, in the kind of helmet used in American footbalI, where the helmet must function time after time without replacement, materials that recover their shape and properties are to be preferred. Where the helmet must perform its protective function just once, but then to maximum effect, a plastic material will be the best. The deformation to the helmet will be permanent, and the helmet suitable only for scrapping.

One of the primary objectives of good helmet design is to maximise the area of padding that can interact with the head during impact. This is because maximising the amount of material used during the impact maximises the absorption of kinetic energy and thereby minimises the transfer of energy to the head. A wellfitting helmet will maximise the contact area between head and padded liner.

No known form of head protection can completely protect the wearer against all foreseeable head impacts. Even the best available padding material has a definite limit to its energy-absorbing capability. No material can crush more than its original thickness, and when a material is nearly fully crushed it will become very stiff and the forces then developed will become very high. At that point the unabsorbed energy will be transferred to the head

Future improvements in head protection will be found with increased padding thickness, increased padding area (especially over the area of the temples), decreased crushing strength of the padding and uniform crushing strength. The first two of these properties will maximise the amount of energy absorbed, and the second two will minimise the force developed. The basic constraint known to all helmet wearers and manufacturers is that there is a practical limit to the thickness of padding in a helmet. Several analysts (for example, Corner et al, 1987; Mills and Gilchrist, 1991; Smith et al, 1993) have suggested that present standards, which employ solid headforms for testing, tend to favour stiff padding. In the real world, softer padding may protect more people from injury, although protection might suffer in the most severe (and rare) of impacts. The subject is still under research.

Helmet design is also complicated by other factors. It must be more or less spherical in shape. The amount of energy that will be delivered in an impact can never be forecast with accuracy. The shape, stiffness and other characteristics of the impacting surface or object cannot be anticipated. The user of the helmet will have specific needs not necessarily relating to safety that will limit options for good designs.

In practice, helmet designers aim to ensure that in a given impact the force that is developed is less than some predetermined value.

4.6 Helmet construction in practice

A protective helmet usually consists of two parts: the outer shell and a padded layer for energy absorption. In the case of motorcycle and car racing helmets, the object of the outer shell is to provide a hard strong outer surface that serves to distribute the impact load over a large area. It also provides protection against penetration against small sharp and high speed objects. It also serves to protect the padding from abrasions and knocks during day to day use. These requirements mean that the shell must be rigid, tough and hard, usually with a smooth exterior finish. The special requirements of bicycle helmets have led to a rethink of the need for an outer shell, and this issue will be discussed later in this report.

It is the liner of the helmet that is responsible for absorbing the energy of impact through its own destruction. Desirably, it should have a well defined relatively constant crushing strength and be essentially plastic in the nature of its crushing. Although in theory there are several materials which would function as liners, in practice the choice for manufacturers falls simply to one or two kinds of expanded plastic foam.

The most common materials used for the outer shell of protective helmets are glass reinforced plastic (GRP) and polycarbonate thermoplastics. Other components may be used for highly specialised applications. For bicycle helmets, thermoplastics are the only materials used for hard shells (becoming increasingly rare) and the much thinner "microshells". Other helmets have thin coatings of epoxy or other plastics, or have no coatings at all.

Because rotational accelerations are associated with brain injury, and because of a theoretical risk to the neck when a helmet catches on the ground or an object and rotates, studies have been undertaken to determine the sliding resistance of different kinds of outer helmet surface. Corner et al (1987) simulated real crashes and (with the helmets of the time) found evidence of severe rotational accelerations. Hodgson (1990) conducted an initial study of the sliding resistance of hard shell and no-shell helmets (foam-only), using a dummy impacting on slanted concrete. He concluded that no-shell (all foam) helmets do have a higher sliding resistance at impact angles of 30 to 45 degrees, and could increase neck loadings in a crash. His tests also predicted the possibility of facial injuries. A follow-up study (Hodgson, 1991) included microshell helmets. He found that both hard-shell and microshell helmets would slide rather than hang up, and sliding reduced the potential for neck injury. At the test impact speeds of 10 to 14 km/h, rotational head motion did not approach dangerous levels of angular acceleration or angular velocity. However, a later Swedish study (Andersson et al, 1993), using impact speeds of 20 to 40 km/h, showed that no-shell helmets did tend to hang up on asphalt surfaces, threatening both the brain and the neck.

It should be stressed that real-world crash experience (reviewed later in this report) shows that none of these laboratory results are reproduced in the field to any measurable extent; in other words, in the real world, rotational acceleration has not shown up as an important problem.

The materials used for liners are either semi-rigid polyurethane foams or expanded polystyrene bead foams. The latter are far more common for bicycle helmets. They are produced by introducing a known amount of pr-eexpanded polystyrene bead into a closed mould and ejecting steam. The beads expand and adhere to one another. Some three years ago General Electric began producing a new polystyrene foam (GESET) which is combined with other resins to mould into a stronger material that will permit lighter helmets with thinner foam to pass test standards, and which holds together better in multiple impacts. It is becoming very popular.

4.7 Standards

Because it is essentially impossible for a consumer to assess the protection afforded by a safety helmet by simply looking at it, in Australia and most other motorised countries there have developed systems of standards which set down requirements for protection and, in some cases, follow-up monitoring and quality control.

All helmet standards for impact performance are much the same in their approach. They entail the following. The helmet is placed on an artificial headform in the way that it would be worn by a real person. Different standards use different headforms, although all try to model the important features of the human head. The helmeted headform is then subjected to an impact. This is supposed to be typical of the type of impact that could be encountered in the specific application for which the helmet is used. Energy level, environmental factors and impact surface characteristics are considered, although no helmet performance standards presently monitor for a helmet's ability to reduce angular acceleration of the test headform. During the test the linear acceleration of the headform is monitored throughout the duration of the impact.

Most standards use a vertical drop test in which the helmeted headform is raised to some pre-determined height and released. At the moment of impact the assembly will have acquired a kinetic energy that is proportional to the drop height and its weight. This energy will be dissipated during collision with the impact, and to pass any particular standard the response of the headform must be within prescribed acceleration limits. A criticism of the solid headform universally used for standards approval is that it does not mimic the deformable characteristics of the human head (especially that of the child). The effect is that to pass the standards tests, the padding has to be firmer than might be desirable on theoretical grounds. However, it has not yet been shown as practicable to use a "soft" headform that produces consistently reproducible results.

Australian Standard AS 2063 covers requirements for lightweight protective helmets for use in pedal cycling, horse riding and other activities requiring similar protection. Part 1 of this standard describes basic performance requirements for impact attenuation, penetration resistance and so on. Some specific requirements for helmets for pedal cyclists that are variations from those basic performance requirements are specified in Part 2 of the standard, Helmets for Pedal Cyclists. This part of the standard includes requirements for ventilation and retention system effectiveness, and distribution of localised loading (a unique Australian requirement).

International comparison of standards is difficult because of differences in detail between test requirements, types of anvil and so on. As noted, the Australian standard has a unique requirement for what is known as localised loading. This was developed as a response to some dissatisfaction about the requirement for penetration resistance that was in the original bicycle helmet standard, because inclusion of such a requirement essentially mandated a hard outer shell. Such a shell, in turn, added weight to the helmet and there were many that questioned its practical effectiveness in the real world (Long et al, undated). The localised loading test ensures that the integrity of a helmet will be maintained when impact is localised over a small area, but without the need for a hard shell to resist penetration by a sharply pointed test impactor. It also has the effect of prohibiting in Australia some helmets sold in overseas markets that have inserts of very hard foam around large ventilation openings.

When subjected to the impact tests specified in the standard, the Australian standard requires that the headform acceleration shall not exceed a peak of 400 g, 200 g for a cumulative duration of three milliseconds and 150 g for a cumulative duration of six milliseconds. Both a flat anvil and a hemispherical anvil are used for the drop testing. Worldwide requirements are of the same order of magnitude, although there are differences in acceptable acceleration levels that are related to the energy imparted during the test.

Bicycle helmets have now been shown to provide such good general protection that related standards can cover helmet use for several similar activities such as in-line skating or skateboarding.

Because good fit is so vital to the protection offered by a helmet it is essential to have straps and fitting pads that are easy to adjust. Some helmets are much easier to fit than others and a wide range of helmet size is helpful. Most helmets have inner soft pads that attach with plastic strips allowing some experimentation with various thicknesses of pad as the helmet is fitted. It is a mistake, however, to compensate for a badly fitted helmet by using thick pads, as protection will be severely compromised.

5
THE EFFECTIVENESS OF BICYCLE HELMETS

5.1 Measurement of effectiveness

The effectiveness of bicycle helmets can be assessed in several different ways. These include field crash investigation including examination of helmets that have been involved in accidents, comparison studies of helmet use and nonuse in similar populations, and statistical studies associated with changes in population helmet usage in association with legislation or otherwise.

The protective effect--"effectiveness"--is usually expressed as the percentage reduction in risk, helmet worn versus helmet not worn. The risk reduction may be in terms of death, overall injury, or injury to a defined part.

5.2 Motorcycle helmet effectiveness

Supplementing more recent studies of the effectiveness of bicycle helmets is a long history of research on the effectiveness of crash helmets for motorcyclists. Parallels are valid because the principles for protection are the same, and as shown by McDermott and Klug (1982) head injuries were once much more prevalent among unhelmeted bicyclists than helmeted motorcyclists.

During the 1939-45 world war, a neurosurgeon of Australian background, Sir Hugh Cairns, advocated the use of helmets to cut the high incidence of head injury among motorcycle dispatch riders. Cairns and Holbourn (1943) concluded that crash helmets reduced the risk of skull fracture by 33 per cent. Jamieson and Kelly (1973) studied patterns of injury in Brisbane before and after the introduction of mandatory crash-helmet wearing laws in Queensland in 1970, and showed a dramatic reduction in both the incidence and severity of head injury. As various American states introduced, repealed, then reintroduced helmet-wearing laws for motorcyclists, analyses showed that helmet use resulted in a 43 per cent reduction in the risk of being killed (Watson et al, 1980) and a 65 per cent reduction in the risk of injury to the head, face and neck (McSwain and Petrucelli, 1984). More recent sophisticated statistical analyses of Evans and Frick (1988) and Wilson (1989) in the United States have found a reduction in the risk of being killed of around 30 per cent.

5.3 The first studies of bicycle helmet effectiveness

One of the first evaluations of the effectiveness of bicycle helmets was conducted by the NHMRC Road Accident Research Unit in Adelaide (Dorsch et al, 1984). In this study 894 cycling enthusiasts were contacted by mail with regard to their most recent bicycle accident and their helmet use at the time. Overall, 197 bicyclists were identified who had experienced an accident within the previous five years and had struck their head or the helmet in the crash. Helmet use fell into different groups: no helmet was used by 75 riders, an old style "hairnet" type of helmet by 69, an unlined solid helmet by 37, and a lined solid helmet by 16. This admittedly fairly crude study showed a consistent and statistically significant relationship between helmet use and reduced severity of head injury. The association persisted after adjustment for age, gender and severity of crash forces. The authors of this study estimated that the risk of death from head injury was three times higher for an unhelmeted rider than for a rider wearing a helmet of poor protective capacity, and ten times higher for an unhelmeted rider compared to one wearing a high standard helmet This 1984 study, although subject to respondent bias (as acknowledged by the authors), provided important support for the moves already under way at that time in Australia to increase the use of protective helmets by bicyclists.

At about the same time in Sweden, hospital accident records on cyclists injured in urban traffic accidents were studied for the years 1983-4 (Kroon et al, 1986). There were 36 helmeted riders in the study, 31 of whom had injuries which were more than minor. Two-thirds of these riders had not collided with another vehicle. A matched pair comparison method was used, with each helmeted rider being matched with an unhelmeted rider who had attended hospital following a similar accident. It was estimated by these authors (in what again was an early stage and rather limited study) that the risk of minor injury would be reduced by a factor of three if a helmet was worn, and for moderate injuries the risk of injury would be halved if a helmet was worn.

5.4 Crash reconstruction and helmet studies

For the Federal Office of Road Safety, Corner et al (1987) in a valuable and comprehensive report documented many aspects of head protection for pedal cyclists. Among other components of their work, they studied a total of 171 bicyclist crashes resulting in head injury. Eighteen of the injured people were wearing helmets. The study showed again that crashes involving other vehicles carry a far higher risk of head injury compared with other types of bicycle crashes, including falls from the bicycle and collisions with fixed objects. Collisions with other vehicles accounted for all of the 14 deaths included in the survey. There was also found to be a high proportion of children with head injury.

Corner et al found in this comparatively early study that bicycle helmets were reducing the severity of head injury, and this was particularly the case when injury resulted from a collision with another vehicle. In these collisions, among helmeted riders 92 per cent sustained minor injury, none had moderate injury, and 8 per cent severe head injury. Among those not wearing helmets 65 per cent sustained minor injury, 7 per cent moderate and 28 per cent severe (three to four times as many). Similar differences were found to exist for injuries sustained in non-collision crashes, although the differences were not so striking.

It was also found in this study that there was a substantial variation in the protection offered by different kinds of helmets, with the best results being found in association with helmets conforming to the then current Australian standard for bicycle helmets. The light "hairnet" type of helmets were associated with a higher degree of head injury.

Some other early studies took the first steps towards reconstruction of actual crashes in which helmets were worn. Hurt and Thom (1985) selected six bicycle accidents, all involving adults and in which helmets were worn. The crashes were reconstructed from observed damage to helmets and vehicles, from known road conditions and hospital records. Based on the accident reconstructions, an equivalent test drop height was found by experiment that would reproduce in a helmet test rig the observed real damage to the helmet for each case. The importance of this early study lay in the finding that there was a very large difference in protective capacity between helmets not complying to the American national standard for helmets and those that did so. Permanent or fatal brain damage could be expected for damage equivalent to a test drop height of around 1.0 metre in the case of the poor helmets, whereas moderate concussion or no injury or fracture could be expected in the case of damage equivalent to drop heights of 1.4-1.8 metres in the case of the good helmets.

Extending this kind of research, an important study was reported by Williams (1991) of the Royal Melbourne Institute of Technology. A group of 64 helmets worn by bicycle riders during crashes was evaluated to examine the level of protection they provided, and to gain some insight into the efficacy of bicycle helmet performance standards. The helmets were obtained during a study of the injuries sustained by 1,892 bicyclists who were admitted as casualties in Victoria during two periods between 1987 and 1989. Of the bicyclists in the study, 432 were wearing helmets at the time of their accidents and 64 of these had sustained an impact to the helmet. The helmets they had been wearing were submitted for evaluation in the context of hospital records of injuries sustained by the riders, and descriptions of the circumstances of the crashes.

The majority of the helmets (95 per cent) consisted of a hard shell with an expanded polystyrene foam liner. Nearly all were designed to meet the requirements of either the original Australian standard or one of its later amendments. A few helmets complied with overseas standards and two were not certified to any standard. Twenty-five of the accidents (39 per cent) involved a single bicycle, and 39 (61 per cent) involved a collision between a bicycle and another road user. Most of these collisions were with a motor vehicle and they resulted in all of the severe head injuries.

The severity of the impact that had been sustained by the helmets was simulated in the laboratory. This was done by dropping sample helmets from progressively greater heights in a test apparatus until the damage observed from the sample matched that observed on a helmet damaged in a real crash. Then, the severity observed in the simulated impacts was compared with the severity of test impacts prescribed in American and Australian performance standards.

The majority of impacts were of low to moderate severity. Sixty-seven percent of the impacts were reproduced at a drop height less than 0.75 m and 90 per cent at a height less than 1.5 m. Ten percent of impacts equated to higher drop heights. The majority of simulated impacts produced transmitted accelerations of between 0 and 100 g, with 90 per cent below 200 g. The results indicated that the helmets designed to the Australian and Snell standards provided a margin of protection greater than their respective standards required.

The study found that all the impacts occurred against flat objects and surfaces. A high proportion of helmets had sustained more than one impact. Most impacts occurred on areas of the helmet which were not tested during certification and many impacts were more severe than those stipulated in performance standards. Injury records showed that the predominant form of head injury was concussion of low severity.

All the serious head injuries occurred when the helmet came off the rider's head and collapsed because of a material defect, or was struck predominantly below the rim of the helmet. A high proportion of helmets worn by young riders had been misused. In summary, therefore, these bicycle helmets protected all their riders from severe head injury as long as they were properly worn and retained on the head during the crash.

A follow-up to this study has recently been reported (Cameron et al, 1994). The earlier helmets studied by Williams were mostly of the hard-shell kind, but these became unpopular and essentially superseded by the lighter thin-shell (microshell) and no-shell (foam only) helmets after the penetration-resistance requirement was dropped from the Australian Standard. This follow-up study therefore was centred on the newer helmets, but otherwise the methodology was comparable. In this data set, 75 per cent of the crashes involved a collision with a motor vehicle, a rather higher proportion than in the earlier data set. In neither data set were any injuries caused by penetration of the helmet.

There were strong, but different, relationships between the impact severity (measured by the drop height) applied to each helmet and the resulting peak acceleration experienced by the headform. The helmets in the more recent set of data, collected in 1991/92, produced lower peak accelerations for a given severity of impact on the helmet's external surface, in comparison with the earlier data set (1987-89). This was true for a range of impact types representative of those that occur in real bicycle crashes, where the majority of impacts are against a blunt surface. The impact surface was a bitumen roadway or concrete surface in two-thirds of the cases studied.

None of the head accelerations in this later data set exceeded 200 g.

There was no evidence of a real difference in protective effects between the newer and older helmets, although the number of helmets tested (38) may have been too low to allow small differences to emerge.

McIntosh and Dowdell (1992) conducted a rather similar study of bicycle helmets that had been involved in accidents in Sydney during early 1991. The accident sample, which was selected and not intended to be representative of the population of bicycle accidents, was drawn from accident and emergency departments, police reports, coroners' courts and direct advertising. An attempt was made to reconstruct the crash events, and the damage suffered in the real-world crashes was duplicated on identical but undamaged helmets in the laboratory.

Forty-two cases were investigated. Consistent with the Victorian data, two-thirds of the accidents involved collisions with another vehicle, predominantly passenger cars. The estimated pre-collision speed of the bicyclists was 0 to 60 km/h, and of opposing vehicles 0 to 110 km/h. Half the helmets had hard shells, 36 per cent soft shells and 14 per cent microshell. All complied to the Australian Standard or to an accepted American standard. Two-thirds of the impacts were to the front and sides of the helmets. All but three of the bicyclists were injured to some extent.

Helmet damage replication showed that the mean peak head acceleration among those who suffered some head injury was 180 g, and 129 g among those who sustained no head injury. Equivalent impact velocities were in the range of 14 to 20 km/h. In all these cases, therefore, head injury in the absence of a helmet would have been likely. In fact, however, in 75 per cent of the cases no brain injury was sustained, and only one head injury case was admitted to hospital because of the head injury. The fatally injured riders were subjected to impacts during collisions of such high energy severity that the head injuries were unpreventable.

An American study of the same general kind was undertaken at the Head Protection Research Laboratory at the University of Southern California. In contrast to the above Australian studies, however, it aimed to investigate collisions that did not result in injury requiring medical attention (Smith et al, 1993). It did this through the use of bicycle helmet manufacturer return programs, which allow helmet users to return helmets which have been damaged in an impact directly to the manufacturer for replacement. The rider is asked to provide a brief description of the crash and injuries along with the damaged helmet. Many of these riders had sustained only minor injuries, spent relatively little or no time in hospital, and no police accident reports existed for most of the crashes. Accordingly these collisions would have been difficult to track and evaluate by any other means. A total of 72 helmets were included as part of this study. All had impact damage. Impact sites were measured and documented in detail, related to injuries and where possible the damaged helmet was classified as having contributed to the prevention of the injuries, reduced injury severity or made injury worse. Some helmets were then selected for replication of the damage in the laboratory including replication of the impact surface in a standard drop test apparatus.

Nearly all met the Snell Memorial Foundation Standard, the retention system had been properly fastened and the helmet remained on the wearer's head for 97 per cent of the cases that were examined. The most prominent impact locations were found to be in the left and right front regions, and although some helmets had suffered more than one impact in only one case the helmet sustained more than one impact at the same location. The majority of primary impacts were onto a flat surface and the material that was most often struck was asphalt.

The impact replication study showed that among the type of impacts sustained by these riders, most of which had been at low speed, it was possible to replicate the impacts with drop heights of less than one metre in most cases. This is generally similar to the Australian findings of Williams (1991) and Cameron et al (1994). The replicated peak headform accelerations were all considerably less than required by current bicycle test standards. This study showed that most bicycle riders would benefit from the use of bicycle helmets under typical impact conditions.

The Bell helmet company in the United States maintains a database for helmets returned to the company through its own replacement program (Fisher and Stern, 1994). Impact locations were found to be distributed in a way similar to that shown in earlier studies, with around 40 per cent in the frontal area, about one quarter at the rear and most of the remaining impacts being on the sides. Yet again, this study showed that the overwhelming majority of impacts were against flat hard surfaces. The average pre-crash speed estimated by the cyclist was 33 km/h. Two-thirds of the riders reported some type of injury, but head injuries with one exception were of a minor nature.

Cameron et al (1994) concluded that the specified drop height of 1.5 metres for the impact attenuation test in the Australian Standard is set too low if the intention is to cover closer to the full range of impact severities experienced by the helmets of cyclists involved in crashes resulting in severe injury. However, Smith et al (1993) note that development of bicycle helmet standards should take into consideration the fact that increases in impact energies designed to reflect high severity impacts could have a significant (and by implication deleterious) effect on the ability of a helmet to protect at low impact levels. This is a similar conclusion to that of Mills and Gilchrist (1991), and a matter promoted for further research by Corner et al, 1987. There will always be a balance required between protection for one population at the expense of another.

5.5 Case comparison studies

The best studies seek to evaluate helmet effectiveness by comparing cases with cases, outcomes with outcomes, and controlling for all variables except for helmet use. These are sometimes known as case-control studies, although strictly speaking it is usually groups of people that are being compared, not individual cases. Most such studies are based on hospital emergency department data, and from study to study the different percentages of those injured (whether wearing helmets or not) is a factor of the kind of patients seen and treated in the particular units. The important difference, of course, is in the risk of head injury between helmet wearers and non-wearers in the same population sample.

There have been two major studies of this kind reported from Australia (one for adults and one for children) and two more in the United States and United Kingdom. They provide the nearest thing to definitive information on helmet effectiveness at the present time.

One of the best and most comprehensive of such studies has been the Australian one by McDermott et al (1993). They measured the effectiveness of bicycle helmets by studying crashes and injuries sustained by 1710 casualties treated at Melbourne and Geelong hospitals in 1987, 1988 and 1989. They compared causes of death and types and severity of injury for injured riders wearing standards-approved helmets (261), non-approved helmets (105) and no helmet (1344). The helmets used were all of the hard-shell type, and it was a subsample that was tested and reported upon by Williams (1991), as noted in section 5.4.

Male casualties outnumbered females four to one. Most casualties struck the ground first, and the second most commonly struck object was a motor vehicle. Five helmets came off because the retention system had not been fastened. Head injuries were significantly less frequent among wearers of Standards Australia-approved helmets (21.1 per cent) than unhelmeted casualties (34.8 per cent).

The relative proportion of head injury represented a reduction in head injury risk of 39 per cent. Excluding the helmets that were dislodged (the amended standard makes dislodgment much less likely for modern helmets) gives an injury risk reduction of 45 per cent. Helmets were found not only to reduce the frequency of head injury but also the severity of injury, with wearers of approved helmets sustaining significantly shorter periods of unconsciousness. This has obvious, and favourable, implications for the prevention of brain injury and subsequent permanent impairment.

The McDermott et al (1993) study used as the control (comparison) group those cyclists who were treated for injury (or who died) when not wearing a helmet. As noted, and as the authors acknowledge, this sampling misses many of those wearers who avoided head injury because the helmet was effective, and therefore underestimates the risk reduction effect to an unknown extent. (This, it might be said, is a failing of nearly all studies of safety equipment that rely only on casualty data.) The 45 per cent reduction in injury risk calculated by McDermott et al should therefore be regarded as at the very bottom of the possible range. In order to overcome this possible bias, an earlier study in Seattle used a population-based control group, consisting of bicyclists who had had accidents, whether or not they were injured or sought hospital care (Thompson et al, 1989). In addition, these authors also used an emergency room control group, similar to that of McDermott et al (1993).

In the Seattle study, the case patients were bicyclists who sought care for a bicycle-related head injury in the emergency room of one of five hospitals in the Seattle area during the study period (December 1, 1986, through November 30, 1987). Head injury was defined as an injury to those areas of the head that a helmet might reasonably be expected to protect: the forehead, scalp, ears, skull, brain, and brain stem. There were 235 bicyclists with head injuries in the study.

As noted above, the study had two separate control groups. The emergency room control group consisted of bicyclists who sought care at the same five emergency rooms for bicycle-related injuries other than head injuries. The population-based control group was designed to sample the population at risk for bicycling injuries; it consisted of cyclists who had had accidents, whether or not they were injured or sought medical care. The use of data from this control group permitted focus on assessing the degree to which helmets protect cyclists in accidents against head injuries that require medical attention.

The two comparisons--with emergency room patients, and the population-based group--gave risk reduction estimates for helmet wearing of 74 per cent and 85 per cent respectively. Comparing this study with the McDermott et al study, the Seattle definition of "head injury" included face injuries, which accounted for 109 cases among their total 235 head-injury cases. McDermott et al adjusted for this and other definitions, and showed that the risk reduction effect shown in the Seattle study using the emergency-room controls is 61 per cent, compared to their 45 per cent. The Australian study included several times as many helmeted cyclists, which makes their conclusions rather more robust. However, the Seattle group's inclusion of a population-based control group gives strength to the overall conclusion that the risk reduction effect of helmets is even greater than shown by the comparisons using emergency-room patients as controls.

These controlled studies provide convincing evidence of the effectiveness of bicycle helmets in preventing injury. They show that a rider not wearing a helmet is between two and three times as likely as a helmet wearer to suffer a head injury in a crash. This level of effectiveness is at least as high, and may be much higher, than that of seat belts in preventing injury to car occupants.

Thompson et al (1989) conceded that they could not completely role out the possibility that more cautious cyclists may have chosen to wear helmets and also had less severe accidents. Spaite et al (1991) compared injuries sustained by helmeted and unhelmeted riders in collisions with motor vehicles, finding that the helmeted riders had less injury to other parts of the body as well as to the head. They concluded that the collisions for these riders had been less severe, because the riders had been more cautious. However, Thompson et al (1989) made adjustments made for age, experience, and accident severity should have largely accounted for such potential differences between case patients and controls. In the large Australian study by McDermott et al, helmeted casualties had more frequent and severe body injuries than unhelmeted casualties, and riders wearing helmets slightly more often hit their helmeted heads or their faces on colliding vehicles. It may be reasonably concluded that the wearing of helmets is not strongly associated, if at all, with more cautious riding.

A study very similar to that by the Seattle group was more recently conducted in Queensland, this time directed specifically at head injuries sustained by children (Thomas et al, 1994). During 1991 and 1992, 445 children aged 14 years or less, presenting with bicycle-related injuries to the two main children's hospitals in Brisbane, were selected for analysis. The case group was composed of 102 children with injuries to the upper head area, including injuries to the skull, forehead and scalp, or loss of consciousness. The control group consisted of the 278 cyclists who were treated for injuries other than to the upper head or face. Information was recorded on the circumstances of the accident and the surface on to which the child fell. The degree of damage to the bicycle was used to assess the severity of the impact. The study was controlled for age, gender, hospital, education, accident cause, collision with a moving vehicle or stationary object and severity of impact.

Three-quarters of those injured in this study were boys. Age, however was not significantly associated with upper head injury. Contact with another moving vehicle was reported by 31 children and significantly more children with upper head injury had crashes involving contact with another moving vehicle. More injuries to the upper head occurred when the children fell on paved surfaces than on gravel, dirt or grass. Most children who were wearing a helmet at the time of the accident had hard shell helmets.

Significantly fewer children with head injury were wearing a helmet at the time of the accident compared with control subjects. Only one-fifth of the children who lost consciousness were wearing a helmet at the time of the accident. The reduction in risk among helmet wearers was 63 per cent for upper head injury and 86 per cent for loss of consciousness. This translates to a risk of injury to the upper head of 2.7 times higher among non-helmet wearers than among helmet wearers. For loss of consciousness, the risk was 7.3 times higher among non-helmet wearers than among helmet wearers. All these benefits were seen to persist after adjustment for possibly confounding variables. This emergency-room study, like all clinical studies, probably underestimates the effectiveness of safety equipment such as helmets, because those saved from any injury would not attend for treatment.

The benefits of helmet use shown by this carefully-conducted study of children were very much in the same order of magnitude as the benefits demonstrated in the Seattle and Melbourne studies.

The Seattle group used their same sample to assess the potential effectiveness of helmets in preventing injuries to the face (Thompson et al, 1990). The study in this case included 212 bicyclists with facial injuries and 319 controls with injuries to other body areas. Controlling for age, gender, education and income, they found no definite effect of helmets on the risk of serious facial injury overall, but there was evidence of an effect in preventing serious injury to the upper parts of the face.

In one of the largest of the comparative studies, Maimaris et al (1994) throughout 1992 collected data on all patients who attended the emergency department of a Cambridge hospital following a bicycle crash. They recorded helmet use and personal injuries with particular regard to head injuries. Head injury was defined as evidence of skull fracture, brain injury as shown by CT scan, and loss of consciousness or post-traumatic amnesia associated with other symptoms. Other minor injuries such as abrasions and bruises were not regarded as head injuries. The study sample included 1,040 patients with complete data, of whom 114 had worn bicycle helmets when the crash occurred.

There were two deaths following collisions with motor vehicles, but in neither case had a helmet been worn. There was no significant difference found between helmet wearers and non-wearers in the types of accidents in which they were involved. Most injuries were to soft tissues only, and mainly to the limbs. Significantly more children wore helmets (16 per cent) than did adults (9 per cent).

There were no significant differences between the two groups of cyclists with respect to the nature and site of injuries sustained, except in the incidence of head injury. Head injury was sustained by four out of 114 (4 per cent) of helmet wearers, compared with 100 out of 928 (11 per cent) of non-wearers. The risk reduction effect was therefore over 60 per cent. The incidence of head injuries sustained in accidents involving motor vehicles was higher than in those not involving motor vehicles.

Statistical analysis showed a protective factor of 3.25 for wearing a helmet. In other words, these authors calculated that non-wearers were over than three times more likely to sustain a head injury than a helmet wearer. There was a strong relationship between head injury and helmet wearing, and between head injury and involvement of a motor vehicle. Not only were the odds of head injury significantly reduced (by a factor of three) by wearing a cycle helmet, but also the protective effect of wearing a helmet was present for all ages and all types of accidents including collisions with motor vehicles. These authors also found that when helmet wearers sustained head injuries, they were less severe. All the patients in the study with skull fractures and severe brain injury, including the two deaths, had not been wearing helmets.

As discussed above, some critics have argued that statistical studies are misleading because helmet wearers are more likely to be cautious than non-wearers and are therefore less at risk of head injury. However, in this British study there was no difference in the types of accidents suffered by the two groups. It has also been argued that cyclists who own a safety helmet are more aware of the risks of cycling than those who do not. If helmet owners are safer riders than non-owners, then in a study of this kind there would be fewer injuries among cyclists who owned a helmet but were not wearing it at the time of the collision. This variable was known. The authors report that there was a similar rate of head injury observed among non-wearing helmet owners and non-owners, and the rate of head injury was much higher than for helmet wearers. The authors also found no difference between helmet wearers and non-wearers in the types of injuries other than head injury that were sustained, or in the areas of the body injured. All these findings, including data from different nations, do not support claims either that helmeted cyclists are more cautious or that they take more risks.

Another criticism that has emerged in association with the vigorous debate on helmet use in Europe and the United States is that helmets can be of little or no use in the severe collisions that occur between bicyclists and other motor vehicles, and that therefore it is the cars that need to be controlled in such a way as to prevent injury to cyclists in collisions. However, McDermott and Lane (1994) have published follow-up data to show that this belief is wrong. Wearing a helmet reduced the frequency of head injury among cyclists who were struck by moving vehicles from over 71 per cent to 50 per cent, a risk reduction of 30 per cent. The severity of those head injuries that did occur was less among those wearing helmets.

5.6 Time series analyses

Trends in the incidence of head injuries and bicycle related injuries in Brisbane have been examined by using injury surveillance data in a population of 600,000 people (Pitt et al, 1994). During the period 1985-1991 the rate of head injury from bicycle accidents fell by more than half. Admissions to hospital with bicycle related injuries other than to the head remained unchanged during the same period. Head injuries from other causes also fell through the whole period but remained unchanged after 1988, which is about the time that helmet wearing became widespread.

Bicycle related head injuries occurring over time have also been tracked in the United States (Sacks et al, 1991) These authors reviewed death certificates and emergency department injury data for 1984 through 1988. Using the Seattle data for effectiveness, they concluded that as many as 2,500 of the 2,985 head-injury deaths (62 per cent of all bicycling deaths) among cyclists in the United States from 1984 through 1988 could have been prevented if helmet-wearing had been universal.

5.7 The effect of legislation

5.7.1 Early promotion of helmet use

Because the proportion of riders using bicycle helmets has risen so substantially over the past decade, it has become possible for epidemiologists to track changing patterns of injury, and especially in relation to head injury match these patterns to helmet use in order to determine helmet effectiveness.

Much of the stimulus that resulted in legislation for the wearing of bicycle helmets in Australia (an internationally unique initiative) came from individuals, professional organisations and government authorities in Victoria. During the mid-1980s there was substantial promotion in the public media of helmet use in Victoria (about $1.3 million dollars in 1986, according to Staysafe, 1988), and there was introduced some public subsidy for the purchase of bicycle helmets for children. Following a heavy media campaign in 1984 the proportion of primary school children using bicycle helmets rose from 4.6 per cent in 1983 to 13.3 per cent in 1984, and to 38.6 per cent in 1985. For secondary school students the wearing rates were lower, going from 1.6 per cent in 1983 to 5.1 per cent in 1984 and 14 per cent in 1985 (Wood and Milne, 1988). An associated examination of changes in the incidence of head injury for pedal cyclists involved in accidents with motor vehicles showed a significant reduction in rate between the years 1982 and 1983 (before) and 1984 (after). This reduction was in the order of 20 per cent, and coincided closely with the above significant increases in the use of bicycle helmets (Healy, 1986).

5. 7.2 The first legislation

On July 1 1990 a law requiring wearing of an approved safety helmet by all bicyclists (unless exempted) came into effect in Victoria. This was the first such regulation in the world. Evaluations of the effect of the regulation in the state have centred on the use of helmets, the use of bicycles and the effect of the law on bicyclists' head injuries.

Regular observational studies have been undertaken in Melbourne since 1983. These have included children and commuter cyclists. Since 1985, similar surveys have been conducted in a selected sample of Victorian country towns, and recreational cyclists have been included both in the city and in the country. Additional surveys were undertaken when the law was introduced. When the law was introduced there was an immediate and substantial increase in helmet wearing by all age groups (Cameron et al, 1992). Through the period the average wearing rates for bicyclists in Victoria rose from 5 per cent in 1982/83 to 31 per cent in 1989/90, and then jumped to 75 per cent in 1990/91 following introduction of the helmet wearing law.

Surveys revealed a 36 per cent decrease in cycling by children between the years 1990 and 1991. Bicycle use decreased by 15 per cent for the 5-11 year olds and dropped by 44 per cent for the teenage group during the period after the law. The use of bicycles by adults, however, had been increasing before this period and maintained an increase which offset the decrease in cycling among children.

5.7.3 Early results in Victoria

To examine the initial effects of the law on bicycle injuries, data from three sources were examined: Transport Accident Commission (TAC) claims for nofault injury compensation, Health Department records, and Victorian Injury Surveillance System records for child cyclists (Cameron et al, 1992).

The number of cyclists killed or admitted to hospital with head injuries in Melbourne fell progressively between July 1981 and June 1990 as the use of helmets increased. Following the introduction of the law the number of head injuries decreased by 41 per cent relative to the corresponding period the year before the law was introduced. The number of cyclists in Melbourne sustaining severe injuries other than to the head increased during the early 1980s, fluctuated about a constant value for a few years, and then decreased by 8 per cent in 1990/91. This last reduction was consistent with the general reduction in all road deaths and hospital admissions in Victoria during that period.

Analysis of bicycle injury data showed a large reduction (37-51 per cent, as measured through different data sources and in different regions) in the number of bicyclists killed or admitted to hospital with head injuries during the first 12 months of the law. There were, however, also substantial (20-24 per cent) reductions in the number of severely injured bicyclists who did not sustain head injury. Nevertheless the percentage of severely injured bicyclists who suffered a head injury during the after-law period was statistically significantly below that which would have been expected had wearing rates seen before the law continued unchanged.

The authors concluded that the reduction in the number of severely injured cyclists with head injury following introduction of the law was achieved by both a reduction in the risk of head injury for cyclists who were severely injured, and a reduction in the number of cyclists involved in crashes resulting in severe injury.

There was some indication from this first study that the increase in helmet wearing after the law was not as effective in reducing the risk of head injury to crash-involved cyclists as might have been predicted from available effectiveness data. The reduced effectiveness appeared to apply particularly to adult cyclists, and to a lesser extent to those in their teens. It was suggested that this might be due to helmets being less securely adjusted or fastened by those cyclists who had not previously worn helmets and were doing so only because the law was introduced.

A follow-up study (Finch et al, 1993) showed that wearing rates through all ages continued to increase to around 83 per cent in Melbourne by the middle of 1992. Also, the number of bicyclists killed or admitted to hospital with a head injury in Melbourne continued to fall and by 1991/92 there were 66 per cent fewer injuries recorded than in the year before the law. Although there was a decline in serious injuries other than to the head the reduction was less dramatic, and by 1991/92 injury cases without head injury had declined to 17 per cent fewer than before the law.

The reduction in the number of severely injured bicyclists with injuries other than to the head, and some of the reduction of those with head injuries during the after law period, may have been due to a reduction in bicycle use as well as to other factors affecting the risk of accident involvement. Observational studies showed that bicycle use among teenagers had decreased by 43 per cent by 1991 and 45 per cent by 1992 relative to 1990. Bicycle use in children aged 5-11 years also decreased over the same period, by 3 per cent in 1991 and 11 per cent in 1992 compared to 1990. However, there was an increase in adult bicycling of 86 per cent by 1991 and a doubling of bicycle use in 1992 when compared with a survey in November 1987. When data for all age groups were combined, the total bicycle usage in 1991 was 9 per cent greater than 1987/88, and by 1992 it had increased by a further 3 per cent.

Turning to head injuries among bicyclists in the two after-law years (1990-91 and 1991-92) the number of bicyclists with head injuries decreased by 48 per cent and 70 per cent respectively, relative to the last year before the law (1989/90). The number of Victorian bicyclists sustaining severe injuries other than to the head fluctuated during the 1980s, decreased by 23 per cent in 1990/91 compared to 1989/90, and in 1991/92 the corresponding drop was 28 per cent. Accordingly, it is clear that the reduction in head injuries among pedal cyclists in Victoria after the law was introduced was first due to a reduction in the risk of head injury for bicyclists who were severely injured and, second, a reduction in the number of bicyclists involved in crashes resulting in severe injury. All "accident injuries" went down during this period in Victoria and for that matter in NSW and other parts of Australia, and it is clear that bicyclists benefited from this general improvement in road safety. Motor vehicle usage was reduced measurably during this period of recession, which would have reduced the risk of injurious collisions between cyclists and cars.

There is some indication that the effectiveness of increased helmet wearing increased in the second year of the legislation compared to the first year. This is arguably because the general standard of helmet wearing and adjustment improved.

Summarising studies undertaken in Victoria since the introduction of the legislation, Cameron et al (1994) highlighted the fact that two years after the introduction of the helmet-wearing law in Victoria, there were 70 per cent fewer cyclist casualties with serious head injuries in collisions, compared with 28 per cent fewer with other injuries. They concluded that the introduction of the law was accompanied by an immediate large reduction in the number of bicyclists with head injuries. This appeared to have been achieved through a reduction in the number of bicyclists involved in crashes plus a reduction in the risk of head injury of bicyclists involved in crashes. These improvements were maintained at least through 1992.

5.7.4 Results in New South Wales

In NSW, although head injury rates have not been tracked so systematically as in Victoria, there have been similar regular observational studies of helmet wearing rates, at least since 1990. The introduction of compulsory helmet wearing for adult cyclists in NSW was on January 1, 1991. Helmet wearing was first monitored in September 1990 in Sydney, Newcastle, Wollongong and ten major rural centres. The average rate of helmet wearing was 26 per cent, but the rates were much lower in outer rural areas (12 per cent), among young adolescents (11 per cent) and secondary school students (9 per cent) (Walker, 1990).

Before the introduction of the regulation the estimated helmet wearing rate for cyclists under 16 years of age was 31.5 per cent. By April 1992 the helmet wearing rate for cyclists under 16 years of age had risen to 75.9 per cent (Walker, 1992). In addition, however, a dramatic increase in helmet wearing among secondary school children was apparent. In April 1991 the lowest helmet wearing rates of all had been observed among secondary school children commuting to school (Sydney 11.4 per cent, rural centres 16.6 per cent). In April 1992, overall, 81.2 per cent of school students were wearing helmets and 71 per cent of students whose estimated age was 1315 years were wearing helmets. A concern noted in the Victorian studies was that the protection afforded by the safety helmet might be compromised by the way in which it is worn. It was observed in NSW that the helmet was not correctly positioned on the head in 16 per cent of cases, the helmet strap was not fastened tightly in 13 per cent, and not fastened at all in 6 per cent.

Among adults over 16 years, helmet wearing increased to 77 per cent in January 1991 after the regulation was introduced, and increased again to 85 per cent by April 1992. The increase was seen to be primarily among recreational cyclists rather than commuters, and occurred among cyclists travelling in the middle of the day and the evening rather than among cyclists observed earlier in the morning.

The most recent observational survey was published in 1993 (Smith and Milthorpe, 1993). The number of adult riders was seen to have increased marginally, but there was a further decrease in the number of children riding on roads and around school sites. However, because of the way that the observation sites were selected the figures should not be used to estimate risk exposure or the extent of riding in the state of NSW, and direct comparisons with the Victorian experience are not appropriate in this regard. Overall, 74 per cent of NSW children under 16 were seen to be wearing helmets, with wide variations in usage by area, age and activity. The lowest rate observed was 10 per cent among riders at a school in the western area of Sydney, and the highest was 99 per cent on the Wollongong bike path. Once again it was observed that up to 20 per cent of children were not being properly protected by their helmets because of the way they were being worn.

In general, helmet compliance by adults has been higher than for teenage children since the legislation was introduced. Now, more than two years after the legislation was introduced, the helmet wearing rate for riders over 16 years of age appears to have reached a plateau at 83 per cent . It has been observed that riders wearing cycling clothes had a consistently much higher level of helmet wearing and it is suggested that these could be regarded as "serious cyclists".

5.7.5 Overseas studies

There have been a few less comprehensive studies of the effectiveness of mandatory helmet laws in the United States (Kedjidjian, 1994). In Howard county Maryland, a mandatory law and education program directed at children under the age of 16 increased helmet use from 4 per cent to 47 per cent. In New Jersey, after the state enacted its mandatory law for cyclists under aged 14, fatalities among bicyclists up to that age dropped by 80 per cent and helmet use rose from 3 per cent to 60 per cent.

5.8 The effect of other strategies for increasing helmet use

During the years following the emergence of bicycle safety helmets on to the market there have been a wide variety of educational approaches used to promote the use of bicycle helmets. These have included classroom instruction, subsidising the purchase of helmets, and promotion of helmets within the health care setting.

A summary conclusion in the United States is that none of these strategies when used on their own have been shown to have any significant impact in increasing helmet use (Graitcer and Kellermann, 1994). However, Australian efforts documented above (Cameron et al, 1994) appear to have been considerably more effective. There has been, in addition, in the United States at least one successful program, in Seattle. This, like the strategies employed in Australia, included a mix of classroom instruction, discount purchase programs, demonstrations, distribution of printed material and intensive promotional efforts by community leaders and the media (Bergman et al, 1990). This broadbrushed approach has raised wearing rates to more than 40 per cent in the Seattle area (Rivara et al, 1994), which is quite similar to wearing rates achieved in Australia before the introduction of legislation.

Dannenberg et al (1993) compared helmet use in three Maryland school districts: one had a helmet law plus a helmet education program, one had only an education program, and one had no formal program. In the district with a law, helmet use rose from 11 per cent to 37 per cent. Where education was used on its own, the rise was from 8 per cent to 11 per cent. With no formal program, the rise was about the same at 7 per cent up to 11 per cent. There are several methodological problems with this study, but it does not support the proposition that education on its own will raise wearing rates.

It seems clear from the Australian experience, and from American legislation affecting helmet use by children in very many states since l990, that legislation is a necessary component in any package of measures intended rapidly to increase wearing rates to 80 per cent or so.

Even where legislation is in force, however, it remains the case that there is a recalcitrant group of cyclists who will not wear safety helmets for one reason or another. The wide disparity in wearing rates area by area and group by group demonstrated in NSW shows that the pressures against wearing helmets are also very different. It is probably the case that these differences are not directly related to legislation as such, but to personal reactions to helmet use and beliefs about helmet effectiveness.

Among older riders, failure to use head protection may at least in part be due to a degree of fear and misunderstanding about the benefits of head protection. For example, many riders mistakenly believe that because they ride "safely", the only real risk they face is being hit by a motorised vehicle, and that in such a case the helmet cannot provide protection. There is strong scientific evidence however, documented in this report, to show that helmets do provide such protection.

There remains a proportion of the riding population who are opposed to legislation requiring the use of helmets on grounds of principle and--as some reduction in the amount of cycling when legislation is introduced has shown--may even prefer not to ride rather than wear a helmet. The fact that the costs of their injury will be borne by others may not be properly appreciated.

Young people and their parents may be dissuaded by the high cost of bicycle helmets. Also, among teenage adolescents, especially males, there is undoubtedly a perception that helmets are "daggy" and lack macho. Peer pressures will act against the ready acceptance of safety equipment by all in such groups. A recent study in Maryland (Gielen et al, 1994) examined the extent to which psychosocial factors in addition to the presence of legislation might be associated with the use of bicycle helmets. It was found that the children's use of helmets was significantly associated with their beliefs in the social consequences of wearing helmets and the extent to which their friends wear helmets. It was concluded that to increase use, issues of style, comfort and social acceptability need to be addressed.

6
SUMMARY AND CONCLUSIONS

Wearing a helmet substantially reduces the risk of head injury to a cyclist in a crash. This has been shown by a raft of strong evidence generated by epidemiological and biomechanical research, and cited in the present report.

6.1 The importance of cycling

Bicycling is a worldwide activity and an important means of transport for millions of people. Worldwide bicycle sales have grown far more rapidly than car sales over the last 20 years, so that the number of new bicycles produced is now three times the number of new cars. Head injuries have emerged as a serious problem for bicyclists involved in accidents, and for the community as a whole because to a large part the community carries the cost of injuries to others. Over the 20 years 1970 to 1990, bicyclist fatality rates per 100,000 people have fallen by an average of 1.0 per cent each year, but this is a rate of fall less than onethird of that shown by other road-user groups.

6.2 Injuries to bicyclists

There is gross under-reporting of non-fatal injuries resulting from bicycle accidents in official road accident statistics. Six times more cyclists are actually admitted to NSW hospitals than police/RTA road accident statistics record as being admitted to hospital. In NSW in 1990, hospital data show that pedal cyclists (2,108) were numerically the road users third most likely to be admitted to hospital as the result of a road crash compared to other road users, after vehicle drivers (3,954) and passengers (2,972) and before pedestrians (1,958) and motorcycle riders (1,792).

Injuries are especially common in children and in males. In NSW in 1993, RTA data show that 102 cyclists aged five to 16 years were killed or seriously injured. This is 36 per cent of all cyclists recorded as killed or seriously injured. Of these 102, 85 were male.

In Australia, recent mass data indicates that 25 per cent of bicyclists admitted to hospital, and 44 per cent of those killed, had head injury as their single most important injury. These figures do not include multiple injuries, among many of which are unrecorded head injuries. Head injury is a cause of death in 80 per cent of cyclists' deaths and 33 per cent of reported injuries in Victoria, and several other studies have shown that, depending how the statistics are collected and analysed, bicycle crashes result in serious head injuries in one-quarter to two-thirds of bicyclists admitted to hospital, and up to 80 per cent if the collisions involved a motor vehicle. Up to 80 per cent of deaths among bicyclists are due to severe head injury.

Bicyclists admitted to hospital with head injuries are 20 times as likely to die as those without. Long-term sequelae have been found to include behavioural disturbance.

Overseas data show that patterns of injury in other countries are similar to those recorded in Australia, including the predominant importance of head injury in causing death and incapacitation.

6.3 Characteristics of bicycle crashes

Bicycle crashes occur mainly during times of heavy traffic, and during daylight. Three-quarters of crash victims are male, with a high proportion being teenagers on school trips and young adults on work trips. Most collisions between bicycles and cars occur at intersections or where cyclists or drivers enter a roadway. The commonest injuries are to the limbs, followed by more injuries to the head.

Collisions between bicycles and motor vehicles result in the worst injuries. The primary impact is with the bicycle and the lower limbs of the cyclist. The body of the cyclist is then thrown up over the front of the car. Impact with the windscreen of the car is common at impact speeds as low as 25 km/h. The cyclist's head almost always hits the hood, the lower centre part of the windscreen or the A pillars that support the ends of the windscreen. The body of the cyclist is further injured by contact with roof structures, and at impact speeds of 55 km/h and over the cyclist is likely to be thrown completely over the car.

6.4 The development of head protection

Because of the predominant importance of head injury, from the earliest stages of accident analysis attention was concentrated on head protection. Early standards for bicycle safety helmets complied with the requirements of safety advocates, but failed the test of consumer acceptance. There then started a long process of education and persuasion, together with detailed modifications to the original Australian standard, aimed at wider acceptance and acceptability of pedal cycle helmets.

6.5 The introduction of legislation

In July 1990 Victoria made the wearing of pedal cycle helmets compulsory, and through 1991 and 1992 NSW and the other states and territories followed suit. Nationwide, official figures show that deaths among pedal bicyclists have fallen from around 100 each year some 10 years ago to about half that number currently. Most of that fall has occurred in the years since 1989.

6.6 Studies of effectiveness

Several scientific studies have now been conducted into the effectiveness of helmets, including laboratory work, field in-depth investigations, and statistical analysis. Among the findings of the better studies are the following:

• The effectiveness of crash helmets for motorcyclists has been studied for decades, and they are known to reduce the risk of severe head injury by about one-third
• The most careful, conservative estimates from good studies show that the reduction in risk of head injury to a bicyclist as a result of wearing a helmet is in the order of 45 per cent. In other words, at the very minimum a helmet halves the risk of head injury.
• Other estimates from controlled studies give even higher risk reduction figures. Depending on the type of impact and the severity of injury, the reduction in the risk of head injury as a result of wearing a helmet has been shown in several studies from all over the world to be in the range of 45 per cent to 85 per cent.
• Those who do not wear helmets are several times more likely to sustain injury to the brain tissue than riders who do.
• For children, an Australian study has shown that the risk of injury is reduced 63 per cent for head injury and 86 per cent for loss of consciousness, when a helmet is worn. For loss of consciousness, the risk is over seven times higher among non-helmet wearers than among helmet wearers.
• In the two years after the compulsory helmet legislation was introduced in Victoria, the number of bicyclists with head injuries decreased by 48 per cent and 70 per cent in each of the two years, relative to the last year before the law.
• In Queensland, the rate of head injury from bicycle crashes fell by more than half following the introduction of a helmet-wearing law; admissions to hospitals with bicycle-related injuries other than to the head remained unchanged over the same period.
• Helmets designed to the Australian and Snell standards provide a margin of protection in the real world greater than the respective standards require.
• Old-style helmets that do not comply with the Australian Standard reduce the risk of head injury by little or nothing.
• The vast majority of head impacts occurring in the real world of traffic are easily survivable if a Standards-approved helmet is worn.
• No studies have come to conclusions contrary to the above.

6.7 The maintenance of effectiveness

It seems clear from the Australian experience, and from American legislation affecting helmet use by children in very many states since 1990, that legislation is the only effective way rapidly to increase wearing rates to 80 per cent or so.

Even where legislation is in force, however, it remains the case that there is a recalcitrant group of cyclists who will not wear safety helmets for one reason or another. There is a wide disparity in wearing rates area by area and group by group in NSW, and this shows that the pressures against wearing helmets are also very different. It is probably the case that these differences are not directly related to legislation as such, but to personal reactions to helmet use and beliefs about helmet effectiveness. Where there are doubts about helmet effectiveness, such beliefs should be corrected as a matter of urgency.

There are other factors that affect the effectiveness of helmets.

Serious head injuries have been found by research to occur when the helmet comes off a rider's head, or the head is struck predominantly below the rim of the helmet.

These injuries are often the result of misuse. In New South Wales and other administrations it has been shown that a high proportion of helmets--especially those being used by young riders--are fitted loosely or otherwise poorly, are placed wrongly on the back of the head, or are worn without the chin straps being fastened. Unless such deficiencies are corrected, neither the helmets nor the laws requiring their use can reach anything like their full effectiveness.

7.
REFERENCES

Anderson P R, Adena M A and Montesin H J, Trends in road crash fatality rates: Australia, 1970-1990, INTSTAT Australia Pty Ltd for the Federal Office of Road Safety, Report No. CR 113, Department of Transport and Communications, Canberra, 1993.

Andersson, Larsson and Sandberg, Chin strap forces in bicycle helmets, Swedish National Testing and Research Institute, SP Report, 1993 (BHSIDOC # 487, Bicycle Helmet Safety Institute, Arlington VA, 1990).

Baker S P, O'Neill, B Ginsburg M J and Li G, The injury Fact Book, Second Edition, New York NY, Oxford University Press, 1992.

Baker S P, Li G Fowler C and Dannenberg A L, Injuries to Bicyclists, a National Perspective, Johns Hopkins Injury Prevention Centre, Johns Hopkins School of Public Health. Baltimore, Maryland, 1993.

Bergman A B, Rivara F P, Richards D D and Rogers L W, The Seattle children's bicycle helmet campaign, American Journal of Diseases of Children, 144:727-731, 1990.

Rodgers G, Bicycle Use and Hazard Patterns in the US, and Options for Injury Reduction US Consumer Product Safety Commission, Washington DC, 1993.

Buntain W L, Bicycle injuries in children, an economic insight and literature review, in Proceedings, 29th Annual Conference, American Association for Automotive Medicine. AAAM 1985.

Cairns H and Holbourn H, Head injuries in motorcyclists with special reference to crash helmets. British Medical Journal, 1 :591, 1943.

Cameron M H, Finch C F and Vulcan A P, The protective performance of bicycle helmets introduced at the same time as the bicycle helmet wearing law in Victoria, Report No. 59, Monash University Accident Research Centre, July 1994.

Cameron M H, Vulcan A P, Finch C F and Newstead S V, Mandatory bicycle helmet use following a decade of helmet promotion in Victoria, Australia: an evaluation, Accident Analysis and Prevention, 26:325-337, 1994.

Cameron M H, Heiman L and Neiger D, Evaluation of the bicycle helmet wearing law in Victoria during its first 12 months, Monash University Accident Research Centre, Report No.32, July 1992.

Corner J P, Whitney C W, O'Rourke N and Morgan D E, Motor cycle and bicycle protective helmets: requirements resulting from a post crash study and experimental research, School of Civil Engineering, Queensland Institute of Technology, for the Federal Office of Road Safety, Department of Transport and Communications, Report CR 55, 1987.

Dannenberg A L, Gielen A C, Beilenson P L, Wilson M E H and DeBoer, Bicycle helmet laws and educational campaigns: an evaluation of strategies to increase children's helmet use, American Journal of Public Health, 83(5), 1993.

Denny-Brown D and Russell W R, Experimental cerebral concussion, Brain, 64:93-164, 1941.

Dorsch M M, Woodward A J and Somers R L, Effect of helmet use in reducing head injury in bicycle accidents, in Proceedings, 28th Annual Conference, American Association for Automotive Medicine, AAAM, 1984 (reprinted in Accident Analysis and Prevention, 19:183, 1987).

Drummond A E and Jee F M, The risks of bicyclist accident involvement, Monash University Accident Research Centre, Report No. 2, October 1988.

Evans L and Frick M C, Helmet effectiveness in preventing motorcycle driver and passenger fatalities. Accident Analysis and Prevention. 20:447-458, 1988.

Federal Office of Road Safety, Road crashes resulting in hospitalisation: Australia 1990: statistical summary, year ended 31 December 1990, Department of Transport and Communications, Canberra,1993.

Fife D, Davis J and Tate L, Fatal injuries to bicyclists: the experience in Dade County, Florida, Journal of Trauma, 23:745-755, 1983.

Finch C F, Newstead S V, Cameron M H and Vulcan A P, Head injury reductions in Victoria two years after introduction of mandatory bicycle helmet use, Monash University Accident Research Centre, Report No.51, July 1993.

Fisher D and Stern T, Results of a study on 1100 bicycle helmets involved in accidents, in Proceedings, 38th Annual Conference, Association for the Advancement of Automotive Medicine, AAAM, 1994.

Gadd C W Use of a weighted impulse criterion for estimating injury hazard, in Proceedings, Tenth Stapp Car Crash Conference, Society of Automotive Engineers, 164:174, 1966.

Gielen A C, Joff E A, Dannenberg A L, Wilson M E H, Beilenson P L and DeBoer M, Psychosocial factors associated with the use of bicycle helmets among children in counties with and without helmet use laws, Journal of Paediatrics, 124:204-210, 1994.

Gennarelli T A and Thibault L E, Biomechanics of acute subdural haematoma, Journal of Trauma, 22: 680-686, 1982.

Gennarelli T A, Thibault L E, Adams J H, Graham D I, Thompson C J and Marcincin R P, Diffuse axonal injury and traumatic coma in the primate, Annals of Neurology, 12:564-574, 1982.

Gilbert K and McCarthy M, Deaths of cyclists in London 1985-92: the hazards of road traffic British Medical Journal. 308: 1534-1537, 1994.

Graitcer P L and Kellermann A L, Is legislation an effective way to promote bicycle helmet use?, in Proceedings, 38th Annual Conference, Association for the Advancement of Automotive Medicine, AAAM 1994.

Guichon D M P and Myles, J J Bicycle injuries: one year sample in Calgary, Journal of Trauma, 15:504-506, 1975.

Healy D J, Trends in helmet usage rates and bicyclist numbers sustaining a head injury, July 1981 to December 1985, Victoria, Report GR/86/11, Road Traffic Authority, Victoria 1986.

Hodgson V R, Impact, skid and retention tests on a representative group of bicycle helmets to determine their headneck protective characteristics, Department of Neurosurgery, Wayne State University, Michigan, 1990 (BHSIDOC # 310, Bicycle Helmet Safety Institute, Arlington VA, 1990).

Hodgson V R, Skid tests on a select group of bicycle helmets to determine their head-neck protective characteristics, Department of Neurosurgery, Wayne State University, Michigan, 1990 (BHSIDOC # 357, Bicycle Helmet Safety Institute, Arlington VA. 1990).

Hoque M M, An analysis of fatal bicycle accidents in Victoria (Australia) with a special reference to nighttime accidents, Accident Analysis and Prevention, 22:1-11, 1990.

Hurt H H and Thom D R, Laboratory tests and accident performance of bicycle safety helmets, in Proceedings 29th Annual Conference, American Association for Automotive Medicine, AAAM, 1985.

Illingsworth C M, Naval D and Bell D, 150 bicycle injuries in children: a comparison with accidents due to other causes, Injury, 13:7-9, 1981.

Jamieson K G and Kelly D, Crash helmets reduce head injuries, Medical Journal of Australia, 2:806-809, 1973.

Kraus J F, Fife D and Conroy C, Incidence, severity and outcomes of brain injuries involving bicycles, American Journal of Public Health, 77:76-78, 1987.

Kedjidjian C B, Bicycle helmets save lives, Traffic Safety, 94(3):16-19, 1994.

Kroon P O, Bunketorp O and Romanus B, in Proceedings, International IRCOBI Conference on the Biomechanics of Impacts, IRCOBI, 1986.

Kruse D L McBeath A A, Bicycle accidents and injuries, a random survey of college population, American Journal of Sports Medicine, 8:342-344, 1980.

Lane J C, Helmets for child bicyclists: some biomedical considerations, Federal Office of Road Safety Report No. CR 47, Department of Transport, Canberra, November 1986.

Larsen L V, The epidemiology of bicyclists collision accidents, Journal of Traffic Medicine, 22:27-31, 1994.

Li G and Baker S P, Alcohol in fatally injured bicyclists, Accident Analysis and Prevention. 26:543-548, 1994.

Lighthall J W, Goshgarian H G and Pinderski C R, Characterisation of axonal injury produced by controlled cortical impact, Journal of Neurotrauma, 7:65-76, 1990.

Lissner H R, Lebow M and Evans F G, Experimental studies on the relation between acceleration and intracranial pressure changes in man, Surgery, Gynaecology and Obstetrics, 111 :329-338, 1960.

Long G J, Dowdell B, and Griffiths M, Development of a localised loading test for pedal cycle helmets, Crashlab, Road Safety Bureau, Roads and Traffic Authority of NSW (undated).

Maimaris C, Sommers C L, Browning C and Palmer C R, Injury patterns in cyclists attending an accident and emergency department: a comparison of helmet wearers and non wearers, British Medical Journal, 308: 1537-1540, 1994.

McDermott F T, Why pedal cyclists should wear safety helmets, Australian Family Physician, 13:284-285, 1984.

McDermott F T and Klug G L, Differences in head injuries of pedal cyclist and motor cyclist casualties in Victoria, Medical Journal of Australia, 2: 30-32, 1982.

McDermott F T and Lane J C, Protection afforded by cycle helmets (letter), British Medical Journal, 309:877, 1994.

McDermott F T, Lane J C, Brazenor G A and Debney A E, The effectiveness of bicyclist helmets: a study of 1710 casualties, Journal of Trauma, 34:834-844, 1993.

McIntosh A and Dowdell B, A field and laboratory study of the performance of pedal cycle helmets in real accidents, in Proceedings, International IRCOBI conference on the Biomechanics of Impact, 51-60, IRCOBI 1992.

McKenna S, Borman B and Flemingh H, Pedal cycle accidents, New Zealand Medical Journal, 97:657-658, 1984.

McSwain N E and Petrucelli E, Medical consequences of motorcycle helmet nonusage, Journal of Trauma, 24:233, 1984.

Melvin J W, Lighthall J W and Ueno K, Brain injury biomechanics, in Nahum A M and Melvin J W, Accidental Injury: Biomechanics and Prevention, Springer-Verlag 1993.

Mills N J, and Gilchrist A, The effectiveness of foams in bicycle and motor cycle helmets. Accident Analysis and Prevention,23:153-163, 1991.

Nakayama D K, Gardner M J and Rogers K D, Disability from bicycle related injuries in children Journal of Trauma, 30:1390-1394, 1990.

O'Connor P J, Road Injury in Australia, Road Injury Information Program, Report Series No.2, National Injury Surveillance Unit, Australian Institute of Health and Welfare. Canberra, 1993.

Ommaya A K and Gennarelli T A, Cerebral concussion and traumatic unconsciousness: correlation of experimental and clinical observations on blunt head injuries, Brain, 97:633-654, 1974.

Otte D, Injury mechanism and crash kinematic of cyclists in accidents: an analysis of real accidents, in Proceedings, 33rd Stapp Car Crash Conference, Society of Automotive Engineers, 1989.

Piggott P, Knuiman M W and Rosman D L, Rates and pattern of injuries from bicycle crashes based on population: police and hospital data, Rood and Transport Research, 3:64-74, 1994.

Pitt W R, Thomas S, Battistutta D, Nixon J, Clark R, and Acton C, Trends in head injuries among child bicyclists, British Medical Journal, 308:177-1994.

Rimel R N, Giordani M A, Barth J T, Boll T J and Jane J A, Disability caused by minor head injury, Neurosurgery, 9(3), 1981.

Rivara F P, Thompson D C and Thompson R S, The Seattle children's bicycle helmet campaign: changes in helmet use and head injury admissions, Paediatrics, 93:567-569, 1994.

Rivara F T, Traumatic deaths of children in the United States: currently available prevention strategies, Paediatrics, 75: 456-462, 1985.

Road Safety Bureau, Rood Traffic Accidents in NSW 1993, Statistical Statement: year ended 31 December 1993, Roads and Traffic Authority of New South Wales, 1994.

Sacks J J, Holmgreen P, Smith S M and Sosim D M, Bicycle associated head injuries and deaths in the United States from 1984 through 1988: how many are preventable? Journal of the American Medical Association, 266:3016-3018, 1991.

Smith T A, Teees D, Thom D R and Hurt H H, Evaluation and replication of impact damage to bicycle helmets, in Proceedings 37th Annual Conference, Association for the Advancement of Automotive Medicine, AAAM 1993.

Smith N C and Milthorpe F W, An observational survey of low compliance and helmet wearing by bicyclists in NSW, 1993, Transport and Network Development Branch, Roads and Traffic Authority NSW, September 1993.

Spaite D W, Murphy M, Criss E A, Valenzuela T D and Meislin H W, A prospective analysis of injury severity among helmeted and non-helmeted bicyclists involved in collisions with motor vehicles, Journal of Trauma, 31:1510-1516, 1991.

Staysafe 12, Bicycle Safety, Joint Standing Committee upon Road Safety, Parliament of NSW, 1988.

Thomas S, Acton C, Nixon J, Battistutta D, Pitt W R and Clark R, Effectiveness of bicycle helmets in preventing head injury in children: case control study, British Medical Journal. 308: 173-176. 1994.

Thompson R, Rivara F and Thompson D, A case control study of the effectiveness of bicycle safety helmets, New England Journal of Medicine, 320: 1361-1367, 1989.

Thompson D C, Thompson R S, Rivara F P and Wolf M E, A casecontrol study of the effectiveness of bicycle safety helmets in preventing facial injury, American Journal of Public Health. 80:1471-1474, 1990.

Thorson J, Pedal cycle accidents, Scandinavian Journal of Social Medicine, 2:121-126, 1974.

Travers Morgan Pty Ltd, Bicycle crashes in Western Australia, for the Federal Office of Road Safety, Report No. CR56, Department of Transport and Communications, Canberra. 1987.

Walker M B, Law compliance among cyclists in NSW, April 1992: a third survey, Network Efficiency Branch, Roads and Traffic Authority NSW, July 1992.

Walker M B, Law compliance and helmet use among cyclists in NSW, Consultant report CR6/90, Road Safety Bureau, Roads and Traffic Authority NSW, December 1990.

Watson G S, Zador P L and Wilks A, The repeal of helmet use laws and increased motorcycle mortality in the United States 1975-1978, American Journal of Public Health, 70:579, 1980.

Williams M, The protective performance of bicyclists' helmets in accidents, Accident Analysis and Prevention, 23: 119-131, 1991.

Wilson D C, The effectiveness of motorcycle helmets in preventing fatalities, report DOT HS807 416, National Highway Traffic Safety Administration, Washington DC, 1989.

Wood T and Milne P, Head injuries to pedal cyclists and the promotion of helmet use in Victoria, Australia, Accident Analysis and Prevention, 20:177-185, 1988.

For those who need to cite this study

The info at the top of the study on this page is actually the full publication information.

We received this study as a printed volume bound with GBC binding (the plastic spine type) from the Motor Accidents Authority of New South Wales, Australia. It is located in Sydney, Australia, and is a government agency. It should be cited as the report of a government agency.

The cover says:
The Effectiveness of Bicycle Helmets - A review
Revised Edition Prepared by Dr. Michael Henderson for the Motor Accidents Authority of NSW
Reorder Number - MAARE-010995
ISBN 0 7310 6435 6


There is no date, but the cover letter which accompanied it was dated September 26, 1995 and signed by Anne Deans, Rehabilitation Manager, Motor Accidents Authority of New South Wales, Level 12 139 Macquarie Street, Sydney, NSW 2000, Australia. Telephone (02) 252-4677. Fax (02) 252-4710. It was Ms. Deans who gave us permission to put it up on the Internet.

In our opinion this study qualifies as primary research even though it is in the form of a literature review. In fact the author digests and summarizes the relevant topics so that the literature review feature becomes just another form of footnoting, done this way because it would be too cumbersome if done as traditional footnotes.