Circumstances and Severity of Bicycle Injuries

Summary Report of Harborview Helmet Studies Sponsored by the Snell Memorial Foundation


Executive Summary

This is the largest study conducted to date on the circumstances of bicycle injuries and the protective effect of helmets. The purpose of this study was to address a number of previously unanswered questions about the effectiveness of bicycle helmets. Injured cyclists were recruited between March 1992 and August 1994 from seven Western Washington hospitals (Harborview Medical Center, Group Health Cooperative Central and Eastside hospitals, Children's Hospital and Medical Center, Mary Bridge Hospital, Overlake Hospital and Medical Center, University of Washington Medical Center), as were cases from two Medical Examiners offices (King County and Pierce County).


Of 3,854 bicyclists who were injured or died during this time period, 3,390 (88 percent) were recruited for this study. This project was based on a case-control design in which individuals with head or brain injuries (cases) were identified and compared to those who were involved in crashes but did not suffer head or brain injuries (controls). Data were collected by self-report questionnaires, abstraction of medical records, and, in some cases, examination of bicycle helmets and measurements of cyclists' heads.


Questionnaires completed by subjects included demographic inquiries, as well as questions on cycling experience, circumstances of the crash, severity of damage to the bicycle, ownership and use of helmets, and self-reported helmet fit. Slightly more than half (50.6 percent) of subjects wore helmets at the time of their crashes.

The study found significant evidence that bicycle helmets prevent head and brain injury. Corollary research questions were also successfully answered.

Major findings include:

Despite the overwhelming protectiveness of helmets, a few helmeted cyclists did suffer head injuries. This may be due to inadequate coverage by the helmet, improper wearing of the helmet because of poor fit or incorrect wearing behavior, movement of the helmet at the time of the crash, or crash forces that exceed the helmet's ultimate protective capacity.


As bicycling grows in popularity as a recreational activity and a means of transportation for adults and children, injuries continue to take a toll on cyclists. Each year crashes involving cyclists cause approximately 900 deaths, 23,000 hospital admissions, 580,000 emergency room visits, and 1.2 million visits to physician offices and clinics in the U.S. These statistics have challenged epidemiologists to gather information on cycling injuries and the effectiveness of methods for preventing them.

Bicycle helmets have been considered the single best means of protecting cyclists from the leading cause of injury and death: head injuries during crashes. Educational and legislative efforts have been successful in increasing the use of bicycle helmets.

Despite previous studies demonstrating that bicycle helmets are effective in preventing head and brain injuries, there remained significant questions about helmet efficacy. This study was designed to answer these questions by addressing the following aims:


This project was based on a case-control design, in which individuals (cases) with the outcome of interest (head injury from bicycling crashes) are compared to a control group (cyclists without head injuries from crashes). Unlike our prior study, no population-based control group was used.

Information was collected on possible differences between cases and controls (e.g., crash severity) that could obscure the central relationship between helmet use and head or brain injury, thus permitting needed adjustments between comparison groups through multivariate analysis. Analyses were also conducted on sub-groups (e.g., different age groups), different circumstances of the crash, and different helmet types.

Subjects were recruited from seven hospitals in Western Washington and from records of the King and Pierce County Medical Examiner's offices. The characteristics of the hospitals are as follows:

To identify injured cyclists, emergency room logs and treatment forms were examined once or twice a week between March 1, 1992 and August 31, 1994. All injured cyclists were eligible for the study. Information was entered into a computer database to track subjects throughout the study. Detailed questionnaires were sent to all subjects, and those who did not respond to the mailed questionnaire were telephoned about two weeks after the initial mailing.

The questionnaire included inquiries regarding demographic characteristics of subjects, cycling experience, severity of damage to the bicycle, ownership and use of helmets, and self-reported helmet fit. Information on injuries was gathered from emergency room, hospital and medical examiner's records.

Injuries were assessed using the Abbreviated Injury Scale (AIS) for injuries in individual body regions, and the Injury Severity Score (ISS) for overall measure of severity. A commercial computer program, TRI-CODE, was used to ensure consistent and accurate coding and injury severity scoring for data gathered from seven hospitals.

For the purposes of this study, head injury, brain injury and severe brain injury are defined as follows:

Study Population

During the 30-month study period, 3,854 eligible subjects were treated in the emergency rooms of the seven hospitals. This includes five subjects identified in medical examiner's records who died from bicycle-related injuries before arriving at an emergency room. Completed questionnaires and injury data were obtained for 3,390 subjects, an 88 percent study response rate. Of the 3,390 subjects, 1,718 (50.6 percent) were helmeted at the time of the crash. The authors believe this is the largest group of injured bicyclists ever to be the subject of a study of this nature.

The figures on these pages describe essential information on the subjects of this study. Subjects were most likely to be male, well educated, relatively affluent, and helmet owners.

More than two-thirds of the study population (72 percent) were male, and 43 percent were under 13 years of age (see Figure 1)new window). The subjects (or their parents, in the case of children) were an educated population, with nearly one-half (49 percent) having college degrees and nearly a quarter (24 percent) having post-graduate degrees (see Figure 2new window). Almost half (48 percent) had incomes of more than $35,000 annually (see Figure 3new window).

Nearly two-thirds (62 percent) of the cyclists reported that they bicycle daily. More than half of the adults (53 percent) rode more than five hours per week, and 37 percent rode more than 50 miles weekly.

Three-fourths of subjects (76 percent) reported they own bicycle helmets, with the rate of ownership lowest for teenagers (67 percent) and highest for cyclists older than 20 (79 percent) (see Figure 4new window). Slightly more than half (51 percent) were wearing helmets at the time time of the crash, with helmet use lowest (32 percent) among teens (see Figure 4new window).

Among helmeted cyclists, hard-shell helmets were most common (49 percent), followed by thin-shell helmets (29 percent) and helmets without shells (19 percent) (see Figure 5new window). When examined by testing standard, helmets were most commonly Snell-approved (54 percent) (see Figure 6new window).

The most common cause of crashes was loss of control by cyclists causing the cyclist to fall to the ground or hit an obstacle. Motor vehicles were involved in only 15 percent of crashes (see Figure 7new window). Crashes occurred most often (77 percent) while cyclists reported riding at speeds 15 mph or less (see Figure 8new window).

The majority of crashes occurred on paved streets (78 percent), followed by dirt, gravel and sand surfaces (see Figure 9new window). Almost half of the bicycles involved in crashes (43 percent) sustained some damage (see Figure 10new window).

The vast majority of injuries sustained by cyclists (92 percent) were in the Injury Severity Score (ISS) range of 0-8. An ISS of 9 or greater indicates moderate to severe injury. Only 6 percent of the injured cyclists with scores of 8 or less were hospitalized compared to 73 percent with scores of 9 or greater.

Facial injuries were most common, found in 34.8 percent of subjects. Head injuries were suffered by 22.3 percent of cyclists, and 6.0 percent received brain injuries (see Figure 11new window). About a tenth (9.7 percent) required hospital admission, nine subjects died in hospitals, and five died before transport to a hospital.

Overall Effectiveness of Helmets

Although prior studies have found that bicycle helmets are effective in reducing head and brain injuries, there have remained several significant questions about helmet efficacy. The authors of this study sought to remedy these deficiencies in understanding the overall effectiveness of bicycle helmets in injury prevention. This study was designed to provide information on helmet effectiveness in four age groups (less than 6 years old, 6-12, 13-19, 20 and older), to evaluate the different helmet types and engineering standards, and to determine helmet effectiveness for crashes involving motor vehicles.


Because of small sample sizes, no previous study has evaluated the effectiveness of helmets in different age groups, particularly children under six years old.

In this study, analysis of the protective effect of helmet use for risk of head injury and brain injury show substantial efficacy in all age groups. Overall, helmets were found to prevent 69 percent of head injuries, 65 percent of brain injuries, and 74 percent of severe brain injuries (see Figure 12new window), with no significant difference in the protective effect for any age group (see Figure 13new window). While the protective effect appeared to be somewhat lower in teenagers, this was not significant. These results are the same as those obtained in that portion of our 1989 study with emergency room controls. As previously mentioned, had it been possible to employ population controls in the present work (e.g., crashing cyclists regardless of medical attention), comparably higher levels of protection would have been obtained.

Of 62 bicyclists with severe brain injuries, only 24 percent were helmeted (15 riders out of the 1,718 who were helmeted), compared to the 57 percent rate of helmet use by the control group (riders with injuries other than head trauma). No helmeted bicyclists who sustained crashes in the youngest group (under 6 years old) and the 13-19 group suffered a severe brain injury. This again demonstrates the excellent protective effect of bicycle helmets.

Helmet Types and Standards

Previous studies have evaluated hard-shell helmets, the only type in wide use at that time. The growing popularity of no-shell and thin-shell helmets suggested the need to examine the effectiveness of these models, as well. There are three helmet standards in the U.S.: Snell Memorial Foundation (Snell), American National Standards Institute (ANSI), and American Society for Testing and Materials (ASTM). No real-world field data have compared the relative effectiveness of the three standards in protecting against head and brain injury.

Of the helmet types used by cyclists in this study, 49 percent were hard-shell, 29 percent where thin-shell, and 19 percent were no-shell (see Figure 5new window). Over half the helmets (54 percent) were Snell- certified, 28 percent had ANSI certification and 0.4 percent had ASTM certification (see Figure 6new window). Since Snell certification is most stringent, helmets meeting this standard would also meet ANSI and ASTM standards.

The protective effect of different helmet types is shown in Figure 14. The protective effect of hard- shell helmets for brain injuries was 73 percent compared to the 58-59 percent for other types. The protective effect against severe brain injuries was 83 percent compared to 70 percent for thin-shell and

Snell and ANSI approved helmets provided similar protection against head and brain injuries. However, Snell helmets decreased the risk of severe brain injuries by 81 percent and ANSI helmets by 72 percent. In order for this difference to be statistically significant the number of people in the study would have had to be 53 times greater.


Motor Vehicle Involvement

Some observers have questioned the effectiveness of helmets in protecting the head against collisions between bicycles and motor vehicles. No previous study has systematically examined this issue.

Of the 518 bicyclists who were involved in motor vehicles crashes, 42 percent were helmeted. A similar level of protection against head injury was found after comparing for the protective effects of helmets in crashes with and without motor vehicle involvement (see Figure 15new window).



Results of this study indicate that helmets are effective for all bicyclists, regardless of age. The levels of protection are the same as were delineated using emergency department controls in our 1989 study. There is no evidence that the youngest children need a different type of helmet since the level of protection in this age group is similar to other ages. Current energy-management capabilities of helmets are effective for all ages.

Helmets are effective in preventing head and brain injuries in all types of crashes, including those involving motor vehicles. The three types of helmets (hard-, thin- and no-shell) offer approximately the same degree of protection. Hard-shell and Snell-certified helmets may provide more protection against severe head injuries. However, since only 15 helmeted riders had severe brain injuries, we were unable to show a statistical difference between the helmet types and certification standards in this study.

Bicycle Helmet Effectiveness in Preventing Serious Facial Injuries

While helmets have been widely recommended as protection against head injuries, questions have remained about the effectiveness of these helmets in the prevention of facial injuries. Unlike motorcycle helmets, which usually have face pieces, bicycle helmets lack this protective feature. As part of the overall study of bicycle helmet effectiveness, we also collected data on facial injuries.

Facial injury was defined as any injury of the jaw, lips, cheeks, nose, ears, eyes, forehead and mouth. Only serious facial injuries--fractures and lacerations--were considered for this study. Minor facial injuries that were found when cyclists were treated for other injuries (e.g., head injuries or broken legs) were not counted. Facial injuries were categorized as occurring in three regions: the lower face (lips, mouth and lower jaw), the middle face (nose and cheeks) and upper face (forehead, orbit, eyes and ears).

Of the 3,388 injured bicyclists analyzed in this study, 34.8 percent had facial injuries, 700 (20.7 percent) of which had serious facial injuries. Lacerations were nine times more frequent than fractures, occurring most often to the chin, lip and forehead (see Figure 16new window). The most common fractures were to the nose and mandible, each occurring in about 4 percent of facial-injury patients. Children aged 5 to 12 were most likely to suffer facial injuries, as 51 percent of all facial injuries occurred to cyclists in this age group, compared to 38 percent of other injuries.

By comparing the injuries suffered by helmeted cyclists with the injuries to those who were unprotected, it was found that helmets reduce the overall risk of serious facial injuries by 50 percent. Helmets were most effective in preventing injuries to the upper and middle facial regions, reducing(see Figure17new window).

This is the first study to demonstrate clearly that bicycle helmets provide protection to the midface region as well as the upper face.

Given the vulnerability of the lower face to serious injury in a bicycle crash, consideration should be given to adding a face bar or chin covering to the present design of bicycle helmets. Injuries in this area can be particularly disfiguring, lending an urgency to the need to develop such a protective modification.

Helmet Fit and the Risk of Head Injury

Studies of the effectiveness of bicyle helmets, including this one, have reported that a number of helmeted cyclists have sustained head injuries. This study found that 29 percent of cyclists who sustain head injuries and 30.5 percent of those who suffer brain injuries were wearing helmets at the time of the crash. In seeking a better understanding of why helmeted riders suffer these injuries, the relationship between helmet fit and the risk of injury was also examined.

Helmet fit was assessed by asking cyclists (or their parents for children younger than 14 years old) to report on helmet snugness, positioning on the head, custom fitting using adjustable foam pads, comfort, adjustability of straps, whether the helmet covered the forehead, and whether the helmet could be removed while the strap was fastened.

Factors in poor helmet fit included helmets worn too far to the back of the head instead of fitting on the center of the head. Cyclists whose helmets came off during a crash were three times more likely to have a head injury compared to those cyclists whose helmets were snugly fastened at the time of the crash.

Helmet fit proved to correlate closely with the degree of protection afforded by helmets during a crash. Overall, a clear dose-response relationship between self-reported fit and head injury emerged. Cyclists who reported that their helmets fit poorly were nearly twice as likely to suffer head injury than cyclists whose helmets fit the best. Cyclists reporting good and fair helmet fits received relatively less protection (see Figure 18).

Self-reported fit appears to be a valid measure of helmet fit and one that has easy applicability for assessing community programs to promote helmet use. There is a strong suggestion that fit, or lack thereof, may contribute to increased risk of head injury by a factor of two. However, since riders or parents were asked about fit after the injury that brought them to the emergency room, it is possible that those with head injuries may have had different recall of fit than those without such injuries. This phenomenon, known as recall bias, may partially explain these results.

In another portion of our study assessments of helmet fit were made by expert study personnel following a standard protocol. Their observations were compared to parent- and self-report of helmet fit. In general, the results indicated that many parents did not have an adequate understanding of "good fit." Based on this empiric experience, we have developed information describing how to test a helmet for proper fit.

Damage to Bicycle Helmets Involved in Crashes

Helmet design, in addition to improper fit, may play a role in head and brain injuries sustained by cyclists who are wearing helmets at the time of the crash. Energy exceeding the threshold of protection and impact occurring outside the covered area of the head are potential reasons for a lack of complete protection.

In this study, helmets were purchased from cyclists who met any of the following criteria: helmeted cyclists who sustained head injury; helmeted cyclists who reported that they hit their head at the time of the crash; and cyclists who reported that their helmets were visibly damaged in the crash.

A total of 527 helmets were forwarded to the Snell Memorial Foundation for laboratory testing. Examiners were blinded to all information on circumstances of the crash and injuries to cyclists. Damage was scored according to the site (i.e., within an inch of the edge, middle or top of helmet) and severity.

The degree of damage was scored as follows:

Score = 0: No visible damage related to crash (40 percent of helmets);

 Score = 1: Minimal damage, but none to liner (20 percent);

 Score = 2: Moderate damage, with evidence of energy attenuation to liner (18 percent);

 Score = 3: Major damage, with more than minimal compression of liner (14 percent);

 Score = 4: Catastrophic damage, in which integrity of shell is lost (7 percent).

Damage occurred most frequently to the helmet front (47.1 percent of damaged helmets) and sides (30.4 percent). There was no difference between frequency of damage to the right and left sides of the helmets, and little damage occurred on the top or in the back of helmets. Damage to the edge of the liner accounted for nearly half of all impacts to the helmets (see Figure 19new window).

Damage score of helmets was found to be associated with the risk of head and brain injury to cyclists. Cyclists with catastrophic helmet damage were five times more likely to have suffered injury to the head (including the scalp, forehead, skull or brain) and nearly eleven more times as likely to have sustained a brain injury (i.e., concussion or worse) (see Figure 20 and Figure 21new window).

A relatively large proportion of helmets examined (39 percent) had damage to the liner. Damaged liners may not offer the same protection as new helmets, and many cyclists may be unaware that damage to the liner has occurred. For these reasons, the offer by many manufacturers of free replacement for helmets that have been involved in crashes appears to be appropriate.

Despite the association between the degree of helmet damage and the risk of head and brain injury, it should not be assumed that a causal relationship exists. Cyclists who have been in high- energy crashes generally have both head injuries and helmet damage. It's quite possible that injury to the head or brain would have been far worse for unhelmeted cyclists involved in the identical crashes. The association does suggest, however, that for certain types of crashes the helmet's energy-absorbing threshold has been exceeded.

Location of helmet damage (primarily to the front and at the edge) suggests that manufacturers should consider building an extra energy-absorbing capacity in the front, designing the helmet to fit as close as possible to the top of the eyes, and improve the retention system to prevent the helmet from rotating back.

The large number of injuries to the forehead suggest that either helmet design provides inadequate coverage or that the helmet is being worn improperly (see "Helmet Fit and the Risk of Head Injury"). The frequency of these injuries indicates the need for further investigation of these issues.

Risk Factors for Serious Injury

Helmet use is not the only factor involved in whether a cyclist is vulnerable to serious injuries. By analyzing the types and severity of all injuries suffered by bicycle riders in the study group, researchers were able to make suggestions on safety measures that may decrease the number of the more serious injuries.


Of the injured cyclists, 52.1 percent sustained one or two injuries, 37.1 percent had three to five injuries, and 10.8 percent had more than five. Injuries to the upper extremities were most common (suffered by 59.6 percent of injured cyclists) followed by injuries to the lower extremities (46.9 percent). Slightly more than one-fifth of the injured cyclists sustained head injuries, and one-third had facial injuries (see Figure 22new window). Neck injuries were quite infrequent.

Injured cyclists most commonly had abrasions, lacerations and contusions, while one-fourth of the study group suffered fractures. Brain injuries (defined as concussion or more severe brain injury) occurred to 6.0 percent of riders. Injuries to internal organs and to blood vessels and nerves were un-common.

Risk Factors

Motor vehicles were involved in 15.3 percent of all crashes. Riders crashed most often after losing control of the bicycle, then hitting the ground or an obstacle (see Figure 7new window).

Researchers correlated circumstances of crashes with injury severity (ISS) to determine the importance of various risk factors. Collisions with motor vehicles increased the risk of severe injury (ISS>8) by 360 percent and markedly increased the risk of fatal injury. Riding at speeds greater than 15 mph increased the risk of severe injury by 40 percent. Children under the age of 10 were most likely to sustain injuries to the head and face, while teenagers and young adults were more apt to suffer injuries to the extremities.

Cyclists who sustained neck injuries (2.7 percent of the study group) tended to be more severely injured, with 22.3 percent having ISS>8, compared to 6.4 percent of cyclists without neck injuries. Neck injuries included sprains, cervical spine fractures, and nerve-cord injuries. There was no correlation between neck injury and helmet use or helmet type.

Of 14 fatal injuries, 10 were suffered by cyclists hit by motor vehicles, and only one, a 6-year-old child crushed by a truck, was helmeted.


Other Potential Safety Measures

The risk of crash involvement with motor vehicles may argue for more bicycle paths, but risk remains at points of intersection between bicycle paths and roads. Injuries suffered by younger cyclists suggest that some children may be riding before they are developmentally ready, that their bicycles may not fit their size, or that the sites they ride may be unsafe or poorly supervised.

Injuries occurring to cyclists riding at greater speeds may argue for separate facilities for these riders or for a version of the leather clothing worn by motorcyclists or the neoprene suits worn by ski racers. The number of facial injuries suffered by such high-risk groups as children and young adults suggests the need for additional facial protection on helmets for these cyclists. Fractures and dislocations to the extremities dictates study of the effectiveness of elbow and knee pads and wrist guards.

The number and severity of injuries to the face and body indicate that bicycle helmets alone are not sufficient to prevent injury to cyclists. Multiple approaches to intervention, including educational programs, product modification and regulation, are clearly warranted as strategies in the overall cycling-safety effort.


Overall Conclusions

This study provides powerful statistical evidence that bicycle helmets, regardless of type, provide protection to cyclists in all circumstances, including crashes involving motor vehicles. Our research clearly indicates that bicycle helmets should be worn by all riders, regardless of their age, experience as cyclists, the distance they plan to ride, or any other factor.

Educational programs can increase the rate of helmet use dramatically, yet there is national and international evidence that the rate of compliance plateaus when helmet use remains voluntary. A legislative approach, similar to the mandate that motor vehicle drivers wear seat belts, would appear to be the most promising next step.

Bicycle helmets cannot protect riders against all trauma, as the findings on neck injuries and extremity fractures, abrasions and contusions indicate. Environmental changes, such as safer roads and separate bike lanes, should be explored as an additional means of reducing injuries to cyclists.

This study of 3,390 injured cyclists, the largest undertaken to date, produced a wealth of data and the inevitable conclusion that bicycle helmets are the single most important protection against head injury and brain injury. Other issues that arose during the course of the study may inspire others to proceed with further investigations. The design of this study can be readily adapted to the evaluation of other questions regarding bicycle injuries.


Research Staff

Harborview Injury Prevention and Research Center
Principal Investigator: Frederick P. Rivara, MD, MPH; Co-Investigators: Diane C. Thompson, MS; Randy Ching, Ph.D.; Martha Nunn, DDS, Ph.D.; Co-Investigators: Sterling Clarren, MD; Susan Astley, Ph.D. (CHMC); Project Coordinators: Alexandra Kaufman, Viviana Rebolledo; Research Assistants: Mitchell Rauh, Jeff Goode, Nichole Walters, Christine Thompson, Linley Vance, Allegra Kim, Jennifer Winder, Sharon Triplett, Christine Greenmun, Jacqueline Mann, Lynda Emel, Nan Lopez, Beatrice Kaufman, Mike Neathery, James Swantz, Bruce Alexander; Interviewers: Matthew Farley, Dennis Bejin, Elizabeth Gaskill, Rosalie Ginnett, Shannon Greer, Jean Purkhiser, Kathy Vickers; Computer Programmer: Bob Soderberg; Administrator: Noreen Schedler; Secretary: Millicent Morrow
Group Health Cooperative of Puget Sound

Co-Principal Investigator: Robert S. Thompson, MD; Research Staff: Esther Normand, Mary Sunderland, Kathryn Reiss, Tom Thompson, Bob Hartl, Gracie Melton-Holman


Snell Memorial Foundation
Daniel J. Thomas, MD, MPH; William C. Chilcott, Ph.D.; Edward Becker, MS; Frank Lin, Ph.D.


Acknowledgment of Participating Institutions

Harborview Medical Center
 Medical Director, Emergency Services: Michael Copass, MD; Emergency Department staff: Chris Martin, RN; Mary Royce, RN; "Ocie" West; Trauma Registry: Susan Pilcher, RN
Children's Hospital and Medical Center
Chief, Emergency Services: Linda Quan, MD; Medical Records: Carol Bettis, Bill Michaelson


University of Washington Medical Center
Associate Director, Emergency Medical Services: Richard Cummins, MD; Medical Records staff


Overlake Hospital and Medical Center
Medical Director, Emergency Services: John Ciliberti, MD; Clinical Manager, Emergency Department: Beth Reid, RN; Medical Records staff
Mary Bridge Children's Hospital and Medical Center
Medical Director, Emergency Services: Ted Walkley, MD; Clinical Manager, Emergency Department: Eileen Newton, RN: Medical Records: Dawn Watson


King County Medical Examiner's Office
 Pierce County Medical Examiner's Office

Funding for this study was provided by a grant from the Snell Memorial Foundation.


For additional copies, contact:

Snell Memorial Foundation or
Tel: 916.331.5073
Fax: 916.331.0359

Editor: Larry Zalin
Designer: Gina Davidson

© 1996 Snell Memorial Foundation. This report may not be reproduced without permission from the Snell Memorial Foundation.