Bicycle-related head injury: a study of 86 cases

Accident Analysis and Prevention 36 (2004) 561 567 Bicycle-related head injury: a study of 86 cases Bart Depreitere a,, Carl Van Lierde b, Sigrid Maene b, Christiaan Plets a, Jos Vander Sloten b, Remy Van Audekercke b, Georges Van der Perre b, Jan Goffin a a Department of Neurosurgery, University Hospital Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium b Division of Biomechanics and Engineering Design, Catholic University of Leuven, Celestijnenlaan 200A, B-3000 Leuven, Belgium Received 8 August 2002; received in revised form 12 March 2003; accepted 27 March 2003 Abstract Within the framework of a bicycle helmet research program, we have set up a database of bicycle accident victims, containing both accident and clinical data. The database consists of a consecutive series of 86 victims of bicycle accidents who underwent a neurosurgical intervention in our hospital between 1990 and 2000. Data were obtained from police files, medical records, computed tomography head scans and a patient questionnaire. In only three victims, the wearing of a helmet was documented. In this study, the head injuries are analysed and the relation between the different types of head injuries and outcome is assessed. Forty-four accidents were collisions with a motor vehicle and 42 accidents were falls. Most impacts occurred at the side (57%) or at the front (27%) of the head. The most frequent injuries were skull fractures (86%) and cerebral contusions (73%). Age was negatively correlated with outcome (P = 0.0002) and positively correlated with the number (P = 0.00002) and volume (P = 0.00005) of contusions and the presence of subdural haematomas (P = 0.0004). The injuries with the strongest negative effect on outcome were: subarachnoid haemorrhage (P = 0.000001), multiple (P = 0.000005) or large (P = 0.0007) contusions, subdural haematoma (P = 0.001) and brain swelling (P = 0.002). A significant coexistence of these four injuries was found. We hypothesise that in many patients the contusions may have been the primary injuries of this complex and should therefore be considered as a main injury determining outcome in this study. We believe that such findings may support a rational approach to optimising pedal cyclist head protection. 2003 Elsevier Ltd. All rights reserved. Keywords: Bicycle accident; Head impact; Head injury; Outcome 1. Introduction Victims of bicycle accidents are prone to head injury. Twenty-one to 61% of the victims of bicycle accidents seeking medical care have a head injury (Collins et al., 1993; Eilert-Petersson and Schelp, 1997; Elsen et al., 1997; Fife et al., 1983; Wood and Milne, 1988). Moreover, head injury is the cause of death in 69 93% of fatal bicycle accidents (Elsen et al., 1997; Fife et al., 1983; Guichon and Myles, 1975; Oström et al., 1993; Wood and Milne, 1988). Bicycle accidents are not uncommon in Belgium, where pedal cycling always has been popular. On a random day in 1999 9.6% of the questioned Belgians made a ride on their bikes and the bicycle comprised 7.3% of the transport means to work and 19.2% of the transport means to school (NIS, 2000). Moreover, pedal cycling as a recreational sport has been gaining more and more popularity. This results in a relatively high number of bicycle accidents: in 2000, 6655 Corresponding author. Tel.: +32-16-34-42-90; fax: +32-16-34-42-85. E-mail address: bart.depreitere@uz.kuleuven.ac.be (B. Depreitere). injured cyclists and 134 cycling fatalities were counted on public roads for a population of 10 million Belgians. As such, pedal cyclists comprised 9.8% of all road traffic accident victims and 9.1% of road traffic deaths (NIS, 2002). This is a high number compared with only 728 pedal cyclist deaths in the United States in 2001, representing 1.7% of road traffic fatalities (National Highway Traffic Safety Administration, 2002). The wearing of a bicycle safety helmet is not mandatory in Belgium, except in cycling competition, and only rarely helmeted cyclists are seen in the streets. However, the authorities recently launched a promotion campaign and helmet use has also been promoted in media and sports events. The protective effect of bicycle helmets has been demonstrated in numerous epidemiological studies (Dorsch et al., 1987; Maimaris et al., 1994; McDermott et al., 1993; Thomas et al., 1994; Thompson et al., 1989). Other authors, however, have outlined that the currently used helmets are still susceptible to improvement (Ching et al., 1997; Gilchrist and Mills, 1996; McIntosh et al., 1995; Williams, 1991). An important shortcoming in the current helmet standards 0001-4575/$ see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016/s0001-4575(03)00062-9

562 B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 is that linear head acceleration is the only injury mechanism being considered, while it has been demonstrated by several investigators in the field that the occurrence of certain head injuries is correlated with other mechanical parameters rather than linear acceleration. Consequently, tolerance criteria which relate the occurrence of all brain injuries exclusively to translational head motions have been criticised for a long time (Hirsch and Ommaya, 1970). Unfortunately, precise lesion-specific tolerance criteria for head injury in humans do not exist. Such criteria, in combination with the knowledge of the relative frequencies of the different types of traumatic brain lesions for each situation where head protection is desirable and information on the importance of the different brain lesions towards outcome, would allow for a systematic approach to improving protective headgear. In the context of a multidisciplinary bicycle helmet research program, we have set up a database of bicycle accidents drawn from our patient population. We have collected data on the accident circumstances, the sustained injuries and outcome. The goal of this study was to gain more information on the mechanical input on the head in bicycle accidents and to perform a profound analysis of the resultant skull and brain injuries. For this purpose, accident reconstruction techniques have been developed and selected cases have been simulated (described in detail in a separate manuscript, in preparation). In the present report, we describe the head injuries sustained by these 86 pedal cyclists with serious head injury. Moreover, we have investigated the relation between the different types of head injuries and outcome, in order to assess which would be the most important lesions to protect against if lesion-specific protection were possible. 2. Methods Between January 1990 and June 2000 86 pedal cyclists underwent a neurosurgical intervention for head injury in the University Hospital of Leuven. The medical records were reviewed and the computer tomography (CT)-scans of the head and the skull X-rays were studied. All possible traumatic skull and brain lesions were studied. Contact loading can lead to skull fractures and cerebral contusions, i.e. haemorrhagic lesions of the brain surface. Contusions of the frontal and temporal lobes may also arise from inertial loading, due to the relative motion of the brain inside the skull. Inertial loading, when provoking high shear stresses within the brain, may produce diffuse axonal injury, characterized by diffuse damage to the brain parenchyma. Most extradural haematomas, which are blood clots between the skull and outer brain membrane (dura mater), are secondary to a skull fracture lacerating a meningeal artery. Acute subdural haematomas, between the outer (dura mater) and middle brain membrane (arachnoid) can be secondary to an underlying contusion, the rupture of a bridging vein traversing the subdural space or the laceration of a superficial cortical blood vessel. Intracerebral haematomas occur when a parenchymal haemorrhagic lesion turns into a blood clot of a certain volume. Most intracerebral haematomas arise from contusions, but they may also be situated in the deep brain nuclei (basal ganglia). As a reaction to a neighbouring lesion or secondary to the trauma itself the brain parenchyma may swell. The exact mechanical mechanism leading to bleeding in the brain ventricles (intraventricular haemorrhage) and in the space between the arachnoid and the inner brain membrane (subarachnoid haemorrhage) is not known. For a precise classification and measurement of the lesions the following rules were applied. The volume of parenchymal haemorrhagic lesions was calculated using a formula by Broderick et al. (1994). If the haemorrhagic lesion appeared as a homogeneous hyperdense mass with a volume of 10 ml or more, it was considered an intracerebral haematoma. The maximum volume of contusions and intracerebral haematomas was defined as the largest volume on the consecutive series of scans and could only be measured when all CT-scans were available. For the diagnosis of diffuse axonal injury two criteria were used: the typical presentation on magnetic resonance images or, when the latter investigation was not performed, the presence of deeply situated haemorrhagic lesions on the CT-scan, as described by Zimmerman et al. (1978). The diagnosis of brain swelling was restricted to the swelling of one entire or both entire hemispheres. A questionnaire asking for details on the circumstances of the bicycle accident was sent to all patients or their relatives. The questionnaire was approved by the ethical committee of our hospital. For survivors, the questionnaire also contained a translated version of the postal form for determination of the Extended Glasgow Outcome Score (GOSE) (Wilson et al., 1998). This is an internationally recognised outcome score for head injury ranging from one (dead) to eight (complete recovery). The questionnaire was sent back in 59 cases. We got permission from the attorney in three districts to study the available police files on the accidents. Fifty-three police files could be studied. 3. Results 3.1. Accident and patient characteristics The age of the pedal cyclists ranged from 7 to 82 years with a mean of 40 years. The male/female ratio was 72/14. The age distribution is shown in Fig. 1. Two peaks can be seen, a narrow peak with teenagers and a broad peak with middle-aged to elderly people. Two types of accidents were considered depending on the involvement of a motor vehicle. In 44 accidents, a motor vehicle was involved. Patients who were struck by a motor vehicle were significantly younger than victims of an accident without motor vehicle (mean ages 34.5 and 45.9, respectively, t-test: P = 0.014). Seventy-five percent of the

B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 563 Fig. 1. Histogram of victim age (N = 86). Dark bars represent victims of a motor vehicle collision (N = 44). Light bars represent victims of a bicycle accident without involvement of a motor vehicle (N = 42). children had a collision with a motor vehicle. The first documented Glasgow Coma Score after resuscitation was known in 81 patients and ranged from 3 to 15 with a mean of 9.9. 3.2. Head injuries The frequency of the different types of injuries is summarised in Table 1. The relation between the different types of injuries and age, gender and type of accident (involvement of a motor vehicle) is also shown in Table 1 by means of the P-value from Student s t-tests (for age) and Fisher exact (FE)-tests (for gender and accident type). Since we had the impression from the analysis that cerebral contusions, acute subdural haematomas, subarachnoid haemorrhage and brain swelling often appeared in combi- Table 2 Coexistence of injuries, expressed as P-values from Fisher exact (FE)-tests Contusion Subarachnoid haemorrhage Brain swelling Acute subdural haematoma 0.0021 0.0003 0.013 Contusion 0.042 0.0001 Subarachnoid haemorrhage 0.0001 nation, FE-tests were performed to examine the association of these injuries (Table 2). All associations were significant. Here, a Bonferroni correction was not carried out because of the inherent interdependence of the significance tests. In particular, the relation between subdural haematomas and contusions appeared to be close. Thirty-one out of 36 subdural haematomas had an underlying contusion. In 12 Table 1 Frequency of head injuries and P-values from t-tests for the relation between age and injuries and from Fisher exact (FE)-tests for the relation between gender or accident type and injuries Injury Total group (N = 86) No motor vehicle involved (N = 42) Collision with motor vehicle (N = 44) Age (t-test) Gender (FE-test) Accident type (FE-test) Skull fracture 74 (86.0%) 39 (92.9%) 35 (79.6%) 0.25 0.38 0.07 Extradural haematoma 34 (39.5%) 17 (40.5%) 17 (38.6%) 0.0045 a 0.78 0.52 Acute subdural haematoma 29 (33.7%) 15 (35.7%) 14 (31.8%) 0.00037 b 0.29 0.44 Contusion 63 (73.2%) 32 (76.2%) 31 (70.5%) 0.000036 b 0.14 0.20 Intracerebral haematoma 15 (17.4%) 11 (26.2%) 4 (9.1%) 0.0036 b 0.06 0.035 d (arising from contusion) Basal ganglia intracerebral 1 (1.2%) 0 (0%) 1 (2.3%) haematoma Gliding contusion 5 (5.8%) 0 (0%) 5 (11.4%) 0.053 0.31 0.031 c Diffuse axonal injury 11 (12.8%) 3 (7.1%) 8 (18.2%) 0.25 0.85 0.11 c Subarachnoid haemorrhage 45 (52.3%) 21 (50.0%) 24 (54.5%) 0.012 b 0.44 0.58 Intraventricular haemorrhage 11 (12.8%) 3 (7.1%) 8 (18.2%) 0.0007 b 0.12 0.11 c Brain swelling 66/80 (82.5%) 28/38 (73.7%) 38/42 (90.5%) 0.14 0.73 0.046 c Significant P-values (P <0.005: Bonferroni correction) are given in bold. a Population with injury is younger. b Population with injury is older. c Injury more frequent in bicycle accidents with a motor vehicle. d Injury more frequent in bicycle accidents without involvement of a motor vehicle.

564 B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 patients, the cause of bleeding of the subdural haematoma operated for was described in the surgeon s report: a contusion in seven, a torn bridging vein in two, the combination of a torn bridging vein and a contusion in one, a lacerated cortical vessel in two. Forty-six out of 192 contusions (24.0%) had an overlying subdural haematoma. The majority of contusions were located in the frontal lobes (55.2%) and temporal lobes (34.4%). Patients that developed brain swelling appeared to have more (t-test: P = 0.0066) and larger (t-test: P = 0.012) contusions. 3.3. Age and head injuries As demonstrated in Table 1, analysis by Student s t-tests showed that patients with extradural haematomas were generally younger and that patients with acute subdural haematomas, contusions, contusions evolving to intracerebral haematomas and intraventricular haemorrhage were older. Moreover, we found a positive correlation between age and the number of contusions (Pearson r = 0.44) and between age and the maximum total volume of contusions (Pearson r = 0.51). These relations were re-investigated in the type of accident subgroups, since age and type of accident were not independent variables. In the subgroup of accidents without motor vehicle, which was the group with a more spreaded age distribution, all relations were confirmed, except for intraventricular bleeding. However, only three patients had intraventricular bleeding in this subgroup. In the subgroup of accidents with a motor vehicle, eight patients had intraventricular bleeding and the relation with higher age was confirmed. Although t-testing revealed no significant age difference between patients with and without diffuse axonal injury in the subgroup of victims struck by a motor vehicle, six of the eight patients with diffuse axonal injury in this subgroup were 16 years or younger. Twenty-seven of the 37 extradural haematomas were overlying the temporal and/or parietal lobe. Twenty-five of these had an overlying parietal/temporal fracture. When patients with a parietal/temporal fracture were considered separately, the relation between extradural haematoma and younger age was maintained (t-test, P = 0.0007). 3.4. Outcome The interval between the time of the accident and the time the outcome was evaluated ranged from 7 months to 10 years. The GOSE could be obtained in 69 cases, but three were excluded from the analysis: one because of a pre-trauma mental handicap and two because of the outcome being predominantly affected by extracranial injuries. The mean GOSE was 4.5 (Fig. 2). Mortality was 33.3%. We found no relation between the involvement of a motor vehicle in the accident and the GOSE (Mann Whitney U-test: P = 0.60). A significant negative correlation was found between age and the GOSE (Spearman r = 0.50). Fig. 2. Histogram of the Extended Glasgow Outcome Score (N = 66). For each type of injury the effect on the GOSE was determined. The injuries with a significant effect are presented in order of importance (with the P-value of the Mann Whitney U-test and with their Spearman correlation coefficient): subarachnoid haemorrhage (P = 0.000001; r = 0.62), acute subdural haematoma (P = 0.001; r = 0.41), brain swelling (P = 0.002; r = 0.41), intraventricular haemorrhage (P = 0.004; r = 0.36), intracerebral haematoma (P = 0.010; r = 0.32) and contusion (P = 0.025; r = 0.28). In addition, a significant negative correlation was found between the number of contusions and the GOSE (Spearman r = 0.53; P = 0.000005) and between the total maximum volume of contusions and the GOSE (Spearman r = 0.50; P = 0.00066). To examine the independence of age and the above-mentioned injuries in their relation to outcome, a multiple regression analysis was carried out with all the relevant injuries and age as regressors and with the GOSE as dependent factor. This produced a significant model, but none of the regressors had a significant partial correlation coefficient. 3.5. Bicycle helmet use The wearing of a bicycle helmet was documented in only three victims. All three had a combination of severe injuries: skull fractures, subdural haematoma, subarachnoid haemorrhage, contusions and brain swelling in all three, an intracerebral haematoma in one and an extradural haematoma in one. All three patients died. Their ages were 50, 65 and 71. In one patient, a bystander reported that the helmet had shifted backwards at the moment of the impact on the forehead. In 56 patients, it was documented that they were not wearing a helmet at the moment of the accident. In the remaining 27 cases, the questionnaire was not returned and no information on the wearing of a bicycle helmet was available. 3.6. Location of head impacts It was tried to deduce the location of head impacts on the basis of the location of soft tissue swelling on the CT-scan, documented scalp injuries and depressed skull fractures. In

B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 565 present analysis is a preparatory study aimed at gaining insight into the craniocerebral lesions resulting from bicycle accidents. 4.2. Craniocerebral lesions in bicycle accidents Fig. 3. Impact sites for victims that sustained a single impact (N = 49). F: frontal; T: temporal; FT: frontotemporal; P: parietal; O: occipital. 63 patients, we were more or less certain of the impact site. In the group of accidents without involvement of a motor vehicle 26 patients had a single impact and 4 had a double impact, whereas in the group of collisions with a motor vehicle 23 had a single impact, 8 a double and 2 a triple impact. Most patients had the impact on the forehead or on the side of the head. The impact sites of patients that are believed to have sustained a single impact are shown in Fig. 3. 4. Discussion 4.1. Methodology Some methodological considerations need to be addressed. This is a study of head-injured pedal cyclists as they were seen in our neurosurgical practice. Only patients that underwent a neurosurgical intervention were included, in order to obtain a more or less uniform group of serious bicycle-related head injury. Our department covers a geographical area of about 45 km diameter for head-injured patients in need of neurosurgical care. However, also in patient referral a selection bias will inevitably have occurred. Therefore, epidemiological conclusions, such as lesion incidence in head injured pedal cyclists, are not to be drawn from this study. The presence of diffuse axonal injury may be slightly underreported in this series, because no autopsies were carried out and magnetic resonance-scans were not systematically done. The clinical manifestations of diffuse axonal injury are not specific. The combined clinical radiological criterion given by Gennarelli coma for more than 24 h without mass lesions (Gennarelli, 1983) was of no use, since it is a policy at our centre to have patients at risk for brain swelling ventilated and sedated for at least some days. It would certainly be interesting to study head injuries in helmeted victims of bicycle accidents if one wants to investigate how bicycle safety helmets can be improved. Unfortunately, this was not possible in our situation. However, it is our opinion that the best approach to improve head protection is the study of lesion-specific tolerance criteria. The From our results, it cannot be stated that bicycle accidents produce a typical pattern of injuries. All types of craniocerebral lesions occurred in the bicycle accident victims. Some of the lesions had a notable high frequency. The high frequency of skull fractures probably means that most of our pedal cyclists had their head impacted by a stiff surface, such as the road surface or a windshield. Cerebral contusions were the second most frequent lesions. When comparing the bicycle accidents in which a motor vehicle was involved to the other accidents, diffuse axonal injury, intraventricular bleeding and gliding contusions were more frequent in the motor vehicle collisions. Apart from this finding, which can be explained by the common pathogenetic mechanism of these injuries based on high shear stresses within the brain parenchyma that are known to be associated with car collisions, the injury patterns in the bicycle accidents with and without involvement of a motor vehicle were quite similar. The higher frequency of intracerebral haematomas in the accidents without a motor vehicle is probably explained by the higher ages in that accident group. When the surgeon described the cause of an acute subdural haematoma operated for, this appeared to be an underlying cerebral contusion in two-thirds. Moreover, a strong association of both lesions was found. In the literature, conflicting ideas exist on what is the most important source of subdural haematomas. Gennarelli and Thibault (1982) have stated that tearing of bridging veins is the most frequent cause and demonstrated its relation to high rate rotational acceleration in the sagittal plane. In contrast, Maxeiner (1997) found that two thirds of subdural haematomas in his autopsy series of 45 cases were caused by contusions. If the association between subdural haematomas and contusions is indeed strong, then it is equally important to understand the biomechanics of contusions for protection against acute subdural haematomas. 4.3. Influence of victim age The predisposition of the younger aged for extradural haematomas has been known for a long time and was confirmed in this study. This relation has been attributed to the stronger adherence of the dura mater to the skull bone in the aged (Gallagher and Browder, 1968). In contrast, the presence of subdural haematomas, contusions, intracerebral haematomas and intraventricular bleeding and the extent of contusions were clearly correlated with higher age. We did not find any reports that mention such relations between age and the occurrence and extent of separate injuries in the literature, although we could find studies in which more severe

566 B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 injuries were seen in older people. In a population-based study of bicycle-related injuries in Sweden, the most severe injuries were seen in the age group over 65 years (Eilert-Petersson and Schelp, 1997). A selection bias may have influenced our findings. We do not think, however, that the patient selection is responsible for all relations between age and injuries in our study and it would certainly be interesting to investigate these relations in a true epidemiological study. The implications of these age-relations are that lesion-specific tolerance criteria, if ever realised, will probably be age-dependent. This may have consequences for head protection. 4.4. Outcome Outcome and age are negatively correlated in our series. An inverse relation between age and outcome has been seen in a lot, though not all, of publications on traumatic brain injury (Kraus et al., 1995). In studies on bicycle-related head injury that account for exposure rates, higher fatality risks are found in older people. Rodgers (1995) found that the fatality risk for pedal cyclists over the age of 65 was 5.8 times that of cyclists younger than 14 years of age. Oström et al. (1993) calculated a relative fatality risk of 11 12 for pedal cyclists over the age of 50 compared to those younger than 50. The absence of a correlation between the type of accident and outcome in our study may be explained by the younger age of victims involved in collisions with a motor vehicle. The most important injury associated with poor outcome was subarachnoid haemorrhage. When cerebral contusions were assessed by the effect of their number and volume on outcome, then they were the second most important injury, followed by acute subdural haematoma and brain swelling. Probably the relation between these injuries and outcome can, at least partially, be explained by the relation between age and contusions and subdural haematomas on the one hand and the inverse relation between age and outcome on the other hand. A multiple regression analysis failed to demonstrate both the independence of age and the independence of these injuries in their relation to outcome. So, both play their role, but the strong intern relations make it hard to distinguish one from another. However, the interference of age does eventually not change the relations found between certain injuries and poor outcome in this study. It is perhaps surprising that diffuse axonal injury was not related to outcome. The number of patients that met the criteria, however, was not large and may be underestimated. Moreover, most of the patients with diffuse axonal injury in this study were young. We do not intend to question the established relation between diffuse axonal injury and poor outcome. However, we think that in this population of pedal cyclists the influence of diffuse axonal injury on outcome was overshadowed by the effect of subarachnoid haemorrhage (large) contusions, subdural haematomas and brain swelling. A significant association of the latter four lesions was seen. We hypothesise that in many patients the contusions may have been the primary injury of these four. Bearing in mind that contusions occurred with a frequency of 73%, the importance of contusions in producing a poor outcome should not be underestimated. 4.5. Recommendations for bicycle helmet design A better insight into bicycle-related head injury may help us to better define the needs an ideal bicycle helmet should fulfil. In the following section, some considerations are summed up. First, it may be worthwhile to improve our knowledge of the biomechanics of cerebral contusions. Most contusions occur at the inferior surface of the frontal lobes and the inferolateral surface of the temporal lobes. There is some evidence from finite element models of the human head that these predilection areas correspond well with areas of high shear stresses in impact simulations (Chu et al., 1994; Huang et al., 2000). Apart from this, the pathogenesis of these frontal and temporal contusions is only poorly understood and much work is yet to be done before human tolerance levels for this type of injury will be available. Second, head coverage should be reconsidered. In the present study, we found that 57% of impacts occurred at the side (parietal, temporal and frontotemporal regions) and 27% of impacts occurred at the front of the head. However, it can easily be observed that the current bicycle helmets do not cover the lower parts of the side and front of the head, i.e. the temporal and supraorbital areas. A preponderance of side and frontal impacts has also been reported in studies in which damaged bicycle helmets were analysed (Ching et al., 1997; McIntosh et al., 1995; Williams, 1991). In addition, these studies report that many impacts occurred at the frontal and lateral lower rim of the helmets. Ching et al. (1997) found damage to the frontal rim of helmets associated with lesions to the forehead and thus suggested that the helmet should be extended further down. McIntosh et al. (1995) reported that 25% of impacts in his study of 42 damaged helmets had occurred on the part of the helmet immediately cranial to the temporal area. Seventy-five percent of these victims suffered from head injuries with a revised Abbreviated Injury Scale score of at least 2. Again, this injury rate can probably be brought down by a larger coverage of the area in question. The protection of the temporal area is all the more important since the thin temporal skull bone is easily fractured and since temporal fractures predispose to lacerations of the medial meningeal artery and subsequent extradural haematomas. Finally, half of the bicycle accident victims admitted to our unit had a collision with a motorised vehicle. It has been reported in literature, including a Belgian study, that one third to one half of bicycle accidents involve a motorised vehicle (Ballham et al., 1985; Elsen et al., 1997; Kraus et al., 1987). The question arising from this is whether bicycle helmets are capable of offering any protection in such collisions. It is certain that the fall heights of 1 2 m in the

B. Depreitere et al. / Accident Analysis and Prevention 36 (2004) 561 567 567 descriptions of the drop-test in the current helmet standards only correspond to low velocity impacts. Mills (1990) concluded from his study on the protective capabilities of helmets that they fail to protect the head in direct impacts at high speed. Of course, unlimited protection in all situations is utopic. However, we feel that some additional research is required to clarify this issue. Meanwhile, it is evident that improving pedal cyclist head protection in Belgium should start with the promotion of the wearing of bicycle helmets. Acknowledgements This study was made possible with sponsorship from KBC Banking and Insurance (Brussels, Belgium) and Wolvenberg NV (Mechelen, Belgium). References Ballham, A., Absoud, E.M., Kotecha, M.B., Bodiwala, G.G., 1985. A study of bicycle accidents. Injury 16, 405 408. Broderick, J.P., Brott, T.G., Grotta, J.C., 1994. Intracerebral haemorrhage volume measurement. Stroke 25, 1081. Ching, R.P., Thompson, D.C., Thompson, R.S., Thomas, D.J., Chilcott, W.C., Rivara, F.P., 1997. Damage to bicycle helmets involved with crashes. Accid. Anal. Prev. 29, 555 562. Chu, C.S., Lin, M.S., Huang, H.M., Lee, M.C., 1994. Finite element analysis of cerebral contusion. J. Biomech. 27 (2), 187 194. Collins, B.A., Langley, J.D., Marshall, S.W., 1993. Injuries to pedal cyclists resulting in death and hospitalisation. N. Z. Med. J. 106, 514 517. Dorsch, M., Woodward, A., Somers, R., 1987. Effect of helmet use in reducing head injury in bicycle accidents. Accid. Anal. Prev. 19, 183 190. Eilert-Petersson, E., Schelp, L., 1997. An epidemiological study of bicycle-related injuries. Accid. Anal. Prev. 29, 363 372. Elsen, M., Janzing, H., Broos, P., Delooz, H., 1997. Hoofdletsels bij fietsongevallen. Tijdschrift voor Geneeskunde 53, 945 952. Fife, D., Davis, J., Tate, L., Wells, J.K., Mohan, D., Williams, A., 1983. Fatal injuries to bicyclists: the experience of Dade County, Florida. J. Trauma 23, 745 755. Gallagher, J., Browder, J., 1968. Extradural hematoma. Experience with 167 patients. J. Neurosurg. 29, 1 12. Gennarelli, T., 1983. Head injury in man and experimental animals: clinical aspects. Acta Neurochirurgica (Wien) Suppl. 32, 1 13. Gennarelli, T.A., Thibault, L.E., 1982. Biomechanics of acute subdural hematoma. J. Trauma 22, 680 686. Gilchrist, A., Mills, N., 1996. Protection of the side of the head. Accid. Anal. Prev. 28 (4), 525 535. Guichon, D., Myles, S., 1975. Bicycle injuries: one-year sample in Calgary. J. Trauma 15, 504. Hirsch, A.E., Ommaya, A.K., 1970. Protection from brain injury: the relative significance of translational and rotational motions of the head after impact. In: Proceedings of the 14th Stapp Car Crash Conference, Ann Arbor, pp. 144 151. Huang, H.M., Lee, M.C., Lee, S.Y., Chiu, W.T., Pan, L.C., Chen, C.T., 2000. Finite element analysis of brain contusion: an indirect impact study. Med. Biol. Eng. Comput. 38, 253 259. Kraus, J.F., Fife, D., Conroy, C., 1987. Incidence, severity and outcomes of brain injuries involving bicycles. Am. J. Public Health 77, 76 78. Kraus, J.F., McArthur, D.L., Silverman, T.A., Jayaraman, M., 1995. Epidemiology of brain injury. In: Narayan, R.K., Wilberger Jr., J.E., Povlishock, J.T. (Eds.), Neurotrauma. McGraw-Hill, New York, pp. 13 30. Maimaris, C., Summers, C., Browning, C., Palmer, C.R., 1994. Injury patterns in cyclists attending an accident and emergency department: a comparison of helmet wearers and non-wearers. Brit. Med. J. 308, 1537 1540. Maxeiner, H., 1997. Detection of ruptured cerebral bridging veins at autopsy. Forensic Sci. Int. 19, 103 110. McDermott, F.T., Lane, J.C., Brazenor, G.A., Debney, E.A., 1993. The effectiveness of bicyclist helmets: a study of 1710 casualties. J. Trauma 34, 834 845. McIntosh, A.S., Kallieris, D., Mattern, R., Svensson, N.L., Dowdell, B., 1995. An evaluation of pedal cycle helmet performance requirements. In: Proceedings of the 39th Stapp Car Crash Conference, San Diego, pp. 111 119. Mills, N., 1990. Protective capabilities of bicycle helmets. Brit. J. Sports Med. 24 (1), 55 60. National Highway Traffic Safety Administration, 2002. Traffic safety facts 2001. National Highway Traffic Safety Administration, US Department of Transportation, Washington, DC. NIS (Statistics Belgium), 2000. Mobiliteit 1998 1999, Brussels. NIS (Statistics Belgium), 2002. Toename van de verkeersdoden verkeersongevallen in 2000, Brussels. Oström, M., Bjornstig, U., Naslund, K., Eriksson, A., 1993. Pedal cycling fatalities in Northern Sweden. Int. J. Epidemiol. 22, 483 488. Rodgers, G.B., 1995. Bicyclist deaths and fatality risk patterns. Accid. Anal. Prev. 27, 215 223. Thomas, S., Acton, C., Nixon, J., Battistutta, D., Pitt, W.R., Clark, R., 1994. Effectiveness of bicycle helmets in preventing head injury in children: case control study. Brit. Med. J. 308, 173 176. Thompson, R.S., Rivara, F.P., Thompson, D.C., 1989. A case control study of the effectiveness of bicycle safety helmets. N. Engl. J. Med. 320, 1361 1367. Williams, M., 1991. The protective performance of bicyclists helmets in accidents. Accid. Anal. Prev. 23, 119 131. Wilson, J.T.L., Pettigrew, L.E.L., Teasdale, G.M., 1998. Structural interviews for the Glasgow Outcome Scale and the Extended Glasgow Outcome Scale: guidelines for their use. J. Neurotrauma 15, 573 585. Wood, T., Milne, P., 1988. Head injuries to pedal cyclists and the promotion of helmet use in Victoria, Australia. Accid. Anal. Prev. 20, 177. Zimmerman, R.A., Bilaniuk, L.T., Gennarelli, T.A., 1978. Computed tomography of shearing injuries of the cerebral white matter. Radiology 127, 393.