The Biomechanics of Gender Difference and Whiplash Injury : Designing Safer Car Seats for Women

Female car users are reported to haae a higher incidence ofsofi tissue neck injuries in lou speed rear-end collisions than males, anl they apparently take longer to recoaer. This paper addresses the whiplash problem b1 deaeloping a biomechanical FEM (Finite Element Method) model of the 50th and the 5th percentile female ceruical spi'nes, based on the earlier publislrcd rutle model created at the Nottingham Trent


Whiplash
The number ofwomen drivers in the UK is growing f,aster than the number of men. According to the latest industry reports, women could soon outnumber men on the roads.
There could be more than 20 million women drivers in the next l0 years if current trends continue [1,2]. In terms of safety, men are more likely than women to be killed in car crashes but female drivers have a higher risk of sustaining injury. This gender difference is apparent at slight injury level where the rate is 49 Vo for females compared to 3l Vofor males. Light injuries art most comnon during tear end accidents and they constitute 82 Vo and 89 Vo of alI sustained injuries for male and female front occupants rcspectively. In comparison, in frontal impacts they are 66 Vo for female and 65 Vo for male percentage of all injuries [3]. Whilst most mild injuries are simply bruises, abrasions and lacerations. a very substantial proportion of spinal injuries are soft tissue neck injuries, so-called whiplash injuries [4]. Although classified as minor, their high incidence l'ate and long-term consequences lead to significant societal cost. The annual cost in the UK is estimated at f2.5 billion [5]. Syrnptoms include neck pain, stiffness, headaches, dizziness, blurrcd vision and numbness and may be associated with damage to the cervical muscles, ligaments, facetjoints, nerar'e roots, vertebral arteries, or blain stem. Howeve4 despite numerous studies on human volunteers, cadavers, and animals, there is no consensus about which specific mechanisms are responsible for the majority of neck injuries to car occupants in rear-end impacts, although several  ioural, and sociological parameters. Although many hypotheses have been proposed, the mechanism of cervical spine whiplash injury is not well understood and more research should be conducted. L 3 Sociological factors lVomen tend to driver smalle6 lighter cars than men and this situation is disadvantageous since the car-mass is a key factor in determining injury outcome. However; Koch et aI [11] reported that the relative risk of injury in smaller str-uck cals was still higher for females than for males, even when the female was the driver. Otte et al [12] also suggested diffelences in sex-specific accident framework conditions and conhrmed that women suffer neck injury in small cars more frequently than men. In the UK, medium cars are driven frequently by both sexes, however, 42Vo of female driver collisions are in small cars comPared to 23 % for males [2]. Furthermore, rnen have lower disability levels than women despite having on average less optimal head restr-aint positions [13]. It was suggested that females tend to sit farther fo}ward in their seats than males so their heads move farther before the headrest is reached [14]. Seating position also can affect spinal kinematics and increase the risk of injury.
Matsumoto et al [15] showed that the percentage of kyphosis position is much higher for females than for males. Spine misalignment as a reason for soft tissue injury forwomen was pointed out by Ono et al [16], who showed that rotational angles of cervical vertebrae were larger at kyphosis for females, producing a higher probability of injury. 1.5 Physiological. and, q,natomical factors Physiological and anatomical diflerences imply that the biomechanical tolerance of the female neck is lower than males and may orplain differences in neck injury frequency.
Temming et al [8] indicated that the risk of whiplash injury for both females and males increases with body height but females have higher risk of injury. Also injury risk is higher for females in each weight group [28], disproving the hypothesis of Kraft et al p9] that women are more vulnerable to soft neck injuries because they are generally lighter. Significant gender differences were noted for depths ofthe superior and inferior endplates of and height of the cervical vertebrae, with those for males being larger [20,21]. Differences in neck musculature between men and women are suggested as an important factor in neck injuries [22]. Cervical muscles can be sources of pain and inlluence neck motion, both passively and actively. Statistically, females have smaller neck circumferences, suggesting this may be the actual risk area.
Furthermore, most muscles in women have a lower cross section than those in the men [23]. States [24] attributed the differences in injury risk to the ratio of head volume to cross sectional area of necks. For 50th percentile males the ratio is l:135 and for the comparable female it is l:151, indicating females have narrower necks relative to head size. Male neck muscles are also stronger than female cewical muscles; the female strengths were 30-40 7o lower than their male counterparts [25] or according to others 20-25Vo lower 1261.
According to Vasavada [27] males have 2-2.5 times greater moment-generating muscle capacities and only l.l-1.3 greater mass and head inertia relative to women, suggesting female muscles work closer to maximum capacity. Muscle activation occurs 5 Vo-15 Vo earlier [28,29] for females than males, which may be another source of higher risk for females. As females tend to have smaller and weaker supporting muscles in the cervical spine and also less body weight to collapse back support it can make them more vulnerable to neck injury [30]. Surprisingly, there is no comprehensive data describing differences between female and male ligaments in terms of geometry (cross area, length) and material properties ffoung's modulus, load/deformation); this is a major shortcoming for any biomechanical analysis.

Experirnental tests
During rear end impact sled experiments with volunteers there were observed differences in head-neck motions. Significant gender differences existed between the peak amplitude and time-to-peak amplitude. Females experience higher and earlier peak accelerations of head, torso, C7:I1 joint relative to earth, but this difference was not present for the head relative to the C7:I joint. Also women undergo smaller and earlier head extension than men [31]. Female volunteers demonstrated smaller rearward horizontal head translation 48 [32]. Generally females presented more rebound motionwith larger thorax flexion angles than males.

Safety tests
In spite of the increasing number of female car users, and the higher incidence of soft neck injuries amongwomen, the motor industry has been slow to recognise the new trend and some car manufacturers still do not take designing for female drivers seriously. Only a few attempts have been made to examine the biomechanical response of female cenical spines during car accidents. There is no 50th percentile female ATD (Anthropomorphic Test Dummy) or FEM dummy model in common use. The population of female drivers and occupants is represented by the 50th percentiie male dummy in conjunction with the 5th percentile female dummy, even though it was shown by Calter [2] that 90 7o of female drivers in the UK are lighter and shorter than the 50th percentile male dummy. Thble I indicates how poorlywomen are represented when designing safety systems. The models were developed in the LS-DYNA code. Because the cervical spine is a complex biomechanical systern, the finite-element method seems rvell suited for parametric analytical study. FEM offers the advantage that it can handle complex geometric configurations and material, contact and geometric nonlinearities. This study deals with the relative head and neck motion in whiplash, focusing on differences between female and male models and aiming to explain the higher incidence of iryury among women. The biomechanical responses from a 50th percentile male dummy and a simple scaled-down 50th percentile male dummy were compared against 50th and 5th percentile female models. The principal comparison is between a 50th percentile female model and a 93 7o scaled-down male during low speed rear--end impact. The typical load scenario in a rear-end collision is as follows: l) The vehicle accelerates forward when struck 2) The torso is pushed forward by the seat 3) The spine starts straightening and the necMorso joint rises 4) The head lags behind the torso due to its inertia 5) The upper cervical spine undergoes flexion while the lower part, undergoes extension, promoting an S-shape 6) The rise of the first thoracic vertebra, in (2) above, Ieads to a "ramping phenomenon" which causes cervical compression 7) The head rotates backward, producing a C-shape with extension of the entire cervical spine. Presence of a head restraint reduces the C-shape.
8) The occupant rebounds out ofthe seat, Ieading to flexion of the cervical spine The basic 50th percentile male neck modelwas crcated by the Biomechanics Group at Nottingharn Tient University [35] and consists of a biomechanical head-neck complex combined with the rigid Hybrid III dummy rnodel in a simplified vehicle seat environment. Bony structures are modelled using shell elements with the geometry modified to achieve better interaction with soft tissue. All ligaments are represented, using a mixed structure of shell and non-linear springs elements, except for the Nuchal Ligament, which is modelled with shell elements only. The force/deformation load curves for discrcte element arc based on experimental results [36].
Shell element stiffness properties werc calculated fiom I % of the breaking force and corresponding deflection. Ligarnent geometry is based on available experirnental data. Muscles are rnodelled by spring elernents, as only passive action is represented, with material properties based on sternocleidomastoid muscles [37]. Interrertebral discs are represented using solid elements of Blatz-Ko nrbber' The first approach in this studywas structural scaling of the male model, defined as overall pure size reduction of the male spine without incorporating the characteristic female features. The model waJscaled to 93 Vo, assuming that 50th percentile females are 93 Vo as tall as males. The second model female was more sophisticated; it was again based on a 93Vo scaled male but allowed for a disproportionately larger female head mass (the 50th percentile female/male body mass ratio is 80 % whereas the correspond-ing head mass ratio is 82 Vo) [34]. Strength properties of ligaments and muscle were modified assuming constant Young's modulus but reduced cross-sectional areas due to scaling. It was also taken into consideration that female vertebrae are more slender Thble 2.

Validation
Experiments made by Kronenberg et al [31], were used to evaluate the models. Linear acceleration of the head and the first thoracic vertebra ffl) were obtained. Head angle and trajectorieswerc filmed. Datawere taken from a subjectwhose measurements matched closely the mass and seating height of 50th percentile UK females and males. The marked increase in head x-acceleration and differences in head-neck kinematics observed for females compared to males in the experiments was confirmed by the cornputational models.
The peak head acceleration is higher and earlier for females than rnales. Howeve4 the acceleration is 10 times higher than experiments because in the FEM nodel an unrealistic rigid seat model is used.
Reasonably good agreernent was found for head totation' In sled tests carried out by Siegmund et al [32] the females experienced smaller and earlier peak head extension than mhles. The FEM models confirmed this even though peak values were always higher than the experiment. In this study only a passive muscle response is modelled. This seems to suggest that muscle contraction plays a signifrcant role in cer.rical spine kinematics, although the muscle onset is developed 80-90 ms after impact [38] and full muscle forces are not developed until 60-70 ms later [39]. It also may explain later peak head extension for women, particularly that they activated their muscles earlier.

Results
The biomechanical model of the female cervical spine is intended to solve the mystery of higher risk of injury for females. The simplified scaled down male model shows similarity in several parameters to male models rather than female. The relative rotation between the head and C3 produces hyperflexion which is considered a potential neck injury mechanism. The flexion is higher for the female model, both the 50th male and 93 7o scaled down male. The curve for the 93 Vo scaled down model in the first second after impact shows the same shape as the 50th percentile male (Iig. 4), There was higher axial force on tectorial membrane (TM) and vertical ligament ffC) in the female model than both male models (Frgs. 5, 6). There are a few hypotheses that assume that injuries of upper level ligaments predominate in whiplash injury due to hypertranslation or combined shear and compression [40,41,42]. Higher axial forces on ligaments: alar (AL), apical (AP) and AAOM (anterior antlanto-occipital membrane) and CP foint capsules) were observed during rebound motion. Because females experience more rebound motion this might be a reason for the higher risk of injury as lkafft et al [9] suggested that injuries occur during the rebound sequence when posterior tissues are stressed and the anterior tissues in the neck are compressed (Figs. 6, 7).
Panjabi and associates [43] propose that the lower cewical spirre is injured in hyperextension when the spine forms an S-shape curve before the neck is fully extended. This intervertebral rotation beyond physiological limits implies the stretching of the anterior and compression of the posterior elements of the lower cervical spine. They obsewed that intervertebral rotation at lower segments exceeded maximum physiological extension. The anterior injury at C5-C6 level was observed. In our model it was noticed during the first phase of loading that the female model experienced higher C5 rotation relative to C6 (Frg. 8) than the male model and also higher axial forces at anterior longitudinal ligament (ALL), Frg. 9. This may be a potential explanation for the higher risk of whiplash injury forwomen.
-'** 50th female -10 -*x--93%male +50th male    It should be noted that for female models there is higher downward translation during rebound motion, followed by an upward shift. The models show significant gender difference in vertebral motions which should not be neglected.

Conclusion
Female neck biomechanics is a complex issue, exacerbated by a distinct lack of biomechanical data. The exact mechanism of so-called whiplash injury is not established and there are several hypotheses about the source of pain. In spite of the observed higher risk of injury for female car occuPants most research has involved male subjects or gender differences were not specified. The 50th and 5th percentile male dummies, both ADT and FEM models, do not represent the average female. The 93 Vo scaled down male model is not adequate to simulate female resPonses even though the scaling constitutes a good height and mass match. The 50th percentile female model was in general agreement with test results considering the lack of data about female neck biomechanical properties. This preliminary female model 52 exhibited a satisfactory correlation with experimental results and the gender differences in kinematics prove the need for a 50th percentile female model. The observed difference in head rotation relative to C3 and C5-C6 relative motion could be potential causes of the higher neck injury in females, and needs further consideration. It was shown that there were higher axial forces in cervical ligaments for female than for male models, supporting the theory that females are more vulnerable to whiplash ir{ury.
Further model developments are needed in the following areas: I. Enhancement of muscle response by modeling active resPonse.
2. Remodeling vertebra geometry to incorporate more detailed gender differences in height and cross section area.