Culver et al. performed 11 superior-inferior head impacts of unembalmed cadavers using a padded impactor [62]. No fractures were observed at the peak head forces below 5.7 kN. Mean head impact force producing cervical injury was 7.58 ± 0.94 kN. Most of the injuries produced were in the posterior elements, which the authors attributed to “compressive arching”. These authors were among the first to suggest that neck position, particularly the degree of initial lordosis, influenced the injury mechanism.

Hodgson and Thomas performed a series of quasi-static and impact tests using embalmed cadavers, with one side of the spine exposed to allow for high-speed imaging [20]. The authors concluded that increased constraint of the head in the padded impact surface increased the likelihood of injury to the neck. They also concluded that compression from crown impacts produces exaggerated regional cervical spine flexion and increased the likelihood of fracture or dislocation between C4 and C7. It was one of the first studies to note that global compression of the head and neck could produce local flexion and concomitant shear and described the complex neck dynamics that can lead to a wide variety of compressive injuries to a number of cervical levels.

Nusholtz et al. performed impacts to the head vertex of 12 unembalmed cadavers with a guided 56 kg impactor [63]. Compressive extension cervical injury occurred with the necks buckling in extension or flexion. Head impact loads ranged from 1.8 to 11.1 kN. Nusholtz et al. also performed free-fall drop tests using eight unembalmed cadavers [60]. Drop heights that varied from 0.8 to 1.8 m produced extensive and varied cervical and thoracic fractures and dislocations. Head impact forces ranged from 3.2 to 10.8 kN in these tests. Nusholtz et al. drew a variety of conclusions from these cadaver tests that still remain quite valid. (1) The initial orientation of the spine is a critical factor in influencing the type of dynamic response and injuries produced. (2) Descriptive motion of the head relative to the torso is not a good indicator of neck damage. (3) Energy-absorbing materials are effective methods for reducing peak impact force but do not necessarily reduce the amount of energy transferred to the head, neck, and torso or the damage produced in a cadaver model.

Yoganandan et al. conducted a study to evaluate the effect of head restraints on spinal injuries with vertical impact [64]. Sixteen human male cadavers were suspended head down and dropped vertically from heights of 0.9 to 1.5 m. In 8 of the 16 specimens, the head was restrained to simulate the effect of muscle tone, producing preflexion of the cervical spine in its initial position. Head-impact forces ranged from 3.0 to 7.1 kN in the unrestrained and from 9.8 to 14.7 kN in the restrained specimens. On average, the peak load in the C5 or C6 vertebral body was 70 % lower than the peak head load. Cervical vertebral body damage was observed most commonly when the cadavers remained in contact with the impacting surface without substantial rotation or rebound. This caused the neck to be the major element in stopping the torso motion.

One important factor, not well controlled in all of these whole cadaver impact studies, is the degree of head constraint. When the head is free to rotate on the occipital condyles, bending of the neck occurs with minimal compression. When head rotation is constrained, bending of the neck is restricted and the system becomes much stiffer. The factors that influence head motion are the stiffness and conformability of the impacting surface and the cadaver head. Another factor that cannot be controlled in whole cadaver experiments is the extent to which failure progresses. A certain amount of kinetic energy is available in these tests and failures progress until this energy is dissipated. Finally, it is technically difficult to record the internal forces without violating the integrity of the structure, and because the dynamics are complex, the head impact loads are very different from the cervical spine loads [64, 65]. It is these factors, as well as lack of muscle action, that makes it difficult to determine injury tolerance and mechanisms from whole cadaver tests.

Experiments on whole cervical spines offer the advantage of measuring the cervical spine loads in situ while preserving the geometry and biomechanics. For this reason they have been the most important experiments for establishing cervical spine injury kinetics and tolerance. One of the first such studies is from Bauze and Ardran, [66]. In 1978, Bauze and Ardran recognized that the biomechanics of the bilateral facet dislocation (BFD) were not well understood and undertook the first systematic approach to producing the injury. This was a very significant study, the implications of which would not become clear until decades later. They were able to produce bilateral facet dislocations in 6 of 14 cadaver cervical spines using an apparatus that applied compression while restricting head rotation. They were the first to describe the resulting buckling deformations and hypothesized that it occurred during compressive impacts. The primary limitations of the study were that it was static, and that a non-physiologic stress riser existed at the point of dislocation to control lower cervical bending. In addition, the loads were applied in discrete steps, and only the maximum compressive load for all specimens was reported (1,422 N).

Myers et al. tested 18 cadaver spines as beam-columns in compression using three different mechanical end conditions for the head: fully constrained, rotationally constrained, and unconstrained [67]. Somewhat to the surprise of the authors, the neck could not be broken in the unconstrained mode, which produced a large amount of flexion bending (>90°) with relatively little compressive force (<300 N). Bilateral facet dislocations could be consistently produced in the lower-cervical spine with a rotationally constrained head end condition. Bilateral facet dislocations occurred with an average of 1,720 N of axial force and 2.9 cm of compressive displacement. The study also examined the fully constrained end condition and found it to be substantially stiffer than the other two. In this mode of loading, wedge and compression fractures were produced in the mid-cervical spine. The average axial force and displacements at failure were 4,810 N and 1.4 cm respectively. Increasing the constraints on head motion resulted in a stiffer response and a greater axial load to failure, but there was no statistically significant difference in the energy to failure between the fully constrained specimens and the rotationally constrained specimens. This suggests that the mode of failure shifted from primarily compressive, to a combined loading (compression-flexion) with significant internal bending.

Yoganandan et al. were the first to examine the dynamics of isolated whole cervical spines in compressive impact [68]. They quantified the resulting complex kinematics and suggested that buckling instability may play a role. They also were the first to observe the decoupling between the head response and the cervical spine response.

Pintar et al., studied the responses of whole head and neck preparations to compressive impacts [69]. In these experiments, the necks were straightened (the lordosis was removed) with between 15 and 30° of head flexion and the crown of the head was impacted with a hydraulic ram. The primary variable was the eccentricity of the occipital condyles over the T1 vertebra, which varied from 0.5 cm posteriorly to 2.5 cm anteriorly. They found that compressive injuries in the mid-cervical column were most likely to occur when the condyles were aligned vertically within 1 cm of T1. However, there was no single biomechanical metric that was predictive of injury. They produced a striking variety of injury patterns and concluded that “for essentially the same loading vector, (crown impact to the head) a variety of different injury mechanisms can be produced depending on the configuration of the spine prior to loading. This implies that without sufficient external data (e.g. scalp laceration) an assumed mechanism of injury from retrospective evaluation of radiographs may not be correct.”

Nightingale et al. studied a variety of head constraints and impact angles in compressive drop tests on cadavers [61, 65, 70]. Compressive failure loads were 1.06 ± 0.27 kN for females with a mean age of 65, and 2.24 ± 0.57 kN for males with a mean age of 62. This difference was statistically significant and indicates that sex should be controlled for in biomechanical testing. The injuries produced included nearly the full spectrum of clinically relevant fractures: burst fractures, Hangman’s fractures, odontoid fractures, bilateral facet dislocations, posterior element fractures and anterior avulsion fractures. The authors also observed first and second order buckling of the cervical spine, which gave rise to complex kinematics that can explain mid-cervical fracture patterns (Fig. 11.10). Another significant finding was the speed with which the fractures occurred – in rigid impacts, the spine failed within 8 ms, and in padded impacts, less than 25 ms. Head rebound was another kinetic observation, leading to the conclusion that impact surface should absorb as much of the kinetic energy of the head as possible so as not to return it to the cervical spine (Fig. 11.11). The study had been designed with the thought that angled impacts (contacts fore or aft of the head vertex) would allow the head and neck to rotate and escape the path of the following torso, but because of the head inertia this frequently did not occur and the cervical column fractured.

Toomey et al., performed drop tests very similar to those on Nightingale et al., except that they were designed to apply combined compression and lateral bending [71]. It was thought that this configuration might be more relevant to rollover injuries. The fracture loads and timing were not affected by the lateral orientation, nor was the expression of the first order-buckling mode. The authors concluded that the asymmetric loading produces similar kinetics and injury patterns to sagittal plane loading.

More recently, Saari et al., conducted impact tests on whole human cervical spines with high-speed cine-radiography [72]. The injuries produced were primarily compressive-extension fractures of the cervical spine, with multiple “teardrop” avulsions of the anterior vertebral body, all from extension. As with prior impact studies, very little head motion was observed at the time of injury. One specimen had a C1–C2 fracture dislocation with C1 anterior ring fracture and odontoid fracture, which was observed to spontaneously reduce on the cine-radiography. Maximum cord compression occurred between 8 and 15 ms after impact, which supports the notion that cord injury occurs rapidly following head impact.

Ivancic et al., subjected five cadaver spine specimens attached to a Hybrid III dummy head to high-energy horizontal impacts to simulate athletic neck injuries [73]. Catastrophic neck injuries were produced at many levels of the cervical spine. Using the ATD load cell inserted into the occipitocervical junction, the authors observed similar forces and times indicating that inertial forces of the individual vertebrae are small relative to the reactions in the neck and validating the use of T1 load cells as a means to study the neck without violating the cranio-cervical junction.

The human neck is fairly strong in compression, but it is also very stiff in that direction and not nearly strong enough to manage the kinetic energy of the torso. Head first impacts from low drop heights reliably produce a variety of cervical spine fractures which arise from various buckling modes in the cervical spine [61, 64, 65, 67, 68, 69, 70, 74, 75].

Differences in axial load to failure have been shown to depend on the injury produced and, therefore, on the degree of constraint imposed by the contact surface. From a tolerance standpoint, these injuries should be evaluated separately. As an example, cadaver experiments by Maiman et al., Myers et al., and Pintar et al. show that compression-flexion and compression-extension injuries occur with smaller axial loads than pure compressive injuries [67, 76, 77]. Bilateral facet dislocations reported by Myers occurred at 1,720 ± 1,230 N, and flexion injuries reported by Maiman occurred at approximately 2,000 N. In contrast, compression injuries reported in these studies occurred at 4,810 ± 1,290 N and 5,970 ± 1,049 N, respectively. Given the importance of the musculature in resisting flexion and extension in the lower cervical spine, the above values should be considered as a lower bound for human tolerance in compression-flexion and compression-extension. In contrast, the compressive loads associated with pure compression injuries are a reasonable estimate of cervical compressive tolerance.

Much of the scatter in the existing compressive tolerance data for the cervical spine is due to differences in experimental technique. Since neck responses are very sensitive to end conditions, inertia, and orientations, it is important to use data from studies that employ similar experimental setups. A synthesis of data that meet the inclusion criteria for a dynamic cervical spine tolerance are presented in Nightingale et al. [70]. The results of studies from two institutions were combined to formulate a tolerance for males and females. Averaging the means of the two studies resulted in the following tolerance values: 1.68 kN for females, and 3.03 kN for males. The tolerance values were scaled to account for the age of the cadaver specimens and the authors suggest a cervical spine tolerance for the young human male in the range of 3.64–3.94 kN. The tolerance is lower for a neutrally positioned spine and higher for a straightened spine; however, these are a reasonable first-order representation of the mean resultant force required for a catastrophic cervical spine injury in any given impact situation. The threshold velocity required to produce catastrophic neck injuries is often quoted to be 3 m/s, but catastrophic injuries can occur at even lower velocities during those circumstances in which the neck geometry and initial torso velocity maximize injury potential (i.e. a forward of vertex impact with a straightened spine and the torso velocity perpendicular to the impact surface).

A number of studies have reported on the total compressive displacements of the cervical spine at the time of injury and this metric has been suggested as a measure of tolerance [69]. Compressive impact tests by Pintar et al. with straightened spines resulted in injuries at a mean compressive displacement of 18 ± 3 mm. Myers et al. reported a mean of 14 ± 4 mm in neutrally positioned (lordotic) spines [67]. Displacements can also be estimated from the Nightingale et al., 1997 from the product of injury time and impact velocity for rigid impacts. This results in displacements of 13 ± 7 mm.

Both the neutrally positioned and the straightened cervical spines are stiff structures. As indicated above, the cervical spine generates a great deal of load very quickly and without much compressive displacement. Pintar et al. (straightened spine) produced neck fracture between 1 and 7 ms after impact. Nightingale et al. (lordotic spines) produced injury at an average of 5 ms for rigid impacts and 18 milliseconds for padded impacts. It appears that a burst fracture is more likely in the straight configuration than in the lordotic configuration. A bilateral facet dislocation is more likely in the lordotic configuration than in the straightened. But either injury can happen in either configuration depending on the initial positions, the rate of loading, and chance. The complex deformations resulting from cervical spine buckling explain, in part, why the complete spectrum of cervical spine injuries can be produced in what appear to be very similar compressive impacts [78].

Because cervical spine fracture happens so quickly, there is not enough time for much head motion to occur before the neck breaks. Head rotations greater than 20° have been shown to occur well after neck injury (as much as 80 ms after) and maximum head rotations occur at least 150 ms after injury [61]. In addition, the direction of head rotation (flexion or extension) is not correlated with the neck injury mechanism. The reason for this is that the inertia of head resists changes in linear and angular momentum. This means that it takes both force and time to change the direction and rotation of the head – much more time, it turns out, than it takes to break the neck.

The mechanism of the bilateral facet dislocation (BFD) was long described with biomechanically ambiguous phrases like “forced” flexion or “hyperflexion” and was thought to occur as a result of motion of the head that forces the chin towards the chest. This “hyperflexion” hypothesis was satisfying to lay intuition and became dogmatic in the orthopedic literature [38, 42, 52, 79]. But as early as 1960, difficulties were reported in producing this “flexion” injury in experiments on cadavers [80, 81]. In 1978, Bauze and Ardran were able to produce the injury using an apparatus that applied compression while restricting head rotation [66]. They called the buckling deformation “ducking” and hypothesized that it occurred during compressive impacts. In 1980, Hodgson and Thomas demonstrated these ducking modes of deformation and concluded that compression from crown impacts produces exaggerated regional cervical spine flexion and increased the likelihood of fracture or dislocation between C4 and C7 [20]. Since the earlier work highlighted the importance of mechanical end-conditions on injury type, Myers et al. performed a controlled parametric study to examine the effects of end-conditions and determined that the neck could not be broken in flexion bending, but could be consistently dislocated in compression with constrained head rotation – an end condition similar to the one applied by Bauze and Ardran. To date, the 1991 Myers study is the only one to consistently produce the elusive BFD and it definitively precluded forced flexion of the head as a mechanism. In this same time frame, Dr. Torg was hypothesizing that compressive buckling in a straightened neck was the BFD mechanism. This was based on his review of injuries and film of football players [17, 82]. In fact, Dr. Torg and his colleagues produced the buckling kinematics and failure in a skeleton model [13]. In the mid 1990s, Nightingale et al. applied varied head constraints in compressive impact tests on cadavers and were able to produce some bilateral facet dislocations. The mechanism for the bilateral facet dislocation is now thought to be combined loading at the segment level secondary to neck compression and buckling (Fig. 11.12). There are one or two anecdotal cases (e.g. a full-nelson hold, performed until muscles fatigued) where static flexion-tension loads have been reported to cause injury but these are a rare exception [83]. The role of “hyperflexion” as a mechanism for BFD is still being debated in the interesting and unusual circumstance of the rugby scrum [83, 84].

Interest in the tensile tolerance of the neck was primarily sparked by a string of severe and fatal injuries to children and adults in low speed frontal crashes with airbag deployments. In 1991, a NHTSA Special Crash Investigation was initiated in response and it was determined that most of the victims were either not wearing belts or were very close to the airbag when it deployed. The mechanism of injury was thought to be tension and extension in the neck caused by interaction with a deploying airbag under the chin. The resulting injuries were basilar skull fractures, occipito-atlantal dislocations, and C1–C2 injuries [92–97].

One of the most interesting findings of the ensuing research on tension was that the upper cervical spine is stronger than the lower cervical spine. However this is not consistent with the field studies above in which upper cervical spine injuries are much more common. As a segmented column, one would expect that the weakest point in the neck would be the site of failure, but the findings of the NHTSA investigation showed a large number of upper cervical spine injuries in both adults and children. Chancey et al., found that this discrepancy is most likely due to the effects of the active musculature [90]. The muscles of the neck share tensile loads with the cervical spine by providing a parallel load path. Such load sharing increases the overall strength and stability of the neck and provides greater protection to the caudal motion segments because of the larger size and number of muscles in the lower cervical spine. So, here is a situation where the cadaver, without the benefit of a biofidelic computational model of the muscles, is not a good surrogate for the living human and this has implications for both tension experiments and the deceleration experiments discussed in Sect. 11.4.4.1.

Finally, the epidemiology and the experimental research show that tensile neck injuries in the lower cervical spine are rare. Since tension produces primarily upper cervical spine injuries, further discussion of these will be left to Sect. 11.4.5.

Myers et al. showed that the lower cervical spine is stronger in torsion than the atlantoaxial joint, mitigating torsion as a primary cause of lower cervical injury [67]. The notion that torsion can predispose the lower cervical spine to injury from other types of loading (i.e., flexion), while a common belief, is clearly not true for catastrophic injury (though it does play a role in whiplash) [103].

The etiology of unilateral facet dislocations has been the subject of interest and controversy. Huelke et al. and Harris et al. reported that unilateral dislocation was the result of combined flexion and rotation [2, 42]. Gosch et al. reported that in purely anterior to posterior deceleration sled tests of primates, axial rotation was observed and unilateral facet dislocation produced [104]. Roaf reported that torsion was required to produce lower cervical ligamentous injury and dislocation and that hyperflexion alone could not produce ligamentous injury [80]. Rogers and White and Panjabi noted that unilateral dislocation resulted from an exaggeration of the normal coupling of lower cervical lateral bending and rotation but did not describe the loads required to produce the injury [38, 105]. Braakman and Vinken suggested that unilateral facet dislocation was the result of combined flexion and rotation [33]. Torg and Otis stated that the lesion was the result of compression and buckling [13, 14, 17]. Bauze and Ardran produced unilateral facet dislocations in a number of specimens loaded in compression and flexion [66]. Their apparatus applied no torque to the specimen, and yet lower cervical rotation was observed prior to unilateral dislocation. Myers et al. demonstrated that while unilateral facet dislocation could be produced by direct torsional loading of the lower cervical spine, the lesion could not be the result of torsional loading of the head because of the comparative weakness of the atlantoaxial joint [100]. The authors currently believe that the lesion is produced by a mechanism similar to the bilateral facet dislocation, with the difference being due to bending out of the plane of symmetry, or the presence of pre-existing facet tropism of other structural asymmetry. Another possibility is that the unilateral facet dislocation is a partially reduced bilateral facet dislocation, which might account for the cohort of these lesions that have complete spinal cord injury [33].

One of the most frequently cited studies is the work of Mertz and Patrick [86, 106]. They summarized data from forward and rearward sled decelerations of one human volunteer (Patrick) and four cadavers with weights added to the head at, above, or below the center of gravity of the head. Despite having extremely limited data, these authors were able to define corridors for the kinematic response and recommended noninjurious tolerance values in dynamic and static loading situations that proved useful in motor vehicle safety standards for many years. Reporting data from both volunteer and cadaver decelerations, the authors determined a lower bound for the risk for neck injury based on the bending moment estimated at the occipital condyles for flexion and extension loading. For extension, the authors report noninjurious loading of 35 ft-lb (47.3 Nm) with ligamentous injury expected above 42 ft-lb (56.7 Nm). For flexion, the authors report initiation of pain at 44.0 ft-lb (59.4 Nm), a maximum voluntary loading of 65 ft-lb (87.8 Nm), and the risk for structural injury above 140 ft-lb (189 Nm) in the cadaver following contact of the chin on the chest. The largest deceleration applied to a cadaver produced calculated moment of 189 Nm with no evidence of injury detected by x-rays. This limit was suggested as the upper bound of ligamentous spine tolerance in flexion, despite the lack of injury. The authors note however, that muscular injury may occur at loading below the latter value.

Bass et al. tested 36 head and neck preparations in frontal decelerations impacts to develop a predictive Neck Injury Index [107]. The tests produced AIS 3+ injuries in 11 of the specimens, with four injuries that were judged to have the potential for cord damage. Most of the injuries were to the soft tissues, with the osseous injuries being relatively minor. Upright orientation of the head and neck produced significantly less tension and shear (1,251 and 1,376 N, respectively) than orientations run with 30° of forward flexion (2,146 and 3,747 N, respectively) despite being run at significantly lower velocities. The authors suggest that this was because the neck was loaded in a stiffer mode when flexed forward (more tensile than bending). There was no significant difference in peak flexion moments.

Pintar et al. [108] subjected five cadavers to belted frontal impacts ranging in severity from 13 to 57 km/h and resulting in sled peak accelerations from 10 to 20 Gs. The instrumentation included accelerometer packs on the head and T1 as well as belt load sensors. Chin-to chest contact occurred in all tests, but after the peak calculated forces. The injuries produced in four of the five specimens were dislocations of the C7-T1 (n = 2) and dislocations of C6–C7 and T1–T2. The four specimens with cervical spine fractures also had multiple rib fractures and assorted fractures of the manubrium, clavicle and sternum. In the high velocity tests, the mean axial force and moment at the base of the neck were 1,424 N and 192 N-m respectively. At the occipital condyles they were 1,570 N and 41 N-m. The resultant force in both locations was not reported, but was substantially higher than the z component.

The clinical finding of a spinal cord injury without radiographic abnormality (SCIWORA) by Taylor and Blackwood prompted interest in extension as a causative agent [109]. In a cadaver study by Taylor, which combined myelography with manual loading of the head in flexion and extension, demonstrated posterior canal defects in “forced” hyperextension. These defects were thought to be the result of bulging of the flaval ligaments into the neural canal and were produced without overt structural damage to the spine. The author postulated, based on this observation, that extension could produce spinal cord injury without trauma to the cervical spine. Further evaluation and the recognition of disc bulging during loading reinforced this as a possible mechanism of SCIWORA [110, 111]. The mechanism has also been observed in the more recent compressive impact testing of Saari et al. [72].

In a study by Gadd et al., four whole cadavers were loaded manually using levers coupled to the head to apply combined shear and bending (flexion-extension or lateral) [112]. Rotation in extension greater than 80° or 60° in lateral bending produced audible sounds in two of the four specimens and resulted in decreased stiffness following reapplication of the same load. Defining these events as minor injury, mean mid-cervical moment at injury was 24.3 ± 12 Nm.

The end condition study by Myers et al. in 1991 applied compressive displacements to whole cervical spines with a free rostral end condition [113]. This was not pure moment loading; however, the tests produced primarily flexion because of the limited ability of the cervical spine to react compression with an unconstrained head. The mean peak axial load was 289 ± 81 N with a mean axial displacement of 8.6 ± 1.2 cm and a mean head rotation of 96 ± 7°. None of the specimens were injured despite the large, anatomically impossible head rotations.

Nightingale et al. examined the flexibility and strength of 52 female motion segments in both flexion and extension [114]. This was followed in 2007 by a similar study of 41 male cervical spines [115]. Their flexibility results were similar to prior studies [116–118] and they were the first to provide statistically meaningful tolerance data in pure bending (Table 11.1). They found that the upper cervical spine was stronger than the lower cervical spine. For both males and females, the lower cervical spine failures were complete joint disruptions (tearing of all the ligaments and the disc) with a few associated avulsion fractures of the vertebral bodies. The most common injuries to the upper cervical spine were odontoid fractures, which occurred primarily in extension, but were also observed in flexion.

Wheeldon et al., 2006 tested whole lower cervical spines in pure flexion and extension [119]. The segments were from C2 to T1 and the range of motion for a 2.5 N-m moment was near 50° for flexion and near 30° for extension. The study was unusual in that the mean age of the donors was only 33 years old. By comparing their results with prior studies, the authors concluded that younger specimens were significantly less stiff in bending than older specimens. This was also the only study to use whole lower cervical spines and the large rotations encountered even without the O-C1 joint illustrate the technical difficulty in applying a pure follower moment to these very flexible structures.

Bending studies on isolated cervical spine have shown that the neck is remarkably flexible in bending and its toughness in this mode of loading is underestimated. Researchers who have tried to fail the whole spine in this mode have not succeeded [113] and they have had to resort to segment tests. Even applying pure moments to isolated upper cervical spine segments is technically difficult and requires very specialized test equipment. Summing the angles at failure for the upper and lower cervical spine gives an estimate of the degree of head flexion required for a pure bending injury of the cervical spine. This value is greater than 120° and is clearly nonphysiologic as shown in Fig. 11.13. It can be stated with confidence that failure of the cervical spine in pure flexion is an anatomical impossibility, validating the hypothesis suggested decades ago by Roaf [66]. This may also be true of extension, although further modeling and experimental work is necessary to state this with similar confidence.