Stability Analysis of Craniovertebral Junction Fixation Techniques

Puttlitz, Christian M. PhD; Melcher, Robert P. MD; Kleinstueck, Frank S. MD; Harms, Juergen MD; Bradford, David S. MD; Lotz, Jeffrey C. PhD

Journal of Bone & Joint Surgery - American Volume:
Scientific Articles
Supplementary Content

Background: Craniovertebral arthrodesis in the upper cervical spine is challenging because of the high degree of mobility afforded by this region. A novel method for achieving atlantoaxial fixation with use of polyaxial screws inserted bilaterally into the lateral masses of C1 and transpedicularly into C2 with longitudinal rod connection has recently been introduced. The question remains as to whether this technique provides adequate stability when extended cephalad to include the occiput. The purpose of this study was to determine the primary stability afforded by this novel construct and compare its stability with the current standard of bilateral longitudinal plates combined with C1-C2 transarticular screws.

Methods: We used ten fresh-frozen human cadaveric cervical spines (C0-C4). Pure moment loads were applied to the occiput, and C4 was constrained during the testing protocol. We evaluated four conditions: (1) intact, (2) destabilized by means of complete odontoidectomy, (3) stabilization with longitudinal plates with C1-C2 transarticular screw fixation, and (4) stabilization with a posterior rod system with C1 lateral mass screws and C2 pedicle screws. Rigid-body three-dimensional rotations were detected by stereophotogrammetry by means of a three-camera system with use of marker triads. The range of motion data (C0-C2) for each fixation scenario was calculated, and a statistical analysis was performed.

Results: Destabilization of the specimen significantly increased C0-C2 motion in both flexion-extension and lateral bending (p < 0.05). Both fixation constructs significantly reduced motion in the destabilized spine by over 90% for all motions tested (p < 0.05). No significant differences were detected between the two constructs in any of the three rotational planes.

Conclusions: Both hardware systems provide equivalent construct stability in the immediate postoperative period when it is critical for the eventual success of a craniovertebral arthrodesis. On the basis of this work, we believe that the decision to use either construct should be determined by clinical rather than biomechanical concerns.

Author Information

1 Department of Orthopaedic Surgery, University of California at San Francisco, 1001 Potrero Avenue, Room 3A36, San Francisco, CA 94110. E-mail address for C.M. Puttlitz:

2 Department of Orthopaedics and Traumatology, Center for Spinal Surgery, Klinikum Karlsbad-Langensteinbach Guttmannstrassel, 76307 Karlsbad-Langensteinbach, Germany

Article Outline

Surgical management of craniovertebral instability usually involves arthrodesis of the occiput and the cervical spine. It is commonly held that the reduction or elimination of motion across fixed spinal segments increases the opportunity for fusion. Achievement of a successful fusion when the craniovertebral junction (C0-C2) is involved is complicated because of the relatively high range of motion across these segments1. In addition, the complex arrangement of the neurovascular anatomy in the upper cervical spine further confounds certain technical aspects of hardware implantation.

In the past decade, numerous constructs have been developed in order to increase stability and facilitate implantation of the hardware2-12. Open surgical stabilization of the craniocervical region was initially performed with use of simple bone-graft onlay techniques and was subsequently combined with posterior wiring across the craniovertebral junction13-16. These constructs evolved to include wiring of both the cervical and occipital regions to contoured rods and pins9,11,17-19. Despite the fact that most of these patients were managed with additional postoperative immobilization (such as a Minerva cast or halo vest), many of the rod constructs resulted in unsatisfactory failure rates (up to 30%20).

Most contemporary methods use screw fixation in both the occiput and cervical regions with bilateral longitudinal plates or rods. Higher rates of occipitocervical fusion have been reported with use of these more rigid constructs21-24. The Magerl technique (C1-C2 transarticular screws) has become a standard for attaining atlantoaxial fixation. In order to achieve craniovertebral fixation, these screws are coupled to longitudinal plates that extend superiorly and are attached to the occiput. This technique provides a relatively high degree of construct stability25 and, accordingly, reduces the number of vertebral segments that must be incorporated into the arthrodesis. Although C1-C2 transarticular screw fixation results in a rigid construct, placement of the screws across the atlantoaxial level is highly technical and includes a substantial risk of vertebral artery violation26.

A novel technique for achieving atlantoaxial fixation, which utilizes direct fixation of polyaxial screws to the lateral masses of C1 and through the pedicle of C2, with connection by means of longitudinal rods, has recently been reported27. The proposed advantages of this technique compared with the use of C1-C2 transarticular screws are a reduced risk of vertebral artery insult and the ability to perform reduction maneuvers after individual screw placement in C1 and C2. Extension of the construct to the occiput to achieve craniovertebral fusion has provided excellent preliminary clinical results. However, we know of no biomechanical studies that have documented and compared the stability afforded by cephalad extension of the C1-C2 technique to include the occiput. Thus, the purpose of this study was to investigate and compare the occipitocervical construct stability afforded by this novel screw-rod construct with that of the more commonly used C1-C2 transarticular screw-plate method and wiring techniques.

Back to Top | Article Outline

Materials and Methods

Ten fresh-frozen human cadaveric cervical spines (C0-C4; mean age at the time of death, 67.7 years) were used in this study. Each specimen was evaluated for osseous deformity and alignment with use of anteroposterior and sagittal plane radiographs. Specimens were stored in double plastic bags at –20°C until ready for preparation and/or testing. All soft tissues were removed from the specimen, with care taken to preserve the joint capsules, ligaments, and osseous structures. The fourth cervical vertebra was potted in polymethylmethacrylate such that the midsagittal plane of the C3-C4 disc was kept in a horizontal orientation1.

The C1-C2 transarticular screw-plate construct (C1-C2 plate; Sofamor Danek, Memphis, Tennessee) consisted of bilateral longitudinal plates with several holes to accommodate the transarticular screws, a smooth region without holes where most of the bending for the craniocervical angle can be performed, and a semicircle-shaped occipital portion with holes for occipital screw attachment (Fig. 1). Occipital attachment was achieved with use of two 6 or 8-mm-long screws for each plate. Transarticular screws were inserted across the atlantoaxial joint according to the original description by Magerl and Seemann28. Before screw placement, proper screw trajectory was verified, with use of Kirschner wires, on sagittal plane radiographs. Screw tightening simultaneously fixed the atlantoaxial joint and the plate to the spine.

The posterior screw-rod fixation (C1-C2 rod) was achieved with use of bilateral longitudinal rods that spanned an occipital plate, lateral mass screws in C1, and pedicle screws in C2 (see Fig. 1). A T-shaped plate was fixed to the occiput with use of two midline screws and two more laterally placed screws (6 to 12 mm in length) between the inferior and superior nuchal lines. The plate allowed for variable placement of the rod attachment in the medial-lateral direction. Polyaxial screws (3.5 mm in diameter) with an 8-mm unthreaded upper portion were inserted bilaterally into the lateral masses of C1 with the entry point located in the middle of the confluence of the posterior arch and the inferior aspect of the lateral mass of C1. The trajectory for placement of transpedicular screws in C2 was oriented as described by Ebraheim et al.29. After alignment of the atlantoaxial joints, the rods (3.0 mm in diameter) were fixed with locking nuts to the occipital plate and the screws in the atlas and axis.

We evaluated stability by determining the moment-rotation relationships for each construct. Pure moment loads were applied with use of a system of weights, pulleys, nylon strings, and rods (Fig. 2). Two loading rods were drilled through the skull. A medial-lateral rod extended through the petrous portion of the temporal bone, and an anterior-posterior rod was drilled under the sella turcica. Moments were applied in flexion and extension, lateral bending to both sides, and left and right axial rotation, up to 1.5 N-m after three preconditioning cycles30. Stereophotogrammetry was used to determine three-dimensional displacements of the vertebral levels during pure moment loading. Marker triads were attached to the occiput, C1, C2, and C3. Rigid-body motion was detected with use of a three-camera system (Motion Analysis, Santa Rosa, California). In previous experiments in our laboratory, it was determined that this measurement system is accurate to within ±0.05°. Intervertebral rotations were calculated as Euler angles, and the range of motion for each fixation scenario was determined from the greatest moment load (1.5 N-m).

Each specimen was tested in its intact condition and after destabilization. Destabilization of the specimen was achieved by removing the odontoid process through the foramen magnum. Odontoidectomy has been used by previous investigators to introduce instability to the C0-C2 region31,32. Because screw trajectories from either fixation scenario would intersect, providing less purchase for the second system, each specimen was assigned to one of the two occipital fixation groups described above (C1-C2 rod or C1-C2 plate), providing five specimens for each construct.

Statistical analysis was performed with use of a one-way analysis of variance (α = 0.05). Individual differences between groups were delineated with use of the Fisher protected least-significant difference post hoc test for multiple comparisons (SigmaStat 2.0; Jandel Scientific, Chicago, Illinois). Thus, we tested each specimen in flexion-extension, lateral bending, and axial rotation to determine the moment-induced motion in the intact, destabilized, and reconstructed conditions. The data are presented as total motion across the craniovertebral junction (C0-C2) and the adjacent, subaxial level (C2-C3).

Back to Top | Article Outline


Flexion-Extension Stability

Flexion-extension total motion for the intact C0-C2 segments was a mean (and standard error) of 52.0° ± 3.5° at a flexion-extension moment of ±1.5 N-m. Destabilization by means of odontoidectomy significantly (p < 0.01) increased this motion to a mean of 69.7° ± 4.4°. Both fixation methods were able to significantly reduce flexion-extension motion compared with the destabilized condition (p < 0.01) (Fig. 3). These motion reductions were also significant compared with the intact condition. Specifically, the C1-C2 plate and C1-C2 rod constructs permitted approximately 3.5% of the total flexion-extension motion afforded by the destabilized construct. No significant difference was found between the total flexion-extension motion of the two constructs (p = 0.98).

Flexion-extension motion across the adjacent segment (C2-C3) was increased by 60% after removal of the odontoid process, although this change was not significant (p = 0.21) (Fig. 4). Placement of the C1-C2 rod system on the destabilized spine restored the adjacent segment sagittal motion to that of the intact condition (p = 0.98). The C1-C2 plate system reduced the mean motion at C2-C3 below the normal or intact level; however, this motion reduction was not significant (p = 0.35).

Back to Top | Article Outline
Lateral Bending Motion

Total lateral bending motion at ±1.5 N-m of lateral bending moment for the intact condition was a mean (and standard error) of 23.6° ± 3.0°. Removal of the odontoid process significantly increased (p < 0.01) lateral bending motion to a mean of 33.9° ± 8.3°. Once again, application of both constructs to the destabilized spine resulted in significant motion reductions (p < 0.01) (Fig. 5). The C1-C2 plate reduced total lateral bending motion to a mean of 1.3° ± 0.3°. After application of the C1-C2 rod construct, the total lateral bending motion was a mean of 2.4° ± 1.1°. The difference in motion reduction between the two constructs was not significant (p = 0.81).

Motion at the adjacent level (C2-C3) was affected by application of the C1-C2 plate system. This construct reduced C2-C3 motion with respect to both the destabilized condition and the C1-C2 rod construct (see Fig. 4); however, these motion differences were not significant (p = 0.06 for both conditions). Both destabilization and the C1-C2 rod constructs increased motion compared with the intact condition; however, these increases were not found to be significant.

Back to Top | Article Outline
Axial Rotation Motion

The intact C0-C2 axial rotation motion at ±1.5 N-m of axial torque was a mean (and standard error) of 82.1° ± 4.7°. Destabilization did not significantly increase axial rotation (mean, 86.7° ± 3.8°). However, application of both the C1-C2 plate and C1-C2 rod systems produced equivalent motion reductions to a mean of 2.6° ± 1.0° and 2.6° ± 1.2°, respectively, across the craniovertebral junction (Fig. 6). These changes were significant with respect to both the intact and destabilized conditions (p < 0.01 for both fixation techniques).

The C1-C2 plate hardware demonstrated lower axial rotation range of motion across the adjacent segment than did the intact, destabilized, and C1-C2 rod conditions, although these differences were not significant (p > 0.08 for all conditions). Both the destabilized and C1-C2 rod conditions demonstrated greater axial rotation than did the intact spine, although the difference was not significant (see Fig. 4).

Back to Top | Article Outline


We used a cadaver model loaded by pure moments in order to evaluate the construct stability afforded by two upper cervical hardware systems. The data for combined motion in C0-C1 and C1-C2 from the intact condition indicate that our model is comparable with other models that have used this testing technique30-32. Specifically, we found that the craniovertebral junction (C0-C2) flexion-extension motion, lateral bending motion, and axial rotation was 52.0°, 23.6°, and 82.1°, respectively. Total flexion-extension motion has been reported to range from a mean of approximately 50° to 61°30,33-36. Panjabi et al.30, using 1.5-N-m pure moment loading, reported mean total C0-C2 lateral bending motion to be 24.4°. Axial rotation, which occurs predominantly at the atlantoaxial level, has been reported to occur across the C0-C2 segment and to range from a mean of 72.4° to 92.2°30-35. Thus, the current protocol reproduced physiologic ranges of motion in all three planes and allowed us to compare our results with those reported previously for other constructs. In addition, removal of the odontoid produced significant increases in sagittal and frontal plane rotation, which is consistent with the findings in previously published investigations that have used this procedure to impart severe instability to the craniovertebral junction31,32.

The data indicated equivalence between the two constructs with respect to the achievement of an acute reduction of motion in a destabilized spine. Both constructs significantly reduced motion in the destabilized spine by >90% for all motions tested. This finding is important because the decision to use either construct shifts from initial concerns about stability to other clinical considerations, such as surgical efficacy and technique. Both constructs, although similar with respect to the amount of stability achieved, are different in terms of their surgical implantation. Using a plate system, the surgeon must accomplish the plate-contouring first, followed by fixation of the various vertebral levels to the plate. Individual screw placement is limited by the predrilled holes in the plate, further restricting the possibility for additional compression or distraction to be applied after the hardware is placed. Placement of the C1-C2 transarticular screws can be technically demanding and requires maintenance of atlantoaxial joint reduction during the procedure. Finally, the proper entry point and trajectory of the C1-C2 screws may be difficult or impossible to access in a severely kyphotic patient. Conversely, the rod system allows for independent placement of the screws in C1, C2, and the occipital plate. A reduction maneuver, if necessary, can be performed after the screws and occipital plate are fixed. Each rod can be individually contoured to fit the alignment achieved by the reduction maneuver. The system also allows for application of either distraction or compression forces by means of the screws. Finally, the medial and more cephalad placement of the C2 screws reduces the opportunity for vertebral artery damage.

Differences in occipital fixation were also delineated during the course of this investigation. The more lateral attachment of the longitudinal plates to the occiput resulted in visible perforations of the inner table of the skull, even when the shortest available screws were used, demonstrating the reduction in osseous thickness in these regions37,38. In contrast, the T-shaped plate allows for a solid midline fixation with use of blunt-ended screws of 10 or 12 mm in length. Two additional screws were fixed immediately lateral to the midline close to the superior nuchal line, an area that has been reported to be an ideal site for occipital screw purchase37.

One important aspect of our model was the inclusion of an adjacent level (C3) that was not part of the affected area (C0-C2). This allowed us to determine not only how destabilization and fixation influenced the area to be fused but also how spinal kinetics would be affected on a more global scale. The data consistently showed that the C1-C2 plate construct tended to reduce motion across the C2-C3 joint (as compared with both the intact and destabilized conditions). Conversely, C2-C3 motion with the C1-C2 rod construct more closely approximated that seen in the intact condition for flexion-extension and axial rotation. One possible explanation for this observation could be that the plate actually provides a motion hindrance of the C2-C3 facet joint by overhanging or covering the C2-C3 facet joint. This is most likely an iatrogenic effect produced by a combination of the relatively cephalad trajectory of the transarticular screw and its entry point, which is situated in proximity to this joint. Impingement of the adjacent facet joint associated with posterior plates in the cervical spine has been previously described. In a series of twenty-nine patients who had combined anteroposterior locking screw-plate fixation for the treatment of traumatic cervical instability, Whitecloud et al.39 found that the posterior plates impinged on the adjacent facet joints in twenty-four patients.

It has been shown that levels adjacent to the fusion mass in the spine are at increased risk for degeneration because of aberrant changes in the spinal loading profile. This has important clinical consequences in the cervical spine because one indication for upper cervical arthrodesis is advanced degenerative changes (such as rheumatoid arthritis). Nonphysiologic motion at adjacent levels due to hardware implantation could increase the rate of degeneration in spines with preexisting degenerative changes at the subaxial levels. Interestingly, according to our data, these adjacent subaxial effects may occur in spines that have been fused or, conversely, in those that have been destabilized. The data indicate that destabilization not only created significant instability in the craniovertebral junction but also resulted in increased C2-C3 intervertebral motion. Restrictions (in the case of fusion) or laxity (instability) of C0-C2 intervertebral motion seem to upset the delicate kinematic and force balance in the craniovertebral junction, resulting in an increased burden on the immediate subaxial level. The change in spinal loading is most likely manifested as increased motion at the C2-C3 level, and it provides a plausible explanation for the higher intervertebral rotations measured across this segment in the present study.

The results of this investigation should be kept within the context of its limitations. We chose to use different cadavera for testing each instrumentation system. This was necessary because of the different screw trajectories used by the two different hardware systems. These trajectories would have intersected one another, possibly reducing the screw purchase of the second system tested. This resulted in a sample size of five for each construct, which clearly reduces the power of our statistical analysis. However, because these two systems provided equivalent mechanical stability and very significant motion reductions with respect to the intact and destabilized conditions, we believe that additional specimens would have only confirmed the C0-C2 results demonstrated with the current series of specimens. It is possible that greater numbers of specimens would have resulted in more statistical significance with respect to the differences in motion seen at the adjacent level (C2-C3). Finally, the current testing protocol allows us to comment on and predict in vivo performance only with respect to acute stability. Fatigue testing is warranted to investigate the time-dependent behavior of either system with respect to loosening or cyclic material failure.

Overall, our data indicate that a construct consisting of C1 and C2 polyaxial screws, occipital screws, and longitudinal rods demonstrates acute mechanical equivalence to a system that utilizes C1-C2 transarticular screws, occipital screws, and longitudinal plates. Each construct was capable of significantly reducing motion in a severely destabilized craniovertebral cadaver model. The decision to use either construct should be made on the basis of the surgical technique and not the acute biomechanical stability.

In support of their research or preparation of this manuscript, one or more of the authors received grants or outside funding from DePuy AcroMed, Inc. None of the authors received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, educational institution, or other charitable or nonprofit organization with which the authors are affiliated or associated.

Investigation performed at the Department of Orthopaedic Surgery, University of California at San Francisco, San Francisco, California

1. 3rd, Panjabi MM. Clinical biomechanics of the spine. 2nd ed. Philadelphia: JB Lippincott; 1990.
2. . Occipitocervical fusion for the unstable rheumatoid neck. Orthopedics. 1989;12: 469-73.
3. , Hayward RD, Harkness WF, Crockard HA. The use of autologous skull bone grafts for posterior fusion of the upper cervical spine in children. Spine. 1995;20: 2217-20.
4. , Ngian KS, Cohen L. Posterior upper cervical fusion in rheumatoid arthritis. Spine. 1992;17: 268-72.
5. , Karapurkar AP. Transoral plate and screw fixation of the craniovertebral region—a preliminary report. Br J Neurosurg. 1994;8: 743-5.
6. , Tsuji H, Katoh Y, Yonezawa T, Kitagawa H. Occipito-cervical fusion reinforced by Luque's segmental spinal instrumentation for rheumatoid diseases. Spine. 1988;13: 1234-8.
7. , Cassidy JR, Davis RF, North RB, Ducker TB. Fusion of the occiput to the upper cervical spine. A review of 37 cases. Spine. 1991;16(10 Suppl): S490-4.
8. , Mittal P, Banerji D, Behari S, Acharya R, Chhabra DK. Posterior occipitoaxial fusion for atlantoaxial dislocation associated with occipitalized atlas. J Neurosurg. 1996;84: 559-64.
9. , Crockard HA, Pozo JL, Thomas NP, Nelson IW. Craniocervical instability treated by contoured loop fixation. J Bone Joint Surg Br. 1986;68: 173-7.
10. , Crockard HA, Moskovich R, Harkey HL, Stevens JM, Kendall BE, Ransford AO, Calder I, Corbett M, Shipley M, Sinclair H, Barls A, Lomborough C, Masters AI. Nystagmus and joint position sensation: their importance in posterior occipitocervical fusion in rheumatoid arthritis. Spine. 1994;19: 16-20.
11. , Kawaida H, Morizono Y, Matsunaga S, Fielding JW. Occipitoatlanto-axial fusion utilizing a rectangular rod. Clin Orthop. 1989;239: 136-44.
12. , Menezes AH. Occipital calvarial bone graft in posterior occipitocervical fusion. Spine. 1998;23: 249-55.
13. , Turner WG. The treatment of fracture-dislocations of the cervical vertebrae by skeletal traction and fusion. J Bone Joint Surg. 1937;19: 584-602.
14. . Skeletal traction in treatment of fractures and dislocations of the cervical spine. Ann Surg. 1938;106: 770-6.
15. . Treatment of fracture-dislocations of the cervical spine. J Bone Joint Surg. 1942;24: 245-58.
16. , Bohlman HH. Occipitocervical fusion. Indications, technique, and long-term results in thirteen patients. J Bone Joint Surg Am. 1987;69: 833-6.
17. , Errico T, Cooper P, Benjamin V, DiBartolo T. Occipitocervical fusion with a five-millimeter malleable rod and segmental fixation. Neurosurgery. 1993;32: 198-208.
18. , Ransford AO, Crockard HA. Treatment of non-rheumatoid occipitocervical instability. Internal fixation with the Hartshill-Ransford loop. J Bone Joint Surg Br. 1994;76: 357-66.
19. , Dickman CA, Golfinos JG, Papadopoulos SM, Sonntag VK. Threaded Steinmann pin fusion of the craniovertebral junction. Spine. 1996; 21: 1630-7.
20. , Sonntag V, Dickman CA. Occipitocervical wiring techniques. In: Spetzler RF, Dickman CA, Sonntag VK, editors. Surgery of the craniovertebral junction. New York: Thieme; 1998. p 795-808.
21. , Jeanneret B, Fischer K, Magerl F. Occipitocervical fusion with posterior plate and screw instrumentation. A long-term follow-up study. Spine. 1994;19: 2364-8.
22. . Posterior rod system of the cervical spine: a new implant allowing optimal screw insertion. Eur Spine J. 1996;5: 350-6.
23. , Laheri V. Plate and screw fixation for atlanto-axial subluxation. Acta Neurochir (Wein). 1994;129: 47-53.
24. , Dvorak J, Panjabi M, Froehlich M, Hayek J. Posterior occipitocervical fusion. A preliminary report of a new technique. Spine. 1991;16(3 Suppl): S17-24.
25. , Goel VK, Traynelis VC, Clark CR. A finite element investigation of upper cervical instrumentation. Spine. 2001;26: 2449-55.
26. , Casey AT, Solanki GA, Tuite G, Veres R, Crockard HA. Radiological and anatomical evaluation of the atlantoaxial transarticular screw fixation technique. J Neurosurg. 1997;86: 961-8.
27. , Melcher RP. Posterior C1-C2 fusion with polyaxial screw and rod fixation. Spine. 2001;15: 2467-71.
28. , Seeman PS. Stable posterior fusion of the atlas and axis by transarticular screw fixation. In: Kehr IP, Weidner A, editors. Cervical Spine I. New York: Springer; 1986. p 322-7.
29. , Rollins JR Jr, Xu R, Jackson WT. Anatomic consideration of C2 pedicle screw placement. Spine. 1996;21: 691-5.
30. , Dvorak J, Duranceau J, Yamamoto I, Gerber M, Rauschning W, Bueff HU. Three-dimensional movements of the upper cervical spine. Spine. 1988;13: 726-30.
31. , Crawford NR, Brantley AG, Sonntag VK. Biomechanical effects of transoral odontoidectomy. Neurosurgery. 1995;36: 1146-53.
32. , Puttlitz CM, Kleinstueck FS, Lotz JC, Harms J, Bradford DS. Biomechanical testing of posterior atlantoaxial fixation techniques. Spine. 2002;27: 2435-40.
33. . Handbuch der Anatomie und Mechanik der Gelenke unter Berucksichtigung der Bewegenden Muskein. Jena: Fischer; 1911.
34. , Charpy A. Traite d'anatomie humaine. Paris: Masson; 1926. p 74-89.
35. . Normal movements of the cervical spine. AJR Am J Roentgenol. 1978;130: 317-26.
36. . Studies in spontaneous atlas dislocation. Acta Orthop Scand. 1957;Suppl 23: 78.
37. , Doherty BJ, Heggeness MH. Quantitative anatomy of the occiput and biomechanics of occipital screw fixation. Spine. 1998;23: 1100-8.
38. , Currier BL, Stone J, Grabowski JJ, Larson DR, Fisher DR, An KN. Biomechanical evaluation of occipital fixation. J Spinal Disord. 2000;13: 336-44.
39. , Ricciardi JE, Werner JG. Bone graft, hardware, and halo fixator-related complications. In: Clark CR, editor. The cervical spine. 3rd ed. Philadelphia: Lippincott-Raven; 1998. p 903-23.
Copyright 2004 by The Journal of Bone and Joint Surgery, Incorporated