Unstable cervical spine fractures (CS-Fx) frequently require instrumented fusion to restore biomechanical stability. Several fixation techniques are available. Compared with lateral mass screw (LMS) fixation, pedicle screw (PS) fixation offers superior biomechanical strength (1, 2). However, traditional PS placement has been associated with an increased risk of injury to neurovascular structures, limiting its widespread use.

The introduction of navigation technologies in spine surgery has facilitated more accurate and safer PS placement, leading to broader adoption of PS fixation in the management of cervical spine fractures (3–5). Recent studies have indicated that navigated PS insertion in CS-Fx is associated with high accuracy and a low incidence of severe complications (6, 7). Nevertheless, most published series include analyses of fewer than 200 patients, providing limited statistical power to assess rare but serious complications (8–11). Larger cohort studies are therefore needed to provide a more robust evidence base for future systematic reviews and meta-analyses.

In this study, we present a consecutive cohort of 345 patients with unstable CS-Fx treated surgically using navigation-guided fixation. A total of 1,347 navigated screws were inserted, including C1 lateral mass screws, C2 pars interarticularis screws, C2 and subaxial PSs, and upper thoracic PSs. The primary outcomes were screw placement accuracy and procedure-related complications associated with navigated PS fixation.

Oslo University Hospital (OUH) serves as the sole neurotrauma center in Southeast Norway, where all trauma-related cervical procedures for this population are performed. The population of Southeast Norway increased from 2.9 million in 2015 to 3.1 million in 2024. Norway has a publicly financed healthcare system characterized by equal access to care.

This study is a retrospective cohort analysis based on prospectively collected data from the Oslo Cervical Spine Fracture Registry, a quality-control database maintained by the Department of Neurosurgery at OUH (DPO approval no. 2014/12304). Since 2015, all residents of the southeastern region of Norway with a Norwegian social security number and an imaging-verified CS-Fx have been prospectively registered (12).

In 2015, we implemented the Brainlab spine navigation system (Brainlab AG, Munich, Germany), integrating preoperative cervical CT scans for navigated PS insertion. The registry includes demographic data, information of fracture morphology and management strategy, surgical details, and information on the use of intraoperative navigation for PS placement, screw placement accuracy, and surgery-related complications.

Between 2015 and 2024, a total of 4,689 patients with CS-Fx were registered in the Oslo Cervical Spine Fracture Registry. Of these, 1,014 (22%) underwent open surgical treatment. Among the surgically treated patients, 973 (96%) underwent surgery at OUH, while 41 (4%) were injured outside our health region and underwent primary surgery elsewhere.

Routine preoperative imaging included cervical CT, cervical MRI, and CT angiography (CTA). Navigation-guided pedicle screw (PS) fixation was used in 347 of 973 (36%) surgically treated CS-Fx patients at OUH (Figure 1). Intraoperative three-dimensional imaging was not available; navigation was performed using surface matching based on preoperative CT scans.

During surgery, preoperative cervical CT images with a slice thickness of 0.625 mm were uploaded into the navigation system. The navigation reference array was rigidly attached to the spinous process at the level planned for pedicle screw (PS) insertion. Level-specific registration was performed using surface matching with multiple points, ensuring adequate spatial distribution and height variation. Registration accuracy was verified by correlating navigated pointer positions with known anatomical landmarks. A navigated drill was then used to create the pedicle entry tract prior to screw insertion. Screw length was determined intraoperatively by advancing the navigation ruler to the pointer tip. This workflow was repeated for each instrumented level. In cases with two adjacent, fully ankylosed, uninjured levels, a single registration was used to place navigated screws at both levels.

Postoperative cervical CT scans were obtained within 24 h for 345 of 347 patients. Two patients died within 24 h of surgery and therefore did not undergo postoperative CT; both were excluded from the cohort, and their deaths were deemed unrelated to the use of navigated PS.

The final study cohort consisted of 345 patients with unstable CS-Fx treated with navigated PS fixation at OUH between 2015 and 2024, all of whom had postoperative CT images available for assessment of screw placement accuracy.

Data were extracted from the registry and supplemented with medical charts and imaging reviews when necessary. The accuracy of PS placement in the thoracolumbar spine is typically assessed using the Gertzbein–Robbins scale (13). However, this scale is less suitable for evaluating screw placement in the smaller pedicles of the cervical spine. Therefore, in this study, PS placement accuracy was graded on postoperative CT scans as follows: Grade 1, pedicle perforation <2 mm; Grade 2, pedicle perforation 2–4 mm; and Grade 3, pedicle perforation >4 mm (Figure 2). This grading system aligns with that used in a recent systematic review and meta-analysis, which included 4,697 navigated cervical PSs from 18 studies (14). The CT-based Grade 1–3 classification was independently performed by five neurosurgeons (JM, TA, VS, EH, and HL) under the supervision of a neuroradiologist (DB). The 345 patients were allocated among the five neurosurgeons for evaluation. In instances of diagnostic uncertainty, the neuroradiologist and the senior spine neurosurgeon (HL) were consulted to reach a consensus. The assessments were not blinded, and inter- or intra-observer reliability analyses were not undertaken.

The primary outcome measures were as follows:

Accuracy of navigated PS placement, assessed on postoperative cervical CT obtained within 24 h after surgery, and

The study cohort comprised 345 patients who underwent posterior fixation for CS-Fx at OUH using navigated PSs, all of whom had postoperative cervical CT images available for accuracy assessment. The median patient age was 68 years (range 5–92 years; IQR, 57–75); 60% were ≥65 years, 75% were male, 48% had severe systemic disease (ASA score ≥3), and 73% had a fall-related injury (Table 1).

In descending order, the most frequent levels of instability were C6/C7 (31%), C5/C6 (25%), and C0–C2 (21%) (Table 1). At admission, 35% of patients had trauma-associated cSCI, 18% had cervical radiculopathy, and 15% had VA injury.

All 345 patients underwent posterior fixation using one or more navigated PSs. Posterior decompression (laminectomy) was performed in 102 patients (30%), and additional anterior fixation was performed in 98 patients (28%).

A total of 1,347 navigated pedicle PSs were inserted across 345 patients, ranging from 1 to 14 screws per patient, with a mean of 3.9 navigated PSs per patient. In most cases, the osteosynthesis construct involved a combination of navigated PS with non-navigated LMS. Navigated PSs were most frequently placed at C7, followed by T1, C2, C1, C6, and T2 (in descending order) (Table 2). At the C3–C5 levels, navigated PSs were used less frequently, as freehand LMSs were preferred due to the small pedicle diameter. The most caudal injured level in this cohort was C7/T1 (Table 1); thus, stabilization rarely extended to T3 or T4, which explains the low number of navigated PSs placed at these levels.

The placement accuracy was Grade 1 for 1,210 of the 1,347 screws (90%), Grade 2 for 108 (8%), and Grade 3 for 29 (2%) (Table 2). Suboptimal accuracy (Grade 2 or 3) was observed in 137 of the 1,347 screws (10.1%). The rate of suboptimal pedicle screw (PS) placement was high in the C3–C6 vertebrae (17%–36%), intermediate in C7 (12%), and low in C1–C2 and Th1–Th4 (≤6%) (Table 2).

A total of 23 of the 345 patients (6.7%) experienced complications (Table 3). Eleven of these complications were directly related to PS placement. Among these, the corresponding screw accuracy was Grade 1 in one patient, Grade 2 in six, and Grade 3 in four (Table 3). The estimated overall risk of a PS-related complication per screw was 0.8% (11/1,347). When stratified by accuracy, the risk of complications was 0.1% (1/1,210) for Grade 1 screws, 6% (6/108) for Grade 2 screws, and 14% (4/29) for Grade 3 screws (Fisher's exact test, p < 0.001).

In our cohort of 345 patients, among the 1,347 inserted navigated PSs, the placement accuracy was satisfactory (Grade 1) for 90%, and suboptimal for 10% (grade 2 for 8%, and grade 3 for 2%). Surgery-related complications occurred in 23/345 (7%) patients. Among the complications, 11/23 were direct PS-associated complications (7 cases of VA injury, 2 cases of nerve root injury, 1 case of perioperative hemorrhage, and 1 case of screw misplacement in the AO joint). The twelve postoperative infections were not directly linked to the PS insertion, but rather to the surgical procedure itself. There were no cases of new-onset SCI and no cases of PS pull-out.

The superiority of PSs compared with LMSs in terms of pullout strength and biomechanical stability has been well-documented in both cadaver and clinical studies (1, 2, 7). Although PS placement has traditionally been associated with technical difficulty and a risk of neurovascular complications, its use has become more common with the introduction of navigation tools (4, 5, 7). Most studies published thus far on neurovascular complications after navigated PS insertion are small and likely underpowered to provide reliable estimates of these rare events. Our single-center cohort is larger than most comparable published series (6–11).

For grading PS placement accuracy, we chose not to use the classical Gertzbein–Robbins scale in our study, as we considered it not ideal for assessing placement in the small pedicles of the cervical spine (13). Instead, we graded PS accuracy as follows: Grade 1, pedicle perforation <2 mm; Grade 2, pedicle perforation 2–4 mm; and Grade 3, pedicle perforation >4 mm. This grading system aligns with a recent systematic review and meta-analysis of 4,697 cervical PSs from 18 studies (14). Thus, our results can be included in future systematic reviews and meta-analyses of navigated PS for unstable CS-Fx, both because of the cohort size and the matching methodology of screw precision grading.

The accuracy of PS placement in our cohort was Grade 1 (satisfactory) for 90% of PSs, Grade 2 (suboptimal) for 8%, and grade 3 (misplaced) for 2%, which is somewhat inferior to that reported in a recent meta-analysis (14). Even studies on free-hand insertion techniques have reported better accuracy for perfectly positioned screws (14). However, focusing only on the accuracy of navigated PS placement may be overly simplistic. The main aim should be to minimize PS-related complications while achieving biomechanically sufficient osteosynthesis. In general, the placement of a PS when feasible results in a stronger biomechanical construct than LMS-only fixation and, in some cases, allows for a shorter construct.

In our patient cohort, surgery-related complications occurred in 7% of patients, but only half of these were directly related to navigated PS use. Complications directly related to navigated PS use were seen only in patients with Grade 2 and Grade 3 screws, except in one patient, in whom we observed profuse bleeding during dissection of the entry point. The estimated overall risk of a PS-associated complication per PS was 0.8%. The risk according to PS placement accuracy was 0.1% for Grade 1 screws, 6% for Grade 2 screws, and 14% for Grade 3 screws. Therefore, accuracy undoubtedly matters.

No patients in our cohort had new-onset SCI, but there were two cases of new-onset radiculopathy due to PS insertion. These patients did not undergo reoperation, and both recovered with no/minor neurological deficits.

The incidence of VA injury and the modalities for detecting PS-associated VA injury vary across studies. A systematic review by Soliman et al. (7) revealed an overall incidence of VA injury of 0.4% for PS and 0% for LMS when various techniques were used. Another systematic review by Yoshihara et al. (15) also demonstrated an increased risk of VA injury using PS compared to LMS, but the difference was small.

In this series, 2% of patients had VA injury secondary to PS insertion, and the risk of VA injury per navigated PS was 0.5%. All VA injuries in our cohort were severe stenosis/occlusions secondary to arterial dissections treated with subcutaneous LMWH in the acute phase, followed by peroral acetylsalicylic acid for at least 3 months. None of the patients had cerebral infarction in the acute phase or during a minimum follow-up period of 3 months. The possibility of massive hemorrhage, various degrees of vessel wall injury, such as dissection and pseudoaneurysm formation, and late-onset complications have been described in the literature (16). The reported risk of cerebral infarction after traumatic or iatrogenic VA injury is low (11, 17–20).

Several authors have described an increased risk of VA injury in the midsection of the cervical spine (C3–C5), characterized by thin pedicles and a medially oriented screw trajectory (21). Tomasino et al. (22) studied individual variations in cervical bony anatomy and the VA course. The mean pedicle diameter increased from 4.9 mm to 6.5 mm from C3 to C7, and the pedicles were larger in male patients. The VA entered the transverse foramen at C6 in 82%, C5 in 13%, C7 in 3%, and C4 and C3 in 1%. Unusual pedicle anatomy or an irregular VA pathway was seen in 24% of patients. This emphasizes the need for a careful evaluation of both CT and CTA before surgery.

Although the risk of neurovascular injury from PS placement is low, this risk must, in our opinion, be considered and balanced individually for each level/screw during preoperative planning. Importantly, the cervical musculature in a tight surgical field tends to force the screwdriver medially and the screw tip laterally, a purely technical factor not eliminated by the use of navigation (23, 24).

Deep surgical site infection is a feared complication with potentially severe consequences for patients and increased costs for society. Known risk factors for surgical site infections in spine surgery include the duration of surgery, instrumentation, CSF leakage, obesity, and chronic steroid use (25). Our infection rate was 3.5%, which is well in line with other publications reporting infection rates ranging from 1.3%–8.8% (26–30). All 12 patients recovered after surgical debridement and antibiotic treatment without implant removal. In the absence of radiological signs of screw loosening, treatment of surgical site infection with irrigation, debridement and antibiotics without implant removal is reported to be a valid and effective option (31).

The suboptimal accuracy of our navigated PS placement for 10% of screws in our study warrants some reflection. The accuracy of spinal navigation based on preoperative CT scans is dependent on reliable surface matching of the posterior aspect of the vertebra. The quality of surface matching depends on the spread of registration points, the surface area of the exposed lamina/spinous process, and the contours of the lamina surface. Thus, a higher risk of error in vertebras C1, C3, C4, C5, and C6 is inherent. Another limitation of this navigation method is that the posterior part of the registered/navigated vertebral level must be intact and attached to the anterior column. If there is any undetected injury separating the anterior and posterior anatomy, this will constitute a major possible source of error. We always register each vertebra, except in total ankylosed spines with two neighboring uninjured vertebral levels, to avoid the “excursion phenomenon” (inaccuracy caused by differences in spinal alignment in the perioperative prone vs. supine position during preoperative CT scans). Finally, the reference clamp can be dislocated during surgery. Preoperative planning for navigated PS should address fracture morphology, bone density, spine degeneration, pedicle size and shape, and the course of the VAs.

There are numerous technical aids for the placement of PSs, ranging from basic free-hand techniques with 2D fluoroscopy to navigation by surface matching based on preoperative CT, intraoperative CT or 3D fluoroscopy, patient-specific 3D-printed templates, and robot-assisted surgery. Freehand 2D fluoroscopy-guided techniques for PS placement in the cervical spine require experience and can be especially demanding when the anatomy is altered by trauma or degeneration, in pediatric patients and in patients with anomalies. 2D fluoroscopy has significant limitations in the caudal cervical segments and cervicothoracic junction (6, 22). Intraoperative CT or 3D fluoroscopy provides a better view of this region. Recent studies comparing freehand and various navigation-assisted techniques have reported that the use of navigation-assisted techniques both increases accuracy and decreases radiation exposure to patients and staff (6, 32).

In this population-based cohort, navigated cervical PS fixation demonstrated high accuracy and a low rate of serious neurovascular complications. Although the placement of 10% of the screws was graded as suboptimal, the overall complication rate was low, and no patients developed new-onset spinal cord injury or screw pull-out. The risk of PS-related complications increased markedly with decreasing screw accuracy, underscoring the critical importance of meticulous preoperative planning and intraoperative technique. Navigation substantially facilitates safe PS placement but cannot eliminate anatomical and technical challenges inherent to the cervical spine. Our results add robust data to the literature and will be valuable for future systematic reviews and meta-analyses evaluating the safety and accuracy of navigated PS fixation for unstable cervical spine fractures.