In our study, postoperative CT images were acquired after a month of pedicle screw fixation surgery. The incidence rate of screw violation identified in previous studies ranged from 0% to 42% [12]. This wide variety was due to various reasons such as the indication for surgery, the involved spinal levels, the operative procedure, the intraoperative-assisted guidance, the surgical experience, and the tool of postoperative assessment [13]. In our study, these issues have been standardized as much as possible.
In this study, the frequency of screw violation was 15.88% which was within the reported range of published studies. Laine et al. [14] reported 35.22% of screw violation, and Castro et al. [15] detected cortical penetration in 40% of their patients. Our frequency of screw violation was lower than that reported in these studies.
On the other side, our frequency was higher than that reported in other studies [16,17,18,19]. This difference may be explained by that many authors considered screw displacement of <2 mm was insignificant, and they did not include this group in the their final frequency. In our study, we included all grades of displacement even those with <2 mm in the total rate of screw displacement. Elyan et al. [20] and El Fiki et al. [21] reported screws displacement rates of 17.3% and 16.45%, respectively, which were similar to that found in our study.
Also, lower rates of screw displacement were reported in a meta-analysis conducted by Schizas et al. [22]. They reported a mean displacement rate of 8.7%. Lonstein et al. [23] also reported a displacement rate of 5.1% that has been related to the use of conventional plain radiographs as a postoperative assessment tool, with underestimation of displacement rate compared to CT scan [24].
Other studies used two dimensional (2D) axial CT images for postoperative assessment of pedicle screws. Adding 3D VR or coronal MPR procedures provided more detailed data in evaluating the placement of pedicle screws [15, 25, 26].
Multi-slice 3D reformatted image displayed the full length of the screw with absence of metallic artifacts and accurate assessment of the relation between the screw and neural structures [24]. In our study, axial images detected 36 (10.59%) out of the displaced screws seen at coronal reconstruction of 48 (14.12%) and 3D reformatted image of 54 (15.88%).
Our results agree with Schizas et al. [22] who compared axial CT images with coronal reconstruction and reported that the incidence of displacement was 23.3% for axial and 30% for coronal images. Also, in the study that was conducted by Celik et al. [27], the authors compared the axial images with 3D reformatted images, and they found lower incidence for axial (9%) compared to that of 3D reformatted images (13.1%). And, they explained that this difference is due to the fact that reformatted images evaluate cranial and caudal placement accuracy in addition to medial and lateral screw placement. Also, the presence of metallic artifacts in axial images affected the image interpretation.
In our study, the rate of screw displacement was higher in dorsal pedicles (31.4%) as compared with that in the lumbo-sacral ones (11.9%). This could be explained by the small size of pedicles in the dorsal spines and increasing variability in their anatomy [11, 21, 24, 28]. This was in accordance with previous studies [19, 21, 29, 30] that reported a higher rate of screw displacement in dorsal than lumbar spines. We revealed the highest incidence of screw displacement was at T11 (40%) followed by T9 (30%). This could be explained by the relatively small diameter of pedicles of dorsal spine to fit the screw.
In our study, the rate of lateral violation was (57.41%) more than the medial one (20.37%). Our results agree with many authors [5, 21, 29, 31] who reported a higher rate of lateral violation, and this is because the surgeons choose a more lateral track away from the more dense medial cortical wall of the pedicle and also to avoid spinal cord injury.
In this study, pedicle penetration was more frequently in lateral and medial walls and less in inferior (12.96%) and superior ones (9.26%). This was because of the pedicle anatomy [32]. These results were in accordance with that shown by previous studies [1, 19, 21, 33].
In our study, we reported neurologic sequels in seven of patients (10%). This was within the range reported in previous studies (0.2% to 11%) [21, 34]. While our results were less than that reported by Lotfnia et al. [31] (15.09%), the higher incidence in the later study might be due to the involvement of all patients with sensory or motor symptoms in spite of the existence or absence of neurological deficits.
Not all patients with displaced screws need revision surgery [21, 31]. This agreed with our results where only three patients out of seven need revision surgery with a revision rate of 1.47% among the inserted screws. Also, our results are in agreement with Parker et al. [29] who reported a 0.8% revision rate and completely agreed with El Fiki et al. [21]. No definitive dural, vascular, or pulmonary injuries were reported in our patients. These were in accordance with El Fiki et al. [21], Rodrigues et al. [30], and Reidy et al. [34].
Sensitivity of CT images in assessing screw position compared with operative findings in five revised screws revealed a sensitivity of 100% and 97.6% and a specificity of 100% and 89.4% for 3D reformatted image and axial image, respectively. These results were in accordance with El Fiki et al. [21], who reported sensitivity of 100% and 95.8% and a specificity of 100% and 88.7% for 3D reformatted image and axial image, respectively. Elyan et al. [20] reported sensitivity of 100%, 66.7%, and 33.3% and a specificity of 100%, 76%, and 38.5% for 3D reformatted, coronal, and axial images, respectively, among 24 cases. The lower values of sensitivity and specificity for coronal and axial images in the later study may be explained by the small number of studied cases.