Evaluation of pedicle screw placement in the deformed spine using intraoperative plain radiographs: a comparison with computerized tomography.

STUDY
DESIGN: A retrospective study.
OBJECTIVE: To develop an accurate and reliable method to detect malpositioned pedicle screws during thoracic and lumbar spinal deformity operations using intraoperative plain radiographs. SUMMARY OF BACKGROUND DATA: The reliability of pedicle screw assessment using plain radiographs is more difficult during scoliosis operations compared to nondeformed spine operations. Methodology is necessary to document and improve the accuracy of interpretation of intraoperative plain radiographs for deformity surgeries.
METHODS: A total of 789 pedicle screws, including 632 thoracic and 157 lumbar, inserted from T1 to L4 in 49 patients with spinal deformity with postoperative computerized tomography (CT) data were investigated. According to the diagnoses, the number of screws placed was 683 for scoliosis in 43 patients and 106 for kyphosis in 6 patients. The position of the pedicle screw inserted was graded with CT as an acceptable screw (n = 724) versus violated screw (n = 65), defined as the central axis of the inserted pedicle screw out of the outer cortex of the pedicle wall. There were 3 plain radiographic criteria used to judge the accuracy of screw position after minor screw tip position adjustment according to the relative length of the screws in the lateral radiograph: (1) violation of the harmonious segmental change of the tips of the inserted screws with reference to vertebral rotation using the posterior upper spinolaminar junction in the posteroanterior (PA) radiograph (medial or lateral out); (2) no crossing of the medial pedicle wall by the tip of the pedicle screw inserted with reference to the vertebral rotation using the posterior upper spinolaminar junction in the PA radiograph (lateral out); and (3) violation of the imaginary midline of the vertebral body using the posterior upper spinolaminar junction in the PA radiograph by the position of the tip of the inserted pedicle screw (medial out).
RESULTS: Comparative analysis of these pedicle screws using postoperative CT and intraoperative plain radiographs confirmed 65 violated pedicle screws, including 15 medial violations and 50 lateral violations. Of 15 pedicle screws with medial wall violation, 13 showed a loss of harmonious segmental change in the screw tips and violation of the imaginary midline of the vertebral body (sensitivity 0.87, specificity 0.97, and accuracy 0.98). One case showed only a loss of harmonious change in the screw tip, and the other one case did not show any significant plain radiograph findings. Of the 50 pedicle screws with lateral wall violation, 47 cases showed a loss of harmonious segmental change in the screw tips and no crossing of medial pedicle wall by the pedicle screw inserted (sensitivity 0.94, specificity 0.90, and accuracy 0.96). Two cases did not show any significant plain radiograph findings, and the other one case showed only violation of the harmonious segmental change in the screw tips.
CONCLUSIONS: Intraoperative plain radiographs alone using 3 radiographic criteria were very sensitive to detect lateral wall pedicle screw violations, specific for assessing for medial wall violations, and highly accurate for both. This result confirms the ability of careful intraoperative plain radiographic assessment after pedicle screw insertion to detect malpositioned screws, to allow for possible revision during the index operation.