In vivo accuracy of thoracic pedicle screws

Study Design. A retrospective observational study of 279 transpedicular thoracic screws using postoperative computed tomography (CT). Objective. To determine the accuracy of transpedicular thoracic screws. Summary of Background Data. Previous studies have reported the importance of properly placed transpedicular thoracic screws. To our knowledge, the in vivo accuracy of pedicle screw placement throughout the entire thoracic spine by CT is unknown. Methods. The accuracy of thoracic screw placement within the pedicle and vertebral body and the resultant transverse screw angle (TSA) were assessed by postoperative CT. Cortical perforations of the pedicle were graded in 2-mm increments. Screws were regionally grouped for analysis. Results. Forty consecutive patients underwent instrumented posterior spinal fusion using 279 titanium thoracic pedicle screws of various diameters (4.5-6.5 mm). The regional distribution of the screws was 39 screws at T1-T4, 77 screws at T5-T8, and 163 screws at T9-T12. Fifty-seven percent of screws were totally confined within the pedicle. Although medial perforation of the pedicle wall occurred in 14% of screws, in <1% there was >2 mm of canal intrusion. Lateral pedicular perforation occurred in 68% of perforating screws and was significantly more common than medial perforation (P < 0.0005). Seventeen screws penetrated the anterior vertebral cortex by an average of 1.7 mm. Screws inserted between T1 and T4 had a decreased incidence of full containment within the pedicle (P < 0.0005) and vertebral body (P = 0.039) compared with T9-T12. The mean TSA for screws localized within the pedicle was 14.6 degrees and was significantly different from screws with either medial (mean 18.0 degrees) or lateral (mean 11.5 degrees) pedicle perforation (P < 0.0005). Anterior vertebral penetration was associated with a smaller mean TSA of 10.1 degrees (P = 0.01) and with lateral pedicle perforation (P < 0.0005). There were no neurologic or vascular complications. Conclusions. Ninety-nine percent of screws were fully contained or were inserted with either less than or equal to2 mm of medial cortical perforation or an acceptable lateral breech using the "in-out-in" technique. Anterior cortical penetration occurred significantly more often with lateral pedicle perforation and with a smaller mean TSA. The incidence of fully contained screws was directly correlated with the region of instrumented thoracic spine.