Inter- and intraobserver reliability of computed tomography in assessment of thoracic pedicle screw placement

Study Design. Reliability study of computed tomography imaging in 12 cadaver specimens instrumented with titanium or stainless steel thoracic pedicle screws. Objective. To evaluate inter- and intraobserver reliability of computed tomography scan in determining the accuracy of thoracic pedicle screw placement and to identify the differences in observers' agreement when viewing stainless steel versus titanium screws. Summary of Background Data. Computed tomography is often used to assess the accuracy of pedicle screw placement. Accuracy of screw placement is important in the thoracic spine where pedicle morphometry increases the difficulty of screw placement (vital structures are at increased risk). The current literature lacks a critical evaluation of computed tomography reliability among observers. Methods. Twelve adult cadavers were instrumented with thoracic pedicle screws. Nine cadavers were instrumented with titanium screws and three with stainless steel screws. The spines were imaged with computed tomography. Three observers used a grading scale to score the extent of pedicle violation and independently scored the placement of each pedicle screw on three separate occasions. Interobserver and intraobserver agreement were determined by using the kappa statistic. Results. The mean kappa score for interobserver agreement for all 12 specimens (including titanium and stainless steel screws) was 0.51, which correlates with a moderate degree of agreement. Although the interobserver kappa statistics for titanium ( k = 0.53) and stainless screws (k = 0.44) showed a moderate degree of agreement, the intraobserver reliability was substantial (k = 0.63). The mean intraobserver kappa for titanium screws was 0.63 and for stainless steel screws was 0.62. Conclusions. Our data show that interobserver agreement is moderate and intraobserver agreement is substantial when computed tomography is used to assess placement of thoracic pedicle screws. We conclude that computed tomography is reliable for evaluating thoracic pedicle screw placement throughout the thoracic spine. [Key words: thoracic pedicle screw, computed tomography, interobserver agreement, intraobserver agreement, reliability, titanium screws, stainless steel screws.