Stepwise Methodology for Plain Radiographic Assessment of Pedicle Screw Placement: A Comparison With Computed Tomography

Abstract

The objective of this study is to evaluate the effectiveness of a specific methodology for plain radiographic assessment of lumbar pedicle screw position. To evaluate the effectiveness of using orthogonal plain radiographs and a systematic method of interpretation, developed by the senior author, in assessing the placement of lumbar and lumbosacral pedicle screws. This was an adult cadaver study of the accuracy of using plain radiographs or computed tomography to assess pedicle screw position. Plain radiographs were performed and compared with computed tomography (CT) scans. Gross anatomic dissections were performed to directly confirm screw position. Variables, including screw material, radiographic view, and screw dimensions, were assessed for their effect on the ability of physicians to determine pedicle screw position. Multiple readers were included in the study, including 1 spine Fellow, 3 experienced orthopedic spine surgeons, and 1 neuroradiologist. Five adult cadaveric spines were instrumented with titanium pedicle screws from L1 to S1. Screws were placed outside the confines of the pedicle in all 4 quadrants or within the pedicle using a Latin-Square design. Each cadaver was imaged with orthogonal radiographs and high-resolution CT scans. The spines were then reimaged after the instrumentation was replaced with stainless steel screws placed in the identical position. Finally, each spine was dissected to assess the exact position of the screws. Images were read in a blinded fashion by 1 spine fellow, 2 staff surgeons, and a staff radiologist. The results were compared with the known screw positions at dissection. In total, 120 pedicle screws were placed, 44 (38%) outside the confines of the pedicle. Sensitivity, defined as the percent of the misplaced screws that were correctly identified, was similar across the 3 diagnostic tests, but markedly improved when all CT formats were considered together. Similarly, specificity, defined as the percent of screws correctly read as being placed within the pedicle, was independent of radiographic examination. Sensitivity of the radiographic technique was 70.1% and specificity was 83.0%, whereas sensitivity for CT scans was 84.7% and specificity was 89.7%. There was an observed association with anatomic level, with a consistently less accuracy in detecting screw position at L1 with plain x-ray (P=0.001). Additionally, correct position of stainless steel screws was more difficult to detect as compared with titanium (P=0.033) using either x-rays or CT. Other variables examined, such as screw length and screw diameter, did not have an effect on the ability to read the positioning. CT scans, often considered the "gold standard" for clinical assessment of pedicle screw placement, have limitations when validated with gross anatomical dissection. The described systematic method for evaluating pedicle screw placement using orthogonal plain radiographs attained accuracy comparable to high-resolution CT scans.