Computed tomography assessment of sacroiliac screw placement relative to the first sacral neuroforamen

The radiographic interpretation of sacroiliac screws relative to the S1 neuroforamen is difficult for orthopedic surgeons and radiologists. Computed tomography (CT) with axial images alone or combined with multiplanar reconstructions are often used to assess screw position. The reliability, reproducibility, and accuracy of orthopedist and radiologist interpretations of axial CT images with and without multiplanar reconstructions was determined using 24 cadaveric hemipelves with known sacroiliac screw position. Interobserver reliability of determining screw position was fair for orthopedists and slight for radiologists regardless of imaging modality or screw composition. Intraobserver reproducibility was moderate for orthopedists regardless of imaging modality or screw type. Reproducibility among radiologists was moderate using axial images of titanium screws and substantial with addition of multiplanar reconstructions. Overall accuracy was similar for orthopedists and radiologists. CT images with multiplanar reconstructions improve accuracy in determining sacroiliac screw position, but not significantly. Current imaging modalities are limited by large inaccuracies and by interobserver and intraobserver variation.