Session I - Pelvic Trauma

The Reliability, Reproducibility and Accuracy of Radiographic Interpretation of Sacroiliac Screw

Benjamin A. Goldberg, MD; Ronald W. Lindsey, MD; Christian Foglar, MD; Thomas D. Hedrick, MD; Theodore Miclau, MD; John L. Haddad, MD; Myrna Kahn, MS, Baylor College of Medicine, Houston, TX, University of California, San Francisco, CA

Twenty-four cannulated sacroiliac screws (12 titanium screws and 12 stainless steel screws) were placed bilaterally into twelve cadaveric pelvi and were imaged using both conventional axial and multiplanar reconstructed computed tomography (CT). Five attending orthopedic traumatologists interpreted the films on two separate occasions. The authors studied: 1) the interobserver reliability, intraobserver reproducibility, and accuracy of axial versus multiplanar reconstruction CT in the assessment of sacroiliac screw position relative to the neuroforamen. 2) the influence of using titanium screws versus stainless steel screws on the reliability, reproducibility and accuracy of the two imaging techniques.

In vitro sacroiliac joint screws were deliberately placed into one of three positions related to the first neuroforamen; Position A-screw was at least five millimeters outside the neuroforamen; Position B-screw was within 3 mm of the neuroforamen or less than 25 percent encroachment on the neuroforamen; and Position C-screw was nearly centered in the neuroforamen. Conventional (axial) images and multiplanar reconstructed images were performed on all specimens. The interobserver reliability, intraobserver reproducibility, and interpretation accuracy was determined for each imaging technique. The influence of titanium versus stainless steel screws on reliability, reproducibility and accuracy was determined.

The interobserver reliability of determining screw position relative to the neuroforamen was slight regardless of type of imaging modality or screw. The intraobserver reproducibility of determining screw position was moderate for both axial and multiplanar CT. However, when titanium screws were imaged with multiplanar CT, the intraobserver reliability increased to substantial levels. Axial CT was accurate in determining screw position relative to the neuroforamen in 27% of titanium screws and 42% of stainless screws. The multiplanar CT was accurate in determining screw position relative to the neuroforamen in 43% of titanium screws and 47% of stainless steel screws.

Although computed tomography with multiplanar reconstruction improves accuracy in determining sacroiliac screw position relative to the neuroforamen, significant inaccuracies, as well as significant interobserver and intraobserver disagreement exist in the radiographic interpretation of screw position. Likewise, the use of titanium versus stainless steel screws does not significantly improve reliability, reproducibility or accuracy in determining sacroiliac screw position.