Accuracy of intraoperative plain radiographs to detect violations of intralaminar screws placed into the C2 vertebrae: A reliability study

STUDY DESIGN. Cadaveric radiographic interpretation analysis. OBJECTIVE. To evaluate the reliability of plain radiographs for detecting potentially serious C2 ventral laminar violations during insertion of intralaminar screws. SUMMARY OF BACKGROUND DATA. With the potential serious complications (due to aberrant course of the vertebral artery) associated with insertion of C1-C2 transarticular screws, coupled with inherent difficulties with inserting either C2 pedicle or pars/isthmus screws, the intralaminar technique has recently gained popularity. METHODS. After visually confirming that we had placed bilateral screws entirely in the lamina in 12 cadaveric specimens, we obtained lateral radiographs. We repeated the screw insertions and radiographs for each of the following permutations: right in/left out, left in/right out, and both screws out. Three experienced spine surgeons blindly/independently assessed all 48 radiographs 3 times. A screw projecting into the lateral mass was read as having perforated the ventral lamina. Accuracy was defined as percent agreement between actual and read position for 288 screws (48 L, 48 R times 3 raters). RESULTS. Accuracy was 77.4%. The highest accuracy was with both screws in (86.1%); the lowest with both out (63.9%). Left in-right out was 77.8%. Right in-left out was 81.9%. There was significantly higher incorrect identification of both out (rated "both in" in all but one; χ = 11.5, P < 0.05). Interrater reliability was moderate (k = 0.42) and sufficient for clinical use. Intrarater reliability was excellent for all three (k = 0.98, 1.00, 0.93). CONCLUSION. With both screws out, the accuracy was only 63.9%, suggesting that in over 36% of bilateral screw perforations, plain radiographs cannot detect the violation. While some of these might not cause clinical symptoms, others might cause subtle radiculopathy, myelopathy, or headaches. The presence of any of these should be worked up with a postoperative CT scan, as plain radiographs are not dependable.