Accuracy of plain radiographs to identify wrong positioned free hand pedicle screw in the deformed spine

The accuracy rate of pedicle screws placement in the deformed spine can be easily assessed on computed tomography (CT), while it is difficult to be evaluated in conventional radiography in the posterior-anterior (PA) and lateral projections, even if they are an essential step to identify wrong positioned screws after surgery. Aim of the study is to evaluate the accuracy of plain radiographs compared with CT in identifying wrong positioned pedicle-screws in the deformed spine.

A total of 1125, pedicle screws implanted with free hand technique in 79 patients surgically treated for scoliosis with intraoperative/postoperative. Plain radiographs and CT of the spine were investigated. The pedicle screws location was evaluated by three independent spine surgeon with more than 10 years' experience, using the method described by Kim in plain radiographs. Other three independent spine surgeon with more than 10 years' experience, unknowing the previous results, evaluate the same pedicle screws using the Rongming Xu criteria in CT scans. When there is a disagreement among the readers is chosen the most common classification. Data were finally compared and analyzed using SPSS® 11.0 software.

Comparative analysis of pedicle screws using postoperative CT and plain radiographs showed: 22 true positives, i.e. pedicle-screws considered as out both in plain radiographs and CT scans; 1048 true negatives, i.e. pedicle-screws evaluated as in both in X-ray and CT scans; 9 false positives, i.e. pedicle-screws considered as out in plain radiographs but defined in in CT scans, and 52 false negatives, i.e. pedicle-screws considered as in in plain-radiographs, but defined out in CT. The accuracy of standard radiographs in detecting the placement of pedicle-screws amounts to 94.6%, with a sensitivity of 71% and a specificity of 95.3%.

Even if only 61 pedicle-screws out (5.4%) were not correctly identified in plain radiographs, none dangerous placement (3 cases) is unrecognized, underling that the CT accuracy is higher only to detect screws with a "safe" wrong placement that, according to literature data, not require revision surgery. Intraoperative X-ray, allowing a possible revision of misplaced screws during surgery, must be considered as the gold standard for pedicle screw evaluation. Even if postoperative CT scanning should not be performed as a routine control measure, it still useful in case of clinically suspected screw wrong positioning.