Introduction Thoracic pedicle screws are increasingly being used for internal fixation. Surgeons and radiologists are often required to make decisions about the position of the screws in relation to the pedicle based on AP and lateral plain radiographs alone. We ventured to assess the value of orthogonal radiographs in determining the position of thoracic pedicle screws in 23 cadaveric thoracic vertebrae

Methods Disarticulated cadaveric thoracic vertebrae were used in this study. Pedicle screws were inserted in three positions: 1) within the pedicle, deliberately violating the 2) lateral cortex of the pedicle and 3) medial cortex of the pedicle. AP (antero-posterior) & lateral radiographs were obtained and presented to 6 readers (4 surgeons & 2 radiologists) in booklets consisting of AP views alone, lateral views alone and both AP & lateral views together in a sequential manner. The readers were asked to indicate the position of the screws and the results of the evaluation were compared to the actual position (axial views).

Results On AP views alone, the accuracy in detecting screws placed out of the pedicle laterally and medially were 93% and 76% respectively, while the accuracy for screws placed inside the pedicle was 85% . On LATERAL views alone, the accuracy for the same screw positions were 69%, 58% and 64% respectively. When AP + LATERAL views were considered together, the accuracy for the same screw positions were 93%, 80% and 87% respectively. Comparing the three groups, it was observed that screw positions were read more accurately in AP + LATERAL views (87%) compared to AP views alone (85%), or LATERAL views alone (64%). The sensitivity of correctly identifying screws placement is highest in AP + LATERAL (90%) views with a specificity of 94%. The specificity of detecting screws placed inside the pedicle is highest in AP (94%). The positive predictive value (PPV) of radiographs in general (AP +LATERAL) in detecting screws placed inside the pedicle, out of the pedicle laterally and medially were 73%, 92% and 86% respectively. The negative predictive value (NPV) of radiographs in general for the same screw locations were 90%, 96% and 76% respectively. On AP and AP + LATERAL views respectively, 25% and 23% of screws placed inside the pedicle were read as medially ‘out’. 10% of screws placed medially ‘out’ were read as ‘in’ on both AP and AP + LATERAL views. Inter-observer difference was substantial. In general, surgeons appeared to have consistently higher accuracy, sensitivity, specificity, PPV and NPV values compared to radiologists and fellows in determining screw position.

Discussion The positions of the screws appear to be most accurately detected when both AP and lateral x-rays are provided compared to AP or lateral alone. Screws that perforated the lateral cortex were the easiest and those that were medially out were the most difficult to identify. Screws passed inside the pedicle may create an unnecessary apprehension that they may be medial and screws passed medially may give a false sense of security that the screw is inside the pedicle. Radiographs are just one component in ensuring accurate pedicle screw placement and surgeon’s experience, in the use of tactile skills and anatomical knowledge continue to be vitally important in the safe placement of thoracic pedicle screws.