Can Postoperative Radiographs Accurately Identify Screw Misplacements?

• Published in: Spine deformity 2017-03, Vol.5 (2), p.109
• Main Authors: Sarwahi, Vishal, Ayan, Saankritya, Amaral, Terry, Wendolowski, Stephen, Gecelter, Rachel, Lo, Yungtai, Thornhill, Beverly
• Format: Article
• Language: English
• ISSN: 2212-1358

Retrospective case series. The objective of this study was to determine the safety of postoperative radiographs to assess screw placement. Previously defined criteria are frequently employed to determine pedicle screw placement on intraoperative supine radiographs. Postoperatively, radiographs are typically used as a precursor to identify screws of concern, and a computed tomographic (CT) is typically ordered to confirm screw safety. First, available postoperative PA and lateral radiographs were reviewed by 6 independently blinded observers. Screw misplacement was assessed using previously defined criteria. A musculoskeletal radiologist assessed all CT scans for screw placement. Pedicle screw position was classified either as acceptable or misplaced. Misplacements were subclassified as medial, lateral, or anterior. One hundred four patients with scoliosis or kyphosis underwent posterior spinal fusion and had postoperative CT scan available were included. In total, 2,034 thoracic and lumbar screws were evaluated. On CT scan, 1,772 screws were found to be acceptable, 142 were laterally misplaced, 30 medially, and 90 anteriorly. Of the 30 medially placed screws, 80% to 87% screws were believed to be in positions other than medial, with a median of 73% (63% to 92%) of these screws presumed to be in normal position. Similarly, of the 142 screws placed laterally, 49% to 81% screws were identified in positions other than lateral, with a median of 77% (59% to 96%) of these screws felt to be in normal position. Of the 90 anteriorly misplaced screws, 16% to 87% screws were identified in positions other than anterior, with 72% (20% to 98%) identified as normal. The criteria produced a median 52% sensitivity, 70% specificity, and 68% accuracy across the 6 observers. Radiograph is a poor diagnostic modality for observing screw position. Level IV.