Risk stratification of pedicle screw malposition in «free-hand» technique in scoliosis surgery. A retrospective study

Background. Posterior correction of pediatric spinal deformities using multisegmental pedicle screw fixation is the gold standard for surgical treatment of scoliosis. However, placement accuracy with the «free-hand» technique varies, and the risk of screw malposition depends on anatomical complexity and deformity parameters, creating the need for a simple preoperative planning tool for risk stratification.

Aim. To develop a preoperative planning tool based on radiographic deformity parameters to predict the risk of pedicle screw malposition with the «free-hand» technique.

Materials and methods. A single-center retrospective study was performed. Forty-five patients who underwent posterior spinal deformity correction with «free-hand» pedicle screw insertion were included; a total of 696 screws were analyzed. Females accounted for 73.3% (n=33) and males for 26.7% (n=12), median age was 14.0 [12.0; 16.0] years. Idiopathic scoliosis predominated (71.1%, n=32). Screw position was assessed on postoperative CT with multiplanar reconstructions, recording malposition and breach direction (medial, lateral, anterior). Preoperatively, vertebral frontal tilt, sagittal tilt, and vertebral rotation were measured at each instrumented level. A predictive model was built using binary logistic regression with Nagelkerke’s R²; discriminative performance was evaluated by ROC analysis (AUC) and the optimal cut-off was determined using the Youden index.

Results. The median preoperative Cobb angle was 69° [61°; 93°]; the postoperative mean Cobb angle was 28±12°, corresponding to a 59% mean correction (p<0.001). CT identified 148 malpositions (21.3% of screws). Malposition rates increased significantly with increasing frontal tilt and vertebral rotation (p<0.001). Lateral malpositions were significantly more frequent with greater frontal tilt (p<0.001), whereas medial malpositions increased with higher rotation compared with anterior and lateral breaches (p<0.001). In the binary logistic model, both coronal tilt and rotation were significant predictors; the model differed from the null model (p<0.001) with a Nagelkerke pseudo-R² of 10.3%. Each 1° increase in coronal tilt increased the odds of malposition by 2.7% (OR 1.027; 95% CI 1.013–1.042), and each 1° increase in rotation increased the odds by 5.0% (OR 1.050, 95% CI 1.030–1.070). The AUC was 0.682 (95% CI 0.631–0.733; p<0.001). The optimal probability threshold was P=0.212, yielding 70.5% sensitivity and 62.0% specificity; P≥0.212 corresponded to a combination of coronal tilt and rotation of approximately 20° or more at the same instrumented level.

Conclusion. The risk of pedicle screw malposition with the «free-hand» technique is significantly associated with vertebral coronal tilt and rotation at the instrumented level, with rotation exerting a stronger effect. When coronal tilt and rotation are approximately 20° or greater, the use of CT-based navigation or additive manufacturing (3D-printing) technologies should be considered, as well as modification of the fixation strategy (e.g., use of hook constructs or omission of screw placement at that level) to improve the safety of pediatric scoliosis correction.