Risk stratification and clinical classification for postoperative neurological complications in post-tuberculosis kyphosis: a retrospective cohort study.

PURPOSE

This study aimed to identify risk factors for postoperative neurological complications in patients with post-tuberculosis kyphosis undergoing posterior corrective surgery and to develop a classification system for preoperative risk stratification.

METHODS

We retrospectively analyzed 51 patients with post-tuberculosis kyphosis who underwent single-stage posterior osteotomy and correction at our institution. Radiographic parameters, including the kyphotic angle, cross-sectional area ratio of the spinal cord (CSAR), and spinal cord angle (SCA), alongside surgical factors such as intraoperative blood loss and osteotomy grade, were evaluated. Postoperative neurological complications were recorded. Univariate and multivariate logistic regression analyses were employed to identify potential influencing factors. Based on the interaction between these factors, a clinical classification system for post-tuberculosis kyphosis was established to stratify the risk of postoperative neurological complications.

RESULTS

A minimum 2-year follow-up was conducted for the 51 patients with post-tuberculosis kyphosis. Multivariate logistic regression analysis revealed that the Baltalimani sign, spinal cord MRI type, Rajasekaran classification, three-column osteotomy, C-reactive protein (CRP), and SCA were significant risk factors for postoperative neurological complications. Two risk prediction models were developed accordingly. Synergistic interactions were identified between spinal cord MRI type and Rajasekaran classification, as well as between CRP and SCA. These findings informed the development of two clinical classification systems: the Mechanical Classification (AUC = 0.751) and the Imaging-Biomarker Classification (AUC = 0.883). The area under the curve (AUC) for both classifications demonstrated good predictive performance, with the DeLong test indicating superior efficacy for the Imaging-Biomarker Classification (P = 0.039).

CONCLUSION

Spinal structural instability, spinal cord compression, osteotomy grade, and elevated CRP levels may all contribute to an increased risk of postoperative neurological complications in patients with post-tuberculosis kyphosis. The clinical classification systems established herein facilitate precise risk stratification, enabling targeted preoperative interventions to mitigate surgical risks and enhance surgical outcomes.