Factors associated with postoperative respiratory failure in hospitalized patients with thoracolumbar fractures: a retrospective cross-sectional study based on the MIMIC-IV database.

Background: Postoperative respiratory failure is a serious complication following thoracolumbar fracture surgery, yet research in this specific population remains scarce. This retrospective cross-sectional study used the Medical Information Mart for Intensive Care IV (MIMIC-IV) database to identify factors associated with postoperative respiratory failure in hospitalized patients with thoracolumbar fractures. Methods: Postoperative respiratory failure was identified using diagnosis codes recorded during hospitalization. The variance inflation factor (VIF) test was used to assess collinearity among the collected variables. Least absolute shrinkage and selection operator (LASSO), random forest, and Adaptive Boosting (AdaBoost) regression analyses were used to rank variable importance. A Venn chart was used to determine the common variables identified by the three screening methods. Univariate and multivariate logistic regression were used to determine the factors associated with postoperative respiratory failure. The receiver operating characteristic (ROC) curve and nomogram analysis were used to evaluate discriminative performance. All statistical tests were two-sided. Results: A total of 726 patients were included in this study, and 110 patients experienced respiratory failure. Among 25 candidate variables, 10 were identified as key candidates for respiratory failure according to their importance. On univariate analysis, pneumonia [odds ratio (OR) =11.489, 95% confidence interval (CI): 7.173-18.401; P<0.001], white blood cell (WBC) count (OR =1.086, 95% CI: 1.050-1.123; P<0.001), bicarbonate (OR =0.899, 95% CI: 0.854-0.946; P<0.001), calcium (OR =0.602, 95% CI: 0.463-0.782; P<0.001), potassium (OR =1.501, 95% CI: 1.164-1.935; P=0.002), and age (OR =1.013, 95% CI: 1.002-1.023; P=0.01) were associated with respiratory failure. In multivariate analyses, pneumonia and calcium remained independently associated with respiratory failure in all three models [for model 3, OR =7.532, 95% CI: 4.447-12.758, P<0.001; and OR =0.610, 95% CI: 0.443-0.841, P=0.003, respectively], whereas WBC count and potassium remained associated in most models. The area under the ROC curve (AUC) of the comprehensive model combining pneumonia, calcium, WBC count, and potassium was 0.798 (95% CI: 0.746-0.845). Conclusions: Pneumonia, calcium, WBC count, and potassium were factors associated with postoperative respiratory failure in patients with thoracolumbar fractures. The combined model based on these factors showed acceptable discrimination.