Association between visceral adiposity index and delayed union/nonunion after tibial or femoral shaft fractures: a single-center retrospective cohort study.

BACKGROUND: Delayed union and nonunion remain clinically important complications after tibial and femoral shaft fractures. Although traditional risk factors such as smoking and diabetes have been widely investigated, the association between visceral adiposity-related metabolic dysfunction and impaired fracture healing remains unclear. This study aimed to examine the association between the visceral adiposity index (VAI) and delayed union/nonunion after tibial or femoral shaft fractures and to explore whether VAI could improve internally validated prediction models. METHODS: We conducted a single-center retrospective cohort study of 485 adults who underwent intramedullary fixation for isolated tibial or femoral shaft fractures between January 2022 and June 2025. VAI was calculated using sex-specific equations incorporating waist circumference, body mass index, triglycerides, and high-density lipoprotein cholesterol. The primary endpoint was delayed union/nonunion at 6 months after surgery. The association between VAI and delayed union/nonunion was assessed using multivariable logistic regression and restricted cubic spline analysis. Discriminative performance was compared using receiver operating characteristic analysis, and exploratory machine learning models were evaluated using internal validation. RESULTS: Delayed union/nonunion occurred in 60/485 patients (12.4%). The median VAI was higher in patients with delayed union/nonunion than in those who achieved fracture union [3.1 (IQR, 2.2-4.5) vs. 1.9 (IQR, 1.3-2.8), P < 0.001]. After full adjustment for demographic, lifestyle, metabolic, and fracture-related covariates, each 1-unit increase in VAI was associated with higher odds of delayed union/nonunion (OR 1.61, 95% CI 1.31-1.98; P < 0.001). Restricted cubic spline analysis suggested an approximately linear association, without significant evidence of nonlinearity. VAI showed higher discrimination than body mass index and waist circumference (AUC 0.785 vs. 0.655 and 0.712, respectively). In exploratory internal validation, L1/Elastic Net-regularized logistic regression showed the highest AUC among the evaluated machine learning models (AUC 0.914, 95% CI 0.864-0.963). CONCLUSION: In this single-center retrospective cohort, higher VAI was associated with an increased risk of delayed union/nonunion after tibial or femoral shaft fractures treated with intramedullary fixation. VAI showed better discrimination than conventional anthropometric measures in internal analyses and may provide additional information for early risk stratification. However, these findings do not establish causality, and external prospective validation is required before clinical implementation.