Construction and Validation of a Risk Prediction Model for Postoperative Constipation in Patients With Osteoporotic Thoracolumbar Fracture Undergoing Percutaneous Kyphoplasty

Objective To develop an instrument for predicting postoperative constipation risks in patients with osteoporotic thoracolumbar fracture (OTLF) who have undergone percutaneous kyphoplasty (PKP).

Methods A total of 858 OTLF patients who underwent PKP surgery between January 2020 and December 2024 were enrolled. The patients were randomly assigned to a training set (n = 600) and a validation set (n = 258) in a 7∶3 ratio. According to whether the patients had postoperative constipation, the training set was divided into a constipation group (n = 205) and a non-constipation group (n = 395), and the validation set was divided into a constipation group (n = 90) and a non-constipation group (n = 168). Logistic regression analysis was conducted to analyze the factors influencing postoperative constipation in OTLF patients after PKP, and a nomogram model was constructed accordingly. The receiver operating characteristic (ROC) curve and the calibration curve of the model were plotted, and the Hosmer-Lemeshow test for goodness of fit was performed.

Results A total of 205 OTLF patients (34.17%) in the training set and 90 OTLF patients (34.88%) in the validation set experienced constipation after PKP. Univariate analysis revealed significant differences between the constipation and non-constipation groups in terms of operative time, postoperative water intake, time to first postoperative meal, postoperative bed rest time, the levels of Bifidobacterium, Lactobacillus, Enterococcus, and Enterobacter, the Nutrition Risk Screening 2002 (NRS-2002) score, and the levels of sodium, potassium, and HbA1c (P < 0.05). Least absolute shrinkage and selection operator (LASSO) regression was performed and operative time, time to first postoperative meal, the levels of Bifidobacterium, Lactobacillus, Enterococcus, and Enterobacter, the NRS-2002 score, and the levels of sodium, potassium, and HbA1c were identified as candidate predictors. Multivariate logistic analysis showed that the time to first postoperative meal, the levels of Bifidobacterium and Lactobacillus, the NRS-2002 score, and the levels of sodium and HbA1c were influencing factors of postoperative constipation in OTLF patients (P < 0.05). The ROC curves showed that the area under the curve (AUC) of the training set was 0.842 (95% CI: 0.793-0.892), while that of the validation set was 0.860 (95% CI: 0.830-0.889). The calibration curves demonstrated good agreement between the prediction curve and the standard curve in both the training set and the validation set.

Conclusion The time to the first postoperative meal, the NRS2002 score, and the levels of Bifidobacterium, Lactobacillus, sodium, and HbA1c are influencing factors of post-PKP constipation in OTLF patients. The nomogram model built based on these factors exhibited good predictive performance.