Construction of nomogram based on clinical factors for the risk prediction of postoperative complications in children with choledochal cyst.

ABSTRACT

Objective: The aim of the study was to develop a prediction nomogram based on clinical factors to assess the risk of postoperative complications in children with congenital choledochal cyst. Methods: The clinical data from 131 children who underwent choledochal cyst resection and Roux-en-Y hepaticojejunostomy in our hospital between January 2016 and December 2022 were retrospectively analyzed. The general information, clinical symptoms, procedure, biochemical indicators, and imaging data were recorded. A prolonged hospital stay induced by postoperative complications or a follow-up over 6 months was assessed as the event outcome. A logistics regression analysis was performed to screen for risk factors with statistical significance in inducing postoperative complications. Then, with the dataset split into the training group and internal validation group, the nomogram for the prediction of postoperative complications was developed based on a computer algorithm. In addition, the receiver operating characteristic (ROC) curve and calibration curve were performed for nomogram verification. Results: Of 131 children, the multivariate logistics regression analysis suggested that age ≤2 years [odds ratio (OR) 0.93; 95% confidence interval (CI) 0.15-5.65; p = 0.938], Todani classification type 1 (OR 36.58; 95% CI 4.14-871.74; p = 0.005), cyst wall thickness >0.4 cm (OR 10.82; 95% CI 2.88-49.13; p < 0.001), with chronic cholecystitis (OR 7.01; 95% CI 1.62-38.52; p = 0.014), and choledochal cyst diameter (OR 1.01; 95% CI 0.99-1.03; p = 0.370) were predictors associated with the postoperative complications of choledochal cysts. The data were randomly divided into the training group (n = 92) and internal validation group (n = 39) to build the prediction nomogram including the appeal factors. The accuracy and discrimination of the model were evaluated using a ROC curve and calibration curve. The results showed that the nomogram area under the ROC curve [area under the curve (AUC) = 0.894; 95% CI 0.822-0.966; p < 0.001], validation (AUC = 0.844; 95% CI 0.804-0.952; p < 0.001), and Brier = 0.120 (95% CI 0.077-0.163p; p < 0.001) were indicative of the good stability and calibration of the predictive nomogram. Conclusion: The prognosis of congenital choledochal cysts was associated with multiple aspects of clinical factors. Combined with the internal validation, the novel prediction nomogram was suitable for evaluating the individualized risk of postoperative complications of choledochal cysts. The prediction nomogram could provide a more accurate strategy of procedure and postoperative follow-up for children with choledochal cysts.