Development and validation of nomogram model for postoperative enterocolitis using LASSO-logistic regression: single-center retrospective study of 424 Hirschsprung disease patients

Hirschsprung-associated enterocolitis (HAEC) in patients with Hirschsprung disease (HSCR) is a topic of concern. The aim of this study was to develop and validate a Nomogram model based on least absolute shrinkage and selection operator (LASSO) and Logistic regression for predicting postoperative HAEC in patients with HSCR. We retrospectively enrolled 424 HSCR patients who underwent one-stage Laparoscopic-assisted pull-through surgery. The cohort was randomly divided into a training cohort ( n = 296) and validation cohort ( n = 128) at a 7:3 ratio. LASSO regression identified significant predictors, followed by multivariable Logistic regression to develop the predictive model and generate a Nomogram. The model’s discrimination, accuracy, and clinical utility were assessed using receiver-operating characteristic (ROC) curves, Hosmer–Lemeshow test, and clinical decision curve analysis. The postoperative HAEC rate of 424 patients was 30.0% (127/424). LASSO regression identified 5 predictive factors: feeding method, preoperative nutritional status, caregiver relationship, L-HSCR, and postoperative complications within 30 days. Multivariable Logistic regression further confirms that feeding method (mixed: OR = 3.077, 95% CI, 1.130–8.276, P = 0.026; formula: OR = 7.517, 95% CI, 3.046–19.17, P < 0.001), preoperative nutritional status (malnutrition risk: OR = 3.812, 95% CI, 1.579–9.217, P = 0.003; malnutrition: OR = 4.385, 95% CI, 1.245–17.14, P = 0.025), non-parental care (OR = 4.618, 95% CI, 1.820–11.87, P = 0.001), L-HSCR (OR = 4.766, 95% CI, 1.511–16.30, P = 0.009), and postoperative complications within 30 days (mild: OR = 4.575, 95% CI, 1.752–12.09, P = 0.002; severe: OR = 9.348, 95% CI, 1.860–57.71, P = 0.010) were independent predictive factors for postoperative HAEC. A Nomogram model was constructed based on these factors. This model demonstrated strong predictive performance with the area under of the ROC curves (AUCs) of 0.910 (training) and 0.872 (validation). Hosmer–Lemeshow test and clinical decision curves verified the accuracy and practicability of the predictive model. This study constructed a Nomogram model for postoperative HAEC risk in HSCR patients. This model demonstrates strong predictive efficacy and has high value in clinical applications. However, multicenter prospective studies are required to further validate its generalizability.