Development of Prediction Model for Hirschsprung-Associated Enterocolitis (HAEC) in Postoperative Hirschsprung Patients.

BACKGROUND

Despite identifying numerous risk factors for postoperative Hirschsprung-associated enterocolitis (HAEC), predicting individual risk remains challenging. This study aimed to develop a clinical prediction model for predicting the probability of postoperative HAEC within 5 years after surgery in Hirschsprung individuals.

METHODS

The study included all children with Hirschsprung disease who underwent definitive surgery at Chiang Mai University Hospital from 2006 to 2021. Concomitant anorectal abnormalities and incomplete data were excluded. A multivariable logistic regression analysis, adjusted for correlated data, was utilized to develop the prediction model.

RESULTS

Of the included 274 patients, 75 patients (27.4%) experienced postoperative HAEC within 5 years, totaling 121 episodes. Based on statistical and theoretical significance, eight parameters were utilized as predictors, which included male (OR1.23,95%CI:0.53-2.86), trisomy21(OR1.34,95%CI:0.21-8.45), weight at the time of surgery (OR0.86,95%CI:0.73-1.02), absence of exclusive breastfeeding (OR1.51,95%CI:0.65-3.51), length of the aganglionic segment (rectosigmoid (OR1.32,95%CI:0.48-3.62), long segment (OR41.39,95%CI:3.00-570.37), and total colonic aganglionosis (OR710.20,95%CI:23.55-21420.72)), preoperative stoma (OR1.72,95%CI:0.34-8.58), surgical approach (Duhamel (OR0.06,95%CI:0.01-0.81) and abdominal assisted trans anal endorectal pull-through (OR0.04,95%CI:0.002-0.65)), and early HAEC before two weeks following surgery (OR1.98,95%CI:0.67-5.82). The derived predictive model exhibited acceptable discriminative performance (AuROC:0.749,95%CI:0.679-0.816). Risk groups were categorized into low and high-risk, with positive likelihood ratios of 0.65 and 10.70, respectively. Recommendations for management and follow-up were generated based on these risk groups. An online application has been developed for calculating individual risk of postoperative HAEC and offering management suggestions with follow-up schedule: [https://w1.med.cmu.ac.th/surgery/personnel/pedsurgerycmu/#HAEC-Calculator].

CONCLUSIONS

This risk predictive model accurately estimates the probability of postoperative HAEC within 5 years after surgery in Hirschsprung patients. It facilitates risk stratification and provides personalized recommendations to parents for the prevention and early detection of postoperative HAEC.

LEVELS OF EVIDENCE

Level II Retrospective cohort study (Prognosis study).