Risk Stratification of Children Being Evaluated for Intussusception

Abstract

CONTEXT: Intussusception is the most common cause of intestinal obstruction in young children, and delayed diagnosis may lead to bowel perforation. OBJECTIVE: To determine predictive clinical criteria and develop a decision tree to risk-stratify children with possible intussusception. DESIGN/METHODS: This is a prospective observational cohort study of children aged 1 month to 6 years who presented with possible intussusception. A data-collection form was completed before knowledge of any advanced imaging. Univariate analysis was performed, and decision trees were developed using recursive partitioning. RESULTS: In the study, 310 patients were enrolled, including 38 (12.3%) with intussusception. The median age was 21.1 months and 61% were male. Univariate predictors of intussusception included age older than 6 months (P = 0.04), male gender (P = .007), history of lethargy (P = .001), and abnormal plain x-ray (P = .0001). Multivariate analysis through recursive partitioning identified decision trees (with and without the result of a plain abdominal x-ray) and allowed identification of patients at low risk. The decision tree based on the results of an abdominal x-ray (negative or positive), age (⩽5 or >5 months), diarrhea (present or absent), and bilious emesis (present or absent) had the best test performance (sensitivity: 97% [95% confidence interval (CI): 86–100]; negative predictive value: 99% [95% CI: 93–100]; negative likelihood ratio: 0.08 [95% CI: 0.01–0.6]). CONCLUSIONS: Among children who were being evaluated for intussusception, we prospectively determined clinical criteria and developed a decision tree to risk-stratify children with possible intussusception.