A scoring system to predict the severity of appendicitis in children.

BACKGROUND It appears that two forms of appendicitis exist. Preoperative distinction between the two is essential to optimize treatment outcome. This study aimed to develop a scoring system to accurately determine the severity of appendicitis in children. MATERIALS AND METHODS Historical cohort study of pediatric patients (aged 0-17 y old) with appendicitis treated between January 2010 and December 2012. Division into simple, complex appendicitis, or another condition based on preset criteria. Multiple logistic regression analysis was used to build the prediction model with subsequent validation. RESULTS There were 64 patients with simple and 66 with complex appendicitis. Five variables explained 64% of the variation. Independent validation of the derived prediction model in a second cohort (55 simple and 10 complex appendicitis patients) demonstrated 90% sensitivity (54-99), 91% specificity (79-97), a positive predictive value of 64% (36-86), and an negative predictive value of 98% (88-100). The likelihood ratio+ was 10 (4.19-23.42), and likelihood ratio- was 0.11 (0.02-0.71). Diagnostic accuracy was 91% (84-98). CONCLUSIONS Our scoring system consisting of five variables can be used to exclude complex appendicitis in clinical practice if the score is <4.