Prediction Model for Failure of Nonoperative Management of Uncomplicated Appendicitis in Adults.

BACKGROUND: Prediction of failure of nonoperative management (NOM) in uncomplicated appendicitis (UA) is difficult. This study aimed to establish a new prediction model for NOM failure in UA. METHODS: We included 141 adults with UA who received NOM as initial treatment. NOM failure was defined as conversion to operation during hospitalization. Independent predictors of NOM failure were identified using logistic regression analysis. A prediction model was established based on these independent predictors. Receiver operating characteristic (ROC) curve analysis and the Hosmer-Lemeshow test were used to assess the discrimination and calibration of the model, respectively, and risk stratification using the model was performed. RESULTS: Among 141 patients, NOM was successful in 120 and unsuccessful in 21. Male sex, maximal diameter of the appendix, and the presence of fecalith were identified as independent predictors of NOM failure for UA. A prediction model with scores ranging from 0 to 3 was established using the three variables (male sex, maximal diameter of the appendix ≥ 15 mm, and the presence of fecalith). The area under the ROC curve for the new prediction model was 0.778, and the model had good calibration (P = 0.476). A score of 2 yielded a sensitivity of 71.4% and a specificity of 90.8%. Patients were stratified into low (0-1), moderate (2), and high (3) risk categories, which had NOM rates of 5.2%, 47.1%, and 77.8%, respectively. CONCLUSIONS: Our prediction model may predict NOM failure in UA with good diagnostic accuracy and help surgeons select appropriate treatments.