Identifying Perforated Appendicitis Preoperatively: A Rule-In Risk Stratification Model

Abstract

Background

Accurately predicting perforated appendicitis (PA) preoperatively remains challenging.

Methods

We retrospectively studied appendectomy patients with histopathologically confirmed acute appendicitis (2022-2023) to identify predictors of perforation. The primary outcome was histopathologically confirmed perforated appendicitis. Bayesian univariate analysis and Bayesian logistic regression were performed to estimate risk probabilities, with frequentist analyses conducted for confirmation. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity.

Results

Of 770 patients with histologically confirmed appendicitis, 155 (20%) had PA. Bayesian univariate analysis demonstrated decisive evidence (BF₁₀ > 100) for several predictors, with C-reactive protein (CRP) (BF₁₀ = 251 079), age (BF₁₀ = 828), and lymphocyte percentage (BF₁₀ = 352) showing the strongest associations. Multivariate Bayesian modeling identified a parsimonious three-variable model comprising CRP, age, and lymphocyte percentage. This model demonstrated good discrimination (AUC 0.78) with high specificity (95.4%) and modest sensitivity (29%). Frequentist analyses confirmed these findings.

Conclusion

A predictive model incorporating CRP, age, and lymphocyte percentage provides a highly specific tool for ruling in perforated appendicitis. This approach may aid in prioritizing surgical urgency and optimizing perioperative management. Prospective validation is warranted.