LAUSR

Background: Caustic ingestion has gained increasing attention worldwide. However, the insight into whether to use esophagogastroduodenoscopy (EGD) or computed tomography (CT) for first-line investigation remains controversial. This study aimed to evaluate a diagnostic and management algorithm that combines EGD and CT for rapid triage. Methods: We established an algorithm for our hospital in 2013, aiming to maximize the benefits and minimize the limitations of EGD and CT. Then, we retrospectively analyzed the 163 enrolled patients treated between 2014 and 2019 and categorized them into 4 groups: A = 3 (1.8%): with perforation signs and directly confirmed by CT, B = 10 (6.1%): clinically suspected perforation but not initially proven by CT, C = 91 (55.8%): initial perforation less favored but with EGD grade ≥ 2b or GI/systemic complications, and D = 59 (36.2%): clinically stable with EGD grade ≤ 2a, according to initial signs/symptoms and EGD/CT grading. The morbidity and mortality of each group were analyzed. The predictive values of EGD and CT were examined by logistic regression analyses and receiver operating characteristic (ROC) curves. Results: The outcomes of such algorithm were reported. CT was imperative for patients with toxic signs and suspected perforation. For non-emergent operations, additional EGD was safe and helpful in identifying surgical necessity. For patients with an initially low perforation risk, EGD alone sufficiently determined admission necessity. Among inpatients, EGD provided excellent discrimination for predicting the risk for signs/symptoms’ deterioration. Routine additional CT was only beneficial for those with deteriorating signs/symptoms. Conclusions: According to the analyses, initial signs/symptoms help to choose EGD or CT as the first-line investigative tool in caustic patients. CT is necessary for seriously injured patients, but it cannot replace EGD for moderate/mild injuries. The severity stratification and patient categorization help to simplify complex scenarios, accelerate decision-making, and prevent unnecessary intervention/therapy. External validation in a larger sample size is further indicated for this algorithm.