Skip to main content
RACS ASC 2024

Predictive criteria for oesophageal perforations in patients with pneumomediastinum – can invasive diagnostic modalities be avoided?

Verbal Presentation

Verbal Presentation

2:50 pm

08 May 2024

Conway 3

RESEARCH PAPERS

Disciplines

Upper GI Surgery

Watch The Presentation

Presentation Description

Institution: Princess Alexandra Hospital - Queensland, Australia

Purpose: This retrospective study aims to study the incidence of oesophageal perforations (OP) in cases of pneumomediastinum (PNM) diagnosed on high-resolution computed tomography (CT). The primary objective is to identify clinical, biochemical, and radiological predictors of OP to help clinicians decide on the need for further investigations and treatment. The study seeks to determine if a set of criteria can reliably predict the incidence of OP. Methodology: A retrospective analysis of 427 adult patients was conducted with index CT scan reports containing 'pneumomediastinum' from January 2016 to December 2022 at a single tertiary institution. Patients with known PNM and recent interventions were excluded from the study. Patient demographics, symptoms, biochemical markers, and radiological findings were all recorded and various statistical methods were used to analyse variables as predictors of OP. Results: 336 patients with PNM were included in this study, from which 22 patients were identified with OP. The study identified several statistically significant risk factors for OP, with the leading presenting complaint of dysphagia having a p-value of 0.003. Patients with no reported symptoms also reliably excluded OP, with a p-value of 0.002. Radiological findings of pleural effusion, mediastinal free fluid, and disruption of oesophageal wall were all highly significant features of OP with a sensitivity exceeding 80%. Conclusion: This study endeavours to enhance clinical decision-making in PNM cases by providing insights into the risk factors for patients with OP. The results may contribute to the formulation of an algorithm that can aid clinicians in risk stratifying patients and minimising unnecessary further investigations.

Speakers

Authors

Authors

Dr Justin Hsieh - , Dr Dong Tony Cheng* - , Dr Adam Frankel - , Dr Iain Thomson -