Esophagography and chest CT for detection of perforated esophagus: what factors influence accuracy?

Abstract

To determine: whether the use of both esophagography (EG) and CT is superior to either study alone in the detection of esophageal injuries and perforations. Paired CT and EG performed for suspected perforated or injured native esophagus (NAE) or neo-esophagus (NEOE) were retrospectively identified and independently scored for likelihood of perforation with a Likert scale. Official reports were scored with the same scale. Retrospective study and official interpretations were assessed separately for overall diagnostic accuracy, for diagnosis of contained versus free perforation. 110 paired exams performed in 107 patients fulfilled inclusion criteria. Overall, combined CT and EG retrospective study interpretation was greater than EG or CT scans alone. Study EG and combined CT and EG interpretations were less sensitive for contained perforations than for free perforations. Evaluations of NAE and NEOE showed similar sensitivities. Receiver operating characteristic (ROC) curve generated from combined official CT and EG interpretations was superior to ROC from combined retrospective study interpretations. Combination of EG and CT can improve sensitivity for detecting perforated intrathoracic viscus, but even with combined studies accurate diagnosis of contained perforations is challenging. Superior performance of official reports suggests that concurrent review using both modalities may improve accuracy.