Background: With the routine use of advanced multi-slice CT scanners, pulmonary embolism (PE) is increasingly detected as an incidental finding among cancer patients. Although this generally leads to therapeutic interventions, the accuracy of diagnosing PE on routinely performed contrast enhanced CT scans is unknown.
Methods: Consecutive cancer patients diagnosed with incidental PE were eligible for inclusion. Their CT images were reassessed in a blinded fashion by two thoracic radiologists. To ensure blindness, a total of 19 cancer staging CT images without PE were included. The inter-observer reliability for the presence of PE was calculated with use of Kappa statistics.
Results: A total of 62 incidental PE patients (mean age 64years, 60% male) were included. All patients received anticoagulant treatment upon diagnosis. Level of agreement between the two expert readers was high: they disagreed on the presence of PE in only two patients (3.2%), resulting in a Kappa statistic of 0.93. After final consensus reading, it was concluded that the CT images of all 62 patients initially diagnosed with incidental PE were indeed positive for PE.
Conclusions: This study indicates that an incidental PE diagnosis is reliable and highly reproducible, despite the suboptimal reading conditions of a non-dedicated scan protocol.
Keywords: Cancer; Diagnosis; Pulmonary embolism; Reproducibility.
Copyright © 2015 Elsevier Ltd. All rights reserved.