Study objective: To describe and prospectively evaluate a new radiologic sign with the potential to increase the diagnostic accuracy of soft-tissue radiography of the neck in the identification of adult epiglottitis.
Methods: We conducted a prospective, before-and-after blinded study at two tertiary care institutions. A convenience sample of four staff emergency physicians, three otolaryngology residents, four radiology residents, and four senior medical students volunteered to participate. We assembled 26 soft-tissue radiographs of the neck from consecutive patients ED with the diagnosis of epiglottitis made on the basic of direct visualization. Twenty-six control radiographs were identified from ED patients who were being evaluated for the presence of foreign bodies or minor cervical trauma. We then randomly mixed the two sets of radiographs. Participants were asked to identify epiglottis among the 52 randomly sequenced radiographs. A standardized 5-minute tutorial on the vallecula sign was presented to all participants after the first interpretation. We then asked the participants to make a second interpretation of the 52 radiographs without knowledge of correct answers from the initial evaluation.
Results: The participants accurately classified 80.5% of all radiographs reviewed before the tutorial and 98.8% after the tutorial (P < .0001). Similarly, sensitivity improved from 78.5% to 98.2% (P < .0001) and specificity improved from 82.8% to 99.5% (P < .0001). We found no significant differences in performance characteristics among the different types of participants.
Conclusion: We have described a new radiographic sign that improves the diagnostic accuracy of soft-tissue radiography of the neck. If reproduced in prospective studies, the absence of the vallecula sign on radiography might obviate the need for routine use of direct visualization as an initial screen.