Interrater agreement of nasal endoscopy for chronic rhinosinusitis

Abstract

Background: Nasal endoscopy is a routine, important diagnostic tool in the evaluation of chronic rhinosinusitis (CRS). Although the procedure is ideally "objective," the subjective nature of endoscopy interpretation and lack of standardization are potential limitations. The goal of this study was to examine the interrater agreement of various categories of nasal endoscopy findings in patients undergoing evaluation for CRS.

Methods: Fourteen patients (28 sides) with CRS underwent clinical evaluation, SNOT-22, sinus computed tomography (CT), and digital video nasal endoscopy. Five academic rhinologists blindly reviewed the endoscopies for structural anatomic issues, inflammatory rhinosinusitis findings, and atypical lesions. Statistical comparison of the endoscopy interpretations was performed using the unweighted Fleiss' kappa statistic (K(f) ).

Results: The mean Lund-Mackay CT scan score was 7.8 (standard deviation [SD] 4.9) and the mean SNOT-22 score was 35.8 (SD 22.7). Significant variability was noted among the raters with respect to the various categories of nasal endoscopy findings. The overall levels of interrater agreement for the various categories were as follows: "almost perfect" for atypical lesions (K(f) = 0.912); "substantial" for nasal polyps (K(f) = 0.693); "moderate" for nasal discharge (K(f) = 0.422) and mucosal inflammatory changes of the middle turbinate (K(f) = 0.413); and "fair" for edema of the middle meatus (K(f) = 0.214), obstruction by nasal septum deviation (K(f) = 0.240), and obstruction by the middle turbinate (K(f) = 0.276).

Conclusion: Significant variability was noted in the interrater agreement for nasal endoscopy findings in this study, with relatively limited agreement on some of the key findings of the procedure. Additional investigation and standardization of nasal endoscopy interpretation is required to improve the clinical utility of the procedure.