Interrater reliability of different scoring systems for drug-induced sleep endoscopy

To explore the interobserver reliability of drug-induced sleep endoscopy (DISE) for patients with obstructive sleep apnea syndrome (OSAS) of two classification systems.

DISE examinations were recorded digitally for all patients and were evaluated independently by five examiners blinded to all patient data. Areas of obstruction were rated using VOTE (velum, oropharynx lateral wall, tongue base, epiglottis) classification and PTLTbE (palate, tonsils, lateral pharyngeal wall, tongue base, epiglottis) classification. Additionally, palatal coupling was assessed during the jaw thrust maneuver. Interobserver reliability was evaluated with Fleiss' kappa for categorical data and linearly weighted kappa for ordinal data.

In total, 123 patients were included in this study, 92 males and 31 females, with a mean (SD) age of 50.7 (12.1) years and a body mass index (BMI) of 28.3 (5.0) kg/m2. The mean apnea-hypopnea index was 22.2 (18) per hour, and the Epworth Sleepiness Scale was 7.3 (2.3). In our cohort, the interobserver reliability for the VOTE classification ranged from 0.32 to 0.59 and for the PTLTbE classification from 0.23 to 0.49 over all assessed levels, indicating fair to moderate interobserver reliability. The weighted kappa for palatal coupling was 0.37. In the VOTE classification, the level with the best agreement was the velum, while in PTLTbE, the best agreement was observed for tonsillar obstruction.

The interrater agreement was fair to moderate for both classifications, with no clear superiority for one classification. The large variability shows the need to objectively quantify upper airway collapse during DISE and natural sleep endoscopy.