Is image interpretation in drug-induced sleep endoscopy that reliable?

Area of Science:

• Otolaryngology
• Sleep Medicine
• Medical Imaging

Background:

• Drug-induced sleep endoscopy (DISE) is utilized to assess upper airway obstruction in sleep-disordered breathing.
• Previous research suggested DISE has interobserver reliability and a learning curve.

Purpose of the Study:

• To evaluate the interobserver agreement and variability of DISE among French ENT physicians with varying experience levels.
• To assess the impact of experience on the reliability of DISE interpretation.

Main Methods:

• A prospective study involving 76 French ENT physicians (69 inexperienced, 7 experienced) observing 7 DISE videos.
• Physicians classified obstruction levels, configurations, and degrees using the VOTE classification.
• Interobserver agreement was quantified using Fleiss' Kappa coefficient (k).

Main Results:

• Interobserver agreement for obstruction level detection ranged from poor to good, with the oropharynx showing the highest agreement.
• Agreement for obstruction configuration and degree was moderate, except for the tongue base, which showed poor concordance.
• No statistically significant difference in agreement was found between experienced and inexperienced physician groups.

Conclusions:

• DISE demonstrates limited interobserver agreement in detecting obstructive sleep apnea sites, even within a group of experienced French ENT specialists.
• The study found no evidence of a learning curve effect, indicating that experience does not significantly improve agreement.
• Current interpretation of DISE may not be entirely reliable for clinical decision-making.