Inter- and Intra-Rater Reliability of Laryngeal Sensation Testing with the Touch Method During Flexible Endoscopic Evaluations of Swallowing

Abstract

Objectives: Sensation is an integral component of laryngeal control for breathing, swallowing, and vocalization. Laryngeal sensation is assessed by elicitation of the laryngeal adductor reflex (LAR), a brainstem-mediated adduction of the true vocal folds. During Flexible Endoscopic Evaluations of Swallowing (FEES), the touch method can be used to elicit the LAR to judge laryngeal sensation. Despite the prevalence of this method in clinical practice and research, prior studies have yet to examine inter- and intra-rater reliability. Methods: Four speech-language pathologists rated 125 randomized video clips for the presence, absence, or inability to rate the LAR. Fifty percent of video clips were re-randomized and re-rated 1 week later. Raters then created guidelines and participated in formal consensus training sessions on a separate set of videos. Ratings were repeated post-training. Results: Overall inter-rater reliability was fair (κ = 0.22) prior to training. Pre-training intra-rater reliability ranged from fair (κ = 0.35) to almost perfect (κ = 0.89). Inter-rater reliability significantly improved after training (κ = 0.42, P < .001), though agreement did not reach prespecified acceptable levels (κ ≥ 0.80). Post-training intra-rater reliability ranged from moderate (κ = 0.49) to almost perfect (κ = 0.85). Conclusion: Adequate inter-rater reliability was not achieved when rating isolated attempts to elicit the LAR. Acceptable within-rater reliability was observed in some raters 1 week after initial ratings, suggesting that ratings may remain consistent within raters over a short period of time. Limitations and considerations for future research using the touch method are discussed.