Prediction of Difficult Videolaryngoscopic Intubation and Development of a Dedicated Glottic View Score: A Multicentre Prospective Study

Background:

Videolaryngoscopy has improved glottic visualization and facilitated tracheal intubation. However, difficulties-including failed intubation-still occur. At present, no prospectively derived classification system exists to assess the difficulty of videolaryngoscopic (VL) intubation across both normal and anticipated difficult airways. Additionally, current glottic view grading systems, designed for direct laryngoscopy, may not adequately capture the specific challenges of VL intubation.

Objectives:

This study aims to:

1. Develop a predictive model for difficult VL intubation in surgical patients with both normal and anticipated difficult airways.
2. Create a glottic view scoring system specifically tailored to videolaryngoscopy.
3. Compare the predictive accuracy of the new scoring system with existing laryngeal view grades in forecasting difficult VL intubation.

Methods:

A prospective cohort of 4,977 patients will be enrolled. Patient and intubation related variables-including VL findings, airway features, clinical parameters, device, and procedural details-will be analyzed. Binary logistic regression will be employed to build the initial predictive model. In parallel, machine learning techniques (Random Forest, Support Vector Machine, XGBoost, LightGBM, etc.) will be applied to evaluate predictive performance. Comparative analysis will be conducted between the machine learning models and the logistic regression baseline.

Expected Impact:

The development of a robust predictive tool and an associated VL-specific glottic view score could enhance clinical decision making, particularly in identifying patients at risk of difficult or failed VL intubation. This may support early consideration of awake tracheal intubation, and use of standardized terminology and reduce complications associated with difficult airway management