Risk patterns of consecutive adverse events in airway management: a Bayesian network analysis

Background: Minor adverse airway events play a pivotal role in the safety of airway management. Changes in airway management strategies can reduce such events, but the broader impact on airway management remains unclear.

Methods: Minor, frequently occurring adverse airway events were audited before and after implementation of changes to airway management strategies. We used two Bayesian networks to examine conditional probabilities of subsequent airway events and to compute the likelihood of certain events given that certain previous events occurred.

Results: Independent of sex, age, and American Society of Anesthesiologists physical status, targeted changes to airway management strategies reduced the risk of a first event. Obese patients were an exception, in whom no risk reduction was achieved. Frequently occurring event sequences were identified, for example the most likely event to follow difficult bag-mask ventilation was a Cormack-Lehane grade ≥3, with a risk of 14.3% (95% credible interval [CI], 11.4-17.2%). An impact of the targeted changes was detected on the likelihood of some event sequences, for example the likelihood of no consecutive event after a tracheal tube-related event increased from 43.3% (95% CI, 39.4-47.6%) to 56.4% (95% CI, 52.0-60.5%).

Conclusions: Identification of risk patterns and typical structures of event sequences provides a clinically relevant perspective on airway incidents. It further provides a means to quantify the impact of targeted airway management changes. These targeted changes can influence some event sequences, but overall, the benefit results from the cumulative effect of improvements in multiple events. Targeted airway management changes with knowledge of risk patterns and event sequences can potentially further improve patient safety in airway management.

Clinical trial registration: NCT02743767.