Multivariate predictors of failed prehospital endotracheal intubation

Conventionally trained out-of-hospital rescuers (such as paramedics) often fail to accomplish endotracheal intubation (ETI) in patients requiring invasive airway management. Previous studies have identified univariate variables associated with failed out-of-hospital ETI but have not examined the interaction between the numerous factors impacting ETI success. This study sought to use multivariate logistic regression to identify a set of factors associated with failed adult out-of-hospital ETI.

The authors obtained clinical and demographic data from the Prehospital Airway Collaborative Evaluation, a prospective, multicentered observational study involving advanced life support (ALS) emergency medical services (EMS) systems in the Commonwealth of Pennsylvania. Providers used standard forms to report details of attempted ETI, including system and patient demographics, methods used, difficulties encountered, and initial outcomes. The authors excluded data from sedation-facilitated and neuromuscular blockade-assisted intubations. The main outcome measure was ETI failure, defined as failure to successfully place an endotracheal tube on the last out-of-hospital laryngoscopy attempt. Logistic regression was performed to develop a multivariate model identifying factors associated with failed ETI.

Data were used from 45 ALS systems on 663 adult ETIs attempted during the period June 1, 2001, to November 30, 2001. There were 89 cases of failed ETI (failure rate 13.4%). Of 61 factors potentially related to ETI failure, multivariate logistic regression revealed the following significant covariates associated with ETI failure (odds ratio; 95% confidence interval; likelihood ratio p-value): presence of clenched jaw/trismus (9.718; 95% CI = 4.594 to 20.558; p < 0.0001); inability to pass the endotracheal tube through the vocal cords (7.653; 95% CI = 3.561 to 16.447; p < 0.0001); inability to visualize the vocal cords (7.638; 95% CI = 3.966 to 14.707; p < 0.0001); intact gag reflex (7.060; 95% CI = 3.552 to 14.033; p < 0.0001); intravenous access established prior to ETI attempt (3.180; 95% CI = 1.640 to 6.164; p = 0.0005); increased weight (ordinal scale) (1.555; 95% CI = 1.242 to 1.947; p = 0.0001); and electrocardiographic monitoring established prior to ETI attempt (0.199; 95% CI = 0.084 to 0.469; p = 0.0003). This model was the most parsimonious of the models evaluated and demonstrated good fit (Hosmer-Lemeshow test p = 0.471) and discrimination (area under ROC curve = 0.906). There were no significant interaction terms.

The authors used multivariate logistic regression to identify a set of factors associated with failure to accomplish ETI in adult out-of-hospital patients. Findings from this analysis could provide the basis for clinical protocols or decision rules aimed at minimizing the incidence of out-of-hospital ETI failure.