Emergency airway management in a tertiary trauma centre (AIRMAN): a one-year prospective longitudinal study

Ethics

Ethics approval for this study was obtained from the University of Manitoba Health Research Ethics Board (Winnipeg, MB, Canada; Ethics #, HS22799 [H2019:164]).

Study design

We performed a single-centre, prospective, observational study, including all adult patients (≥ 17 yr old) that were intubated by emergency medicine or critical care medicine teams. We collected data on all consecutive emergent orotracheal intubations over a seven-month period at the Health Sciences Centre, a tertiary care trauma centre.

Data collection

The respiratory therapy department assists at all intubations that occur outside of the operating room. As such, the respiratory therapist liaised with the physician responsible for the intubation to complete a case report form after the patient had been intubated and stabilized. We collected additional data via chart review retrospectively.

Power and sample size

Several previous studies reported first-pass success rates of 80–85% for intubations performed in the ICU and ED setting.1,2,3,4 Based on these reported values, using an alpha of 0.05 and a power of 80%, 113 patients were needed to accurately report incidence of first-pass success in 75% of studied patients. The estimated 75% rate of first-pass success was based on the assumption that there would be a high prevalence of trainees performing the initial intubation attempts.

Outcome measures and definitions

The primary outcome was the incidence of first-pass success. A first-pass success was defined as the successful intubation of the trachea with the first insertion of a laryngoscope blade. If a laryngoscope blade is inserted and then withdrawn, with no attempt at intubation, this was defined as a failed intubation attempt.3,17

Factors previously identified to affect first-pass success rate such as anatomically difficult airway (defined as two or more anatomic features known to contribute to difficult intubations)9,14,18 and physiologically difficult airways (defined as patient instability leading to time pressure characterized by oxygen saturation < 90% despite intervention or systolic blood pressure < 90 mm Hg prior to the intubation attempt)11,12 were collected.

Secondary outcomes were postintubation hypoxia (defined as an oxygen saturation that begins above and subsequently drops below 90% during intubation) and postintubation hypotension (defined as a systolic blood pressure that begins above 90 mm Hg and subsequently drops below 90 mm Hg or a decrease in mean arterial pressure [MAP] to less than 60 mm Hg).

Both the medical and surgical ICUs are staffed with residents from internal medicine, surgical specialties, anesthesia, and emergency medicine from years R1 to R7, as well as in-house medical officers. Night-time coverage is provided by two in-house physicians in each unit, with critical care attending back-up. At least one of the two physicians has previous experience with intubation. The ED has at least two attending emergency physicians on a 24-hr basis in-house along with resident staff. An experienced intubator was defined as any attending staff person in critical care or emergency medicine or any third-, fourth- or fifth-year anesthesia or emergency medicine resident or critical care fellow.

Statistical analysis

Descriptive statistics are reported as mean and standard deviation for normally distributed data. Categorical variables are reported as frequencies and percent values. We used univariate logistic regression models to explore the unadjusted relationships between each predictor and the odds of a failed intubation attempt, hypoxemia, and hypotension. Results are presented as odds ratios and their 95% confidence intervals (CIs) and P values. A P value less than 0.05 was considered significant.

For our primary outcome, we created a parsimonious multivariate logistic regression model identifying predictors of first-pass success. We used the group least absolute shrinkage and selection operator (LASSO) estimation method, which penalizes the model for complexity according to the sum of the absolute value of the regression coefficients. This in turn shrinks the coefficients, some of them to exactly zero, thereby enabling variable selection that occurs alongside the optimization of the model likelihood. Group LASSO is a particular variation in which the levels of categorical variables are selected or excluded together as a group, which aids interpretation. The LASSO method has been shown to have superior properties to certain ad hoc methods, univariable screening in particular.19 We analyzed the secondary outcomes hypoxia and hypotension using univariate analyses only because of insufficient events and problems with multiple comparisons between the adverse events. We feel that the multivariate analyses predicting adverse events would be underpowered and would therefore not help in drawing conclusions.

Multiple LASSO models were fit, each with differing penalty weights for model complexity, and the model with the best Akaike Information Criterion (AIC) was selected. PROC LOGISTIC of SAS version 9.4 (SAS Institute, Cary, NC, USA) was used to estimate the univariate logistic regression models. PROC HPGENSELECT was used to estimate the Group LASSO variable subset for first-pass intubation. Participants with missing information were omitted from the models. We excluded body mass index from multivariate modelling because of excessive missing information. Data from this multivariate model are presented as odds ratios without CIs as these are not possible with LASSO. A receiver operating characteristic (ROC) curve was established to estimate the area under the curve (AUC) of the model.

留言 (0)

沒有登入
gif