Peri-intubation adverse events and clinical outcomes in emergency department patients: the BARCO study

Abstract

Background

Emergency tracheal intubation in critically ill patients carries a high risk of complications, and practices vary substantially across different settings. Identifying risk factors and understanding how peri-intubation adverse events affect patient outcomes may guide standardization of care and improve survival.

Methods

This prospective cohort study involved 18 emergency departments in Brazil (March 2022–April 2024). We included adults (≥ 18 years) undergoing emergency intubation and excluded patients intubated electively or for cardiac arrest. We defined major peri-intubation adverse events as severe hypoxemia, new hemodynamic instability, or cardiac arrest occurring within 30 min of initiating intubation. The primary outcome was 28-day mortality. Multivariable regression analyses assessed associations between adverse events and mortality, controlling for potential confounders.

Results

Among 2846 patients, major adverse events occurred in 919 (32.3%) intubations, most frequently new hemodynamic instability (20.0%), followed by severe hypoxemia (12.5%) and cardiac arrest (3.5%). The overall 28-day mortality was 45.1%. Patients experiencing any major adverse event had a significantly higher 28-day mortality (57.6 vs 39.2%; aHR 1.43, 95% CI 1.26–1.62; p < 0.001). Sensitivity analyses confirmed these findings. Successful first-attempt intubation was associated with a reduced likelihood of major adverse events (aOR 0.52; 95% CI 0.41–0.65; p < 0.001).

Conclusion

One in three patients undergoing emergency intubation experienced a major peri-intubation adverse event, which was associated with higher 28-day mortality. These results underscore the importance of optimizing intubation strategies to reduce complications and potentially improve patient outcomes in critically ill patients.

Background

Tracheal intubation (TI) is a critical procedure performed in emergency departments (EDs) and intensive care units (ICUs) worldwide [1]. While essential for managing critically ill patients, it carries significant risks, particularly in resource-limited settings [1,2,3]. Major adverse events (MAEs), such as hemodynamic instability, severe hypoxemia, and cardiac arrest, are frequently observed in the peri-intubation period and have been associated with worse patient outcomes [3,4,5,6].

Critically ill patients frequently present with a ‘physiologically difficult airway,’ characterized by acute hemodynamic instability, compromised oxygenation, and metabolic disturbances that increase the risk of peri-intubation adverse events, even in the absence of anatomical airway difficulty [7, 8]. An international cohort study involving 29 countries found that the incidence of MAEs during emergency intubations exceeded 40% [4]. The study reported a first-attempt intubation success rate of nearly 80%, with only 4.5% of patients requiring more than two attempts. This high incidence of MAEs, despite relatively high first-attempt success, suggests that factors beyond anatomic challenges contribute to adverse events [4, 9,10,11].

Understanding potential causal pathways in airway management is crucial for improving outcomes in critically ill patients undergoing emergency intubations globally [12,13,14]. However, the global burden of these complications remains poorly understood in low- and middle-income countries (LMICs), since prospective studies examining the association between peri-intubation adverse events and 28-day mortality are limited [4,5,6].

Therefore, we established the Brazilian Airway Registry Cooperation (BARCO), the first multicenter registry of emergency intubations in a middle-income country. This study aims to determine the incidence of MAEs and their association with 28-day mortality in critically ill patients undergoing emergency intubations, offering essential data from a resource-limited setting.

Methods

Study design and setting

We conducted a prospective cohort study across 18 EDs in Brazil, spanning four regions, as part of the BARCO network. We report these results in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [15]. This study was approved by the ethics committees of all participating centres. Due to the observational nature of the study and the use of de-identified data, a waiver of individual patient consent was granted by each institution’s ethics committee.

Participants

We enrolled adults (age ≥ 18 years) undergoing TI in the ED. Exclusion criteria were intubations performed for elective procedures or during cardiac arrest. Tracheal intubations were performed by clinical staff working in the ED. Specific procedural aspects, including the choice of medications, equipment, and intubation techniques, were determined according to clinicians preferences and standard practices at each participating center. To ensure comprehensive data collection and avoid selective reporting, each site investigator submitted a study compliance plan approved by the coordinating center (Hospital das Clínicas de São Paulo, SP, Brazil). This plan detailed the process for identifying consecutive ED intubations and ensuring at least 80% were recorded in the BARCO database. Monthly compliance reports were submitted and reviewed by BARCO coordinators for quality control.

Exposures and outcomes

Our primary exposures were MAEs, defined as the presence of one or more of the following within 30 min from the start of the intubation procedure: severe hypoxemia (peripheral oxygen saturation < 80%), new hemodynamic instability (systolic arterial pressure < 65 mmHg recorded at least once, new requirement for or increase the dose of vasopressors, or administration of a fluid bolus > 15 mL/kg to maintain target blood pressure), or cardiac arrest [4]. Our primary outcome was 28-day mortality after TI, assessed through electronic health record reviews, with follow-up until hospital discharge, death, or 28 days, whichever occurred first. Secondary outcomes included the incidence of MAEs, difficult intubations (defined as three or more attempts), the first-attempt success, transient hypotension, defined as a single episode of systolic blood pressure < 90 mmHg or mean arterial pressure < 65 mmHg that does not fulfill the criteria for new hemodynamic instability, and esophageal intubation.

Data collection

Intubation observers completed a structured, web-based data collection form using a survey on REDCap®. [16, 17] We required the form to be completed within 30 min of tracheal intubation confirmation. A site investigator at each center trained staff on completing the form and designated a trained observer rather than the clinician who performed the intubation to complete the form.

We collected variables representing patient characteristics, illness severity, preprocedural physiology, and intubation characteristics. Definitions for all collected variables are available in eMethods1. (Appendix File).

Statistical analysis

As this study was designed to be a prospective observational airway management registry, we did not calculate a target sample size. Given the inclusion of 18 centers with a minimum of 40 intubations to be included, the minimum sample size would be 720 intubations, but we estimated that 3–5 times more participants would be included given the center’s volumes of inclusion, which would allow sufficient power to estimate the rate of adverse events, 28-day mortality and to explore their association.

We present descriptive statistics as mean ± SD or median (interquartile range [IQR]) for continuous variables, frequency and proportion for categorical variables, stratified by peri-intubation MAEs. We performed bivariate analyses with χ² or Fisher’s exact test for categorical variables, and the Mann–Whitney or t-test for continuous variables, as appropriate. We plotted the 28-day survival with a Kaplan–Meier survival curve, stratified by each combination of MAEs. Patients transferred or with unknown outcome were excluded from survival analysis.

We utilized Cox proportional hazards models to assess the association between the occurrence of MAEs and 28-day mortality. Recognizing that adverse events during emergency intubations and 28-day mortality may have shared underlying causes, we identified potential confounders using a directed acyclic graph (DAG) (Supplemental eFigure 3). This method allowed us to account for causal pathways while avoiding mediators, open backdoor paths, and collider bias [18]. The Cox model accounted for clustering with center-specific shared frailties.

We developed three models: (1) an unadjusted model, (2) a model adjusted for patient characteristics and preprocedural physiology (age, sex, BMI, Charlson comorbidity score, shock index, SOFA score, and pre-intubation SpO₂), and (3) a model adjusted for patient characteristics, preprocedural physiology and intubation characteristics (first-attempt success, Cormack-Lehane classification, subjective impression of difficulty, and use of intubation drug agents such as analgesics, hypnotics, and paralytics). We report hazard ratios (HRs) with 95% confidence intervals (CIs) for these analyses. In the main analysis, patients discharged home were assumed to be alive at 28 days.

We performed sensitivity analyses for the Cox model to account for potential informative censoring. We analyzed the results with (a) a logistic regression model using in-hospital mortality as the outcome, (b) a Cox model censoring patients at hospital discharge, (c) a Fine and Gray model accounting for hospital discharge as a competing outcome, and (d) a worst-case scenario analysis. Furthermore, we performed sensitivity analyses for unmeasured confounding with E-values, which evaluate the strength of the association of an unknown confounder that would neutralize the observed associations [19].

We performed supplemental exploratory analyses to identify factors associated with first-attempt success and the incidence of MAE (Supplemental eMethods 3). Additionally, we assessed the association between first-attempt success and the number of intubation attempts with MAE.

All statistical tests were two-sided, and p-values less than 0.05 were considered significant. All analyses were conducted using R version 4.2.2 and Stata SE 18.0.

Results

From March 1, 2022, to April 30, 2024, we enrolled patients at 18 EDs, 4 community and 14 academic hospitals (eTable 1, Appendix File). We screened 3618 patients who underwent TI during the 2-year study period, of whom 2846 were included in the final analysis (Fig. 1). Additional details on enrollment numbers by centers characteristics, and missing intubation data are described in Supplement eTables 1 and Fig. 1, respectively.

Fig. 1

Enrollment flow diagram. *Others reasons for no inclusion intubations were “forgot to fill out”, “doesn’t remember completing”, or “invalid record number”). HR: Health Records; ICU: Intensive Care Unit

Patient baseline characteristics

The median age was 63 years (IQR, 49–73), and most were male (58.0%). Patients who experienced MAEs were older (median age 65 vs 62 years; p < 0.001), had a higher pre-intubation shock index (0.87 vs 0.74; p < 0.001), and were more likely to have been intubated for acute respiratory failure (47.9% vs 30.6%, p < 0.001). Table 1 and eTable 2 detail the characteristics of included patients.

Table 1 Patient demographics and intubation characteristics

Intubation characteristics

First-attempt success occurred in 2116 (74.3%) cases. Direct laryngoscopy (DL) was used for the first intubation attempt in 2294 (80.6%) cases (Table 1). The median number of attempts was 1 (IQR, 1–2), with difficult intubations occurring in 198 (7.0%) patients. Pre-treatment with fentanyl was used in 630 (22.1%) intubations. Etomidate was the most used induction agent (n = 1655 [58.2%]), followed by ketamine (n = 751 [26.4%]). Propofol was rarely used (n = 89 [3.1%]). Rapid sequence intubation (RSI) was employed in 2462 (86.5%) of the intubations. Residents in Emergency Medicine and Internal Medicine performed the first intubation attempt in 1097 (38.5%) and 1125 (39.5%) cases, respectively. As first operators, anesthesiologists participated in only 3 intubations (0.1%). First-year residents were the first operators in 1351 (47.5%) cases. A surgical airway was performed on only 1 (0.04%).

Incidence of MAEs

Of 2846 patients, 919 (32.3%) experienced at least one MAEs. New hemodynamic instability was the most common MAE (569, 20.0%), followed by severe hypoxemia in 356 (12.5%) intubations, and peri-intubation cardiac arrest in 100 (3.5%) intubations, with 73 (73.0%) of these achieving the return of spontaneous circulation. Table 2 presents other complications. At least one episode of transient hypotension occurred in 345 (12.1%) intubations, and esophageal intubation was reported in 86 (3.0%) intubations.

Table 2 Frequency of peri-intubation adverse events

Association of MAEs with 28-day mortality

Overall, 28-day mortality was 45.1%. Mortality was higher among patients experiencing MAEs compared to those who did not (57.6 vs 39.2%, p < 0.001). Figures 2 and 3 presents the association of MAEs and their subcomponents with 28-day mortality. MAEs were associated with increased 28-day mortality (aHR 1.43, 95% CI 1.26–1.62, p < 0.001). All subcomponents were also associated with increased 28-day mortality, including increased HR from hemodynamic instability (aHR 1.28, 95% CI 1.11–1.48, p < 0.001) followed by severe hypoxemia (adjHR 1.39, 95% CI, 1.16–1.66, p < 0.001) and cardiac arrest (aHR 2.52, 95% CI 1.86–3.40, p < 0.001). Finally, patients experiencing both hemodynamic instability and severe hypoxemia had an increased risk of mortality compared to those with neither (aHR 1.97, 95% CI 1.38–2.81, p < 0.001, Appendix file, eTable 7), which was higher than either hemodynamic instability or severe hypoxemia alone.

Fig. 2

Association between the occurrence of major adverse events and 28-day mortality. HR: Hazard Ratio; OR: Odds Ratio; CI: Confidence Interval; MAE: Major Adverse Event; Major adverse events peri-intubation were defined as events during or thirty minutes after the intubation process, as listed below: hemodynamic Instability: Systolic Blood Pressure < 65 mmHg, need for starting or increase in vasopressor dose or fluid resuscitation; Severe hypoxemia: Oxygen peripheral saturation less than 80%; Cardiac arrest: Presence of cardiac arrest signs peri-intubation

Fig. 3

Mortality rate by days after intubations, stratified by major adverse events. Patients who were discharged were considered alive through the 28-day follow-up period for this analysis

All sensitivity analyses to the Cox model assumptions demonstrated similar results (Fig. 2 and Appendix, eTables 8–11), suggesting the robustness of model assumptions. E-values for the point estimate and its lower confidence interval were, respectively, 1.88 and 1.63 for all MAEs; 1.66 and 1.36 for hemodynamic instability; 1.82 and 1.45 for severe hypoxemia; and 3.18 and 2.44 for cardiac arrests (Appendix file, eFigures 4–7).

Factors associated with MAEs and first-attempt success

Patients who experienced MAEs had lower observed first-attempt success (66.3% vs 78.2%, p < 0.001). Successful first-attempt intubation was associated with a lower likelihood of experiencing MAEs (aOR 0.52, 95% CI 0.41–0.65, p < 0.001), while each additional attempt was associated with higher odds of MAEs (aOR 1.65, 95% CI: 1.45–1.88, p < 0.001; Appendix file, eTable 5). Moreover, each additional intubation attempt markedly increased the risk of severe hypoxemia (aOR 2.28, 95% CI 1.95–2.65, p < 0.001, eTable 5). Increasing age, elevated shock index, higher SOFA score, lower pre-intubation oxygen saturation, and acute respiratory failure as an indication for intubation were significantly associated with peri-intubation MAEs in the multivariable analysis (Table 3). Performing an intubation checklist was significantly protective against peri-intubation MAEs (aOR, 0.75; 95% CI 0.57–0.98; p = 0.036). Patients receiving ketamine experienced a higher unadjusted rate of MAEs compared to those receiving etomidate (29.8 vs 24.8%, p < 0.001), however, no significant difference in MAEs was observed among different induction agents in multivariable analysis (Table 3). The specialty of the first operator was associated with first-attempt success, with internal medicine (OR 0.61, 95% CI 0.44–0.83, p = 0.002) specialists having lower odds of first-attempt success compared to emergency medicine specialists. Intubation with a bougie was associated with higher odds of first-attempt success (OR 1.57, 95% CI 1.06–2.34, p = 0.025, eTable 6).

Table 3 Association between patient and intubation characteristics with peri-intubation major adverse events

Discussion

In this multicenter prospective cohort study of critically ill adults undergoing emergency tracheal intubation, MAEs occurred in 32.3% of patients: hemodynamic instability, in 20.0%; severe hypoxemia, in 12.5%; and cardiac arrest, in 3.5%. The overall 28-day mortality was 45.1%. Patients who experienced MAEs had significantly higher mortality, even after adjustment for patient-level and intubation-related characteristics. These associations remained consistent across multiple sensitivity analyses. Factors associated with MAEs included both nonmodifiable characteristics and potentially modifiable elements, such as first-attempt success, pre-intubation hemodynamic status, and pre-intubation peripheral oxygen saturation.

In 2021, INTUBE study analyzed 2964 intubations in critically ill patients and observed MAEs in 45.2% of cases, a higher incidence compared to 32.3% in our study. They reported new hemodynamic instability in 42.6% of patients, severe hypoxemia in 9.3%, and cardiac arrest during intubation in 3.1% [4]. In contrast, our cohort experienced less new hemodynamic instability (20.0%), and more severe hypoxemia (12.3%) and cardiac arrest (3.5%). In a different population, with slight differences in the definition of MAE, the PREPARE II trial reported findings that were similar to our observations, with an incidence of cardiovascular collapse at 21.0% [20]. In our cohort, the utilization of propofol as an induction agent (3%) and opioids (18%) prior to intubation was notably lower compared to INTUBE cohort (45% and 51%, respectively) [4]. Propofol, while widely used for its rapid onset and short duration of action, is associated with a higher incidence of hemodynamic instability [21, 22]. In a secondary analysis from INTUBE study, propofol was identified as a modifiable factor linked to increased peri-intubation hemodynamic instability, with an aOR of 1.28 [2]. Furthermore, the practice of using opioids as a pre-treatment during emergency intubation is controversial. North American registry cohorts with 15,776 intubations show that opioids were used in only 2.9% of patients [6]. Despite limitations, Ferguson I et al. raised questions about the risk of hypotension when fentanyl is used as pre-treatment during emergency intubation [23]. Differences in drug preferences likely reflect global variations in clinical practice, as illustrated by the significantly higher proportion of anesthesiologists performing intubations in INTUBE cohort (54%) compared to our study (0.4%) [4]. Although the Brazilian model of outsourcing anesthesia services differs from that in Europe, in many North American centers, anesthesiologists do not commonly perform intubations outside the operating room either [14]. The generalization of this finding depends on the contextual factors of each country regarding airway management operators and specialties.

In our sample, patients experiencing MAEs had a markedly higher 28-day mortality, with an aHR of 1.43 (95% CI 1.26–1.62). These findings align with those reported by Russoto et al., who also found an association between MAEs and 28-day mortality (aOR 1.44, 95% CI 1.19–1.74) [4]. While recent trials have addressed modifiable factors to improve pre-oxygenation and avoid hypoxemia [12] or enhance first-attempt success [14, 24], factors to prevent cardiovascular collapse have not been extensively investigated or were not effective [20, 25]. In our cohort, patients who received fluid resuscitation or vasopressors prior to intubation were more likely to experience MAEs, suggesting a more severe clinical profile at baseline. Although these findings highlight the importance of hemodynamic optimization before intubation, it is also possible that hemodynamic instability serves as a surrogate marker of illness severity not fully captured by the available covariates. This interpretation is supported by the lower E-values observed for hemodynamic instability compared with other MAEs, indicating greater susceptibility to unmeasured confounding.

Recent consensus guidelines from an international Delphi study highlight structured strategies for managing the physiologically difficult airway, emphasizing pre-intubation assessment, hemodynamic stabilization, optimized pre-oxygenation, and careful induction agent selection to minimize peri-intubation adverse events [8]. We identified several key risk factors for MAEs during intubation, consistent with previous research. De Jong et al. highlighted hypotension, hypoxemia, lack of pre-oxygenation, obesity, and age over 75 years as critical risk factors for cardiac arrest [26]. In our cohort, markers of illness severity, including shock index, the number of organ dysfunctions, and vasopressor use, were strongly associated with MAEs. Therefore, targeted efforts to identify and stabilize patients presenting with shock or respiratory failure before intubation are essential. Early optimization of hemodynamic status and oxygenation may reduce the risk of peri-intubation adverse events. Two ongoing international trials, the FLUVA (NCT05318066) and PREVENTION (NCT05014581) trials, are evaluating if pre-emptive vasopressor use can reduce cardiovascular collapse during intubation in critically ill adults and may help clarify the role of hemodynamic optimization in this context. Furthermore, capnography use was limited in our cohort (25%), similar to INTUBE study⁴, and likely reflects restricted access to waveform capnography in many resource-limited settings. In contrast, use of a pre-intubation checklist was associated with a lower risk of MAEs (aOR, 0.75), supporting its value as a simple, high-impact intervention to improve the safety of emergency airway management.

Our findings highlight the protective effect of first-attempt intubation success in reducing MAEs (aOR, 0.52), with each additional intubation attempt significantly increasing the odds of severe hypoxemia (aOR, 2.28). Operator specialty was significantly associated with first-attempt success, emergency medicine residents achieved higher success rates, emphasizing the importance of specialized training in emergency airway management. Although anesthesiologists are considered airway, they performed very few intubations in our cohort (3 first-attempt intubations and 11 attempts overall), limiting meaningful comparisons with other specialties. Despite evidence supporting the use of videolaryngoscopes and bougies to improve first-attempt success [27, 28] these devices were underutilized in our cohort, similar to the INTUBE⁴ cohort in which videolaryngoscopy was used in only 17% of cases. This limited use, along with the high proportion of novice first operators, may have contributed to the lower first-attempt success rate observed in our study (74.3%). The higher incidence of severe hypoxemia (12.5%) compared with Russotto et al⁴ may also be explained by this lower first-attempt success rate, as well as the predominant use of bag-valve-mask ventilation for preoxygenation, rather than noninvasive ventilation or high-flow nasal oxygen. Although evidence suggests that stylet use may reduce complications and improve first-attempt success, 15% of patients in our cohort were intubated without any auxiliary devices [29]. Future efforts should focus on improving access to these devices in LMICs.

The choice of induction agent for intubation remains an important area of investigation. In this study, etomidate was associated with fewer MAEs in univariate analysis; however, the association was not significant in the multivariable analysis. The survival benefit of etomidate remains uncertain, and recent evidence, including a meta-analysis by Kotani et al., has reported findings favoring ketamine [30,31,32,33]. Further research is needed to clarify the comparative effectiveness of induction agents in critically ill patients..

Strengths and limitations

Our study has several strengths. As the first prospective airway cohort established in a low- and middle-income country, this research captures variability in practice beyond what is typically observed in high-income countries. Additionally, we adhered to current best practices in causal inference, carefully selecting confounders and conducting sensitivity analyses to address both modeling assumptions and unmeasured confounding, which enhances the interpretability of our findings.

However, our study has limitations. First, we did not follow up with patients after hospital discharge, which may result in an underestimation of mortality. To address this, we conducted multiple sensitivity analyses to address potential informative censoring, and the effect estimates remained consistent. We did not assess longer-term outcomes either, which would be a gap for future research. Second, despite adjusting for multiple confounders for the association of MAEs with 28-day mortality, the possibility of residual confounding cannot be entirely excluded, particularly in the context of hemodynamic instability, which was more susceptible to unmeasured confounding. Importantly, the other observed associations of multivariable analyses should be interpreted as exploratory to guide future research. Third, although each center had a case manager, not all consecutive intubations were captured in the participating centers. Nonetheless, the incidence of cardiac arrest among non-enrolled patients was comparable to that among enrolled patients, suggesting minimal selection bias. Fourth, while the center enrollment process was broad, most participating centers were academic institutions with emergency residency programs, which may have led to a more selected sample with potentially higher standards of care compared to other centers in Brazil. Fifth, we did not collect specific data on withdrawal of care, which limits our ability to evaluate its impact on 28-day mortality, although there is no strong reason to assume that withdrawal decisions were directly influenced by peri-intubation adverse events, except in cases of peri-intubation cardiac arrest. Furthermore, we did not record the annual rate of intubations performed at each site, limiting our ability to assess how intubation volume or operator experience could have influenced outcomes. Moreover, the inclusion of patients exclusively from emergency departments may be a limitation, as the findings may not fully reflect outcomes among all critically ill patients, particularly regarding the incidence of adverse events and 28-day mortality. However, in our setting, critically ill patients are often intubated in the ED before an ICU bed becomes available. As such, these results provide important insights into the quality of pre-ICU care and highlight opportunities for improvement during this critical period. Lastly, a few centers were rural units without ICU beds, so patients were transferred after stabilization, and we did not have access to 28-day hospital outcomes for these patients.

Conclusions

In conclusion, one in three patients experienced a peri-intubation major adverse event, which may increase 28-day mortality. First-attempt success, pre-intubation hemodynamics, pre-oxygenation, and sedative choices are potentially modifiable factors that may reduce the risk of MAEs. These findings highlight an urgent need for targeted interventions to mitigate peri-intubation adverse events in resource-constrained settings.

Ethics approval and consent to participate

This study was approved by the ethics committees of all participating centres. Due to the observational nature of the study and the use of de-identified data, a waiver of individual patient consent was granted by each institution’s ethics committee. Certificate of Presentation for Ethical Consideration at Coordinator Center: CAE-52424821.1.0000.0068.

Consent for publication

Not applicable.