Tracheal intubation in critically ill patients: a comprehensive systematic review of randomized trials

Critically ill patients frequently require tracheal intubation in the intensive care unit (ICU), in the emergency department (ED), or during in-hospital emergency in general wards [1‐3]. Critically ill patients differ from elective surgical patients intubated in the operating theatre: often they present with severe respiratory failure, hemodynamic instability, increased sensitivity to the side effects of sedatives, recent food intake, and cardiac or cerebrovascular diseases [2]. Airway management outside the operating theatre has a high rate of major complications, such as severe hypoxia, hemodynamic collapse, cardiac arrest, and death [1, 3‐6]. Moreover, the incidence of difficult intubation is high when compared to elective intubation in the operating room [1, 7, 8]. Lack of training, supervision and assistance, failure to identify patients at risk, failure to plan and carry out a backup strategy if required, and deficiency in available equipment, all are the most relevant modifiable risk factors [6]. Other factors that could increase the incidence of adverse events include the emergent requirement of tracheal intubation preventing adequate preparation, and the commonly limited physical space around ICU beds. Accordingly, interventions to improve everyday practice, in particular pre-oxygenation and first-attempt success rate, have been proposed and evaluated individually or as combined in bundles [2, 3, 9]. However, these protocols are based mostly on expert opinions, low-quality retrospective or before-after studies, or are derived from guidelines developed for elective intubation in the operating theatre. So far, there has not been a systematic review focused on randomized controlled trials (RCTs).

We performed a systematic review of RCTs evaluating any drug, technique, or device aimed at improving the success rate or the safety of tracheal intubation in critically ill patients performed in ICU, ED or general ward settings. When feasible, we also performed a meta-analytic assessment of these findings.

Database searches and scanning of references yielded 880 articles. From these we finally identified 22 randomized clinical trials for inclusion in seven areas of interest (Additional file 1: Figure S1): use of a pre-procedure check-list (1 study) [10], pre-oxygenation and apneic oxygenation (6 studies) [11‐16], sedatives (3 studies) [17‐19], neuromuscular blocking agents (NMBA) (1 study) [20], patient’s position (1 study) [21], videolaryngoscopy (9 studies) [22‐30], and post-intubation recruitment maneuver (RM) (1 study) [31].

One multicenter RCT evaluated the efficacy of a verbally performed 10-item pre-intubation checklist [10] compared to no checklist in 262 enrolled critically ill patients: no difference was found in any outcome (lowest peripheral oxygen saturation (SpO2), lowest systolic blood pressure, number and length of intubation attempts, life-threatening episodes, or in-hospital mortality).

The six RCTs focusing on pre-oxygenation and apneic oxygenation were heterogeneous in treatment and comparator groups [11‐16]. Pre-oxygenation refers to the administration of oxygen before induction (though some oxygen delivery devices used for pre-oxygenation can then be left on after induction or even during laryngoscopy). Apneic oxygenation refers to oxygen applied to a patient who is not spontaneously breathing (i.e. during induction-to-laryngoscopy and laryngoscopy-to-intubation periods). The characteristics and findings of these six RCTs (563 patients overall) are summarized in Table 1. The two studies applying non-invasive ventilation (NIV) for pre-oxygenation compared to standard pre-oxygenation showed positive results [11, 13] In a meta-analysis of the five studies applying high-flow nasal cannula (HFNC) [12‐16], stratified for the application as pre-oxygenation tool and/or as apneic oxygenation tool, limited evidence suggests that HFNC is ineffective if used for apneic oxygenation, while it might have some efficacy in improving the levels of lower oxygen saturation, but without improving the incidence of severe desaturation if used for pre-oxygenation (Additional file 1: Figures S2 and S3).

Three studies compared sedatives. Sivilotti et al. compared thiopental, fentanyl and midazolam (together with NMBA) during rapid sequence induction in the ED in 86 critically ill patients [17]: thiopental slightly shortened the time to intubation but was associated with lower reduction in systolic blood pressure compared to fentanyl and midazolam, while midazolam was associated with an increase in heart rate compared to fentanyl and thiopental. In the second study, alfentanil, sufentanil, and fentanyl were compared during rapid sequence intubation (RSI) in the ED in 90 trauma patients [18]: no significant difference in hemodynamic parameters was observed. Finally, Jabre et al. evaluated etomidate versus ketamine in 469 critically ill patients in 12 EDs and 65 ICUs in France [19]: no difference in intubation conditions or intubation-related adverse events was found.

In the only RCT on NMBA, succinylcholine was compared to rocuronium for RSI in 401 ICU patients [20]: no difference was observed in intubation conditions, rate of success of first attempt, and oxygen desaturation episodes, but the duration of the intubation sequence was on average 14 seconds shorter with succinylcholine.

One multicenter trial compared the sniffing position (entire bed flat, with the patient’s head elevated) and ramped position (upper half of the bed raised at an angle of 25° and the neck extended to have the patient’s face parallel to the ceiling) during laryngoscopy in 260 patients [21]. The sniffing position allowed a better view of the glottis across the full range of body mass indices and prior level of experience, and reduced the number of intubation attempts; however, it did not improve oxygenation, hemodynamic, or other clinical outcomes.

Nine studies compared different models of videolaryngoscopy to direct, traditional laryngoscopy in different conditions and settings in 2069 patients (Table 2) [22‐30]. Time to intubation and first-attempt success rate were the most relevant reported outcomes. Videolaryngoscopy did not shorten the time to intubation (Fig. 1a), nor the first-pass success rate (Fig. 1b), even when evaluating the studies according to the greater or lesser experience of the operators, or according to the setting (ICU versus ED), or the model used (hyper-angulated vs non-hyper-angulated) (Additional file 1: Figures S4 − S6). Four studies [22, 25, 27, 29] analyzed the subgroups with anticipated difficult airways: no study found a difference in the outcomes. The two largest trials found an increased incidence of severe complications in post-hoc analyses when videolaryngoscopy was employed: Yeatts et al. reported longer duration of the intubation procedure, greater incidence of severe desaturation episodes and highest mortality rate in the group with severe head injury, while Lascarrou et al. reported an increased incidence of life-threatening complications [22, 29].

Finally, Constantin JM et al. evaluated the efficacy of an RM (40 cmH₂O for 30 seconds) immediately after successful RSI in 40 ICU hypoxemic patients [31]. The RM group had significantly higher arterial pO₂ both 5 and 30 minutes after the RM; no difference in hemodynamic parameters was observed, although one RM was interrupted due to hypotension.

We identified and meta-analyzed 22 RCTs in seven different areas, evaluating drugs, techniques or devices aimed at improving the success rate or the safety of tracheal intubation in critically ill patients. The main findings were a possible beneficial role of pre-oxygenation with NIV and/or HFNC, the effect of the ramped position in increasing the number of intubation attempts, the negative impact of thiopental on blood pressure, and the possible efficacy of post-intubation RM in increasing arterial PaO₂. No other significant effect was found in the use of a checklist, choice of drugs, neuromuscular blockers, and use of videolaryngoscopy (the latter being associated with severe adverse effects in four trials). Further research in this poorly explored area is required.