Airway Management in Intensive Care Medicine

Defining the Problem

Tracheal intubation is the most commonly performed procedure in the intensive care unit (ICU) and is associated with significant mortality and morbidity, ^, the incidence of which is higher in the ICU than in the operating room (OR). Nonetheless, some ICUs still manage tracheal intubation in a relatively primitive manner. ^, Few studies address how many ICU patients have a difficult airway (DA) and caseloads vary, but Astin and colleagues provided useful context for the intensivist: in general ICUs 4.1% of patients had been admitted to the unit for a primarily airway-related reason, and overall 6.3% were judged to have a DA. Studies quantifying airway difficulties in the ICU are also rare. Martin and colleagues reported 3423 emergency tracheal intubations performed by anesthesia residents with 60% occurring in the ICU. Difficult intubation (DI) (Cormack and Lehane [CL] grade 3 or 4, or >2 attempts) occurred in 10%, and complications were observed in 4.2% of intubations. This is approximately twice the incidence of DI seen in the OR (5.8%) and is consistent with other ICU studies reporting DI rates of 7% to 13%. ^{, , , ,} Mort reported a failed intubation rate of 1 in 10 to 20 attempts outside the OR, and Jaber and colleagues reported death as a complication in 0.8% of intubation attempts in the ICU. Thus, the DA in the ICU is potentially lethal and dramatically increases the risk of life-threatening complications (51% vs 36% with non-DAs).

The Fourth National Audit Project (NAP4) of the Royal College of Anaesthetists was a nationwide, prospective observational audit conducted over 1 year in the United Kingdom (UK). ^, It included data from the OR, emergency department (ED), and ICU. Inclusion criteria were airway-related mortality, brain damage, emergent surgical access, ICU admission, or prolongation of ICU stay. There were 38 airway-related deaths: 16 in the OR, 4 in the ED, and 18 in the ICU. Considering the denominator values were 2.9 million anesthetics in the OR, 20,000 tracheal intubations in the ED, and 58,000 episodes of advanced respiratory support in ICU, airway-related death is approximately 58 times more common per patient in the ICU than in the OR. The rates of death or brain damage seen when DAs are encountered in the ICU were 61%, compared with 14% in the OR. This suggests airway-related mortality rate of 1 in 2700 ICU patients versus 1 in 180,000 in the OR. NAP4 assessors stated that many of these ICU deaths were avoidable and found a higher incidence of poor and mixed-quality airway care than in the OR. One crucial difference is when ICU-related airway problems occur: The UK National Reporting and Learning System ICU airway incidents database documents 18% occurring at tracheal intubation but 82% after the airway had been secured, and 25% of these contributed to mortality. Other studies and closed claims analyses are consistent, and all suggest that many of these incidents are preventable. An updated analysis from 2019 of the Anesthesia Closed Claims Project demonstrated that 76% of cases had at least one predictor of DI, and 41% had at least two.

Why Is ICU Airway Management So Hazardous?

Common to all reports that evaluate ICU airway management is failed tracheal intubation, delayed recognition of esophageal intubation, training deficiencies, poor communication, and poor judgment (late recognition of evolving crises and slow escalation strategies). Institutional preparedness is often poor (e.g., inconsistent equipment provision, meager supervision of residents, or few local algorithms). The commonalities can be distilled into failure to evaluate and plan for the DA and failure to recognize and rescue the failed airway. There are repeated problems relating to displaced airways, particularly in obese patients and on patient movement or transfer. Often, patients with a DA are not recognized, and even when potential problems are identified early, coherent plans to manage crises are not formulated, including proper equipment and expertise being immediately available in the ICU. When difficulty arises, rescue techniques fail more frequently in the ICU than in other clinical settings. ^,

There are several reasons why the ICU is associated with such high risks. ICU patients are physiologically compromised, having significant ventilation/perfusion (V˙/Q˙) abnormalities and lower functional residual capacities (FRCs), which renders preoxygenation less effective and leads to reduced or nonexistent tolerance of apnea. Patients are frequently not fasted or have delayed gastric emptying. Patients are hemodynamically compromised, with exaggerated physiological responses to vasodilation from induction agents and the hemodynamic consequences of the transition to positive-pressure ventilation, especially in the setting of right ventricular strain. Severe metabolic acidosis means apnea and rising carbon dioxide (CO ₂ ) levels are poorly tolerated. Intubation in the ICU is often urgent, with little time for assessment or preparation, and minimally conscious or unconscious patients prohibit intubation options that are available in awake and cooperative patients. Apnea and hypercarbia are detrimental in patients with raised intracranial pressure (ICP). Airway assessment in the critically ill is problematic, so difficulty is often unanticipated. These issues are collectively referred to as “the physiologically difficult airway.”

Furthermore, ICU patients may have artificial airways in situ for days or months and are cared for by staff whose expertise is not primarily airway management. Patients’ airways are manipulated by these staff during oral hygiene and insertion of feeding tubes or other devices. They are positioned prone and transferred to remote areas of the hospital, during which artificial airways are extremely vulnerable to displacement. Often, critically ill patients are agitated and liable to self-extubate. Although sedation holds are vital to optimal ventilation, ^, they may lead to self-extubation. The fluid status of critically ill patients is problematic as well; patients can be hemodynamically compromised and may poorly tolerate sedatives and raised intrathoracic pressure, or may have significantly positive fluid balances with generalized edema, including of the airway, especially if prone ventilation is used. Even normal patients become difficult in such circumstances. ^,

In the ICU, access to the patient is challenging. Beds rather than operating tables, poor lighting, and lack of space make intervention difficult. ^, A commonly used approach during failed management in the OR is simply to allow the patient to awaken, but when intubation is indicated for respiratory, cardiovascular, or neurologic failure, this is not an option. Airway rescue with supraglottic airway devices (SGAs) and invasive access, especially needle cricothyrotomy, fails more commonly in the ICU. ICU equipment is often different from that in other areas of the hospital and deficient in important ways. The most striking of the NAP4 practice deficiencies, and one also seen around the world, is the poor use and understanding of capnography in the ICU. ^, Although the availability of life-saving airway devices is crucial, this should be team appropriate. Simply providing a plethora of devices reveals that unit leaders have failed to agree on how airways should be managed and have no clear strategies developed. Additionally, this paradox of choice hampers team performance because expertise can only be achieved in a relatively small number of techniques. Airway equipment should be purchased with the least experienced user in mind. ^,

Airway-related problems in the ICU occur after hours more commonly than in the OR (46% vs 31%). When difficulty with airway management arises, senior medical help or experienced airway surgeons are not available to help residents; some of the residents interviewed following NAP4 incidents did not recognize terms such as backward, upward, rightward pressure (BURP) or Combitube , revealing a lack of basic airway knowledge. The expertise of the doctors is important. ^, Joffe and colleagues conducted a national survey of airway training in US internal medicine ICU fellowships, and only 58% had a designated program. Actual experience varied markedly: Before graduation, 67% of fellows reported performing less than 50 direct laryngoscopies (DLs), 73% had used an SGA fewer than 10 times, 60% had used an intubating stylet fewer than 25 times, and 73% had used video-assisted laryngoscopy (VAL) fewer than 30 times. An SGA learning curve is more than 15 uses. Minimal competence in many other techniques is achieved with perhaps 30 to 60 uses, and expertise with more than 100 uses. ^, This problem is global. ^{, ,} Only two-thirds of ICUs in the UK have a resident in possession of the initial assessments of airway competency, equivalent to 3 months of dedicated airway training, and in only 27% of ICUs had the resident received specific training in how to manage a displaced airway. In 72% of Australian ICUs, the night doctor is not required to have prior anesthetic or airway training; in 97% of them, the senior airway doctor at night is either not stationed in the unit or may have duties elsewhere. Only 15% of the Australian units have a doctor who is present in the hospital overnight, who does not have other clinical commitments, and always has formal airway training. The skill set of ICU doctors is rarely specified: For instance, the US Accreditation Council for Graduate Medical Education simply states that a fellow in an internal medicine critical care fellowship “must demonstrate competence in airway management” but does not define this, specify a target number of procedures, or describe methods for formal competency assessment.

Evidence-Based Improvements

Evidence suggests that training and experience in airway management are likely the most important factors for improving airway-related outcomes in the ICU. A study with data from 22 of the 24 ICUs in Scotland suggested that airway care led by experienced practitioners can result in higher first-attempt tracheal intubation success rates (90% vs 63%–75%) and lower complication rates than in many other studies. ^{, , , ,} In this study, 74% of ICU physicians had more than 24 months of anesthesia training, and only 10% had fewer than 6 months of such training (all but one of the novice tracheal intubations were supervised by a senior). Success rate in tracheal intubation was related to duration of anesthesia training ( p < 0.001). In contrast, Griesdale did not demonstrate improved outcomes with experts rather than nonexperts, but in this study, 92% of nonexperts were immediately supervised by experts: The impact of a lack of expertise was therefore unclear, other than to confirm that nonexperts need significantly more attempts to achieve tracheal intubation, and more than two attempts is associated with more severe complications. Tellingly, Jaber and colleagues found that the presence of two operators (a junior and a senior) was protective against tracheal intubation complications in the ICU. ^, De Jong and colleagues reported that most studies are inadequately powered to analyze the effect of operator experience and that 24 months of anesthesia training was significantly associated with a lower incidence of DI. However, experience is not always protective. A 2018 study of patients intubated in the ICU within 30 days of an elective operation found that, despite being intubated by the same anesthesiologists, there was worsened glottic visualization, an 8% lower first-attempt success, a 7% increase in moderate or difficult intubation (16% overall), and a 31% increase in complications when intubated in the ICU.

Nontechnical factors are extremely important for DA-related outcomes. Pathologic thought processes occur during airway crises—practitioners overanalyze the situation, become task-fixated on tube placement, and neglect rescue oxygenation. An idealized, as opposed to pragmatic, solution is sought by the practitioner, which leads to delay and perseveration; the operator’s appreciation of how urgently they must deal with the situation fails. A loss of situational awareness means timely progression through airway algorithms slows or stops. This is especially true in the ICU, where high-risk physiologic derangements and the emergent need to secure an airway simply overload even experienced practitioners’ cognitive strategies. Crisis management can be taught and improved. Although there are similarities between OR and ICU airway management, there are sufficiently important differences to justify ICU-specific guidelines, as there are in obstetric and pediatric anesthesia practices.

There is no single innovation that can mitigate these risks. Like all demanding areas of human activity, performance is improved by the aggregation of marginal gains—achieving small improvements in every element of the task, whether relating to individuals, teams, or institutions. Quality improvement initiatives incorporating many of the following components discernibly improve performance metrics ^, : the use of intubation bundles to minimize procedural difficulty (e.g., checklists, standards of practice), optimizing human factors, adequate training and supervision, continuous waveform capnography, sound assessment and backup planning, using appropriate equipment with which the team is familiar and practiced, immediate availability of appropriate rescue devices, deploying the most appropriate rescue techniques when faced with difficulty, attention to detail once the airway has been secured, and carefully planned extubation.

Indication-Specific Points Relevant to Airway Management

Patients Who Fail Noninvasive Strategies: Implications for Intubation

In patients with respiratory failure, NIPPV reduces the risk of requiring intubation, but when NIPPV fails, the risk of death is increased. The source of this increased mortality is the delay in tracheal intubation, cardiorespiratory stress, patient self-inflicted lung injury, as well as the complications associated with emergent tracheal intubation. ^{, , ,} The increased risk of mortality with failure of HFNO does not appear to be the same, suggesting that NIPPV may be intrinsically injurious through patient self-inflicted lung injury. ^, The physiologic ramifications of delayed intubation may make complications of tracheal intubation more likely or more severe. Mosier and colleagues studied patients intubated after failed NIPPV and compared composite airway outcomes (hypoxemia, hypotension, and aspiration) with patients who were similarly intubated but had no prior NIPPV. A propensity-adjusted model for factors affecting the decision to use NIPPV showed an adjusted odds ratio for a complication of intubation in patients failing NIPPV of 2.20 (95% confidence interval [CI]: 1.14–4.25). Furthermore, when one of these complications occurred, the odds ratio of death in the ICU was 1.79 (1.03–3.12). There was no difference in the incidence of DI or number of attempts, and all tracheal intubations were supervised by an experienced senior. Certainly, the decision regarding timing of intubation when NIPPV fails is of vital importance. As patients receiving a trial of NIPPV fail, tachypnea, tachycardia, hypoxemia, acidosis, agitation, poor mask tolerance, loss of consciousness, and hemodynamic instability develop. If these become apparent by 1 to 2 hours after NIPPV is initiated, then even worse deterioration should be avoided by prompt intubation. This is especially true when the indication for NIPPV was weak in the first place: for instance, pneumonia or hypoxemic respiratory failure. For patients on HFNO, a ROX index (ratio of oxygen saturation [Spo ₂ ]/ Fio ₂ to respiratory rate) <4.88 at 2, 6, and 12 hours is an accurate predictor of patients who will likely fail and require intubation.

Assessment

It is advantageous to identify DA patients to avoid prolonged, complicated, unanticipated DA management and to facilitate better planning. Within the anesthetic literature, prediction of difficulty is credited with a reduction in mortality and morbidity, but this is not certain. Standard airway assessments all suffer from variably poor sensitivity and poor-to-moderate specificity and translate only partially to the critical care arena. ^, Emergent ICU intubation usually precludes detailed imaging investigation. A common approach in emergency patients is the LEMON (look, evaluate, Mallampati, obstruction, neck mobility) mnemonic. ^, Of the “look” criteria, Reed and colleagues found that only large incisors, reduced inter-incisor gap, and reduced thyrohyoid distance were associated with DI. Reduced thyromental distance, obstruction, and Mallampati score trended to significance. Only 57% of subjects in this study were suitable for Mallampati scoring. Similarly, Levitan found that only 32% of ED patients could follow simple commands (needed to perform Mallampati) while also not having cervical spine protection in place (permitting a cervical spine mobility assessment). However, Levitan and colleagues used the classic Mallampati assessment with the patient sitting up, opening the mouth fully, and maximally protruding the tongue, although it can be scored in the supine position in cooperative patients. They questioned LEMON’s utility because, even if it was able to better guide practitioners, few options for management other than rapid sequence intubation (RSI) or primary surgical airway exist in the ED. They stated that awake intubation has almost no role in uncooperative patients, those needing immediate intubation, or those with secretions, blood, or vomit in the airway.

Importantly, De Jong and colleagues developed the only ICU-specific assessment tool: The MACOCHA score ( Table 43.1 ). It is a seven-item assessment with ICU-specific risk factors for DI. A score of 3 or greater is recommended as the cutoff value. Its discriminant value is high. The MACOCHA predictive score includes two ICU-specific criteria: hypoxemia and coma before intervention. These are important because hypoxia permits less time for preparation, leads to quicker desaturation, and may increase operator stress; coma makes assessment more difficult and is associated with greater laryngeal contamination. Uniquely, MACOCHA scores operator experience: specifically, anesthesia training of at least 24 months. MACOCHA makes it explicit that OSA, rather than obesity per se, causes DAs. ^, The Mallampati score in this study was determined in the recumbent position and obtained in 77% of subjects. The best predictor of a DA is documentary evidence of previous difficulty, but documentation is typically poor. If a history is suggestive of upper airway obstruction, flexible nasal endoscopy is a valuable tool. Even when a DI is not predicted, unanticipated difficulty is still possible. Predictors of difficult mask ventilation (DVM) and SGA ventilation have been described but not validated in the ICU setting.

Awake Intubation

Awake intubation is regarded as the gold standard for management of the DA, and there is certainly a place for it in the ICU. Advantages include maintaining airway patency and spontaneous ventilation. Awake intubation includes flexible scope intubation (FSI), VAL, tracheostomy, cricothyrotomy, retrograde intubation, and intubation via an SGA. Combined upper and lower respiratory tract pathology renders ICU patients with DAs particularly challenging. ^{, , , ,} Contraindications to awake intubation in the ICU include noncooperative patient, inexperienced operator, need for immediate intubation, absolute refusal by patient, raised ICP, and dependency on NIPPV (although HFNO may significantly mitigate this contraindication).

Some patients have physiologic abnormalities that make RSI risky because of limited or no safe apnea time. These patients, usually suffering refractory hypoxemia, have reduced FRC, as well as high shunt fractions that create conditions unsuitable for maintaining hemoglobin saturation during apnea. In these patients, awake intubation can sometimes be advantageous to maintain spontaneous breathing if oxygen saturation is adequate using HFNO until they can be immediately transitioned to positive pressure ventilation. ^, No definitive studies favor awake intubation or RSI. Some authorities believe PEEP dependency prior to intubation favors RSI, especially if postinduction bag-mask ventilation eliminates the apneic phase. Significant hemodynamic compromise before intubation may lead to cardiovascular collapse with adequate doses of induction agents: Awake techniques should be considered.

Intubation Care Bundle

No single intervention can improve airway safety in the ICU, and so care bundles are required. Jaber and colleagues demonstrated that implementation of a 10-component bundle reduced life-threatening complications compared with controls. The incidence of Sp o ₂ less than 80% reduced from 25% to 10% using the bundle, as did hypotension (defined as systolic blood pressure [SBP] <65 mm Hg) from 26% to 15%. The incidence of esophageal intubation did not decrease, but the incidence of esophageal intubation with hypoxia reduced from 50% to 0%. The protocol, specifically aimed at reducing airway-related complications, included five evidence-based elements: the use of NIPPV, ^, the presence of two operators, RSI, capnography, and protective ventilation; plus five elements that were regarded as good practice based on experience: fluid loading in absence of cardiogenic pulmonary edema, preparation of long-term sedation, use of the Sellick maneuver, norepinephrine for diastolic hypotension, and prompt initiation long-term sedation. Which individual components resulted in improvement is open to question. Marshall and Mehra demonstrated that cognitive aids and checklists improve team performance and help clinicians complete tasks without freezing or panicking. Importantly, the checklist should cover all the areas that must be addressed prior to induction—namely, preparation of the patient, preparation of all equipment/medications that might be required, and preparation of the whole intubation team rather than just the operator. In addition, the checklist guides the team in how to verbalize preparation for DI, if it arises. ⁶⁸ A recent pragmatic trial by Janz and colleagues showed that a checklist prior to intubation did not improve outcomes. We advocate for using a bundle, which is proposed in Box 43.2 and borrows from Jaber’s work. This approach is amenable to simulation training and checklist-style deployment, which should be embedded into ongoing training. ^{, ,}

Rapid Sequence Induction and Intubation

ICU patients are rarely fasted and often have intraabdominal pathology or functional ileus; as a result, they are at high risk of regurgitation of gastric contents and pulmonary aspiration. The definition of RSI varies, but it is a technique to reduce the risk of pulmonary aspiration during intubation by minimizing the time interval between drug-induced loss of intrinsic airway reflexes and restoration of airway protection using an endotracheal tube (ETT) with its cuff inflated. The adoption of RSI by anesthesiologists and emergency physicians has improved success rates for the emergency tracheal intubation and has decreased complications, with similar results in the ICU. ^, Critical care physicians should learn to perform and become comfortable with all aspects of RSI.

Preintubation and Periintubation Oxygenation

Critically ill patients tolerate upper airway intervention and apnea poorly. Mort found that all ICU patients who needed three or more tracheal intubations developed hypoxemia with an Sp o ₂ of less than 70%. Saturation this low risks arrhythmias, cardiovascular collapse, brain damage, cardiac arrest, and death. Severe hypoxemia complicated 26% of ICU intubation attempts in Jaber’s study, and 1.6% sustained cardiac arrest. The underlying reasons for the profound drop in Sp o ₂ are a combination of intrapulmonary shunt, low mixed venous saturation (as a result of low cardiac output, anemia, and hypermetabolic states), and apnea/hypoventilation. The oxygenation maneuvers described here address these issues.