Inhaled anesthetic induction with maintenance of spontaneous ventilation is a method for management of patients with a potentially difficult airway.1–9 One of its proposed advantages is that if airway patency is lost, inhaled drug delivery ceases and redistribution will result in spontaneous awakening and airway recovery. Sevoflurane has largely supplanted halothane in the developed world for inhaled induction, but halothane may still be used in austere medical environments, and past studies have used halothane as a point of comparison with sevoflurane. Inhaled inductions are most often performed in the pediatric population, but have also been described in adults and other populations with or without anticipated difficult airways, including obstetric anesthesia.2–6
There are little, and conflicting, clinical data on factors affecting spontaneous recovery from inhaled induction after airway obstruction, especially in differing physiologic conditions and patient populations.10,11 The event of cannot-ventilate cannot-intubate is rare, potentially catastrophic, and difficult to study. Computer modeling and simulation provide methods for exploring these situations. In this study, we used a computational simulation of the kinetics of sevoflurane and halothane after airway obstruction during inhaled induction to investigate factors that affect time to spontaneous recovery.
We used GasMan® version 4.1.5022 (Med Man Simulations Inc., Chestnut Hill, MA) to perform our simulations. GasMan is a computer simulation program based on a physiologic multicompartment model of inhaled anesthetic uptake and distribution. GasMan has been used for the teaching of inhaled anesthetic kinetics,12,13 as well as the accurate modeling of volatile anesthetic concentrations during induction and emergence.12–18
GasMan is based on a 6-compartment model: alveolar functional residual capacity (FRC), vessel-rich group that includes the brain, muscle, arterial blood, venous blood, and fat compartments with an additional compartment for the anesthetic circuit. Equilibration among these compartments is assumed to follow first-order kinetics for partial pressures or tensions based on the blood/gas and tissue/gas solubility and tissue perfusion with arterial blood. Intertissue diffusion, drug metabolism, and ventilation/perfusion variation in the lungs are not simulated.12,13 Table 1 displays the blood/gas and tissue/gas partition coefficients used in this study. Further details can be found in the User Manual for GasMan.a
We constructed our simulation model with the following simplifying assumptions:
- Complete and unrecoverable airway obstruction occurs at an unpredicted and discrete anesthetic level in the brain. This is represented in our simulation by an airway obstruction threshold level in the vessel-rich group. Because there is no known airway obstruction threshold in the literature, we chose a large range of values to explore the effect on time to spontaneous recovery.
- In our simulation, airway recovery occurs only when the vessel-rich group’s anesthetic level decreases below the airway obstruction threshold. Airway obstruction is complete and cannot be resolved with common maneuvers such as chin lift or insertion of airway devices.
- The effects of the studied parameters are sufficiently independent, such that the behavior of the system can be understood by varying only 1 parameter at a time while holding all other parameters constant.
At the initiation of each computer simulation, the semiclosed circuit was primed and the vaporizer set to the induction concentration of anesthetic specified for that simulation. Fresh gas flow was set at 10 L/min. Alveolar ventilation was then initiated. When the vessel-rich group’s anesthetic level reached the airway obstruction threshold specified for that simulation, alveolar ventilation was reduced to 0 and the circuit was flushed to produce zero inspired concentration. The simulation was continued with alveolar ventilation at 0 until the vessel-rich group’s decreased below the airway obstruction threshold. This time was recorded as time to spontaneous recovery. Figure 1 shows a screen capture of a running simulation.
Each specific simulation was performed once. Two simulations were performed at baseline parameters once with sevoflurane and once with halothane as detailed in the central column of Table 2. A simulation was performed for each single-parameter change from the baseline values as shown in the flanking columns of Table 2. Sevoflurane set concentrations of 2, 3, and 4 minimum alveolar concentration (MAC) were examined representing a wide, but clinically useful range. Halothane set concentrations of 4 and 6 MAC were simulated. Lower halothane concentrations were not examined because they would result in impractically slow inductions. Cardiac outputs between 2.5 and 10 L/min were arbitrarily chosen as a broad but reasonable range around a normal output of 5 L/min to explore system behavior. Similarly, the range in FRC was chosen broadly around a normal value of 2.5 L, and organ perfusion around a normal organ perfusion distribution of 76% in the vessel-rich group, 18% in the muscle group, and 6% in the fat group. This resulted in 25 simulations, including baseline parameters and a single variation from baseline parameters.
Combinations were simulated between variations in airway obstruction thresholds and set anesthetic induction concentrations for sevoflurane and halothane, because these are potentially clinician-modifiable factors. Further simulations examining the interaction between cardiac output and induction concentration were performed because large variations in cardiac output can occur within the same patient and this may influence the induction concentration used by the clinician. Variations in interactions of FRC and relative vessel-rich group perfusion with sevoflurane were simulated because these parameters are generally unmodifiable patient physiologic parameters.
To provide contrast to previous human studies comparing sevoflurane and halothane using end-tidal concentration, 4 additional simulations were performed. In these simulations, alveolar instead of vessel-rich group anesthetic concentration were used as an end point for obstruction. The first 2 were performed with halothane or sevoflurane with a set induction concentration of 4 MAC and airway obstruction threshold of 1 MAC. For the next 2 simulations, the airway obstruction threshold was 2 MAC. The set induction concentration of sevoflurane was 3 MAC, and set induction concentration for halothane was 6 MAC with an open circuit. These settings for halothane were chosen to achieve an increase in alveolar concentration in a similar timeframe as 3 MAC set induction sevoflurane concentration. Physiologic model parameters were otherwise at baseline. These simulation parameters were chosen to illustrate the qualitative difference between sevoflurane and halothane when alveolar concentration end points are used.
Figure 2 displays anesthetic level plotted over time of an example sevoflurane and halothane induction simulation where the set induction concentration was 4 MAC and the airway obstruction threshold was 1 MAC with the baseline physiologic model. As can be seen, there appear to be 3 general phases:
- The initial alveolar and vessel-rich group wash-in phase where a gradient develops between the alveolar and vessel-rich group compartments. The gradient is much larger, and induction occurs more rapidly with sevoflurane because of the rapid increase in alveolar level.
- At cessation of ventilation, there is an overshoot of the vessel-rich group concentration as it equilibrates with the alveolar concentration and alveolar concentration decreases rapidly. This overshoot is much larger for sevoflurane because of the higher alveolar concentration and its lower blood/gas solubility.
- Finally, after the rapid alveolar and vessel-rich group equilibration, there is a slow redistribution out of the vessel-rich group that is at approximately the same rate regardless of induction agent.
Notably, in the short-time course of inhaled induction, there is no significant buildup in the peripheral muscle. Even more so, anesthetic levels in the fat compartment are effectively 0 and are not shown.
Overall, time to spontaneous recovery in these simulations for sevoflurane ranged from 35 (at 2 MAC induction concentration with air obstruction threshold of 1.5 MAC) to 749 seconds (at 4 MAC induction concentrations with airway obstruction threshold of 1 and 10 L/min cardiac output). For halothane, time to spontaneous recovery ranged from 13 (at 4 MAC induction concentration with an airway obstruction threshold of 2 MAC) to 222 seconds (with 6 MAC induction concentration with airway obstruction threshold of 1 MAC and cardiac output of 2.5 L/min).
Table 3 shows the effect of single-parameter changes relative to baseline parameters on time to spontaneous recovery. Sevoflurane had a much longer time to spontaneous recovery than halothane under all equivalent conditions, 289 vs 85 seconds, respectively, at baseline parameters. The relative magnitude of effect on time to spontaneous recovery of induction concentration, airway obstruction threshold, cardiac output, FRC, and vessel-rich group is very similar between sevoflurane and halothane, seen in parenthetical data of Table 3.
Figure 3 displays the effects of different induction concentrations and airway obstruction thresholds on time to spontaneous recovery. As induction concentration is increased, there is a prolongation of the time to spontaneous recovery at all airway obstruction thresholds. As airway obstruction threshold is decreased, there is increasing prolongation of time to spontaneous recovery. There appears to be a synergistic interaction between the effect of airway obstruction threshold and the set anesthetic induction concentration. Direct multiplication of the magnitude of the single-parameter changes from Table 3 tend to underpredict time to spontaneous recovery at low airway obstruction thresholds and overpredict at high airway obstruction thresholds.
Figure 4 demonstrates the effect of cardiac output with different set induction concentrations on time to spontaneous recovery. With increasing cardiac output, time to spontaneous recovery decreases at a decreasing rate. At very low cardiac output, time to spontaneous recovery is extremely prolonged especially when high induction concentrations are used. However, there does not appear to be a synergistic effect between cardiac output and induction concentration because direct multiplication of the magnitude of the single-parameter changes from Table 3 predicts time to spontaneous recovery for sevoflurane induction at 2 MAC across the simulated cardiac output.
Figure 5 shows the effects of changes in FRC and changes in relative vessel-rich group perfusion. This graph demonstrates that increasing the relative perfusion of the vessel-rich group prolongs time to spontaneous recovery. Increasing FRC also prolongs time to spontaneous recovery. These effects are relatively small compared with other factors in the context of the large physiologic range explored in these simulations: Vessel-rich group perfusion ranges from 67% to 85% of cardiac output and 1.5 to 3.5 L for FRC. There does not appear to be a synergistic effect between the changes in FRC and vessel-rich group perfusion because the times to spontaneous recovery are well predicted by direct multiplication of the magnitude of the single-parameter changes from Table 3.
Figure 6 illustrates the simulations using alveolar anesthetic concentration as an end point instead of vessel-rich group anesthetic levels. Shown in the upper graph of Figure 6, when delivered in equipotent induction concentrations, halothane takes significantly longer to achieve the same alveolar concentration because of its higher solubility. Although both anesthetics reach 1 MAC, the vessel-rich group saturation is higher for halothane. After obstruction, halothane has a much faster second-phase decrease in alveolar concentration, but this is limited by the higher vessel-rich group level, and the sevoflurane alveolar concentrations eventually decrease to a lower level than halothane.
In contrast, the lower graph in Figure 6 illustrates the situation of delivering a higher induction concentration of halothane to achieve 2 MAC in a similar time period to sevoflurane. In this case, vessel-rich group levels of both anesthetics are similar when alveolar concentrations reach 2 MAC. After obstruction, alveolar halothane decreases more rapidly. Alveolar sevoflurane has a slower second-phase decrease as vessel-rich group levels continue to increase, and both alveolar and vessel-rich group levels of sevoflurane are higher than halothane in the third phase.
These results show that spontaneous recovery from inhaled induction after complete airway obstruction within a clinical timeframe is plausible, but highly variable. Time to spontaneous recovery varied widely depending on the clinical and physiologic circumstances, between 13 seconds and > 12 minutes in these simulations. There are multiple clinician-controlled and patient factors that influence these results.
The most easily controlled factor in shortening time to spontaneous recovery from inhaled induction is reducing induction concentration or slowing induction. This reduces vessel-rich group overshoot resulting from the larger alveolar to vessel-rich group gradient after airway obstruction. This is more pronounced with a less-soluble agent, such as sevoflurane. These effects are especially significant in patients with low cardiac output as seen in Figure 4. A lower airway obstruction threshold results in significantly longer time to spontaneous recovery. When inhaled induction is used for a potential difficulty airway, these results would advise against the concomitant use of adjuvants that may lower the obstruction threshold, such as opioids or other sedatives. They would also suggest that additional caution be taken in patients such as the elderly or those with obstructive sleep apnea if spontaneous recovery after obstruction is a consideration.
Patient-specific physiologic variations also affect time to spontaneous recovery. Low cardiac output prolongs time to spontaneous recovery by both increasing the alveolar to vessel-rich group gradient and slowing redistribution away from the vessel-rich group. This results in a larger and longer overshoot of the vessel-rich group and slower constant rate of washout after cessation of alveolar ventilation. Patients with relatively increased perfusion to the vessel-rich group, such as infants, will likely demonstrate a prolonged time to spontaneous recovery after airway obstruction, and additional caution should be observed in these patient populations. In obese, pregnant, or other patient populations with reduced FRC, there would be a lesser but notable effect, shortening time to spontaneous recovery. However, clinically this benefit is likely lost in the concurrent reduction in apneic reserve until hypoxia.
Most patients will have a combination of these physiologic perturbations. For example, the obese patient could have decreased FRC, decreased fractional vessel-rich group perfusion, and increased cardiac output that would accelerate spontaneous recovery. However, their lower airway obstruction threshold from associated sleep apnea would result in a slower spontaneous recovery time. Sepsis is another example that can result in a variety of physiologic disturbances. Sepsis might result in an increased cardiac output and decreased relative vessel-rich group perfusion, shortening time to spontaneous recovery. However, sepsis can also reduce cardiac output and reduce anesthetic levels required for airway obstruction. In these complex combinations, the exact magnitude of these changes in each patient could be difficult to predict but would have a significant effect on the spontaneous recovery time.
Two previous studies in human subjects tried to examine the relative speed of decrease in end-tidal sevoflurane and halothane during simulated airway obstruction with conflicting conclusions. In the study by Girgis et al.,11 patients were induced to an end-tidal concentration of 2 MAC, and then the airway was occluded for 3 minutes after which the end-tidal concentration was measured. In that experiment, sevoflurane was found to have a lower end-tidal concentration after the 3-minute occlusion. Talbot et al.10 had healthy volunteers breathe halothane and sevoflurane of 0.1 MAC end-tidal concentration, then initiated rebreathing for 90 seconds while measuring end-tidal concentrations. In contrast to the results of the study by Girgis et al., Talbot et al. found that halothane had a much more rapid decay in end-tidal concentrations.
The results of our simulation reconcile these previous studies by emphasizing that induction is a non–steady-state situation, and there is a significant delay between the alveolar concentration and the vessel-rich group. Thus, using end-tidal concentration end points would result in different tissue levels depending on the anesthetic agent and induction. Sevoflurane, because of its lower solubility and faster increase in alveolar concentration, results in lower saturation of the vessel-rich group. This allows for a larger decrease in end-tidal concentration from redistribution after airway obstruction. In contrast, because of its higher solubility, halothane has a more rapid immediate redistribution, but it is limited by higher saturation of the vessel-rich group during the slower wash-in phase. This is illustrated in Figure 6. This effect is more pronounced at higher end-tidal end point concentrations such as 2 MAC vs 0.1 MAC, but it is not actually clinically relevant because the vessel-rich group is the site of anesthetic action, and not the alveoli.
Thus, although the end-tidal concentrations in these previous human experiments were controlled, the actual vessel-rich group anesthetic levels, which is the effect site of anesthetics, were likely different between anesthetic agents. Future human experiments should not depend on end-tidal concentration as an end point unless relative steady-state circumstance can be ensured. Rather, the effect site should be modeled, and the results were then analyzed. Modeling of effect site has been shown in other studies to significantly improve the predictive accuracy.19
Also notable, the decay rate of anesthetic level from the vessel-rich group was very similar between sevoflurane and halothane after the vessel-rich group has equilibrated with the alveolar compartment. This can be seen in Figures 2 and 6 and is because the rate of redistribution from the vessel-rich group is predominantly controlled by cardiac output and relative perfusion of the other compartments. The difference in anesthetic solubility is minor. Thus, although our results show halothane to provide a more rapid time to spontaneous recovery, this was because of the slower induction with a smaller alveolar to vessel-rich group gradient leading to less overshoot after cessation of alveolar ventilation.
In the broader context of difficult airway management, the 4th National Audit Project of the Royal College of Anaesthetists and the Difficult Airway Society. Major Complications of Airway Management in the United Kingdom 8 included multiple case reports of failed inhaled induction with loss of airway and without spontaneous recovery. The results of this study explain the clinical observation of that study. Herein, we have shown that spontaneous redistribution of inhaled anesthetic after airway obstruction is highly variable and depends on the multiple physiologic and clinical parameters. Furthermore, patients would likely become hypoxic and hypercarbic before redistribution of the anesthetic. In these cases, spontaneous recovery of the airway cannot be relied on, and other decisive steps must be taken to ensure oxygenation and ventilation.
The limitations of this study are inherent in computation simulations and the assumptions of the model used. Our simulations used arbitrarily designated values for airway obstruction threshold based on a fraction of MAC because there is no known clinical value for this threshold. Thus, we used a wide range of these designated values to demonstrate the scope of time to spontaneous recovery behavior. Our model involved complete airway obstruction, but maneuvers that even partially recover the airway would lead to kinetics more similar to traditional emergence from inhaled anesthetics and more rapid recovery. Our model ignored metabolism and intercompartment diffusion. However, these are unlikely to be important in the short timeframe of an inhaled induction. More significantly, we assumed airway obstruction occurs perfectly abruptly at a specific vessel-rich group level and recovery occurs at the same vessel-rich group level. We also assumed that there would be no difference in propensity to cause airway obstruction between the different anesthetic agents. Furthermore, we ignored the dynamic changes in physiologic parameters, such as cardiac output, relative tissue perfusion, FRC, and alveolar ventilation. All these factors would have an impact on spontaneous recovery time but would be complicated to model simultaneously and would likely lead to model overspecification. Regardless, the results of this study, with its controlled assumptions, provide a general framework for future clinical research.
Future research could include repetition of human studies similar to those by Talbot et al.10 or Grigis et al.11 but could apply the data to effect-site modeling or computer simulation instead of end-tidal concentration as an end point. These results could be compared in differing patient populations, such as obese versus nonobese. This would validate the physiologic model and assumptions used in this study and provide clinical credence to the results so that they may be extrapolated to other patient populations not directly tested. From a broader standpoint, computational simulation can be used to explore other difficult-to-study clinical situations, such as reanesthetization from hypoventilation after emergence.19
In conclusion, within its limitations, this computational simulation study provides insights into inhaled induction for difficult airways. Spontaneous recovery based on the anesthetic redistribution during airway obstruction is plausible, but highly variable depending on the clinical and physiologic parameters and cannot be relied on consistently. This study emphasizes the importance of effect-site modeling and also demonstrates that computer simulation offers an exploratory approach to anesthetic scenarios that are difficult to study in practice. The results of this study can be used to provide insights into clinical practice and guide future clinical research.
Name: Alexander S. Kuo, MS, MD.
Contribution: This author helped design the study and prepare the manuscript.
Attestation: Alexander S. Kuo approved the final manuscript, attests to the integrity of the original data and the analysis reported in this manuscript, and is the archival author.
Conflicts of Interest: Alexander S. Kuo declares no conflicts of interest..
Name: Mary A. Vijjeswarapu, MD.
Contribution: This author helped analyze data and prepare the manuscript.
Attestation: Mary A. Vijjeswarapu approved the final manuscript and attests to the integrity of the original data and the analysis reported in this manuscript.
Conflicts of Interest: Mary A. Vijjeswarapu declares no conflicts of interest.
Name: James H. Philip, ME(E), MD.
Contribution: This author helped design the study and prepare the manuscript.
Attestation: James H. Philip approved the final manuscript and attests to the integrity of the original data and the analysis reported in this manuscript.
Conflicts of Interest: James H. Philip is the creator of GasMan. GasMan is owned and distributed by a 501(C)(3) Charitable Organization, Med Man Simulations, Inc., in which James H. Philip is President. James H. Philip received no financial benefit from GasMan during the period of this study.
This manuscript was handled by: Ken B. Johnson, MD.
a User Manual for GasMan®. Available at: http://www.gasmanweb.com/software.html. Accessed July 5, 2012.
1. Patel A, Pearce A. Progress in management of the obstructed airway. Anaesthesia. 2011;66(suppl 2):93–100
2. Mostafa SM, Atherton AM. Sevoflurane for difficult tracheal intubation. Br J Anaesth. 1997;79:392–3
3. MacIntyre PA, Ansari KA. Sevoflurane for predicted difficult tracheal intubation. Eur J Anaesthesiol. 1998;15:462–6
4. Kandasamy R, Sivalingam P. Use of sevoflurane in difficult airways. Acta Anaesthesiol Scand. 2000;44:627–9
5. Péan D, Floch H, Beliard C, Piot B, Testa S, Bazin V, Lejus C, Asehnoune K. Propofol versus sevoflurane for fiberoptic intubation under spontaneous breathing anesthesia in patients difficult to intubate. Minerva Anestesiol. 2010;76:780–6
6. Nakazawa K, Ikeda D, Ishikawa S, Makita K. A case of difficult airway due to lingual tonsillar hypertrophy in a patient with Down’s syndrome. Anesth Analg. 2003;97:704–5
7. Ratner EF, Cohen SE, El Sayed Y, Druzin M. Mask induction with sevoflurane in a parturient with severe tracheal stenosis. Anesthesiology. 2001;95:553–5
8. Cook T, Woodall N, Frerk C 4th National Audit Project of the Royal College of Anaesthetists and the Difficult Airway Society. Major Complications of Airway Management in the United Kingdom. 2011 London RCoA:143–54
9. Philip BK, Lombard LL, Roaf ER, Drager LR, Calalang I, Philip JH. Comparison of vital capacity induction with sevoflurane to intravenous induction with propofol for adult ambulatory anesthesia. Anesth Analg. 1999;89:623–7
10. Talbot NP, Farmery AD, Dorrington KL. End-tidal sevoflurane and halothane concentrations during simulated airway occlusion in healthy humans. Anesthesiology. 2009;111:287–92
11. Girgis Y, Frerk CM, Pigott D. Redistribution of halothane and sevoflurane under simulated conditions of acute airway obstruction. Anaesthesia. 2001;56:613–5
12. Philip JH. Gas Man—an example of goal oriented computer-assisted teaching which results in learning. Int J Clin Monit Comput. 1986;3:165–73
13. Eger EI II, Shafer SL. Tutorial: context-sensitive decrement times for inhaled anesthetics. Anesth Analg. 2005;101:688–96
14. Bouillon T, Shafer SL. Hot air or full steam ahead? An empirical pharmacokinetic model of potent inhalational agents. Br J Anaesth. 2000;84:429–31
15. De Wolf AM, Van Zundert TC, De Cooman S, Hendrickx JF. Theoretical effect of hyperventilation on speed of recovery and risk of rehypnotization following recovery—a GasMan® simulation. BMC Anesthesiol. 2012;12:22
16. Hendrickx JF, Lemmens HJ, Shafer SL. Do distribution volumes and clearances relate to tissue volumes and blood flows? A computer simulation. BMC Anesthesiol. 2006;6:7
17. Van Zundert T, Hendrickx J, Brebels A, De Cooman S, Gatt S, De Wolf A. Effect of the mode of administration of inhaled anaesthetics on the interpretation of the F(A)/F(I) curve—a GasMan simulation. Anaesth Intensive Care. 2010;38:76–81
18. Leeson S, Roberson RS, Philip JH. Hypoventilation after inhaled anesthesia results in reanesthetization. Anesth Analg. 2014;119:829–35
19. Johnson KB, Syroid ND, Gupta DK, Manyam SC, Pace NL, LaPierre CD, Egan TD, White JL, Tyler D, Westenskow DR. An evaluation of remifentanil-sevoflurane response surface models in patients emerging from anesthesia: model improvement using effect-site sevoflurane concentrations. Anesth Analg. 2010;111:387–94