Secondary Logo

Journal Logo

Anesthetic Pharmacology: Research Report

The Effect of Anesthetic Choice (Sevoflurane Versus Desflurane) and Neuromuscular Management on Speed of Airway Reflex Recovery

McKay, Rachel Eshima MD*; Hall, Kathryn T. BA, MPH*; Hills, Nancy PhD†‡

Author Information
doi: 10.1213/ANE.0000000000001022
  • Free

Protective airway reflexes may be impaired during early recovery from anesthesia, after a patient awakens, follows commands, and meets common extubation criteria. Postoperative airway reflex status is partly influenced by the choice of maintenance anesthetic, based on data from multiple studies of nonintubated patients receiving sevoflurane versus desflurane without muscle relaxant.1–3 The faster return of airway reflexes observed after desflurane anesthesia compared with sevoflurane anesthesia is probably attributable to desflurane’s lower tissue solubility, a property that allows less anesthetic uptake during induction and maintenance and faster decline in central nervous system partial pressure during and after emergence.4

An understanding of factors that might facilitate rapid, predictable airway reflex recovery is of greater clinical importance for safe management of intubated patients compared with those managed with LMAs™ because the intubated population includes patients with greater medical complexity (e.g., full stomach, fragile neurologic or pulmonary conditions) and thus have a greater risk of adverse events resulting from airway reflex impairment. Because intubated patients typically receive a muscle relaxant during anesthesia, the extent to which choice of maintenance anesthetic, as opposed to neuromuscular management, affects airway reflex recovery is unclear. Numerous correlations have been observed between inadequate neuromuscular recovery and absent protective airway reflexes, impaired ventilatory function, and adverse outcomes.5–9 It would therefore seem reasonable to expect a higher incidence of early postoperative airway reflex impairment among intubated compared with nonintubated patients. These correlations between incomplete neuromuscular recovery and adverse outcomes may be profound enough to call into question the clinical significance of a chosen anesthetic, in terms of its role in airway reflex recovery among intubated patients because the effects of variable neuromuscular management and recovery might predominate.

However, because neuromuscular block is intensified in the presence of potent inhaled anesthetics, and neuromuscular recovery depends on concurrent elimination of small residual minimum alveolar concentration (MAC) fractions of volatile anesthetic, we hypothesized finding faster, more predictable airway reflex recovery in intubated patients after desflurane versus sevoflurane anesthesia because of desflurane’s lower tissue solubility.10,11 Furthermore, we anticipated finding this faster recovery after desflurane versus sevoflurane, despite the substantial variability in neuromuscular recovery rates expected between patients receiving rocuronium.12,13 To distinguish between the influence of a chosen anesthetic, neuromuscular management would need to follow a standard protocol of monitoring, maintenance, and reversal as much as possible. Additional important information would concern the rates of neuromuscular recovery in each patient and adherence to the neuromuscular protocol by the clinician. With those considerations, we set out to discover whether the impact of the chosen anesthetic (sevoflurane versus desflurane) on airway reflex recovery would be small and clinically unimportant or independently significant within the context of antagonized neuromuscular block.

Accordingly, in patients intubated for laparoscopic surgery, given rocuronium titrated to standard depth of relaxation and subsequent neostigmine antagonism, we compared the 2 intervention groups with the specific aims of (1) measuring timing intervals related to discontinuation of anesthetic, first response to command, and first ability to swallow; and (2) determining the proportion of subjects demonstrating airway reflex recovery adequate to allow for successfully passing a swallow test at 2 minutes after response to command. To standardize muscle relaxant conditions as much as possible among subjects and to avoid excessive or insufficient intraoperative neuromuscular block, a block management protocol was written for this study that included quantitative twitch measurement, rocuronium titration to specific parameters, and attainment of TOF ≥0.7 before extubation. We examined the influence of the clinician’s adherence to the neuromuscular protocol on airway reflex recovery as defined by the ability to swallow after first response to command, in patients overall, and stratified by adherence/nonadherence.


This study was registered with, with identifier NCT01199237, on August 23, 2010 (REM, principal investigator), and approval was obtained from the University of California San Francisco Committee on Human Research. We recruited 81 ASA physical class I to II patients, aged 18 to 65 years and with body mass index (BMI) <35 kg/m2, who were scheduled to undergo surgery of 2 to 3 hours’ duration requiring general anesthesia with intubation and muscle relaxation. Patients were randomized to receive either sevoflurane or desflurane for the maintenance of anesthesia. Written consent was obtained from all participants, and the research technician was blinded to the identity of the assigned anesthetic by placement of an opaque boundary over the vaporizers on the anesthesia machine and the end-tidal anesthetic readings.

All patients were administered and passed the preanesthetic 20-mL swallowing test as previously described.11,12 Patients with any of the following conditions were excluded: pre-existing neuromuscular or central nervous system disorder, conditions interfering with gastric emptying, asthma or reactive airways disease, known liver disease, serum creatinine >1.5 mg/dL, or a previous diagnosis of obstructive sleep apnea. We additionally excluded patients using neuroleptic medications or opioids, or with a contraindication or previous adverse response to any of the study drugs. Patients whose planned surgical procedures involved the head and neck, as well as those who would be unable to assume a postoperative upright seated position or in whom coughing would be contraindicated were also excluded from the study.

The clinical anesthesia team received a copy of the randomization assignment in a sealed envelope. All patients received 25 µg/kg midazolam (maximum 2 mg) before being transported to the operating room, where they were connected to standard American Society of Anesthesiologists monitors and to a neuromuscular monitor, the TOF-Watch SX (Oss, the Netherlands), which measures neuromuscular function by accelerometry. The clinical outcome of interest during neuromuscular block and recovery is force transduction. Because neuromuscular monitoring is a crucial aspect of this study, a brief discussion of the topic is included in Appendix 1.

Each patient received 1.5 µg/kg fentanyl, 1 mg/kg lidocaine, and 1.0 to 3.5 mg/kg propofol at induction, after which the neuromuscular monitor was placed over the right or the left ulnar nerve and calibrated to baseline function of the adductor pollicis muscle, with 2-Hz stimulation for 2 seconds delivered once every minute. Additional propofol was permitted as necessary during calibration of the TOF-Watch. The ability to ventilate via facemask was determined before 1 mg/kg rocuronium was administered to facilitate intubation. Endotracheal tube sizes were standardized to 6.5-mm internal diameter for female and 7.0-mm internal diameter for male patients. Controlled ventilation was established with target end-tidal CO2 32 to 36 mm Hg, and each patient’s anesthesia was maintained with either sevoflurane or desflurane (in 80% oxygen/20% nitrogen in 2 L/min fresh gas flow) as randomized.

Neuromuscular Monitoring and Management

When accelerometry response exceeded 10% of baseline value, if further relaxation was clinically desired, rocuronium infusion was started at 0.5 mg/kg/h. The protocol directed the clinician to maintain first twitch height between 10% and 15% of baseline value, with output from TOF stimulation provided every 1 minute. Fentanyl infusion at 1 µg/kg/h was initiated 15 minutes after induction. Signs of inadequate analgesia were treated with additional boluses of 0.5 µg/kg fentanyl, and potent inhaled anesthetic was adjusted as clinically indicated. During wound closure, the clinician was permitted to decrease this anesthetic to as low as 0.5 MAC end-tidal partial pressure (sevoflurane ≥0.9% or desflurane ≥3.0%). Fentanyl and rocuronium infusions were discontinued 20 to 30 minutes before anticipated conclusion of surgery. Neostigmine 70 µg/kg + glycopyrrolate 14 µg/kg were administered when (1) the response to first stimulation reached ≥25% of baseline value, or (2) presence of at least 1 response was palpable at the adductor pollicis in response to ulnar TOF stimulation.

When TOF reached ≥0.7, the potent inhaled anesthetic was turned off, fresh gas flow was increased to 15 L/min, and controlled ventilation continued with a target end-tidal CO2 36 to 44 mm Hg. The commands “open your eyes” and “squeeze my hand” were given at 30-second intervals thereafter, and the patient’s trachea was extubated at the time deemed appropriate by the clinical anesthesiologist. After extubation, the upper body was elevated 60° with the axis of the head between 70° and 90° upright, unless excessive sedation, circulatory instability, or other adverse symptoms precluded doing so in the judgment of the clinical anesthesiologist.

Administration of the Swallowing Test: The Effect of Anesthetic (Sevoflurane Versus Desflurane) on Interval Between Awakening and First Ability to Swallow

At 2 minutes after first appropriate response to command, the patient was asked to swallow 20 mL of water from a paper cup, and the research technician, blinded to anesthetic assignment, assessed the swallowing effort (passing the test was defined as swallowing the entire volume of water in one effort without drooling from the lips, pooling in the hypopharynx, gagging, or coughing). This test was repeated at 6, 14, 22, 30, and 60 minutes after the time of first appropriate response to command. All patients were evaluated for 1 hour continuously after first appropriate response to command by the research technician. The protocol indicated that (1) any clinician responsible for the patient at any point during this hour (anesthesiologist, certified registered nurse anesthetist, or postanesthesia care unit nurse) could request that the research associate abstain from administering the swallowing test at any of the predetermined time intervals if he/she judged that the test would impose a safety risk or unreasonable discomfort to the patient, and (2) the patient could refuse this test himself/herself at any point. If at any time point the test was not given, the technician noted the reason.

A series of time intervals were measured (in minutes) for the purposes of comparing desflurane and sevoflurane, including “T1”: time elapsed from discontinuation of anesthetic until first response to command; “T2”: time from first response to command until first ability to swallow (2, 6, 14, 22, 30, 60, and >60 minutes); and “T3”: time from anesthetic discontinuation until first ability to swallow (T3 = T1 + T2). All 3 time intervals were compared between patients receiving either desflurane or sevoflurane, the primary predictor.

We also compared the ability to swallow at 2 minutes after response to command. This was first done in all patients, with those unable to take the swallow test at 2 minutes for any reason counted as “failures.” We then repeated the analysis in a subgroup of 71 patients categorizing those who had responded appropriately to command but were too somnolent at 2 minutes to take the test as failures and excluding those unable to take the test at this time for unrelated reasons (n = 10). We also tested differences in ability to swallow at 2 minutes after response to command overall and between the desflurane and sevoflurane groups separately in patients whose care did and did not adhere to the use of the TOF-Watch data to guide the administration of rocuronium or who did not receive full-dose (70 µg/kg) neostigmine. This was done in the entire cohort and in the subgroup described earlier.

Additional data were collected on compliance with the administration of full reversal dose (versus underdosing), length of surgery, duration of anesthesia, MAC hours, and weight-adjusted total intraoperative fentanyl and rocuronium use. Information on patient demographics, such as age, BMI, and ASA physical status, was also collected. A single research technician administered all swallowing tests, recorded all relevant time stamps, and collected TOF-Watch data, as well as nausea and vomiting scores.

Statistical Analysis and Hypotheses

Demographic characteristics between the 2 groups were compared using χ2 tests for categorical variables. Normally distributed continuous variables were compared using t tests, whereas the nonparametric Wilcoxon rank sum test was used for continuous variables about which no distributional assumptions could be made.

Our null hypothesis was that airway reflex recovery, indicated by passing the swallowing test, would not differ according to randomized maintenance anesthetic (sevoflurane versus desflurane). We hypothesized that the presence of antagonized neuromuscular block, itself a possible delaying influence on airway reflex recovery, would exert an effect on each subject’s recovery that would minimize or obscure the contributions related to the maintenance anesthetic. The alternative hypothesis was that swallowing ability after first response to command would differ according to maintenance anesthetic, with patients receiving desflurane showing more rapid rates of recovery than patients receiving sevoflurane in terms of measured time intervals as described earlier and ability to swallow at 2 minutes after first response to command.

Sample size calculation was based on our previous research in nonintubated patients, which showed that, among 80 subjects whose BMI was <35 kg/m2, 47% (19/40) of subjects receiving sevoflurane and 100% (40/40) of patient receiving desflurane passed the swallowing test 2 minutes after first appropriate response to command (χ2 = 26.8). Because it seemed reasonable to anticipate a smaller effect size attributable to the anesthetic, because the impact of the rocuronium would be present in all patients, we assigned an airway reflex recovery probability of 0.6 in the sevoflurane group and 0.95 in the desflurane group, calculating the need for a minimum of 37 subjects in each arm for β = 0.9 and 2-sided α = 0.05. To be conservative, we chose to enroll 81 patients in total. All statistical analysis was performed with Stata version 13® (College Station, TX).

Differences in time intervals not normally distributed were analyzed by nonparametric testing, with P values for differences calculated using Wilcoxon-Mann-Whitney (WMW) tests. WMWodds ratios and 95% confidence intervals were calculated for these comparisons.14,15 In the case that a P value derived from a Wilcoxon rank sum test was significant at a level between 0.01 and 0.10, simulation techniques were used to generate more reliable P values. In univariate generalized linear regression models fit using iteratively reweighted least squares (maximum quasi-likelihood) regression to estimate relative risks, we examined, in addition to anesthetic assignment, several other predictors potentially associated with the ability to swallow at 2 minutes after first response to command.

To compare the proportion in each intervention group passing the swallow test at 2 minutes, χ2 tests were used to calculate P values, except in cases where observations in at least one cell numbered <5, in which case the Fisher exact test was used.

The effect of compliance with protocol was tested by comparing rates of passing the swallow test at 2 minutes in compliant versus noncompliant groups overall using a χ2 test. Likewise, rates in the 2 groups were compared among only those actually taking the test at 2 minutes and among those actually taking the test in addition to those too somnolent (counted as failures), using χ2 tests. Time to first successful swallowing was compared in compliant versus noncompliant groups using the WMW test; WMWodds and confidence intervals were generated, and the P value for this test was simulated. We then compared the rates of passing at 2 minutes in the desflurane and sevoflurane groups stratified by whether or not there was compliance with the neuromuscular protocol. This was done using χ2 and Fisher exact tests.

We used logistic regression techniques to test potential associations between the ability to pass the swallow test at 2 minutes after first response to command and various predictors. Because of the high rate of outcome in our study, we estimated relative risks rather than odds ratios, using generalized linear models.16 Univariate analysis was first conducted with our primary predictor (desflurane versus sevoflurane) and with other potential predictors. Multivariable models were then run adjusting for protocol adherence, using extensions of generalized linear models to achieve convergence. Models were run for all 81 patients who were randomized (categorizing all those who did not take the swallow test at 2 minutes as having failed the test) and in the tested subset of 71 patients as described earlier. All statistical analyses were performed with Stata version 13.


Data Collection and Clinical Care

Anesthesia was administered by numerous providers in the following configurations: (1) faculty anesthesiologists working alone; (2) faculty anesthesiologists supervising residents; and (3) faculty anesthesiologists medically directing CRNAs. In all, there were 33 faculty members, 17 CRNAs, and 11 residents involved in clinical care. There were no discernable patterns of anesthetic dosing or other care patterns among these individuals. An individual blinded to anesthetic assignment, a nonclinician and member of the study team, collected data.

Patient Characteristics

Demographic information for the entire cohort is included in Table 1. Many baseline characteristics were similar in the sevoflurane and desflurane arms, including age, ASA physical status, anesthetic duration, and use of rocuronium. Patients receiving desflurane had slightly lower BMIs (P = 0.02) and received fewer anesthetic MAC hours (P = 0.03) but more fentanyl overall (P = 0.004) than those in the sevoflurane group (although fentanyl administered during the final 30 minutes of anesthesia did not differ significantly between groups; P = 0.22). The rate of adherence to the neuromuscular monitoring and reversal protocol was also lower in the desflurane group (P = 0.03; Table 1). In cases where patients could not take the swallowing test at any particular time interval, the condition cited most frequently for deferring the test to the subsequent time interval was drowsiness, followed by dizziness, nausea, and pain (Fig. 1). Data on the subset of 71 patients (excluding those with irrelevant reasons for failing to take swallow test at 2 minutes) are presented.

Table 1:
Baseline Patient Characteristics
Figure 1:
Percentage of subjects who underwent swallowing test at each time period after first response to command, by anesthetic. This population included patients who were not managed according to neuromuscular protocol. Reasons for failure to administer test were coded (1 = patient too sedated; 2 = postural, patient light-headed or clinician concerned about circulatory stability for patient to assume upright position; 3 = patient refusing because of pain; 4 = patient refusing because of nausea; 5 = patient unavailable; 6 = other causes). At 2 min after response to command, 26 of 81 patients did not undergo testing (17 patients receiving sevoflurane, 9 patients receiving desflurane). Of the 26 patients untested at 2 min: 16 were “too sedated” (14 received sevoflurane, 2 received desflurane); 4 were postural (3 received desflurane, 1 received sevoflurane); 1 refused because of pain (received desflurane); 2 refused because of nausea (both received desflurane); 2 were unavailable, being transported from OR to postanesthesia care unit (received sevoflurane); and 1 was coughing after extubation before swallowing test (received desflurane).
Table 2:
Time from Anesthetic Discontinuation to First Response to Command and to Airway Reflex Recovery: Comparison of Sevoflurane Versus Desflurane Anesthesia

Median (interquartile range [IQR]) time from anesthetic discontinuation to first appropriate response to command (T1; Table 2) was significantly shorter among patients receiving desflurane compared with patients receiving sevoflurane (P = 0.0001). Median (IQR) time from anesthesia discontinuation to first ability to swallow (T2; Table 2) was also shorter after desflurane (P = 0.0007). Median time from first response to command to first ability to swallow (T3) was 120 seconds (120, 360) in the desflurane groups compared with 360 seconds (120, 360) in the sevoflurane group, although their IQRs overlapped (P = 0.054, P value calculated using simulation). WMWodds and 95% confidence intervals for comparisons between treatment groups are presented in Table 2.

The Effect of Anesthetic (Sevoflurane Versus Desflurane) on the Interval Between Awakening and First Ability to Swallow)

Table 3:
Comparison of Ability to Pass the Swallowing Test at 2 Minutes After First Response to Command Between Patients Receiving Sevoflurane Versus Desflurane
Table 4:
Factors Associated with Passing the Swallowing Test at 2 Minutes: Univariate Logistic Regression
Figure 2:
A, Percentage of subjects passing the water swallowing test by time (min) after first appropriate response to command by anesthetic. This includes all subjects tested, regardless of clinician adherance to the neuromuscular protocol. Subjects judged as “too sedated” to be tested were coded as failures in this analysis. Subjects not test for reasons other than somnolence (n = 10, citing nausea, pain, refusal) were not counted. The total number of patients considered in this analysis at each test point following first response to command: 71 at 2 minutes; 76 at 6 minutes; 72 at 14 minutes; 74 at 22 minutes; 74 at 30 minutes; 76 at 60 minutes. B, Percentage of subjects passing the water swallowing test by time (minutes) after first appropriate response to command, by anesthetic, excluding patients whose management was not adherent to the neuromuscular protocol (n = 66). Subjects judged as “too sedated” to be tested were coded as failures in this analysis (n = 16 at 2 min). Excluding the patients whose neuromuscular management deviated from protocol, virtually all patients receiving desflurane passed the swallowing test consistently at each recovery time point.

We also compared the 2 groups with respect to rates of swallowing at 2 minutes after first response to command. Patients who were tested or who were judged to be too somnolent to undergo testing (n = 71) are presented in Table 3. Those in the desflurane group were significantly more likely to pass than those in the sevoflurane group when all 81 patients were included (P = 0.04), as well as when this was tested in the subgroup of 71 patients (P = 0.006; Table 4). Figure 2A shows the percentage of patients who passed the test at each time point, excluding patients who were not tested for reasons other than somnolence.

Protocol Adherence: Neuromuscular Monitoring and Administration of Reversal

The TOF-Watch device was applied to all 81 subjects at the beginning of anesthesia. Overall, the neuromuscular protocol was followed without deviation in 77.8% of the entire cohort (63/81 subjects) and 77.5% (55/71) of the smaller cohort comprising subjects who were tested at 2 minutes after response to command. Deviations from protocol involved one or more of the following: (1) failure to capture TOF-Watch data for the duration of surgery, with empiric rocuronium administration and reliance on tactile measurement of TOF and tetanus at the conclusion of surgery (7 patients); (2) proceeding to extubate the patient using subjective criteria, despite availability of TOF-Watch data showing TOF ratio <0.7 (8 patients); or (3) decision made on the part of the clinician to administer <70 µg/kg of neostigmine based on his/her own clinical comfort level (4 patients, 3 from first category and 1 from second category above). Figure 2B shows the percentage of patients who passed the test at each time point, excluding patients who were not tested for reasons other than somnolence and those whose neuromuscular management deviated from protocol.

Recovery data from these subjects whose neuromuscular management deviated from protocol as described earlier were analyzed in comparison with subjects whose management followed the protocol. In the entire cohort, regardless of maintenance anesthetic, those whose neuromuscular management was compliant with standard protocol were significantly more likely to pass the swallow test at 2 minutes (83.7% vs 41.7%; P = 0.003; results not shown). Patients whose neuromuscular management complied with protocol also demonstrated earlier successful swallowing than did those who management deviated from protocol (WMWodds 1.91; 95% CI, 1.09–3.80; P = 0.329; results not shown; P value calculated using simulation). Of the 18 patients in the group whose neuromuscular management deviated from protocol, 13 were tested at the 2-minute point, with 8 of 13 (61.5%) failing the test. This rate increased to 68.8% (11/16) when patients who responded to command but were too somnolent to take the test were included as failures. In comparison, among the remaining 44 subjects actually tested at 2 minutes of the 63 overall whose management adhered to protocol, only 8 of 44 (18.2%) failed, a rate still only raised to 34.5% if somnolent patients are counted as failures (19/55) (χ2 results, P = 0.01 for only those actually tested; P = 0.015 with somnolent patients included). Comparative results at other times are shown in Figure 3.

Figure 3:
Percentage of subjects passing the swallowing test by time (minutes) after first appropriate response to command, according to whether the neuromuscular protocol, was followed by the clinician.

We then tested in both groups whether desflurane had a different effect than sevoflurane in patients when stratified by adherence to protocol. In both cohorts, among patients whose neuromuscular management either did or did not adhere to protocol, those in the desflurane group performed better than those in the sevoflurane group, although this difference was significant in the protocol-adherent group of the entire cohort (P = 0.004) and in the protocol-adherent group of the subcohort (P < 0.0001) but not significant in the protocol nonadherent group in either case, although we likely lacked the power to detect this difference in such small groups (n = 18/81 and n = 16/71 who were included in the analysis; Table 3).

The Effect of Protocol Adherence on Airway Reflex Recovery

Models were created, which included all 81 subjects (with all those unable to take the test at 2 minutes defined as having failed), and in the subset of 71 subjects evaluated at 2 minutes after response to command, which excluded those who were unable to take the test for reasons other than somnolence after having first responded to command. We found that treatment with desflurane was significantly associated with the ability to swallow at 2 minutes when all 81 patients were included (P = 0.04) and when only the subset of 71 patients (as described earlier) were included (P = 0.006). Failure to pass at 2 minutes was associated with nonadherence to neuromuscular protocol in the univariate model, in all 81 enrolled patients, as well as in the tested cohort (Table 4).

The Combined Effect of Anesthetic and Adherence to Neuromuscular Protocol

Table 5:
Factors Associated with Passing the Swallowing Test at 2 Minutes: Multivariate Logistic Regression

When inhaled anesthetic assignment was adjusted in a multivariable model by neuromuscular protocol adherence (using an extension of the generalized linear model), both factors significantly affected a patient’s ability to pass the swallow test at 2 minutes, whether we examined the entire cohort of 81 or the subgroup of 71 (Table 5). We also constructed models adjusting for patient age, MAC hours, BMI, fentanyl, and rocuronium; however, the significant impact of anesthetic and neuromuscular protocol adherence persisted, and a modest impact of MAC hours was seen, while none of the other factors contributed significantly (analysis not shown).


This study shows that desflurane is associated with faster and more consistent recovery of protective airway reflexes compared with sevoflurane in intubated patients who receive rocuronium, with subsequent antagonism by neostigmine. The predictability of airway reflex recovery after desflurane anesthesia is consistent with its low tissue solubility. A separate component of patient care, that of adherence to a standardized protocol for titration, monitoring, and antagonism of the neuromuscular block, exerted both a clinically and a statistically significant influence on protective airway reflex recovery that may be of equal or greater magnitude than that of the chosen anesthetic.

Although airway reflex recovery requires full neuromuscular recovery to TOF ≥0.9, complete neuromuscular recovery does not guarantee complete recovery of protective airway reflexes. Because airway reflex recovery cannot be tested and confirmed before extubation, we sought to examine factors that would predict that the reflexes are most likely to be intact. Although our findings confirm the relationship among a less soluble anesthetic, timely awakening, and airway reflex recovery, the management of neuromuscular block plays a significant, possibly equivalent, or greater role that is independent of the chosen anesthetic (Table 5; Fig. 3). The comparative contributions of neuromuscular management and chosen anesthetic cannot be determined on the basis of our findings. However, we argue that most adverse clinical outcomes are related to multiple factors, including choice and dose of drug, vigilance and appropriateness of monitoring, and clinical decision making, and that their relative contributions vary in every clinical scenario. Using desflurane, as opposed to sevoflurane, may not be financially feasible for all clinicians; a more predictable neuromuscular reversal drug (sugammadex) is expensive and not currently Food and Drug Administration approved, and quantitative train-of-four monitoring is not currently part of standard practice in the United States. The consistent recovery of swallowing at 2 minutes among patients receiving desflurane, managed according to neuromuscular protocol, implies a property unique to desflurane, probably its lower tissue solubility, that allows rapid decline in anesthetic partial pressure well below that where airway reflexes are impaired.

We considered the possibility that clinicians may have titrated anesthetic differently near the conclusion of surgery between groups, thereby affecting partial pressure near the time of emergence in a manner that could favor recovery in patients receiving desflurane. In fact, analysis of end-tidal MAC fraction during the final 15-minute epoch of anesthesia showed a trend toward greater MAC fraction in the desflurane group during the final 15-minute epoch that was not statistically significant (median [IQR]): desflurane (0.55 [0.19, 0.75]) versus sevoflurane (0.27 [0.22, 0.65]), P = 0.1702. We also considered the possibility that differing opioid delivery near the conclusion of the case might affect airway reflex recover, a difference not captured by simple reporting of overall fentanyl delivery per hour of anesthesia. However, analysis of fentanyl delivered to patients in the final 30 minutes before response to command did not differ significantly according to maintenance anesthetic.

There are several limitations to this study. First, the lack of uniform adherence to protocol makes a pure association between anesthetic and outcome difficult to establish. Second, airway reflex status of patients who are not tested, even when the cited reason for not testing is drowsiness, cannot be determined with certainty. Restricting inclusion to only those subjects who took the swallow test at 2 minutes would have resulted in a substantial loss of power. However, we feel that including patients who had responded appropriately to command before the 2-minute test, but were too somnolent at 2 minutes to take the test, as failures, is justifiable. Protective airway reflexes have been shown to be impaired frequently at levels of anesthesia as low as 25% of MAC awake, when patients are concurrently responsive to command.13 Despite these limitations, we conclude that anesthetic choice and neuromuscular management contribute to differences in early recovery of protective airway reflexes. Differences in these very early time periods are of clinical importance when we consider that emergence is a time of substantial risk of aspiration and that transport of patients from the operating room to recovery areas also occurs during this time—a time when suction is usually unavailable and the clinician may be distracted by tasks related to record keeping, moving a gurney, and verbally communicating with other care providers.


Neuromuscular function, as measured by the TOF-Watch device, uses accelerometry rather than direct transduction of force. Force (mass × acceleration) exerted by skeletal muscle is the physical entity that is of clinical relevance during recovery that ensures adequate ventilatory function and airway protection. A force transducer requires a strain gauge with adequate resting tension against the muscle being tested and for that reason may be very sensitive to changes in hand position or preloading conditions. Accelerometers use an array piezoelectric crystals which move in concert with the thumb during adductor pollicis contraction when ulnar nerve stimulation is applied by the TOF-Watch. The displacement creates an electrical signal that is proportional to the displacement velocity, and an assumption is made that the mass of the thumb is constant. Previous investigation has shown excellent agreement between measurements made by accelerometers compared side-by-side with conventional force transducers.17 Therefore, we have taken the liberty to consider accelerometry measurements to be directly correlated with force exerted by the adductor pollicis muscle.


Name: Rachel Eshima McKay, MD.

Contribution: This author helped design the study, conduct the study, analyze the data, and prepare the manuscript.

Attestation: Rachel Eshima McKay approved the final manuscript, reviewed the original study data and data analysis, and attests to the integrity of the original data and the analysis reported in this manuscript. Rachel Eshima McKay is the archival author.

Conflicts of Interest: Rachel Eshima McKay received research support for Baxter Healthcare Corporation.

Name: Kathryn T. Hall, BA, MPH.

Contribution: This author helped conduct the study and collect the data.

Attestation: Kathryn T. Hall approved the final manuscript, reviewed the original study data and data analysis, and attests to the integrity of the original data and the analysis reported in this manuscript.

Conflicts of Interest: None.

Name: Nancy Hills, PhD.

Contribution: This author helped in statistical analysis and editing the manuscript.

Attestation: Nancy Hills approved the final manuscript, reviewed the original study data and data analysis, and attests to the integrity of the original data and the analysis reported in this manuscript.

Conflicts of Interest: None.

This manuscript was handled by: Ken B. Johnson, MD.


1. McKay RE, Large MJ, Balea MC, McKay WR. Airway reflexes return more rapidly after desflurane anesthesia than after sevoflurane anesthesia. Anesth Analg. 2005;100:697–700
2. McKay RE, Malhotra A, Cakmakkaya OS, Hall KT, McKay WR, Apfel CC. Effect of increased body mass index and anaesthetic duration on recovery of protective airway reflexes after sevoflurane vs desflurane. Br J Anaesth. 2010;104:175–82
3. Sundman E, Witt H, Sandin R, Kuylenstierna R, Bodén K, Ekberg O, Eriksson LI. Pharyngeal function and airway protection during subhypnotic concentrations of propofol, isoflurane, and sevoflurane: volunteers examined by pharyngeal videoradiography and simultaneous manometry. Anesthesiology. 2001;95:1125–32
4. Eger EI II, Bowland T, Ionescu P, Laster MJ, Fang Z, Gong D, Sonner J, Weiskopf RB. Recovery and kinetic characteristics of desflurane and sevoflurane in volunteers after 8-h exposure, including kinetics of degradation products. Anesthesiology. 1997;87:517–26
5. Eriksson LI, Sundman E, Olsson R, Nilsson L, Witt H, Ekberg O, Kuylenstierna R. Functional assessment of the pharynx at rest and during swallowing in partially paralyzed humans: simultaneous videomanometry and mechanomyography of awake human volunteers. Anesthesiology. 1997;87:1035–43
6. Sundman E, Witt H, Olsson R, Ekberg O, Kuylenstierna R, Eriksson LI. The incidence and mechanisms of pharyngeal and upper esophageal dysfunction in partially paralyzed humans: pharyngeal videoradiography and simultaneous manometry after atracurium. Anesthesiology. 2000;92:977–84
7. Eikermann M, Blobner M, Groeben H, Rex C, Grote T, Neuhäuser M, Beiderlinden M, Peters J. Postoperative upper airway obstruction after recovery of the train of four ratio of the adductor pollicis muscle from neuromuscular blockade. Anesth Analg. 2006;102:937–42
8. Murphy GS, Szokol JW, Marymont JH, Greenberg SB, Avram MJ, Vender JS. Residual neuromuscular blockade and critical respiratory events in the postanesthesia care unit. Anesth Analg. 2008;107:130–7
9. Berg H, Roed J, Viby-Mogensen J, Mortensen CR, Engbaek J, Skovgaard LT, Krintel JJ. Residual neuromuscular block is a risk factor for postoperative pulmonary complications. A prospective, randomised, and blinded study of postoperative pulmonary complications after atracurium, vecuronium and pancuronium. Acta Anaesthesiol Scand. 1997;41:1095–103
10. Wright PM, Hart P, Lau M, Brown R, Sharma ML, Gruenke L, Fisher DM. The magnitude and time course of vecuronium potentiation by desflurane versus isoflurane. Anesthesiology. 1995;82:404–11
11. Lowry DW, Mirakhur RK, McCarthy GJ, Carroll MT, McCourt KC. Neuromuscular effects of rocuronium during sevoflurane, isoflurane, and intravenous anesthesia. Anesth Analg. 1998;87:936–40
12. Debaene B, Plaud B, Dilly MP, Donati F. Residual paralysis in the PACU after a single intubating dose of nondepolarizing muscle relaxant with an intermediate duration of action. Anesthesiology. 2003;98:1042–8
13. Sacan O, White PF, Tufanogullari B, Klein K. Sugammadex reversal of rocuronium-induced neuromuscular blockade: a comparison with neostigmine-glycopyrrolate and edrophonium-atropine. Anesth Analg. 2007;104:569–74
14. Divine G, Norton HJ, Hunt R, Dienemann J. Statistical grand rounds: a review of analysis and sample size calculation considerations for Wilcoxon tests. Anesth Analg. 2013;117:699–710
15. Dexter F. Wilcoxon-Mann-Whitney test used for data that are not normally distributed. Anesth Analg. 2013;117:537–8
16. Wilber ST, Fu R. Risk ratios and odds ratios for common events in cross-sectional and cohort studies. Acad Emerg Med. 2010;17:649–51
17. Itagaki T, Tai K, Katsumata N, Suzuki H. Comparison between a new acceleration transducer and a conventional force transducer in the evaluation of twitch responses. Acta Anaesthesiol Scand. 1988;32:347–9
© 2016 International Anesthesia Research Society