Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Neostigmine Versus Sugammadex for Reversal of Neuromuscular Blockade and Effects on Reintubation for Respiratory Failure or Newly Initiated Noninvasive Ventilation: An Interrupted Time Series Design

Krause, Martin MD*; McWilliams, Shannon K. MA; Bullard, Kenneth J. BS*; Mayes, Lena M. MD*; Jameson, Leslie C. MD*; Mikulich-Gilbertson, Susan K. PhD†,‡; Fernandez-Bustamante, Ana MD, PhD*; Bartels, Karsten MD, PhD*,§

Author Information
doi: 10.1213/ANE.0000000000004505

Abstract

See Article, p 137

KEY POINTS

  • Question: Were proportions of reintubation for respiratory failure or initiation of
  • new noninvasive ventilation reduced after a system-wide transition from neostigmine to sugammadex for reversal of rocuronium or vecuronium?
  • Findings: In this interrupted time series study design that included 7316 patients, the composite primary outcome of reintubation for respiratory failure or new noninvasive ventilation occurred in 6.1% of the presugammadex group and 4.2% of the postsugammadex group.
  • Meaning: A systemwide transition of the standard pharmacologic reversal agent from neostigmine to sugammadex was associated with a decrease in the odds of adverse postoperative respiratory outcomes.

Postoperative pulmonary complications (PPCs) are associated with increased perioperative mortality.1 Residual neuromuscular blockade (NMB) from pharmacologic muscle relaxation results in decreased functional residual capacity and laryngeal and pharyngeal dysfunction. This is clinically associated with hypoventilation, airway collapse, and impaired airway protection, which all contribute to PPCs.2–5 Optimal reversal pharmacologic muscle relaxation is therefore critical to avoid PPCs.6,7

For many years, cholinesterase inhibition has been the only reversal mechanism for nondepolarizing muscle relaxants with neostigmine being the preferred agent. Neostigmine reverses the effects of nondepolarizing muscle relaxants by increasing the concentration of acetylcholine at the synaptic cleft and displacing the nondepolarizing muscle relaxants from nicotinic acetylcholine receptors.8 Significant muscarinic side effects such as bradycardia, double vision, and postoperative nausea and vomiting are common. Vagolytic drugs such as glycopyrrolate or atropine are used to counteract the muscarinic side effects but have their own spectrum of side effects including tachycardia and xerostomia.9–11 Furthermore, the peak effect of neostigmine occurs approximately 10 minutes after injection with a duration of action of about 20–30 minutes.8 These pharmacodynamic properties can lead to residual paralysis in the postoperative period.

Sugammadex has been recently introduced into clinical practice in the United States. It irreversibly binds to and encapsulates the muscle relaxants rocuronium and vecuronium.12 Given its mechanism of action, muscarinic side effects are rare.9,11,13 Current literature consistently reports that sugammadex has a plasma elimination half-life of 2.2 hours and reverses the effects of rocuronium in ≤2.2 minutes independent of the depth of NMB.8,14–17 Complete reversal, defined as a train of 4 ratio >0.9, occurs at least 2.5 times faster than any neostigmine initiated reversal.18,19 Consequently, sugammadex led to a significant reduction in signs of postoperative residual paralysis compared to neostigmine (relative risk of 0.40 [95% confidence interval {CI}, 0.28–0.57]).11 Despite the faster reversal of NMB and fewer signs of residual paralysis with sugammadex compared to neostigmine, a decrease in the incidence of severe PPC, defined as newly initiated noninvasive ventilation (NIV) or reintubation for respiratory failure, has not yet been demonstrated.11,20,21

In 2016, our tertiary academic medical center changed the standard NMB reversal agent from neostigmine to sugammadex. Using an interrupted time series (ITS) study design, we examined the effects of such a system-wide change on the incidence of severe PPCs after procedures requiring general anesthesia. We hypothesized that the transition from neostigmine to sugammadex would be associated with an immediate and persistent reduction in the odds of the composite adverse outcome: reintubation for respiratory failure or initiation of new NIV.

METHODS

Study Design

The Colorado Multiple Institutional Review Board approved this study and waived the requirement for informed consent. The manuscript was written according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.22 An ITS design was chosen and groups were determined according to the date of surgery: August 15, 2015 to May 10, 2016 (presugammadex), and August 15, 2016 to May 11, 2017 (postsugammadex). The period from May 11, 2016 to August 14, 2016 marked the institutional transition (washout/wash-in) from neostigmine to sugammadex.

Sample

Figure 1.
Figure 1.:
Study flow diagram. ASA physical status: I, normal healthy patient; II, patient with mild systemic disease; III, patient with severe systemic disease; IV patient with severe systemic disease that is a constant threat to life; V, moribund patient who is not expected to survive without the operation; and VI, cadaveric organ donor. ASA indicates American Society of Anesthesiologists.

Adult patients undergoing a procedure with general anesthesia at the University of Colorado Anschutz Medical Center that included administration of neostigmine or sugammadex and hospital admission ≥1 night were eligible. Exclusion criteria included patients not receiving any reversal agent or receiving both reversal agents; patients not receiving rocuronium or vecuronium; patients with missing American Society of Anesthesiologists (ASA) physical status, ASA physical status V (patients not expected to survive the procedure), or ASA VI (declared brain-dead organ donors); and patients with >1 procedure requiring neostigmine or sugammadex during the time period (Figure 1).

Data Extraction and Outcomes

The following data were collected immediately before surgery: age, weight, gender, race, ethnicity, ASA physical status, and formal diagnosis of obstructive sleep apnea (OSA) or sleep apnea documented in the standardized anesthesia preoperative assessment. Intraoperative data included amounts of opioids administered normalized as oral morphine equivalents, surgical subspecialty, epidural anesthesia, duration of surgery, final train of 4 counts or sustained tetanus, and administration of rocuronium, vecuronium, neostigmine, and sugammadex as well as the dose of neostigmine and sugammadex.

The primary outcome was a composite of either reintubation for respiratory failure or the need for new NIV. Secondary outcomes included postoperative intensive care unit (ICU) admission, postoperative hypoxic events, and in-hospital mortality. ICU admission was defined as postoperative admission to the ICU at any time during hospitalization. Hypoxia was defined as any duration of verified peripheral blood oxygen saturation by pulse oximetry (Spo2) <90% in the medical record following transfer out of the postanesthesia care unit (PACU) during hospitalization.

To screen patients who were likely to have been reintubated after the procedure, the electronic medical record (EMR) was first searched for patients who had an order for mechanical ventilation placed after leaving the operating room or an intubation order placed following an order to extubate. Individual chart review of all patients identified in this fashion was performed to confirm that patients had, in fact, been extubated in the operating room, and reintubation had occurred for respiratory failure postoperatively. NIV was identified as recorded by nursing staff or respiratory therapists in the PACU, floor, or ICU EMR flow sheets. In our hospital, NIV is administered using the following modes: continuous positive airway pressure, bilevel positive airway pressure, average volume-assured pressure support, or pressure-control ventilation. Only patients not identified on a standardized preoperative pulmonary comorbidity questionnaire as using NIV preoperatively (eg, for OSA) were classified with the outcome new NIV.

Statistical Analysis

Statistical analyses were conducted using SAS software version 9.4 software (SAS Institute, Cary, NC). Event proportions were parsed into 27 ten-day intervals in each cohort and trend lines were fitted using graphical software for visual comparison. Kolmogorov-Smirnov tests were conducted within groups for all continuous variables, and median values and interquartile ranges were reported for distributions that departed appreciably from normality. Variables were compared presugammadex and postsugammadex with χ2 and Mann-Whitney U tests as appropriate.

Proportions of each primary and secondary outcome were plotted over time, together with lines fit to the observed data and predicted lines from logistic regressions incorporating significant group, time, and the group by time interaction effects (but no potential confounders) to visually inform the modeling process. Segmented logistic regression models appropriate for an ITS design were then utilized to evaluate the impact of the implementation of sugammadex.23,24 All models included the same set of 11 prespecified individual-level demographic and clinical variables to control for potential confounding: age, gender, ASA physical status, weight, duration of surgery, intraoperative oral morphine equivalents, race, ethnicity, OSA, surgical subspecialty, and epidural anesthesia. “Success” of the intervention was defined as a reduction in the mean (proportion) of the composite respiratory outcome, in a model with no evidence of nonzero slope in either period.

The following parameters were estimated from the segmented logistic regression corresponding to the ITS design (these parameter estimates can be exponentiated to provide estimates of the corresponding odds ratios [OR]):

β0: logit of outcome at time 0 for the preintervention period (intercept).

β1: preintervention slope of outcome (ie, the time effect before sugammadex introduction).

β2: log-OR of outcome at the beginning of the intervention period compared to end of the preintervention period (ie, the immediate effect of sugammadex).

β3: difference between preintervention and postintervention slopes (ie, change in slope of outcome over time after introducing sugammadex).

Two regression models are presented: a “full” and a “parsimonious” model. Per Mascha and Sessler,25 we fit a full model that included all of the β estimates above, regardless of their statistical significance. Our a priori statistical analysis plan included a parsimonious model for which all parameters were initially estimated, and then nonsignificant β values were sequentially removed.24 An α level of .05 for 2-sided tests was utilized to identify increases or decreases in outcomes postsugammadex transition.

Before the study, a simple power analysis evaluating power for comparing unadjusted proportions preintroduction and postintroduction of sugammadex was conducted instead of a power analysis that captured the more complex ITS design. The baseline incidence of the primary composite outcome was estimated at 5.5%.26 Based on the previously reported reduction in the risk for signs of postoperative residual paralysis, we conservatively estimated the incidence of the primary respiratory outcome after the introduction of sugammadex at 3.85%.11 Following preliminary data extraction for the 2 time periods, we estimated that a sample size of 7000 patients would yield 90% power with a 2-sided α of .05 (G*Power 3.1.9.2).27

RESULTS

Of 13,031 screened patients undergoing surgery during the specified time intervals, 7316 patients were included, with 3420 patients in the presugammadex group and 3896 patients in the postsugammadex group. Baseline characteristics and perioperative variables are summarized in Table 1. Sugammadex (P < .001) and rocuronium use (P = .003), sugammadex dosing (P < .001), and ASA physical status (P = .013) were higher in the postsugammadex group. Intraoperative oral morphine equivalents (P < .001), supplementary epidural anesthesia (P = .001), neostigmine (P < .001), vecuronium use (P < .001), and neostigmine dosing (P < .001) were higher in the presugammadex group. Distributions for race (P < .001) and surgical subspecialty (P < .001) were statistically different between the presugammadex and postsugammadex groups. There were no other significant differences in baseline characteristics and perioperative variables.

Table 1. - Patient and Demographic Characteristics
Characteristic Presugammadex Group (N = 3420) Postsugammadex Group (N = 3896) P
Demographic characteristics
 Age (y) .119
  Median (q1, q3) 56 (43, 67) 57 (43, 68)
 Weight (kg) .159
  Median (q1, q3) 77.8 (65.8, 93.3) 78.5 (65.8, 94.8)
 Women, no. (%) 1905 (55.7) 2197 (56.4) .555
 Race, no. (%) <.001
  Caucasian 2373 (69.4) 2446 (62.8)
  African American 239 (7.0) 284 (7.3)
  Asian 55 (1.6) 84 (2.2)
  AI, AN, NH, OPI 21 (0.6) 35 (0.9)
  Other or >1 352 (10.3) 363 (9.3)
  Not provided 380 (11.1) 684 (17.6)
 Hispanic ethnicity, no. (%) 446 (13.0) 481 (12.3) .379
 ASA physical status .013
  Median (q1, q3) 3 (2, 3) 3 (2, 3)
 OSA, no. (%) 733 (21.4) 844 (22.7) .204
Perioperative characteristics
 Intraoperative OME (mg) <.001
  Median (q1, q3) 76.0 (55.0, 105.0) 75.0 (52.6, 103.0)
 Duration of surgery (min) .899
  Median (q1, q3) 199 (146, 274) 199 (145, 279)
 Surgical subspecialty, no. (%) <.001
  General surgery 999 (29.2) 1001 (25.7)
  Neurosurgery 449 (13.1) 511 (13.1)
  Orthopedic surgery 510 (14.9) 634 (16.3)
  Cardiothoracic surgery 331 (9.7) 325 (8.3)
  Urology 294 (8.6) 329 (8.4)
  IR/GI/cardiology/others 179 (5.2) 298 (7.6)
  ENT/ophthalmology/OMF/plastic/dental 186 (5.4) 234 (6.0)
  Vascular/transplant surgery 177 (5.2) 231 (5.9)
  Obstetrics/gynecology 295 (8.6) 333 (8.5)
 Epidural, no. (%) 433 (12.7) 400 (10.3) .001
 Rocuronium, no. (%) 3393 (99.2) 3885 (99.7) .003
 Vecuronium, no. (%) 145 (4.2) 83 (2.1) <.001
 Sugammadex, no. (%) 1 (0) 3873 (99.4) <.001
 Sugammadex, dosing (mg) <.001
  Median (q1, q3) 100 (75, 160)
 Neostigmine, no. (%) 3419 (100.0) 23 (0.6) <.001
 Neostigmine, dosing (mg) <.001
  Median (q1, q3) 3.0 (2.0, 3.5) 3.0 (2.0, 4.0)
 Final TOF count/ST, no. (%) .225
  0/4 19 (0.6) 56 (1.4)
  1/4 48 (1.4) 105 (2.7)
  2/4 51 (1.5) 95 (2.4)
  3/4 50 (1.5) 75 (1.9)
  4/4 850 (24.9) 1194 (30.6)
  Sustained tetanus 2133 (62.4) 2241 (57.5)
  No peripheral nerve stimulation recorded 269 (7.9) 130 (3.3)
The Kolmogorov-Smirnov test was significant (P < .001) within groups of all continuous variables; Mann-Whitney U test was used. The P value for the TOF count/ST signifies χ2 test results for 4/4 or ST vs 1/4, 2/4, 3/4, or not recorded.
Abbreviations: AI, American Indian; AN, Alaska Native; ASA, American Society of Anesthesiologists (physical status ranks from I to VI: a status of I presents a healthy person; IV presents a person with severe systemic disease that is a constant threat to life); ENT, ear, nose and throat surgery; GI, gastroenterology; IR, interventional radiology; NH, Native Hawaiian; OME, oral morphine equivalent; OMF, oral and maxillofacial surgery; OPI, other Pacific Islander; OSA, obstructive sleep apnea (formal diagnosis or sleep apnea documented in standardized anesthesia preoperative assessment); ST, sustained tetanus; TOF, train of 4 count; q1, first quartile; q3, third quartile.

Calculation of unadjusted proportions indicated that the primary outcome of reintubation for new respiratory failure or NIV occurred in 6.1% of the presugammadex group and 4.2% of the postsugammadex group (Table 2, left panel). This difference was driven by more NIV in the presugammadex (5.9%) compared to postsugammadex groups (3.9%).

Table 2. - Primary and Secondary Outcomes by Group: Unadjusted Rates and Model-Based Estimates From F and P Models
Outcome Unadjusted Rates Model Estimates From Model Adjusted for 11 Patient Characteristics
Presugammadex Group Postsugammadex Group Parameters Odds Ratio
Exp (β)
95% Confidence Limits P
Reintubation for RF or new NIV, no. (%) 209 (6.1) 164 (4.2) F β 0 <0.001 ≤0.001 <.001
β 1 1.001 0.982–1.019 .939
β 2 0.795 0.523–1.208 .893
β 3 0.986 0.959–1.013 .312
P β 0 <0.001 ≤0.001 <.001
β 2 0.667 0.536–0.830 <.001
ICU admission, no. (%) 1071 (31.3) 1163 (29.9) F β 0 0.003 0.002–0.005 <.001
β 1 1.002 0.991–1.014 .683
β 2 0.894 0.702–1.139 .365
β 3 0.998 0.982–1.014 .067
P β 0 0.003 0.002–0.006 <.001
β 2 0.925 0.818–1.047 .220
Spo 2 <90%, no. (%) 2014 (58.9) 2176 (55.9) F β 0 0.032 0.021–0.048 <.001
β 1 1.011 1.001–1.020 .032
β 2 0.717 0.584–0.880 .002
β 3 0.995 0.982–1.009 .501
P β 0 0.033 0.022–0.049 <.001
β 1 1.008 1.002–1.015 .015
β 2 0.718 0.585–0.882 .002
In-hospital mortality, no. (%) 28 (0.82) 18 (0.46) F β 0 <0.001 ≤0.001 <.001
β 1 1.018 0.969–1.071 .478
β 2 0.759 0.245–2.350 .633
β 3 0.946 0.875–1.023 .162
P β 0 <0.001 ≤0.001 <.001
β 2 0.568 0.307–1.050 .071
F: full model including all parameter estimates (β0, β1, β2, β3); P: parsimonious model including only significant parameter estimates; β0: logit of outcome at time 0 for preintervention (intercept); β1: preintervention slope is the change in odds of outcome (ie, odds ratio) per 10-d interval; β2: change in odds of outcome at the start of postintervention compared to end of preintervention (change in intercept, immediate effect); β3: difference between periods in slopes of outcome over time (postintervention – preintervention).
Abbreviations: ICU, intensive care unit; new NIV, initiation of noninvasive ventilation in patients without a history of noninvasive ventilation at home; RF, respiratory failure; Spo2, peripheral blood oxygen saturation by pulse oximetry.

Figure 2 illustrates the observed proportions of the composite respiratory outcome and the secondary outcome hypoxic events before and after the institutional transition from neostigmine to sugammadex. Predicted lines fit from full (dashed line) and parsimonious (solid line) logistic regression models incorporating significant group, time, and the group by time interaction effects (but no potential confounders) are depicted.

Figure 2.
Figure 2.:
Outcomes. Observed proportions before and after the institutional transition from neostigmine to sugammadex, together with predicted lines fit from full and parsimonious logistic regression models for 4 outcome variables: composite respiratory outcome (reintubation for respiratory failure or newly initiated noninvasive ventilation) (A); and hypoxic events defined as Spo 2 <90% after transfer out of the postanesthesia care unit at any time during hospitalization (B). The red dotted line indicates the 3-month institutional transition period (washout/wash-in) from neostigmine to sugammadex. The y-axis shows the proportions corresponding to values on a logit scale. Spo 2 indicates peripheral blood oxygen saturation by pulse oximetry.

Table 2 (right panel) contains adjusted OR, 95% confidence limits, and P values from the full and selected parsimonious models for each primary and secondary outcome, incorporating 11 covariates. Neither for β2, the immediate effect of the introduction of sugammadex on the composite respiratory outcome (P = .893, OR = 0.795 [95% CI, 0.523–1.208]), nor for β3, the change in slope of the composite respiratory outcome over time after introducing sugammadex (P = .312, OR = 0.986 [95% CI, 0.959–1.013]), was statistical significance detected in the full model. Although nonsignificant, the slope in the intervention period is trending negative, numerically (β3) and visually (Figure 2A).

The estimate of β2 was significant in the parsimonious model (P < .001, OR = 0.667 [95% CI, 0.536–0.830]) after the nonsignificant estimates (β1, β3) were removed. The slope in the intervention period, β1, was not found to differ from 0 (P = .939, OR = 1.001 [95% CI, 0.982–1.019]).

For the composite respiratory outcome, we also report estimates for the covariates from the full and selected parsimonious models in Table 3. Six covariates were similarly significantly related to the composite respiratory outcome in both models: ASA physical status (P < .001), weight (P = .002), age (P = .005), men versus women (P = .023), duration of surgery (P < .001), and cardiothoracic surgery (P < .001).

Table 3. - Multiple Logistic Regression of Primary Outcome by Group Adjusting for Other Characteristics
Characteristic Model Odds Ratio (95% Confidence Limits) P
β 0 F <0.001 (≤0.001) <.001
P <0.001 (≤0.001) <.001
β 1 F 1.001 (0.982–1.019) .939
P n/a n/a
β 2 F 0.795 (0.523–1.208) .283
P 0.667 (0.536–0.830) <.001
β 3 F 0.986 (0.959–1.013) .3119
P n/a n/a
Age F 1.011 (1.003–1.019) .005
P 1.011 (1.003–1.019) .005
Weight F 1.008 (1.003–1.013) .002
P 1.008 (1.003–1.013) .002
Men versus women F 1.309 (1.039–1.650) .023
P 1.309 (1.039–1.650) .023
ASA physical status F 2.731 (2.225–3.315) <.001
P 2.726 (2.246–3.309) <.001
Hispanic F 1.069 (0.668–1.679) .808
P 1.066 (0.672–1.691) .786
OSA F 1.090 (0.842–1.412) .512
P 1.094 (0.845–1.417) .494
Intraoperative oral morphine equivalents F 0.997 (0.995–1.000) .051
P 0.997 (0.995–1.000) .051
Racea
Not provided F 0.826 (0.583–1.170) .282
P 0.826 (0.583–1.169) .280
Other or >1 race F 1.312 (0.785–2.194) .301
P 1.304 (0.780–2.181) .311
African American F 1.082 (0.518–2.262) .833
P 1.099 (0.526–2.294) .802
Asian, AI, AN, NH, OPI F 1.383 (0.934–2.049) .106
P 1.386 (0.936–2.053) .103
Duration of surgery (min) F 1.002 (1.001–1.003) <.001
P 1.002 (1.001–1.003) <.001
Surgical subspecialtyb
Cardiothoracic surgery F 3.274 (1.700–6.306) <.001
P 3.288 (1.708–6.330) <.001
Orthopedic surgery F 1.486 (0.762–2.900) .245
P 1.495 (0.766–2.917) .238
General surgery F 1.510 (0.803–2.837) .201
P 1.518 (0.808–2.852) .194
Interventional radiology, GI, cardiology, others F 1.767 (0.875–3.568) .113
P 1.771 (0.877–3.575) .111
Vascular surgery, transplant surgery F 1.607 (0.765–3.376) .210
P 1.602 (0.763–3.364) .213
ENT, ophthalmology, plastic, OMF, dental F 2.049 (0.980–4.281) .057
P 2.061 (0.987–4.305) .054
Urology F 1.210 (0.581–2.517) .611
P 1.212 (0.583–2.522) .607
Neurosurgery F 1.337 (0.676–2.641) .404
P 1.342 (0.680–2.652) .396
Epidural F 1.079 (0.766–1.520) .664
P 1.079 (0.776–1.519) .664
Abbreviations: AI, American Indian; AN, Alaska Native; ASA, American Society of Anesthesiologists (physical status ranks from I to VI: a status of I presents a healthy person; IV presents a person with severe systemic disease that is a constant threat to life); ENT, ear, nose and throat surgery; F, full model; GI, gastroenterology; IR, interventional radiology; NH, Native Hawaiian; OB/Gyn, obstetrics/gynecology; OME, oral morphine equivalent; OMF, oral and maxillofacial surgery; OPI, other Pacific Islander; OSA, obstructive sleep apnea (formal diagnosis or sleep apnea documented in standardized anesthesia preoperative assessment; P, parsimonious model.
aRace was compared to Caucasian.
bSurgical subspecialties were compared to OB/Gyn.

Results from full and parsimonious regression models for secondary outcomes presugammadex and postsugammadex introduction are summarized in Table 2. The odds of ICU admission did not differ significantly between groups, regardless of whether the full or parsimonious model was used. The odds of a hypoxic event were less common postsugammadex compared to before in the full model (β2: P = .002, OR = 0.717 [95% CI, 0.584–0.880]) and the parsimonious model (β2: P = .002, OR = 0.718 [95% CI, 0.585–0.882], Figure 2B). The odds for in-hospital mortality were not significantly lower in the postsugammadex group in the full or the parsimonious model. Overall low mortality rates led to possible conversion issues. However, simplified models without the previously included 11 confounders produced results similar to those reported in Table 2.

DISCUSSION

Unadjusted results revealed a 31% reduction in the incidence of composite respiratory outcome with a 34% reduction in newly initiated NIV in patients undergoing general anesthesia after a system-wide change to sugammadex from neostigmine. The reduction in intercept/level from presugammadex to postsugammadex was not significant in a full logistic regression model that includes all parameter estimates, regardless of statistical significance. However, our observation is supported by nonsignificant within-group time trends and a significant reduction in intercept/level from presugammadex to postsugammadex in a parsimonious logistic regression model adjusting for covariates. In addition, regardless of the full or parsimonious modeling approach, reversal with sugammadex was associated with an immediate reduction in the odds for hypoxic events.

While previously published literature has reported a decrease in signs of residual or recurrent muscle paralysis, it did not demonstrate a reduction in reintubation or NIV: a meta-analysis from 2015 showed no difference in critical respiratory events, which included among others hypoxemic events defined as Spo2 <90% despite oxygen therapy, reintubation, or NIV (relative risk of 0.13 [95% CI, 0.02–1.06], P = .06).20 A 2016 meta-analysis reported a significant difference in respiratory adverse events (OR = 0.36 [95% CI, 0.14–0.95], P = .04) but did not separate between mild or severe PPCs.21 These findings were mostly driven by 1 prospective study which found a decreased proportion of hypoxia.28 In 2017, a Cochrane analysis found no significant differences in overall serious adverse events including respiratory depression and respiratory failure (relative risk of 0.54 [95% CI, 0.13–2.25], P = .71).11 Notably, the largest study reporting on composite serious adverse events incorporated in the Cochrane analysis included 154 patients.11,29 In our study, only 4.9% of all patients required newly initiated NIV, even fewer patients required reintubation for respiratory failure (0.8%). Hence, other studies comparing neostigmine with sugammadex were likely insufficiently powered to detect a significant difference in the incidence of PPCs.

Our study included 7316 patients and, to our knowledge, is the first of its size to evaluate the associations between NIV or reintubation and a system-wide change from neostigmine to sugammadex. The switch after the washout wash-in period was near-complete: only 1 patient in the presugammadex cohort received sugammadex, and only 23 patients in the postsugammadex group (0.6%) received neostigmine. A systematic transition to sugammadex is therefore feasible and could likely be reproduced elsewhere. The decreased odds of the composite respiratory outcome identified by the parsimonious model can likely be explained by the faster onset and longer-lasting reversal of NMB with sugammadex compared to neostigmine.8 A decreased rate of residual or recurrent paralysis in the PACU is associated with improved airway protection and could have translated into a lower incidence of PPCs.2,3,30 Considering the long-term consequences of PPCs including 46,200 related deaths and 4.8 million additional hospitalization days in the United States alone, our results are relevant to patient safety and public health.31

Although the odds for the secondary outcome postoperative hypoxic events were lower postsugammadex, the odds for ICU admission were nonsignificantly different and, no significant change in the in-hospital mortality odds was detected. The reduction of the primary outcome was driven by a decline in the necessity for NIV, not reintubation for respiratory failure. Given the incidence of 0.8% for reintubation, our study was not powered to detect a difference for such an infrequent outcome. Patient and procedural characteristics which were associated with increased likelihood of the composite respiratory outcome in our study included increasing ASA physical status, weight, age, duration of surgery, and cardiothoracic surgery. Indeed, these findings are consistent with previously published study.26,32–34 Other known predictors for reintubation, which were tested in our study but did not show an association with the composite respiratory outcome, included vascular, abdominal, neurological, and transplant surgeries.34

Our study has several limitations: first, a before-after study is not designed as a randomized controlled trial and unaccounted confounding remains possible.35 Investigating relatively uncommon outcomes requires large sample sizes and performing a comparably sized randomized controlled trial would be extremely resource intensive. To mitigate for seasonality and other time-varying confounders, the time periods for both groups started in August and ended in May.23,24 We attempted to further address this limitation with an ITS design, which reflects the next most robust research design to study longitudinal effects of nonrandomized interventions.23 However, approaches for analyzing data from an ITS design vary, and in this case, these different approaches led to different conclusions regarding the composite respiratory outcome. Because we hypothesized that the overall odds for the composite respiratory outcome would decrease with sugammadex administration (Figure 2A), we followed recommendations to exclude the nonsignificant estimates for slope and interaction terms.24 However, others contend that groups usually do not depict outcomes parallel over time and other (unaccounted) factors do typically change over time. Hence, even a slight interaction, regardless of statistical significance, should not be ignored, but rather considered in the inference.25 Therefore, we chose to present both our originally planned parsimonious model and a full logistic regression model (including the β1 and β3 estimates) to permit a complete interpretation of the data. Interestingly, although the β2 estimate changes from nonsignificant to highly significant when one compares the results of the full to the parsimonious model, all other covariates remain essentially unchanged (Table 3). We would argue that the estimated (and very nonsignificant) difference in slopes is “overpowered” by a much larger and clear difference in means between the periods (Figure 2A) and hence would favor the parsimonious model in this case. Regardless, part of the difference in estimates between the 2 models is that the standard error was cut in half for the parsimonious model, in addition to the increase in the log-OR.

Second, although group assignment was not randomized in this ITS design but determined by the date of surgery, demographic and perioperative characteristics were evenly distributed, including weight and gender, which are known risk factors for residual paralysis.36 The only relevant exceptions included the ASA physical status (P = .013), which was higher in the postsugammadex group, supplementary epidural anesthesia (P = .001), which was more frequent in the presugammadex group and the intraoperative dose of oral morphine equivalents (P < .001), which was higher in the presugammadex group. We speculate that the lower opioid use in the postsugammadex group could have been due to institutional and nation-wide effort to lower the use of opioids (eg, in patients with OSA).37 The different composition of performed procedures may also explain why patients in the presugammadex group received a higher dose of oral morphine equivalents: the presugammadex group had a higher rate of procedures performed by general surgery (29.2% vs 25.7%) and cardiothoracic surgery (9.7% vs 8.3%), but a lower rate of less invasive procedures performed by interventional radiology, gastroenterology, cardiology, and other minor procedures (5.2% vs 7.6%). Hence, the procedure performed and the amounts of opioids administered were included in the logistic regression model. Some of the aforementioned differences in demographic and perioperative characteristics are small and therefore of clinically questionable relevance. However, a higher frequency of the composite respiratory outcome would be expected in patients with higher ASA physical status, patients not receiving epidural anesthesia, and patients receiving more opioids. Indeed, in the multivariable logistic regression analysis, a higher ASA physical status was associated with higher odds of the composite respiratory outcome, which correlates with previous findings showing an association between ASA physical status and PPCs.38 The total dose of intraoperative opioids or supplementation of epidural anesthesia, on the other hand, had no effect on the composite respiratory outcome (P = .051 and P = .664, respectively). Finally, it should be mentioned that other demographic or intraoperative predictors of postoperative respiratory complications such as the rate of congestive heart failure, other chronic pulmonary disease, and emergency procedures were not included in our regression models, and we can therefore not guarantee that all relevant confounders were captured.34,39

Third, our composite primary outcome was limited to reintubation for respiratory failure and new NIV. Yet, mortality is increased also with other adverse PPCs such as acute respiratory distress syndrome (50.0%), pneumonia (9.1%), or pleural effusions (7.0%).26 Given our sample size and data collection approach using EMR sources, reliable detection of these outcomes was not feasible. The high postoperative mortality associated with newly initiated NIV and reintubation provided the rationale for choosing our composite outcome.26

Finally, NIV is not always prescribed for respiratory failure. Although we defined new NIV exclusive of patients with known preoperative NIV use, we cannot exclude that NIV was initiated for newly suspected OSA.40 However, the incidence of patients with a diagnosis of OSA or at high risk for OSA which were identified during the preoperative anesthesia assessment showed no significant differences between groups.

CONCLUSIONS

This ITS study of 7316 patients undergoing general anesthesia with pharmacologic reversal of NMB reveals that the institutional transition from neostigmine to sugammadex was associated with a decrease in the odds of a composite primary outcome of either new NIV or reintubation for respiratory failure. A system-wide transition from neostigmine to sugammadex as the preferred reversal agent for rocuronium-induced and vecuronium-induced NMB may lead to a reduction of adverse respiratory outcomes.

DISCLOSURES

Name: Martin Krause, MD.

Contribution: This author helped with data acquisition for the study, wrote the manuscript, revised it critically for important intellectual content, and approved the final version to be published.

Name: Shannon K. McWilliams, MA.

Contribution: This author helped with data analysis for the study, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Kenneth J. Bullard, BS.

Contribution: This author helped with data acquisition for the study, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Lena M. Mayes, MD.

Contribution: This author helped with study design and data interpretation, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Leslie C. Jameson, MD.

Contribution: This author helped with study design and data interpretation, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Susan K. Mikulich-Gilbertson, PhD.

Contribution: This author helped with study design and analysis and data interpretation, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Ana Fernandez-Bustamante, MD, PhD.

Contribution: This author helped with study design and data interpretation, revised the study critically for important intellectual content, and approved the final version to be published.

Name: Karsten Bartels, MD, PhD.

Contribution: This author helped with study design, data analysis, and interpretation, obtained study funding, revised the study critically for important intellectual content, and approved the final version to be published.

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

FOOTNOTES

GLOSSARY

ASA = = American Society of Anesthesiologists;

CI = = confidence interval;

EMR = = electronic medical record;

ICU = = intensive care unit;

ITS = = interrupted time series;

NIH = = National Institutes of Health;

NMB = = neuromuscular blockade;

NIV = = noninvasive ventilation;

OR = = odds ratio;

OSA = = obstructive sleep apnea;

PACU = = postanesthesia care unit;

PPC = = postoperative pulmonary complication;

RF = = respiratory failure

Spo2 = = peripheral blood oxygen saturation by pulse oximetry

STROBE = = Strengthening the Reporting of Observational Studies in Epidemiology

REFERENCES

1. Khuri SF, Henderson WG, DePalma RG, Mosca C, Healey NA, Kumbhani DJ; Participants in the VA National Surgical Quality Improvement Program. Determinants of long-term survival after major surgery and the adverse effect of postoperative complications. Ann Surg. 2005;242:326–341.
2. Cedborg AI, Sundman E, Bodén K, et al. Pharyngeal function and breathing pattern during partial neuromuscular block in the elderly: effects on airway protection. Anesthesiology. 2014;120:312–325.
3. Plaud B, Debaene B, Donati F, Marty J. Residual paralysis after emergence from anesthesia. Anesthesiology. 2010;112:1013–1022.
4. Vidal Melo MF, Musch G, Kaczka DW. Pulmonary pathophysiology and lung mechanics in anesthesiology: a case-based overview. Anesthesiol Clin. 2012;30:759–784.
5. Mirzakhani H, Williams JN, Mello J, et al. Muscle weakness predicts pharyngeal dysfunction and symptomatic aspiration in long-term ventilated patients. Anesthesiology. 2013;119:389–397.
6. McLean DJ, Diaz-Gil D, Farhan HN, Ladha KS, Kurth T, Eikermann M. Dose-dependent association between intermediate-acting neuromuscular-blocking agents and postoperative respiratory complications. Anesthesiology. 2015;122:1201–1213.
7. Karcz M, Papadakos PJ. Respiratory complications in the postanesthesia care unit: a review of pathophysiological mechanisms. Can J Respir Ther. 2013;49:21–29.
8. Srivastava A, Hunter JM. Reversal of neuromuscular block. Br J Anaesth. 2009;103:115–129.
9. Paech MJ, Kaye R, Baber C, Nathan EA. Recovery characteristics of patients receiving either sugammadex or neostigmine and glycopyrrolate for reversal of neuromuscular block: a randomised controlled trial. Anaesthesia. 2018;73:340–347.
10. Khuenl-Brady KS, Wattwil M, Vanacker BF, Lora-Tamayo JI, Rietbergen H, Alvarez-Gómez JA. Sugammadex provides faster reversal of vecuronium-induced neuromuscular blockade compared with neostigmine: a multicenter, randomized, controlled trial. Anesth Analg. 2010;110:64–73.
11. Hristovska AM, Duch P, Allingstrup M, Afshari A. Efficacy and safety of sugammadex versus neostigmine in reversing neuromuscular blockade in adults. Cochrane Database Syst Rev. 2017;8:CD012763.
12. Stair C, Fernandez-Bustamante A. Sugammadex, the first selective relaxant binding agent for neuromuscular block reversal. Drugs Today (Barc). 2012;48:405–413.
13. Kopman AF. Sugammadex: a revolutionary approach to neuromuscular antagonism. Anesthesiology. 2006;104:631–633.
14. Germano Filho PA, Cavalcanti IL, Barrucand L, Verçosa N. Effect of magnesium sulphate on sugammadex reversal time for neuromuscular blockade: a randomised controlled study. Anaesthesia. 2015;70:956–961.
15. Schaller SJ, Fink H, Ulm K, Blobner M. Sugammadex and neostigmine dose-finding study for reversal of shallow residual neuromuscular block. Anesthesiology. 2010;113:1054–1060.
16. Czarnetzki C, Tassonyi E, Lysakowski C, Elia N, Tramèr MR. Efficacy of sugammadex for the reversal of moderate and deep rocuronium-induced neuromuscular block in patients pretreated with intravenous magnesium: a randomized controlled trial. Anesthesiology. 2014;121:59–67.
17. Lee C, Jahr JS, Candiotti KA, Warriner B, Zornow MH, Naguib M. Reversal of profound neuromuscular block by sugammadex administered three minutes after rocuronium: a comparison with spontaneous recovery from succinylcholine. Anesthesiology. 2009;110:1020–1025.
18. Geldner G, Niskanen M, Laurila P, et al. A randomised controlled trial comparing sugammadex and neostigmine at different depths of neuromuscular blockade in patients undergoing laparoscopic surgery. Anaesthesia. 2012;67:991–998.
19. Illman HL, Laurila P, Antila H, Meretoja OA, Alahuhta S, Olkkola KT. The duration of residual neuromuscular block after administration of neostigmine or sugammadex at two visible twitches during train-of-four monitoring. Anesth Analg. 2011;112:63–68.
20. Abad-Gurumeta A, Ripollés-Melchor J, Casans-Francés R, et al.; Evidence Anaesthesia Review Group. A systematic review of sugammadex vs neostigmine for reversal of neuromuscular blockade. Anaesthesia. 2015;70:1441–1452.
21. Carron M, Zarantonello F, Tellaroli P, Ori C. Efficacy and safety of sugammadex compared to neostigmine for reversal of neuromuscular blockade: a meta-analysis of randomized controlled trials. J Clin Anesth. 2016;35:1–12.
22. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335:806–808.
23. Vetter TR. Magic mirror, on the wall-which is the Right Study design of them all? Part I. Anesth Analg. 2017;124:2068–2073.
24. Bernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2017;46:348–355.
25. Mascha EJ, Sessler DI. Segmented regression and difference-in-difference methods: assessing the impact of systemic changes in health care. Anesth Analg. 2019;129:618–633.
26. Fernandez-Bustamante A, Frendl G, Sprung J, et al. Postoperative pulmonary complications, early mortality, and hospital stay following noncardiothoracic surgery: a Multicenter Study by the perioperative research network investigators. JAMA Surg. 2017;152:157–166.
27. Faul F, Erdfelder E, Buchner A, Lang AG. Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav Res Methods. 2009;41:1149–1160.
28. Carron M, Veronese S, Foletto M, Ori C. Sugammadex allows fast-track bariatric surgery. Obes Surg. 2013;23:1558–1563.
29. Brueckmann B, Sasaki N, Grobara P, et al. Effects of sugammadex on incidence of postoperative residual neuromuscular blockade: a randomized, controlled study. Br J Anaesth. 2015;115:743–751.
30. Kopman AF, Yee PS, Neuman GG. Relationship of the train-of-four fade ratio to clinical signs and symptoms of residual paralysis in awake volunteers. Anesthesiology. 1997;86:765–771.
31. Shander A, Fleisher LA, Barie PS, Bigatello LM, Sladen RN, Watson CB. Clinical and economic burden of postoperative pulmonary complications: patient safety summit on definition, risk-reducing interventions, and preventive strategies. Crit Care Med. 2011;39:2163–2172.
32. Weissman C. Pulmonary complications after cardiac surgery. Semin Cardiothorac Vasc Anesth. 2004;8:185–211.
33. Gupta PK, Gupta H, Kaushik M, et al. Predictors of pulmonary complications after bariatric surgery. Surg Obes Relat Dis. 2012;8:574–581.
34. Brueckmann B, Villa-Uribe JL, Bateman BT, et al. Development and validation of a score for prediction of postoperative respiratory complications. Anesthesiology. 2013;118:1276–1285.
35. Ho AMH, Phelan R, Mizubuti GB, et al. Bias in before-after studies: narrative overview for anesthesiologists. Anesth Analg. 2018;126:1755–1762.
36. Rudolph MI, Ng PY, Deng H, et al. Comparison of a novel clinical score to estimate the risk of REsidual neuromuscular block Prediction Score and the last train-of-four count documented in the electronic anaesthesia record: a retrospective cohort study of electronic data on file. Eur J Anaesthesiol. 2018;35:883–892.
37. Cozowicz C, Chung F, Doufas AG, Nagappa M, Memtsoudis SG. Opioids for acute pain management in patients with obstructive sleep apnea: a systematic review. Anesth Analg. 2018;127:988–1001.
38. Canet J, Gallart L, Gomar C, et al.; ARISCAT Group. Prediction of postoperative pulmonary complications in a population-based surgical cohort. Anesthesiology. 2010;113:1338–1350.
39. Rudolph MI, Chitilian HV, Ng PY, et al. Implementation of a new strategy to improve the peri-operative management of neuromuscular blockade and its effects on postoperative pulmonary complications. Anaesthesia. 2018;73:1067–1078.
40. Fernandez-Bustamante A, Bartels K, Clavijo C, et al. Preoperatively screened obstructive sleep apnea is associated with worse postoperative outcomes than previously diagnosed obstructive sleep apnea. Anesth Analg. 2017;125:593–602.
Copyright © 2020 International Anesthesia Research Society