Residual paralysis is common after general anesthesia involving administration of neuromuscular blocking drugs (NMBDs). Management of NMBDs and reversal is frequently guided by train-of-four (TOF) monitoring. We hypothesized that monitoring of eye muscles is associated with more frequent residual paralysis than monitoring at the adductor pollicis.
This prospective cohort study enrolled 180 patients scheduled for elective surgery with anticipated use of NMBDs. Collected variables included monitoring site, age, gender, weight, body mass index, American Society of Anesthesiologists physical status class, type and duration of surgery, type of NMBDs, last and total dose administered, TOF count at time of reversal, dose of neostigmine, and time interval between last dose of NMBDs to quantitative measurement. Upon postanesthesia care unit admission, we measured TOF ratios by acceleromyography at the adductor pollicis. Residual paralysis was defined as a TOF ratio less than 90%. Multivariable logistic regression was used to account for unbalances between the two groups and to adjust for covariates.
150 patients received NMBDs and were included in the analysis. Patients with intraoperative TOF monitoring of eye muscles had significantly greater incidence of residual paralysis than patients monitored at the adductor pollicis (P < 0.01). Residual paralysis was observed in 51/99 (52%) and 11/51 (22%) of patients, respectively. The crude odds ratio was 3.9 (95% CI: 1.8-8.4), and the adjusted odds ratio was 5.5 (95% CI: 2.1-14.5).
Patients having qualitative TOF monitoring of eye muscles had a greater than 5-fold higher risk of postoperative residual paralysis than those monitored at the adductor pollicis.
What We Already Know about This Topic
Assessing the degree of spontaneous recovery at the time of reversal from neuromuscular blockade is important, but has only been performed at the adductor pollicis
What This Article Tells Us That Is New
In 150 patients receiving neuromuscular blockers during surgery, qualitative assessment of train-of-four monitoring of eye muscles led to a 5-fold increased risk of postoperative residual paralysis compared with the adductor pollicis
RESIDUAL paralysis is frequently present after general anesthesia involving the administration of neuromuscular blocking drugs (NMBDs), and is associated with potential morbidity.1,2Residual paralysis is commonly defined as a train-of-four (TOF) ratio less than 90% at the adductor pollicis and has been reported to occur in 38–64% of cases in which conventional qualitative or subjective evaluation of the TOF monitoring was employed.3–6This type of assessment is the predominant mode of intraoperative monitoring in the United States.7The nerve stimulator may be applied at various anatomical sites, such as at the wrist with stimulation of the ulnar nerve and evaluation of the response of the adductor pollicis, or on the face with stimulation of the facial nerve and evaluation of the twitch response of the orbicularis oculi or corrugator supercilii muscles (henceforth referred to as eye muscles). Although residual paralysis is defined as TOF ratio less than 90% at the adductor pollicis, intraoperative monitoring of the eye muscles may be chosen when the thumb is not easily accessible because of surgical positioning. However, several previous studies have documented that the twitch response recovers more promptly at the level of the orbicularis oculi muscle compared with the adductor pollicis.8–16Published reports that provide information applicable to the guidance of reversal of NMBDs with anticholinesterase inhibitors have documented the importance of assessing the degree of spontaneous recovery at time of reversal and have reported the twitch response of the adductor pollicis only.17–23Similar studies using the twitch response of muscles surrounding the eye to guide reversal are lacking. Moreover, guidelines on good clinical research practice in neuromuscular research emphasize that alternative muscles cannot be used interchangeably with adductor pollicis responses.24We designed the present study to test the hypothesis that intraoperative monitoring at the eye muscles is associated with an increased risk of residual paralysis compared with monitoring at the adductor pollicis.
Materials and Methods
The study was approved by the University of Washington Institutional Review Board (Seattle, Washington), and written informed consent was obtained from all patients. This prospective, observational cohort study was conducted at two teaching hospitals, the University of Washington Medical Center and Harborview Medical Center. Between June 16, 2011, and Aug 20, 2011, we enrolled 180 patients with American Society of Anesthesiologists (ASA) physical status I–IV who were free from underlying neuromuscular disorders and scheduled to undergo elective surgery with anticipated use of nondepolarizing muscle relaxants, and who gave written informed consent. Exclusion criteria included age less than 18 yr or greater than 80 yr.
The primary providers of anesthesia were certified registered nurse anesthetists and residents with attending staff supervision. Because the study was not blinded, to minimize awareness of the study, anesthesia providers were not informed about the details and purpose of the study and whether a particular patient was actually enrolled in the study. Patients received usual anesthesia care including standard monitoring, such as electrocardiography, pulse oximetry, end-tidal carbon dioxide concentration, and noninvasive blood pressure monitoring. Each operating room is equipped with a conventional, qualitative, peripheral nerve stimulator for TOF monitoring (Digistim II Nerve Stimulator; Neuro Technology, CCR Medical, Inc., St. Petersburg, FL). These monitors are also capable of delivering 50 Hz and 100 Hz tetanic stimulation. Although no written policy exists, the use of this conventional nerve stimulator is part of routine practice at our hospitals. At the time of the study, quantitative monitors were not available for clinical practice in our hospitals.
The primary endpoint was a quantitative assessment of neuromxuscular block performed by acceleromyography (TOF-Watch SX®, Bluestar Enterprises, Omaha, NE) within 5 min of arrival to the postanesthesia care unit (PACU). Two standard electrocardiogram electrodes, placed 3 cm apart, were applied on alcohol-cleansed skin over the ulnar nerve at the wrist. The TOF-Watch Hand Adapter®was used to apply the acceleration transducer to the distal phalanx of the thumb and to apply the temperature probe to the palm. We used 50-mA TOF stimulation without calibration, i.e. , four pulses of 0.2-ms duration at a frequency of 2 Hz, and averaged two measurements that were separated by 20 s. If the initial two measurements were not within 10% agreement, then we performed additional measurements separated by 20-s intervals (up to four total measurements) and averaged the two closest TOF ratios. We defined residual paralysis as a TOF ratio less than 90%. We noted the time of the last valid measurement. We measured hand temperature at the thenar aspect of the palm.
For each patient, we collected information on multiple other variables with potential relevance to recovery from neuromuscular block. Demographic variables included age, gender, height, weight, and ASA physical status. Clinical variables obtained by way of electronic medical record review included type of surgery, duration of surgery, type of muscle relaxant and whether provided by bolus or continuous infusion, total amount of muscle relaxant, time and amount of last dose of muscle relaxant, time and amount of neostigmine, type of preoperative antibiotic, agent used for maintenance of general anesthesia, and core temperature. We converted all doses of NMBDs to multiples of ED95with the following conversion factors: 0.3 mg/kg rocuronium, 0.05 mg/kg vecuronium, and 0.05 mg/kg cisatracurium. Obesity needs to be considered when dosing NMBDs because rocuronium, vecuronium, and cisatracurium have been reported to have prolonged duration of action if dosing is based on total body weight.25–27We calculated body mass index (BMI = kg/m2) and three different weights for each patient. We categorized patients to three categories based on BMI, normal weight (BMI less than 25 kg/m2), overweight (BMI of 25 kg/m2or more but less than 30 kg/m2), and obese (BMI of 30 kg/m2or more). The calculated weights were: ideal body weight (IBW) was defined as 50.0 + 2.3 kg per inch over 5 feet for males and 45.5 + 2.3 kg per inch over 5 feet for females, adjusted body weight (ABW) 20 was defined as IBW + 20% of excess weight, and ABW40 was defined as IBW + 40% of excess weight. We calculated adjusted doses of NMBDs by using these calculated weights. When a patient received more than one type of muscle relaxant, we categorized this patient based on the last dose.
Additional clinical variables were obtained by way of interviewing the anesthesia provider in the PACU: site of intraoperative qualitative neuromuscular blockade monitoring, response of qualitative monitoring at time of reversal (TOF count and presence of fade when four twitches were present), and duration from reversal with neostigmine to extubation.
Normally distributed data are reported as a mean with SD and were compared using two-sample Student t test. Categorical data are reported as counts and frequency distributions and were compared using the chi-square test. The main study endpoint, residual paralysis, was compared between patients monitored at the eye muscles and those monitored at the adductor pollicis as the difference in proportions, using the chi-square statistic. We used multivariable logistic regression to account for potential unbalances between the two groups and adjust the association of monitoring site with residual paralysis for other explanatory variables. The covariates used include monitoring site (eye muscles or adductor pollicis), gender (male or female), hospital (University of Washington Medical Center or Harborview Medical Center), ASA class (I to II or III to IV, i.e. , two categories), BMI (less than 25, BMI of 25 or more but less than 30, and BMI 30 kg/m2or more), and type of surgical procedure (two categories, thoracic and abdominal or other procedure) as categorical variables. The remaining variables were used as continuous variables: age, last dose of NMBD (expressed as ED95) and neostigmine (mcg/kg), duration of surgery, and the time interval from last dose of NMBD to assessment in the PACU. For the primary analysis, last dose of NMBD was adjusted using ABW20, and a sensitivity analysis was performed using total body weight, IBW, and ABW40.
Covariate balance was investigated by performing bivariate analyses between the two levels of monitoring site and potential confounding variables. Absolute standardized differences were also calculated. We conducted a propensity score analysis (for monitoring site) as a sensitivity analysis to confirm the multivariable logistic regression model provided adequate adjustment for potential confounding.
Regarding sample size considerations, we assumed a frequency of 30% residual paralysis in the group conventionally monitored at adductor pollicis, based on a recent report by Murphy et al. 28We hypothesized that monitoring of eye muscles was associated with a higher frequency of residual paralysis and considered that a frequency of 50% in this group would be a clinically relevant difference. Assuming equal size groups, a two-sided α level of 0.05, and 80% power, the required sample would be 206 (103 per group). By the end of the projected study period, the study had enrolled 180 patients. Because of unavailability of supplemental funds, recruitment could not be continued, and an analysis was conducted. After accounting for exclusions, 150 patients were available for the analysis. Because the null hypothesis was rejected with a two-sided α level less than 0.01, the study was not reopened to enrolment.
All analyses were conducted using Stata 12 (Stata Corporation, College Station, TX).
All hypothesis tests and their associated P values are two-sided.
At the end of the study period, we had enrolled 180 patients who gave written, informed consent. Of those, 150 patients received intraoperative NMBDs, had intraoperative TOF monitoring of either eye muscles or adductor pollicis, and had postoperative quantitative measurements made. Reasons for exclusion included: measurements not obtained because of unavailability of research personnel (13), no use of NMBDs (six), no monitoring at one of the two sites (seven), and case cancellation (four). With the exception of hospital and type of surgical procedure, there were no significant differences between groups in gender, age, BMI, and ASA physical status (table 1). Likewise, the two groups did not differ significantly in perioperative variables, including surgery duration, TOF response at time of reversal (TOF count and fade), total dose of NMBD and last dose of NMBD (doses compared as multiples of ED95), dose of neostigmine (mcg/kg), core and hand temperatures, and type of muscle relaxants. There were also no significant differences between the two groups in the following measured time intervals: last dose of NMBD to TOF ratio measurement in the PACU, neostigmine to extubation, and extubation to PACU arrival. Rocuronium, vecuronium, and cisatracurium were administered to 81%, 11%, and 6% of patients, respectively (table 2). No patient received NMBD by continuous infusion. Maintenance of anesthesia was achieved with inhalational agents (sevoflurane, isoflurane, or desflurane) in 149 patients. Use of perioperative antibiotics was as follows: cefazolin (127), clindamycin (seven), vancomycin (three), fluoroquinolones (three), gentamicin (one), none (three) and unknown (six).
Patients who had intraoperative TOF monitoring of eye muscles had significantly greater incidence of residual paralysis than patients monitored at the adductor pollicis, 51/99 (52%) at the eye muscles and 11/51 (22%) at the adductor pollicis, P < 0.01. The distribution of TOF ratios in each study group is shown in figure 1. The mean ± SD TOF ratios were 0.86 ± 0.22 (median 0.90) for the patients monitored at eye muscles and 0.93 ± 0.18 (median 1.01) for patients monitored at the adductor pollicis. The ranges of values of TOF ratios were 0 to 1.15 and 0.40 to 1.16 for the two groups, respectively. TOF ratios more than 1.0 were present in 27% and 51% of patients monitored at the eye muscles and adductor pollicis, respectively. 17.2% of patients monitored at the eye muscles had TOF ratios less than 0.7, and 34.3% had TOF ratios in the range 0.7 to less than 0.9. Comparative results for patients monitored at the adductor pollicis were 11.8% and 9.8%, respectively. The time interval from neostigmine administration to TOF ratio measurement was not significantly different between the two groups (table 2).
The crude odds ratio (OR) for the association of site of monitoring and residual paralysis was 3.9 (95% CI: 1.8–8.4) comparing monitoring at the eye muscles with monitoring at the adductor pollicis. Multivariable logistic regression adjusting for age, gender, hospital, ASA class, time interval from last dose of NMBD to measurement in the PACU, last dose of NMBD (expressed as ED95), amount of neostigmine, duration of surgery, type of surgery, and BMI yielded an adjusted OR of 5.5 (95% CI: 2.1–14.5, table 3). The variables that were significantly associated with residual paralysis were monitoring site, time interval from last dose of NMBD to time of measurement in the PACU, and BMI. For each 10 additional minutes between last dose of NMBD and measurement, there was 10% lower odds of postoperative residual paralysis (OR: 0.90; 95% CI: 0.83 to 0.99, P = 0.03), irrespective of monitoring site. We performed a sensitivity analysis by using total body weight, IBW, and ABW40 for calculations of the dose of NMBD and found that this did not change the results compared with our primary analysis, which used dose calculations according to ABW20.
The results of the propensity score model were substantially unchanged from the final multivariable logistic regression (OR: 4.7, 95%: CI 1.8–12.4).
A total of 113 patients had multiple doses of NMBDs, and 37 patients had a single dose of NMBDs. We conducted a secondary subgroup analysis on these two subgroups. The crude OR for the association of site of monitoring and residual paralysis in the group that received multiple doses was 4.38 (95% CI: 1.82–10.50), whereas in the group having received a single dose the association was not statistically significant (OR: 2.81, 95% CI: 0.5–15.77).
All patients had a hand temperature of at least 32.3°C.
The SD for neostigmine dose was large; in table 2, the dose is listed according to total body weight. We also calculated the weight-based neostigmine doses according to the other weights (IBW, ABW20, and ABW40) and the standard deviations remained high with the different weight adjustments. In the final multivariable logistic regression model (table 3), neostigmine dose was calculated according to total body weight. Sixteen patients did not receive neostigmine; there was not a statistically significant association between not receiving neostigmine and residual paralysis.
Eighteen patients were not extubated before TOF ratio measurement in the PACU. All were ventilating spontaneously and were planned for extubation in the PACU. The majority (13/18 patients) had undergone orthopedic procedures and were kept anesthetized with a propofol infusion until a peripheral nerve block for postoperative analgesia had been performed. All were extubated in the PACU. For patients who were extubated in the operating room, the average time interval between extubation and arrival to the PACU was 6.5 min (SD: 4.7).
No serious complications, such as airway obstruction or need for assisted ventilation, were identified in either monitoring site group. Two patients had a TOF ratio of 0. Both of these patients showed signs of weakness, but they did not require intervention beyond close monitoring.
In this cohort of 150 patients who were administered NMBDs as part of anesthesia care for a variety of surgical procedures, patients who were monitored by facial nerve stimulation and evaluation of muscles surrounding the eye had a significantly higher incidence of residual paralysis than patients monitored by ulnar nerve stimulation and evaluation of the response of the adductor pollicis.
This finding is consistent with previous reports documenting that muscles surrounding the eye are relatively resistant to NMBDs. Multiple previous studies have consistently documented an earlier recovery of twitches with facial nerve stimulation and evaluation of the orbicularis oculi compared with when the ulnar nerve is stimulated and the response of the adductor pollicis is evaluated.8–16A more recent report by Plaud et al. evaluated the difference between various muscles surrounding the eye, and observed that the corrugator supercilii, and not the orbicularis oculi, appears to be the muscle responsible for the documented relative resistance to NMBDs.29Other studies have also attempted to distinguish between the orbicularis oculi and the corrugator supercilii,30,31although such distinction may be difficult. Gätke, in her detailed report on technical aspects of acceleromyographic monitoring of the orbicularis oculi, concluded: “when monitoring in the face, with its many small nerves and muscles, it is difficult to ensure that only one single nerve is being stimulated and accordingly that only one muscle is contracting.” It may be quite difficult for anesthesia providers to distinguish between twitches of different muscles surrounding the eye. This in turn may lead to twitches of the relatively resistent corrugator supercilii muscle to be counted, even in the absence of a twitch by the orbicularis oculi muscle. Therefore, in this report we refer to monitoring of the eye muscles.
According to our multivariable logistic model, the time interval from last dose of NMBD to acceleromyographic measurement was a significant predictor of residual paralysis. We believe this is an important factor in clinical practice, although it may not always be easily controlled (e.g. , when a surgeon requests paralysis near the end of the surgical procedure).
The lack of a statistically significant association between monitoring site and residual paralysis in the subgroup given a single dose of NMBD is not surprising, because intraoperative monitoring may not substantially impact on management in cases where a single dose of muscle relaxant is given for intubation.
We do not believe that the patients who arrived to the PACU intubated (all of whom were planned for extubation in the PACU) had their NMBDs managed differently, because this would have been contrary to our institutional routines.
More than 70% of patients in both groups had four twitches without fade, and more than 85% in both groups were reported to have had four twitches at time of reversal (table 2). Although omission of reversal medication based on clinical and qualitative monitoring is associated with residual paralysis,6the requirement for neostigmine may be modest when the residual block is small.32A TOF count of four without fade at the adductor pollicis has been reported to be consistent with a TOF ratio of approximately 0.4.33Fuchs-Buder reported that a neostigmine dose of 10 mcg/kg had a high probability of providing successful reversal from a TOF ratio of 0.4. However, when monitoring is performed at the eye muscles, the provider may be challenged to confirm that the residual block is indeed small.
We found that neostigmine doses (measured in mcg/kg) were highly variable, as reflected by their SD. The most common doses were 3 mg and 4 mg, which were administered to 48 and 24 patients, respectively. It is possible that providers do not calculate the dose based on weight, TOF count, and presence or absence of fade. It is also possible that there is substantial variation between providers in how they dose neostigmine. We do not believe that our hospitals have unique practice patterns in this regard. In this study, neostigmine dose was not associated with residual paralysis.
A finding of higher total doses of NMBDs provided to patients with monitoring of the eye muscles would have supported the presumed mechanism behind the association of monitoring site and residual paralysis, namely by an underestimation of the extent of neuromuscular block. However, this increased use of NMBDs was not apparent in this study. A recent report by Murphy et al. ,34in which patients who were randomized to intraoperative acceleromyography monitoring, had reduced symptoms of muscle weakness, and improved quality of recovery, also did not find differences in total amount of NMBDs between study groups. It is possible that quite small differences in management, especially at the end of surgery, including small differences in amount and timing of NMBDs, made a significant difference in the risk of residual paralysis. Our study, and the one by Murphy et al. , may not have sufficient statistical power to document such small differences.
This is an observational study that should be interpreted in the context of its design. We are not able to establish causality, only an association. Supporting a potentially causal relation is the biologic plausibility of these findings and the strength of the association. There were some imbalances between groups with regard to type of surgical procedures and hospital, and such imbalances could lead to confounding. We believe that we have been able to adjust for these measured differences by applying appropriate statistical methods and thus corrected for the presence of confounding in the data. However, there is still the possibility of residual confounding or bias by preferences of the individual practitioners. There may be variations in practices related to dosing, monitoring, and reversal between different institutions, and it is possible that our two participating hospitals have unique practice patterns. However, we believe our practice is representative of a large proportion of U.S. hospitals; the same anesthesia residents rotate at the two hospitals and the training they receive in management of NMBDs is consistent with current Accreditation Council for Graduate Medical Education requirements. We are not aware of any differences in training, procedures, or equipment between our two hospitals.
Our study was not designed to evaluate monitoring of specific muscles surrounding the eye (e.g. , the corrugator supercilii vs. the orbicularis oculi), and we do not know which muscle that anesthesia providers were evaluating; however, we believe that our providers use monitoring at the eye muscles in a manner that is typical for anesthesia providers in the United States and many other countries. We also do not know how many of the providers understood that the study was ongoing or the purpose of the study; 78 different providers were involved in the management of 157 patients recruited to the study (this includes seven patients who were excluded).
When acceleromyography is used without calibration, as was done in our study, there is a tendency to overestimate the TOF ratio.35Therefore, our method may have underestimated the true incidence of residual paralysis. Although multiple previous studies have used the same method that we used to identify residual paralysis in the PACU,6,28,36,37it should be noted that the accuracy of accelerography in awake patients has been questioned.38
Recovery from paralysis is important at time of extubation, and measurements should ideally be made at that time. Although the average time from extubation to the PACU was limited to 6.5 min, measurements separated by this interval are not equivalent. A previous study confirmed that there is a significant decrease in residual paralysis when comparing time of extubation with time of arrival to the PACU.39The average difference between the two measurements was 11 min in that study. Any postextubation delay of measurement is undesirable in studies on residual paralysis.
Our findings are relevant to clinical practice, and we believe that recommendations on NMBDs should include site of monitoring when discussing reversal of muscle relaxation. In addition, providers should be vigilant to the time between last administration of NMBDs and administration of reversal. Considering that both the application of acceleromyography, as used in our study, and the postextubation delay in measurement contribute to a decreased estimate of residual paralysis, it is concerning that the incidence of residual paralysis was found to be 22% and 52% among patients with monitoring of the adductor pollicis or eye muscles, respectively.
There are a variety of possible approaches to mitigate the risk factor related to site of monitoring. An apparently excellent approach would be to substitute use of quantitative monitoring at “the gold standard site,”i.e. , the adductor pollicis, even when this is not readily accessible for tactile evaluation by the anesthesia provider. Several technologies that may be useful for this purpose have been described but are not yet widely adopted.40–42Another approach, although likely more time-consuming, may be to defer reversal and extubation until the adductor pollicis becomes available to the anesthesia provider at the conclusion of a surgical procedure. Further research may be indicated to more precisely evaluate the usefulness of monitoring at sites other than the adductor pollicis.
Our study is a reminder of the limitations of monitoring with conventional nerve stimulators. These monitors do not allow us to diagnose residual paralysis, not even when used at the more reliable site, which is the adductor pollicis. When used for monitoring of eye muscles, they perform worse. Understandably, several authorities on clinical use of muscle relaxants have suggested that routine use of quantitative monitoring should be implemented.43–45
The authors thank Aaron F. Kopman, M.D., Clinical Professor of Anesthesiology (retired), Weill Cornell Medical College, New York, New York, for his insight on the study design and thoughtful review of the manuscript.