INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although no randomized trials have been performed, neuromuscular blockade has proved to be a useful adjunct in the care of critically ill patients with respiratory failure (1). However, optimal monitoring of neuromuscular paralysis in the intensive care unit (ICU) remains a controversial subject. When used in the operating room, paralytics are carefully monitored so that patients can be safely intubated, will not move during surgery, and can be rapidly extubated when the surgery is complete. The monitoring devices in use involve the application of an electrical impulse that is sensed at the neuromuscular junction and translated into muscular movement when the neuromuscular junction is intact. Because the extent of neuromuscular blockade is easily measured with repeated electrical stimulation and muscular movement that rapidly extinguishes in the presence of paralytics, the train-of-four (TOF) stimulation has become the most common evoked response used to assess the depth of blockade (2). Portable, battery-operated units are available for bedside ICU use.

Neuromuscular blockade in the ICU is different in many respects from blockade in the operating room. The major differences are in duration of neuromuscular blockade in patients with slowly resolving lung injury and a decrease in the intensity of monitoring because of the many other factors necessary in the care of the critically ill patient. The depth of neuromuscular blockade required for ICU patients remains controversial and depends on the expected patient outcome and indications for paralysis. Many concomitant diseases and drugs such as corticosteroids and aminoglycosides are present in the ICU patient that influence the pharmacokinetics and complications of paralytics. For all of the above reasons, studies of neuromuscular blockade in the ICU are difficult to perform, are difficult to interpret, and are complicated by the bias of clinicians with different levels of skill in the use of paralytic drugs.

Many complications of neuromuscular paralysis in the ICU have been described, the most common of which is prolonged paralysis (3). Prolonged paralysis in most patients has been associated with a lack of neuromuscular junction monitoring, resulting in either pharmacologic drug overdose or accumulation of toxic or active drug metabolites, particularly in the setting of hepatic or renal insufficiency. Because of a flurry of case reports documenting prolonged paralysis in 1992 (3, 4), many experts have advanced a recommendation that TOF monitoring should be extended to the critical care nurse's armamentarium (5).

A 1991 United States survey documented that TOF monitoring occurred regularly in only 4% of ICUs in the United States and was never used in 79% of ICUs, despite the fact that 98% used neuromuscular blocking drugs (NMB) (1). The low use is likely due to the difficulty in training a large number of nursing staff who use these drugs only intermittently. In addition to the technical problems with TOF monitoring, there are few objective studies that demonstrate that TOF monitoring decreases side effects (6, 7), costs, or amount of drug infused. The current prospective study was begun to examine whether TOF monitoring provided any appreciable benefit to patients receiving continuous atracurium infusion in a general medical ICU.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A neuromuscular blockade protocol was established at the Medical Intensive Care Unit (MICU) at the Medical University of South Carolina (MUSC) in January of 1993. Under protocol, all patients requiring neuromuscular blockade for more than an intubation or brief procedure were initiated with an atracurium bolus of 0.5 mg/kg that was followed by continuous infusion of 8 µg/kg/min atracurium that was titrated every 5 min to clinical effect. All patients received benzodiazepines concomitantly including midazolam (0.1 mg/kg bolus followed by 0.06 mg/kg/h by continuous infusion for most patients, 0.03 mg/kg/h for patients with liver dysfunction or age > 55 yr) or lorazepam (0.05 mg/kg bolus followed by 0.03 mg/kg/h by continuous infusion for most patients, 0.015 mg/kg/h for patients with liver dysfunction or age > 55 yr). Morphine or propofol was administered at the judgment of the attending physician. A daily wakeup period was also left to the discretion of the attending physician.

The original protocol provided for prospective randomization between TOF monitoring or best clinical assessment to guide atracurium dosing. This protocol was granted approval by the MUSC Institutional Review Board for Human Subjects provided surrogate consent could be obtained within 24 h of enrollment. However, after surrogates refused consent on three of the first six patients, dual protocols for TOF or best clinical assessment were made available in the MICU. Each of seven attending physicians was able to choose between the protocols for his patient. At the end of 3 yr, charts for all patients who had received either protocol were reviewed and form the basis of this report. The MUSC Institutional Review Board approved the chart review process. Every physician excepting one enrolled at least one patient in each study arm.

The TOF monitoring protocols were preceded by intensive mandatory education of all MICU nurses by attending anesthesiologists and pulmonologists using a Ministem III (Professional Instruments, Houston, TX) battery-powered device. Monitoring was begun after benzodiazepine sedation but before paralysis by finding the optimal amplitude through skin electrodes over the ulnar nerve to achieve a TOF signal on the adductor pollicis muscle using a standard 2-Hz stimulus every 0.5 s for four bursts (8). In the event that the adductor pollicis could not be used because of an arterial line, dysfunctional nerve, or arm edema on both arms, a TOF signal was obtained by EKG electrodes at the facial nerve monitoring the orbicularis oculi. The amplitude sufficient to evoke the maximal TOF response was recorded and used on subsequent days. Monitoring was scheduled every 4 h with a goal of three twitches. After any dose changes, more frequent monitoring over 30 min was used to find the correct level of paralysis for the subsequent 4 h.

The best clinical assessment patients were allowed to be paralyzed sufficient to maintain ventilator synchrony and prevent clinical movement. At least once every 12 h, the drip threshold sufficient to allow patient movement was found and patients were maintained slightly more paralyzed than this level.

Data were recorded for patient diagnoses, Apache III score at the time of ICU admission and at the time of initial paralysis, and presence of organ failures. Hepatic insufficiency was considered present if criteria for Child-Pugh Class A disease was present. Renal insufficiency was considered present if serum creatinine exceeded 2.0 mg/dl. Days of ventilation before, during, and after paralytic administration were recorded. The indication for paralytic administration was recorded. An hourly tally of atracurium doses was kept that enabled calculation of duration of atracurium in hours and mean µg/kg/min of atracurium infused. Concomitant sedative medications were recorded. All patients with < 24 h of atracurium administration were excluded from the study. The use of any dose of corticosteroid or aminoglycoside during atracurium infusion was noted.

When atracurium was discontinued, the protocols requested documentation of the onset of first muscular movement. Recovery times were taken from the medical record. When mental status allowed meaningful conversation, the patients were asked if they had any memory of the paralysis episode. The patients were followed until they left the hospital, with mortality recorded.

Statistics were analyzed using a computerized program (Statview 4.02; Abacus Concepts, Berkeley, CA). Data are expressed as mean ± SEM where applicable. A two-tailed unpaired Student's t test was performed on continuous variables and a chi-squared test on categorical variables to compare the two patient groups, with p < 0.05 considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Forty-six patients from the MICU census of 1,352 patients (3.4%) received continuous-infusion NMB at the Medical University of South Carolina Hospital over 3 yr beginning in January 1993. Eight patients had infusions stopped before 24 h. Two patients transferred to the ICU on vecuronium infusions and switched to atracurium on arrival were excluded from analysis. The other 36 patients are the subject of this study.

Twenty patients had received succinylcholine and four patients had received vecuronium for short periods with clinical recovery before initiation of atracurium infusions. The demographics of the 20 patients in the TOF group and the 16 patients in the best clinical assessment group are shown in Table 1. There were no differences between groups.

The TOF and clinical assessment groups had a similar mean duration of paralysis (125 ± 19 h versus 125 ± 38 h) (Table 2). There was no difference between groups in the total dose infused (10,460 ± 2,409 versus 9,201 ± 3,237 mg) or in the mean dose indexed to body weight (15.2 ± 1.5 versus 12.0 ± 1.1 µg/kg/min).

The mean time to clinical recovery was no different between groups (50 ± 10 versus 45 ± 12 min). Complications occurred in two patients in the TOF group, with pulmonary emboli despite prophylaxis and an unrecognized cerebrovascular accident (CVA). Death before hospital discharge occurred in 10 of 20 versus four of 16 patients in this critically ill population. No survivor of mechanical ventilation remembered the time during paralysis; the nonsurvivors could not be questioned. Most patients demonstrated atracurium tolerance over time (Figure 1). Mean atracurium requirements increased from 12.0 ± 1.1 to 17.1 ± 2.7 µg/kg/min on Day 10. Small patient numbers beyond Day 10 limit further data.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Concentrations of continuous-infusion atracurium necessary for paralysis increase sequentially over time. *Mean ± SEM.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The data from this prospective study suggest that the use of a TOF monitoring device in the ICU did not decrease either the amount of atracurium infused, the time from cessation of drug to neuromuscular recovery, or the side-effect profile of comparable groups of critically ill patients in a medical ICU. The findings are contrary to published recommendations on NMB monitoring (5).

The TOF monitoring devices are relatively easy to use in a normal patient; however, those patients do not get admitted to an ICU. The high-tech world of the ICU has historically introduced technology before its effects on outcome could be studied. The issues of training time, nursing competency, and nursing administration time have never been studied. In a recent survey of ICU nurses who use TOF monitoring, 42% interrupted their other care every 30 to 60 min to administer peripheral nerve stimulation (9).

Adding to the controversy are the technical difficulties of TOF monitoring in the ICU. Poorly trained nurses often count flexion of the fingers as a response instead of recognizing this effect of direct muscle stimulation which is independent of neuromuscular junction blockade. Other technical difficulties include changing threshold amplitudes associated with changes in arm edema (10), difficulty in visual assessment (11), differences in paralytic effect between muscle groups, differences in the evoked response depending on stimulus amplitude (12), imperfect correlation between depth of blockade with clinical endpoints (13), and altered TOF signals with some clinical diseases (14, 15).

In the group who received best clinical assessment, we chose a clinical endpoint sufficient for patient-ventilator synchrony and optimization of oxygenation. Anecdotally, in a single patient, we found no decrease in oxygen consumption once muscular movement and the ability to assist the ventilator had been suppressed. In many TOF patients, this level of clinical paralysis occurred with four twitches monitored on the TOF, yet drug levels were increased by protocol to three twitches. The resulting trend toward higher drug utilization was thus expected.

Most recommendations for TOF monitoring suggest drug titration to one or two twitches (10, 16), which is the current practice of most critical care nurses using TOF monitoring (9). Whether the lesser depth of paralysis in our patients was responsible for the clean side-effect profile remains unknown. Although a TOF of three twitches generally corresponds to an 80% muscular blockade, the resultant muscular weakness was sufficient to assure patient-ventilator synchrony and lower airway pressures and to optimize oxygen delivery in most patients. We would suggest that future ICU protocols that use TOF monitoring insist upon drug dosing just beyond four twitches, recognizing that the resultant 50% blockade is sufficient to prevent even a headlift in 70% of patients (17).

Responding to multiple reports of prolonged paralysis in patients receiving continuous-infusion NMB drugs, a 1995 executive summary from the Society of Critical Care Medicine and the American College of Critical Care Medicine recommended TOF monitoring for patients receiving sustained NMB (5). These 40+ experts recognized that direct observation for ventilatory effort is the most common (1, 9) method of monitoring, even among anesthesiologist intensivists (18), while conceding that TOF recommendations were not supported by adequate scientific data. Unfortunately, direct observation for ventilatory effect, although simple, does not prevent overparalysis unless the drug is lightened intermittently.

Although the current study would suggest that these recommendations be altered, we would temper that suggestion by the fact that our clinical assessment patients were intensively monitored by a well-educated nursing staff who continuously searched for the threshold of paralysis sufficient to maintain the desired clinical endpoints. This level of education likely is not prevalent in most ICUs (19). A single report of the improved safety profile of a monitoring program in a surgical ICU has been published (6); however, our study would suggest that those benefits were the result of nursing education and not the peripheral nerve stimulator.

Unlike the steroidal paralytics vecuronium and pancuronium, atracurium is degraded by Hoffman degradation, an organ-independent process. After these long infusions, we found no episodes of prolonged paralysis in our patients with hepatic or renal insufficiency. Because of the differences in pharmacokinetics and pharmacodynamics of these drugs, the ability to safely extrapolate our findings about TOF monitoring to other drugs remains unknown.

The concentrations of atracurium required for neuromuscular paralysis in our patients were much higher than those reported in the product monograph (20) but similar to dose requirements in comparable ICU patients (21). Increased requirements for vecuronium have been reported in similar patients (22). One possible explanation is the proliferation of extrajunctional cholinergic receptors that has been reported for a variety of critical illnesses and neuromuscular diseases, including recent third-degree burns, intra-abdominal infections, major crush injuries, spinal cord transsections, and upper motor neuron diseases (23). Despite the increased dose required for clinical effect, we found no evidence of toxicity in the form of prolonged paralysis or seizures, a postulated side effect of laudanosine (21), the major degradation product of atracurium.

Tolerance to atracurium required increasing concentrations during the course of a long ICU stay, a phenomenon previously reported (21, 24). We can only speculate about the cause but suspect that prolonged occupancy of the cholinergic receptor by a competitive antagonist induces proliferation of extrajunctional cholinergenic receptors that then require more drug (25). Despite the frequent use of corticosteroids and aminoglycosides that are known to potentiate the effects of NMB, there was no consistent pattern of association that cleanly explained the phenomenon.

Another important aspect of our study was the outcome of our benzodiazepine schedules. The study was designed to use a low-dose, short-acting benzodiazepine so that benzodiazepines would not interfere with the determination of optimal NMB dosing or recovery. Midazolam alone was used in the majority of patients, with a small subset receiving midazolam plus narcotics when pain was suspected. In the last few months of the study, efforts at cost containment made lorazepam infusion the ICU standard that was used in only three patients. No patient on the low-dose midazolam infusion in our protocol was able to recall any time during paralysis, a testament to the amnestic qualities of this drug. Furthermore, the dosage schedules of midazolam for amnesia were much lower in most patients than concentrations often required for sedation in the absence of paralysis. Despite the long infusions often lasting more than 10 d, we found no episodes of prolonged sedation that have been noted at higher concentrations of this drug, likely due to accumulation of drug or breakdown products (26).

There are limitations to our study. Because no previous data have been collected on the value of TOF monitoring, a power analysis was not performed before study initiation. While we cannot exclude a type II statistical error, a power analysis performed on our study with  = 0.05 and  = 0.05 would suggest that 876 patients would be needed to further test the possibility that TOF monitoring is beneficial in lessening the amount of drug infused or in shortening the clinical recovery time.

Although originally designed as a prospective, randomized, unblinded trial, the reality of obtaining consent for research from distraught family members when the patient was near death proved prohibitive. Rather than sacrifice a comprehensive database in a situation where complications of NMB are intermittent, we chose to stop randomization and allow individual ICU physicians to choose one of two protocols. Because of the possibility of bias in patient selection, the results of this trial should be considered nonrandomized. This study would have benefited by proposed emergency research guidelines (27) that would define a concept of minimum risk (that is externally reviewed) and allow randomized research to proceed without informed consent. This option is not available in current federal guidelines in which surrogate consent must be obtained if temporally feasible (28).

Additional limitations include the fact that the time for neuromuscular recovery was measured by clinical endpoints of muscular movement that required chart documentation. Often this meant the ability to assist the mechanical ventilator or blink the eyes, features that might not have shown full muscular recovery on more formal testing. Because our patients were paralyzed only when death was the alternative choice, these patients still required continued mechanical ventilation for a mean of 5.8 ± 1.7 d after paralytics were withdrawn as lung injury continued to improve. The trend toward longer postparalytic ventilation in the clinical assessment group is therefore likely independent of muscular weakness given the trends toward lesser drug use.

The patients in our study were from a group of critically ill medical patients with a high incidence of multiple organ failure. Nevertheless, no surgical or trauma patients were included. We recognize that TOF monitoring might still be useful in subsets of patients difficult to assess by clinical means, although we identified no patient group in our MICU cohort.

Our study was initiated before the myopathy associated with NMB achieved national prominence (4). Although patients did not receive electromyography in this study, we found no patients in either group who clinically had more than deconditioning after a prolonged ICU stay. Furthermore, we made no diagnoses of critical care neuropathy (29, 30), corticosteroid myopathy (31), or rhabdomyolysis, although creatine phosphokinase was not routinely monitored until the third year of our study.

It should be noted that no patients with respiratory failure due to status asthmaticus were paralyzed during the study interval. Clinical reports have documented the association between status asthmaticus and NMB myopathy that is presumed secondary to the interaction between the steroidal NMBs (pancuronium and vecuronium) and the high doses of corticosteroids used for asthma therapy (4). Because reports of myopathy secondary to atracurium have been documented in patients on corticosteroids (32), an effort to not use NMB in asthma patients was prevalent in the physician group caring for these patients. It is interesting that the bias did not extend to corticosteroid use in general, with 17 of 36 patients using these drugs, often at high doses.

Many complications of paralysis have been previously reported (35). The two recognized complications in our cohort included pulmonary emboli despite prophylaxis and an unrecognized CVA, which both occurred in the TOF group. With the low autopsy rate in our cohort, other unrecognized deep venous thromboses might have occurred. Whether more intensive anticoagulation or the addition of sequential compression devices to low-dose anticoagulation is required for these patients needs additional study.

The incidence of patients who were completely awakened by their attending physicians every day sufficient to assess mental status was very small (two of 36). The anecdotal reason for the lack of daily wakeup was that these patients were often hemodynamically unstable with failing oxygenation just prior to paralysis. Nevertheless, one patient was awakened completely after 5 d of paralysis only to find a hemispheric CVA of indeterminate age. The second reason to provide a daily wakeup if possible is to find the patient who has been overparalyzed. Many patients received wakeup periods every second or third day that also assisted in proper atracurium titration by assuring that prolonged paralysis did not occur. The longest clinical recovery on any of these drug discontinuations was 180 min.

It is gratifying to find that good critical care nursing assessment can supplant technology in these days of cost containment. Because the ICU nurses felt training in the use of the nerve stimulator helped understand the important issues in NMB safety, we do not advocate removal of the TOF from the ICU. Rather, the nerve stimulator can be reserved as an adjunct in difficult patients, particularly in situations where deficiencies in the knowledge base about paralytics exist.

We conclude that careful titration of atracurium using clinical bedside markers should be the standard of care for monitoring this drug. Titration of the drug to an endpoint just beyond that necessary to prevent muscular movement and ventilatory assist is optimal. Because the pharmacokinetics of cisatracurium are similar, the results will likely extend to this drug. Further study is necessary to confirm the efficacy of good clinical assessment when other paralytics are used.