Historic Perspective

One of the first documented uses of a neuromuscular blocking agent (NMBA) to aid in surgical closure of the abdomen occurred in 1912 in Germany.¹ In the 1940s, the use of NMBAs during surgery started to become commonplace. Since the 1940s, the safety profile of NMBAs has significantly improved, and their use in the operating room (OR) has become routine. The introduction of mechanical ventilation in the intensive care unit (ICU) was closely followed by the use of NMBAs in the ICU. Most of our clinical experience with the use of NMBAs comes from the care of relatively healthy patients for a finite period of time in the OR. In contrast to patients in the OR, patients in the ICU are typically critically ill with multiple organ systems failing, require prolonged neuromuscular paralysis, and are receiving a large number of concomitant medications. Clinical studies have only recently started to examine indications for and complications of NMBAs in the ICU.

Current Epidemiology

Retrospective surveys reported by intensivists in 1991 and 1992 revealed that 10 patients per ICU per month required prolonged neuromuscular blockade.² However, a study by Murray et al.³ in 1993 showed the probability of receiving a NMBA ranged from 0% in the neurosurgical ICU to 14% in the neonatal ICU. Several different situations have been identified that require short- or long-term neuromuscular paralysis.

Indications

Despite adequate sedation, some patients will not be able to tolerate mechanical ventilation, which can lead to unsafe peak inspiratory airway pressures. In a retrospective survey, Klessing⁴ reported that 89% of the cases requiring the use of NMBAs were related to the facilitation of mechanical ventilation, making it the most common reason for the use of neuromuscular paralysis. Other indications for NMBA use are the facilitation of endotracheal intubation, control of increased intracranial pressure (ICP), decrease of high muscle tone in certain medical conditions, and facilitation of needed medical procedures or diagnostic studies⁵ (Table 1). Short-acting NMBAs such as succinylcholine can be used to quickly secure an unprotected airway. However, many critically ill patients can have an airway secured by endotracheal intubation without the use of NMBAs. In patients with increased ICP, agitation or coughing caused by tracheobronchial suctioning can cause dangerous increases in ICP. Werba et al.⁶ demonstrated that pretreatment with vecuronium attenuated transient increases in ICP during suctioning.

Table 1 Indications for Use of Neuromuscular Blocking Agents in Intensive Care Unit

Mode of Action

In general, NMBAs function by competing with acetylcholine (ACh) at the nicotinic cholinergic receptor (nAChR) at the neuromuscular motor end plate (Figure 1). NMBAs are classified as either depolarizing or nondepolarizing agents. The nondepolarizing agents are further divided into short, intermediate, and long acting, and classified according to their chemical structure as aminosteroidal or benzyl-isoquinoline compounds.⁹ The ideal muscle relaxant would have rapid onset, with predictable and controllable duration of action, and lack adverse effects on hemodynamics or significant toxicity. Its elimination should be independent of liver and renal function without accumulation over time. The ideal paralytic agent would have no interactions with other medications, as well as being cost effective with a long shelf life.

Figure 1 Neuromuscular end plate. Nerve impulse travels down the nerve (A) to the axon terminal (B), which triggers the release of ACh into the synaptic cleft (C). The ACh molecules traverse the synaptic cleft and bind the nicotinic ACh receptors (nAChRs), which result in a muscle contraction. When succinylcholine binds to the nAChRs (D) a muscle contraction results; however, succinylcholine does not readily dissociate from the receptor as ACh does. As a result, the receptors are unable to repolarize, preventing further muscle contractions. In contrast to succinylcholine, the binding of nondepolarizing muscle relaxants such as rocuronium (E) to the receptor does not cause a muscle contraction and blocks the binding of ACh to the receptors, preventing a muscle contraction. ACh, Acetylcholine.

Monitoring of Neuromuscular Blockade

The primary use of NMBAs in the ICU is to facilitate mechanical ventilation.⁴ The degree of neuromuscular blockade required to reach this clinical goal will vary from patient to patient. The first set of guidelines detailing the use of NMBAs in the ICU was published in 1995 and later reevaluated in 2002. The current guidelines state that any patient receiving a NMBA should be assessed both clinically and by train-of-four (TOF) monitoring, with the clinical goal of NMBA titration to one or two twitches.¹¹ Daily clinical assessment involves the observation of skeletal muscle movement and respiratory effort by the patient. Peripheral nerve stimulation by four equal electrical charges delivered at 0.5-second intervals can be evaluated with visual, tactile, or mechanical means (Figure 2). The TOF count is the observed number of twitches out of four. A TOF count of one out of four signifies that 90%–95% of the nAChRs are occupied. A TOF count of three out of four signifies that 75%–80% of the nAChRs are occupied. At 50% occupancy of the nAChRs, the patient can be safely extubated.¹² This translates clinically to a sustained head lift for 10 seconds or to a train of four of four plus sustained tetanus to a 5-second stimulation with 100 Hz. Complete resolution of neuromuscular blockade is important not only to ensure adequate respiratory muscle power to sustain the work of breathing without ventilator support, but even more for the patient to protect the airway by unimpaired activity the bulbar muscles. Ulnar nerve innervation of the adductor pollicis muscle of the thumb is the most frequently used site to monitor TOF. The electrical stimulus should flow through the nerve, reach the neuromuscular junction, and release ACh, leading to a muscle contraction. If the electrodes are too close to the muscle tested, the current may directly cause a false-positive contraction. Hence any nerve/muscle pair that is at a certain distance apart can be used, such as the facialis/orbicularis oculi or the peroneus nerve/tibialis anterior. The monitoring of the degree of neuromuscular blockade may allow for the lowest NMBA dose and may minimize the adverse effects of prolonged NMBA use.¹¹

Figure 2 Patient’s hand with electrodes of nerve stimulator attached on volar side of forearm over the ulnar nerve. After stimulation of the ulnar nerve, the gloved hand of the medical practitioner senses the adduction of the thumb to the index. This is the response of interest to monitor the neuromuscular junction. Caution: the flexors of the hand often contract in response to direct stimulation of the muscles, even in the presence of neuromuscular blockers.

Analgesic Agents and Their Advantages

Morphine and fentanyl are the most widely used and appropriate opioid analgesics for use in the critical care setting (Table 4). Both are cleared by the liver. Although morphine is less costly, it has an active metabolite, morphine-6-glucoronide, that is renally excreted and can hence accumulate leading to renal failure. Fentanyl is eliminated by the liver only, and hence is safe even for long infusions in renal failure. Fentanyl, being very lipophilic, also has a more rapid onset compared to morphine, making it very suitable for the titration of acute pain at the bedside. Caution is warranted, however, as its rapid offset after a single dose is due to redistribution, not metabolism: This means that a few single small dose of fentanyl will only last for about 30 minutes, but with a continuous infusion fentanyl will behave similarly to other long acting narcotics like morphine and may take several hours to wear off. Hydromorphine may be preferable in patients with hemodynamic compromise, because of morphine’s potential for histamine release or preload reduction. Meperidine and the mixed agonist/antagonist opioids should be avoided. The former because of its renally cleared proconvulsant metabolite, normeperidine. The latter because they can precipitate withdrawal in opioid habituated patients.

Indications and Patient-Controlled Analgesia

Narcotics are indicated for pain relief and their beneficial or adverse side effects include sedation, depression of respiratory drive, and euphoria. Patient-controlled analgesia (PCA) is superior in pain control and reduces both side effects and total medication administered, because the patient can administer only the pain medication needed. Two features limit the risk of patient overdose: the PCA device lockout is set to the time of peak effect of the drug chosen (at least 10 minutes for morphine, and at least 2 minutes for fentanyl), so the patient will experience the full drug effect, before being able to initiate a second dose. And obviously, an oversedated patient can no longer trigger the device. However, this limits the use of PCA to patients who can cooperate. While in chronic opioid users, converting regular (methadone) use to a baseline infusion may be a good idea, anesthesiologists normally advise against the use of a baseline background infusion in a PCA, because the pain will gradually wane (reducing the need for analgesia), while the background infusion is constant, and in the heat of action the junior staff may omit adjusting it. This would lead to gradual overdose and could result in severe respiratory depression and even death. Patients generally prefer PCA because they feel that they are in control.

Epidural and Regional Analgesia

To administer analgesics only where the pain originates makes intuitive sense. Besides, local anesthetics lack the respiratory and sedative side effects of their narcotic counterparts. By blocking the conduction of nerves reporting the nociception to the CNS, superior analgesia can be achieved (Figure 3). Limitations are the need for expertise in performing regional blocks or epidural anesthesia. Currently, duration of action of local anesthetic agents is only a matter of hours, making repeated blocks necessary. While longer-acting agents in depot formulations are on the horizon, they are not yet clinically available. Leaving catheters in place for continuous infusions may work well, but is fraught with infectious risks, particularly in septic patients. In addition, in the current litigious climate many anesthesiologists are wary about performing procedures on sedated or unconscious patients who cannot give feedback during block performance. Many critically ill patients are anticoagulated or coagulopathic. In particular, in the tight epidural space a hematoma can have devastating neurological outcomes. Hence, epidural and regional blocks and catheters can only be placed in time windows when the benefits outweigh the bleeding risk or when the anticoagulation is held for a few hours. Once the catheter is in place, prophylactic anticoagulation may be resumed, and indeed may even become superfluous. This is because epidural anesthesia has intrinsic antithrombotic effects by its action on the autonomic system (and consequently blood flow), and because regional anesthesia may significantly enhance patient mobilization. In particular, for the pain of rib fractures and thoracotomies, the superior epidural analgesia in the absence of respiratory depression will convey a significant survival benefit in the elderly critically ill and in patients with respiratory comorbidities like chronic obstructive lung disease or pulmonary fibrosis.

Figure 3 Placement of thoracic epidural catheter. Tip of Tuohy needle is introduced into the epidural space. Once the location of the tip is confirmed by loss of resistance of either water or air, the epidural catheter is advanced into position as shown. Through the catheter the epidural space is infused with local anesthetics that will bathe the spinal nerve roots, blocking the transmission of pain via the lateral spinothalamic tract.

Sedatives

Benzodiazepines and propofol have anxiolytic, sedative, and anticonvulsive properties, but no analgesic properties. Indeed, good pain control will significantly reduce, if not eliminate, the need for sedation. The sedative agent is chosen based on the clinical need for short-term reversible or prolonged sedation, taking into account comorbidities such as renal failure that might prolong the agent’s duration of action (Table 5).

Table 5 Sedatives for Use in the Intensive Care Unit

If there is a need for frequent neurological assessment and swift titration of sedation in intubated patients, propofol is a good choice. Typical doses range from 30 mcg/kg/hr for sedation to 120 mcg/kg/hr for general anesthesia and very invasive procedures (calculations are based on lean body weight). Propofol reduces preload and afterload and can lead to hemodynamic compromise, in particular in underresuscitated patients or in those with septic shock. While infusions over hours and days can lead to slower recovery because of the accumulation of propofol in fat tissue, its elimination in renal or hepatic failure is not reduced. Propofol is unsafe in patients without a secured airway, except under constant observation by personnel experienced with the drug and airway management, because propofol can lead to rapid respiratory depression and airway obstruction. Propofol’s lipid carrier emulsion is similar to the lipid component of total parenteral nutrition. As such, it supports bacterial growth, has similar effects on the liver and pancreas, and needs to be factored in for nutritional assessments.

Midazolam is favored for short-term sedation. But, particularly in the context of renal failure, the active metabolite 1-hydroxymidazolam can accumulate. In contrast, lorazepam has less potential for accumulation and a faster recovery profile after longer infusions, but also slower onset. Lorazepam is also the cheapest of the most frequently used sedatives. Consequently, a lorazepam drip and subsequent dose increases are best initiated with a bolus, such as 2 mg. A bolus helps prevent oversedation by an initial rate set too high due to the delayed onset of lorazepam. All benzodiazepines can be competitively antagonized at the GABA receptor by flumazenil (0.2 mg intravenously over 15 seconds). However, flumazenil does not reverse opioid overdose, and can lower the seizure threshold.

Haloperidol is the drug of choice for treating ICU psychosis. It is a neuroleptic with strong antidopaminergic activity, and hence patients need to be monitored for extrapyramidal side effects, neuroleptic malignant syndrome, and QT prolongation on the electrocardiogram. Neuroleptic malignant syndrome (NMS) is a rare potentially fatal central neural disorder characterized by hyperthermia, rigidity, autonomic instability, and altered mental status, mediated by central and postganglionic dopamine agonism. NMS is different from the anesthesia-related muscular disorder malignant hyperthermia that is triggered exclusively by succinylcholine and volatile anesthetics such as halothane.

Initial doses of haloperidol are in the order of milligrams, but because of its good safety profile and its lack of cardiovascular or respiratory depression, it can be acutely titrated on occasions to cumulative doses up to 1 mg/kg. It is also antiemetic and dissociative and may be used in conjunction with morphine for terminal sedation.

Dexmedetomidine is a new centrally active alpha-2 agonist, available for intravenous administration (Figure 4). Unlike all other sedative agents discussed previously, it has analgesic properties, yet unlike opioids it does not depress the respiratory drive. Unlike most other sedative and analgesic medications, it can be continued during trials of spontaneous ventilation and even during the actual extubation itself. Many self-extubations occur precisely when the analgesic and anxiolytic coverage is reduced to assess the patient’s suitability for extubation. On the other hand, dexmedetomidine is sometimes by itself not strong enough in trauma patients to control pain, or in patients habituated to ethanol, benzodiazepine, or other recreational drugs, to achieve adequate sedation, and hence supplementation may be necessary (Figure 5). Theoretically, dexmedetomidine might be a suitable drug for analgesia and sedation by nonanesthesia personal for minor procedures in patients with an unsecured airway like dressing changes in burn patients. However, dexmedetomidine is not approved by the Food and Drug Administration (FDA) for this indication, nor can this approach be recommended as long as there is no more evidence in the literature supporting its safety profile under these circumstances. Dexmedetomidine is started as an infusion with or without initial bolus. Depending on the dose, dexmedetomidine can cause hypertension as well as hypotension, and may lead to bradycardia. The latter is often a desired effect in patients with a history of coronary artery disease, as this prolongs the time of myocardial perfusion during diastole and reduces myocardial oxygen demand, but it can reduce cardiac output in patients depending on heart rate for cardiac output.

Figure 4 α₂-receptor subtypes: physiology of α₂ adrenoceptors.

(Adapted from Kamibayashi T, Maze M: Clinical uses of α₂-adrenergic agonists. Anesthesiology 93:1346, 2000.)

Figure 5 Dexmedetomidine activates natural sleep pathways.

Benefits of Bispectral Index in Critical Care Setting

Sleep disruption in critically ill patients is a well-documented phenomenon. It has been shown to increase the state of confusion, more eloquently termed ICU psychosis, in addition to contributing to respiratory dysfunction. A review by Eveloff²³ and Gabor et al.²⁴ concluded that sleep disruption in the ICU can cause increased respiratory muscle fatigue and decreased ventilatory responsiveness to hypercapnia, which could prolong mechanical ventilation. Another study by Rundshagen and colleagues²⁵ found that recall of nightmares and hallucinations in sedated and ventilated patients discharged from the ICU approached 9.3% and 6.6%, respectively. Sleep disruption with decreased slow wave patterns has also been suggested to predispose for post-traumatic stress disorder. The common element in these investigations is the value of attaining near normal sleep patterns, while preventing awareness of stressful stimuli during the ICU period, both of which could be assessed by the BIS monitor.

Minimization of drug withdrawal symptoms is another unforeseen benefit of the BIS index in the ICU. In a study by Cammarano and colleagues,²⁶ 9 of 28 patients mechanically ventilated for more than 7 days were found to exhibit withdrawal reactions including insomnia and sleep disturbances. This association was attributed to high-dose opioids and benzodiazepines. The implementation of sedation scores theoretically should minimize the risk of excessive doses; however, the major drawback to the traditional scoring systems is their reliance on the subjectivity of the practitioner. Several studies have shown that nurses rely on previous experiences with sedation for their population of patients, which is further complicated by a lack of scientific evidence for a particular sedative practice. In addition, most cues used by critical care nurses to judge sedation are inadequate. The importance of objective data is therefore paramount to the optimal management of sedation and convalescence in the critically ill. With the BIS monitor, an objective method for the titration of sedation transforms the realm of abstract theory to tangible and practical clinical application (Table 6).

Table 6 BIS Values and EEG Patterns Correlated to Hypnotic States at Administration of Sedative Agents

BIS, Bispectral index; EEG, electroencephalograph.

Computing the Bispectral Index

The BIS is a single dimensionless number that incorporates information of electroencephalograph (EEG) power and frequency along with beta activation, burst suppression, and bicoherence. During sleep, a portion of the cortical EEG reflects activity in deeper structures. The bispectral analysis filters out the components of the EEG that provide information on the interaction between cortical and subcortical neural generators. The frequencies are derived from the EEG signals using a mathematical technique known as Fourier analysis. As sedation increases, more in-phase coupling occurs within signal frequencies. The degree of coupling represents bicoherence patterns. With increasing amounts of hypnotic drugs, changes in unique bicoherence patterns occur, which serve as markers for sedation level. These patterns are collected and analyzed by the monitor and compared with a database that contains a multivariate statistical model. From the comparison, the monitor assigns a number that designates a hypnotic state.

The BIS is derived empirically based on a statistical model; it is not a physiologic parameter. It was attained by analyzing a large database of EEGs from healthy volunteers who had received one or more of the most commonly used hypnotic agents (Figure 7). EEG features were identified that characterized some portion of the EEG spectrum that changed as the subjects went from the awake state to anesthetized (Figure 8). Multivariate statistical models were then used to derive the optimum combination of these features, which were then transformed into a dimensionless 0–100 linear scale. Details of the algorithm and statistical model are proprietary information. However, a brief four-step model of the process is described in the following (Figure 9):

Figure 7 BIS algorithm development process. Key steps used during development phase of the BIS algorithm. The importance of statistical analysis and modeling to identify and combine key EEG parameters is noted. The circular path shows the “iterative” process by which the BIS algorithm was prospectively tuned and improved to maximize the clinical correlation of the EEG analysis. BIS, Bispectral index; EEG, electroencephalogram.

(Courtesy of Aspect Medical Systems.)

Figure 8 Illustration of electroencephalogram patterns from waking state to sleep sleep.

(Modified from Hauri P: The sleep disorders. Kalamazoo, MI, 1977.)

Figure 9 Schematic diagram of signal processing paths integral to generating a single BIS value. Original EEG epochs (following digitization and artifact processing) undergo three primary paths of analysis—power spectral analysis, bispectral analysis, and time-based analysis for suppression/near-suppression—to look for key EEG features. The BIS algorithm, based on statistical modeling, combines the contribution of each of the identified features to generate the scaled BIS. BIS, Bispectral index; EEG, electroencephalogram.

(Courtesy of Aspect Medical Systems.)

With a BIS number designation, an assessment can then be made on the hypnotic state of the patient.

Several trials were undertaken that compared the effectiveness of the BIS in measuring hypnotic drug effect. In each trial, healthy patients were given increased doses of propofol, midazolam, isoflurane, or the combinations midazolam-alfentanil, propofol-alfentanil, or propofol-nitrous oxide. The agents were increased and decreased systematically to various steady-state plasma concentrations. Clinical measurements of sedation, hypnosis, and memory were simultaneously recorded. The results supported the BIS index as a better predictor of hypnotic state compared to measured or targeted drug concentrations.

Limitations

The BIS monitor, although novel in its approach on quantifying the level of sedation, is not impervious to the setbacks of derivation and statistical models (Table 7). Namely, the instrument’s ability to assess a given level of sedation is only as good as the database with which it compares the filtered EEG. Recalling the aforementioned development process, it is reasonable to presume that the differences between healthy volunteers and critically ill patients undergoing sedation may increase the margin of error, especially when considering that a portion of these patients have neurologic dysfunction or injury. Comparing these EEGs to a database of healthy patients does not seem to be an ideal scenario for assessing the hypnotic state in these patients.

Table 7 Potential Sources of Erroneous BIS Values

Note: This table organizes the most common reasons for erroneous BIS-Monitor readings. Included are the potential root causes and a brief description of why they occur.

BIS, Bispectral index; EEG, electroencephalograph; EMG, electromyogram.

Some authors have even expressed caution about using the BIS as a predictor of response. Because the BIS number is a derivation of EEG data from 15–30 seconds previously, its number reflects the state prior to the reading. Therefore, strong stimulation common during procedures in the ICU setting could lead to rapidly changing brain states. These changes can result in a response to stimulation despite a previously low BIS score.

Other authors have scrutinized the heterogeneity of the clinical endpoints and the subtleties of recall as a cause of the poor correlation between movement response and cerebrally derived parameters. The endpoints typically used include hemodynamic response to noxious stimulus, movement in response to stimulus, and response to command and recall. These endpoints are independent of one another. To add further complications, explicit (conscious) and implicit (unconscious) memory are not readily stratified on a continuum. Therefore, unconsciousness and inability to respond to commands does not equate to loss of recall.

Because BIS monitors the state of the brain and not the concentration of the drug, value ranges will vary among anesthetic regimens. This may lead, for example, to a BIS of 60 with the use of propofol having similar effects to a BIS of 40 with morphine and midazolam when assessing response to stimulation. The difference in values could lead to overestimation or underestimation in sedative dosing if the numbers were followed without taking the drug regimen into consideration.

Finally, electromyographic activity as seen in nonparalyzed sedated patients has been shown to increase BIS values. Studies have shown that BIS values in sedated patients are further decreased with the administration of muscle relaxants. One hypothesis is that the BIS inevitably interprets a portion of muscle activity as cerebral EEG activity, which causes a higher calculated number to be displayed. This is but one source of artifacts that does not include the possibility of nonphysiologic artifacts derived from lighting, pacemakers, monitors, infusion pumps, and radios. These variables contribute to what might seem an insurmountable challenge to overcoming the limitations of sedation monitoring.

Prospective Uses

The BIS monitor, an instrument designed for the quantification of sedation, has prospective uses as both an endpoint determinant of pentobarbital therapy and as a diagnostic tool. Traumatic head injury, cerebral hemorrhagic events, and neoplasms are significant causes of increased intracranial pressures. Pentobarbital-induced coma is a second-line treatment for intracranial hypertension refractory to osmotic or diuretic agents and cerebrospinal fluid drainage. Up to one-third of medical centers within the United States use barbiturates as a treatment modality for ICP. Pentobarbital-induced coma can be confirmed by an isoelectric pattern with intermittent “bursts” of activity—burst suppression—on the EEG. This pattern serves as a guide to therapy that minimizes barbiturate dosage and the likelihood of toxicity. The BIS monitor has the capacity to collect raw EEG data and display burst suppression ratios as a percentage of isoelectric episodes over a 63-second period. The size of the monitor and its ability to express burst suppression ratios as a numerical value make it an ideal bedside monitor in the ICU. Riker and colleagues’²⁷ prospective study of 12 patients comparing EEG monitoring values to BIS values found that BIS values correlated well with the standard EEG-based method used to titrate barbiturate therapy. With the BIS monitor, titration of a pentobarbital infusion could occur on a minute-to-minute basis without the use of special technicians, bulky equipment, or interpretation of the EEG by a neurologist hours later. This could significantly reduce overall cost and improve care of patients with increased ICP.

Conclusion

The BIS monitor has its limitations in terms of quantifying the hypnotic state. However, it does provide an objective method for the titration of sedation, which could lead to improved patient comfort, decreased sleep disruption, decreased withdrawal symptoms, and hopefully lower incidence of post-traumatic stress disorder. As more attention is focused on defining the endpoints to therapy, monitors that can provide insight on physiologic effect on a minute-to-minute basis will become a growing trend. Such is the case now for the BIS monitor and its potential uses as a guide to barbiturate therapy in patients with increased intracranial pressure.