Brain Function Monitors: Are We Just “Chasing” the Numbers?

Electroencephalogram-based brain function monitors have been advocated since 19371 to more effectively titrate anesthetic depth during routine general anesthesia. In principle, such a monitor would better account for interindividual variability in dose/concentration/effect relationships, and help anesthetists avoid either insufficient anesthesia and awareness or excessive anesthesia with delayed recovery. The use of electroencephalogram-based monitoring has gained widespread support over the past 20 years, and in November 2012, the UK National Institute for Health and Care Excellence published a guidance document recommending their use and discussing 3 commercially available “depth of anesthesia” monitors.2 While most clinicians are familiar with the underlying scientific principles, benefits, and limitations of currently available devices, many are not aware of technical issues involved in electroencephalogram processing such as suppression ratio, artifact rejection, or contamination by electromyographic components that affect the accuracy and utility of electroencephalogram-based monitors. In addition, clinicians may not understand how electroencephalogram monitoring relates to different components of the anesthetic state. Recognizing the advantages and limitations of electroencephalogram monitoring may thus improve their clinical efficacy. WHAT IS “ANESTHESIA”? In a 2014 editorial, Pandit3 noted that identifying targets for monitoring the brain for “anesthesia” should begin by asking what exactly is meant by “anesthesia.”3 He added that “intraoperative awareness is a spectrum of brain states.” The results of the 2014 Fifth National Audit Project study on intraoperative awareness4 clearly support the argument of Pandit,3 noting that a wide spectrum of experiences is associated with “not being properly anesthetized,” and that approximately 1/3 involve distress or pain. Shafer and Stanski5 have also argued that anesthetic depth can be regarded as separately attaining a number of different end points relevant to anesthesia. Sleigh6 has described anesthesia as being like the process of switching off a set of different light switches, each related to a different function such as pain, memory, autonomic responses, etc. This conceptualization of anesthesia as an amalgam of different pharmacological effects suggests that our approach to unconsciousness should resemble our approach to other multifaceted physiological systems. For example, we already accept that the cardiovascular system is composed of several basic elements such as cardiac output, cardiac filling, blood pressure, and venous return, all with different monitoring techniques and normal parameters.6 The same may be true for “unconsciousness.” For instance, when an “isolated forearm technique” was used to shield the effects of neuromuscular blocking agents from the arm, deeply anesthetized patients positively responded (hand squeezing), without recall, with median bispectral index (BIS) values of 37 at induction, 45 at laryngoscopy, and 46 at intubation.7 What then should we conclude? Movement to stimulation may indicate (un)consciousness (as Pandit3 suggests), but also accidental awareness under general anesthesia, amnesia explicit recall, suppression of movement responses, attenuation of sympathetic system responses to noxious stimulation, or prediction of future anesthetic events. Processed electroencephalogram monitors at this time are incapable of reliably and separately assessing each of these anesthetic end points. IT IS NOT CLEAR WHAT THE NUMBERS MEAN Although electroencephalogram monitoring devices are calibrated to clinical end points such as loss of eyelash reflex or response to command, they are not well calibrated to brain states once those end points have been met.2 A brain monitor that yields a single numerical value to reflect the state of “consciousness” is, at best, only providing a general estimate about the probability of “consciousness.” It is not telling us which specific element contributing to consciousness is lost and which is retained.3 If, in fact, brain states are comprised of multiple elements contributing to anesthesia, then these “numerical ranges” may accurately measure some of these elements but not all. It then follows that different “numerical ranges,” each calibrated for a specific end point representing a separate brain function such as memory loss, accidental awareness under general anesthesia, nociceptive arousal, etc,6 are likely to be more beneficial than just 1.3 In fact, a BIS range of 40–60, widely recommended to prevent accidental awareness under general anesthesia,8,9 is not derived from, or based on, an integral part of the BIS algorithm.10 And, as is well documented; one BIS range may not “fit all endpoints” so a BIS range between 40 and 60 for accidental awareness under general anesthesia may not be an appropriate range for preventing movement responses. THIRD, WHAT ARE THE CONFOUNDERS? One challenge with brain function monitor assessment is that no conventional comparator technology exists. The comparative standard chosen by the National Institute for Health and Care Excellence was “standard clinical monitoring,”2 which clearly may not be accurate in assessing anesthetic depth. In addition, brain function monitor indices may not correspond with a clinically judged anesthetic depth due to patients’ own underlying physiology which may alter the electroencephalogram response to anesthesia. Brain function monitor indices are electroencephalogram-derived parameters and any condition that can “modify” the raw electroencephalogram signal during general anesthesia is likely to reduce the ability of these monitors to truly reflect the anesthetic state. Instead, such electroencephalogram changes cause brain function monitor values to reflect unrelated electroencephalogram consequences of patient-dependent pathophysiological states like hypothermia, hypovolemia, hypoglycemia, hepatic encephalopathy, acid-base abnormalities, and neurological disorders such as cerebral ischemia or seizures. By exerting their own effect on the electroencephalogram, all of these conditions may alter brain function monitor readings.11 FINALLY, BRAIN FUNCTION MONITORS MUST BE USED PROPERLY The 2012 National Institute for Health and Care Excellence guidance document concluded that clinicians need “training” and “experience” to properly use brain function monitors.2 Anesthesia outcomes likely depend not only on accurate monitors but also on the anesthetist’s own response to monitoring values. Managing this interaction between man and machine is a critical skill for the anesthetist.12 Although published depth of anesthesia studies have not addressed the issue of training and experience, such education in the use of these devices is clearly important.2 A 2008 study13 demonstrated that despite anesthetist’s experience with BIS monitoring and their subjective appreciation of BIS value, mean anesthesia doses were not altered when they used the BIS monitor nor did BIS values differ when clinicians were blinded to BIS values.13 The utility of brain function monitor for guiding anesthetic management thus depends not only on the accuracy of the specific monitor (efficacy) but also to a large extent on practitioner decisions in response to the monitor.14 Moreover, just because a monitor can function (ie, is efficacious) to achieve a desired outcome such as reducing the risk of accidental awareness under general anesthesia, this outcome may not be fully achieved when the monitor is incorporated into routine anesthesia practice and that incorporating a brain function monitor into routine anesthesia practice is cost-effective.13 Interestingly, although previous studies had reported a lower incidence of anesthesia awareness with BIS monitoring,8,9 Avidan et al15,16 found no difference in anesthesia awareness between BIS-guided (40–60) and end-tidal gas monitoring, possibly because they included BIS-guided patients whose BIS values were still above 60 (55.0% of their patients) in an intention to treat (ie, effectiveness) analysis.15 Whether due to difficulty adhering to the recommended protocol or reluctance on the part of practitioners to follow the protocol, excluding all patients above BIS 60 in a per protocol (ie, efficacy) approach may have produced different results.15 Regardless, it is clear that a BIS device may effectively identify an undesired anesthetic state, but clinicians may not be able to “skillfully maintain BIS within the 40–60 range.” ARE WE JUST “CHASING” THE NUMBERS? Due to the proprietary nature of brain function monitor algorithms, the clinician knows only that values from brain function monitor mostly reflect a “head-related” biosignal derived from retrospective observations of electroencephalogram changes in response to increasing incremental doses of certain γ-aminobutyric acid–mediated anesthetic agents and are thus not a “true” reflection of a unified depth of anesthesia nor a measure of brain function independent of the anesthetic agent.17 Although metaphorically apt,18 the concept of “anesthetic depth” is quantitatively hollow because “depth of anesthesia” monitors do not independently measure either “depth” or consciousness. “Depth of anesthesia monitoring” is instead a misnomer; as equi-minimum alveolar concentrations of different volatile anesthetic agents yield different BIS values. A more appropriate terminology would thus be “depth of anesthetics” monitors and not “depth of anesthesia” monitors. The contemporary use of these processed electroencephalogram monitors that somehow “compress” the complex electroencephalogram waveforms recorded from 20 sites into a single number may be too simplistic. Clinicians might be better off interpreting the brain state under anesthesia in the same way as the hemodynamic state, with multiple parameters, multiple targets, and sufficient clinical experience to integrate the 2.19 ACKNOWLEDGMENTS The author thanks Eugenia Lamont, his Medical English Editor, for streamlining and language editing of the manuscript. DISCLOSURES Name: Ashraf A. Dahaba, MD, MSc, PhD. Contribution: This author is the sole author of this manuscript. This manuscript was handled by: Avery Tung, MD, FCCM.