Home > Subjects > Monitoring (Non-cardiac) > Accuracy of the Composite Variability Index as a Measure of...
Anesthesia & Analgesia:
doi: 10.1213/ANE.0000000000000274
Anesthetic Pharmacology: Research Report

Accuracy of the Composite Variability Index as a Measure of the Balance Between Nociception and Antinociception During Anesthesia

Sahinovic, Marco M. MD*; Eleveld, Douglas J. PhD*; Kalmar, Alain F. MD, PhD*; Heeremans, Eleonora H. MD*; De Smet, Tom Ir, MSc; Seshagiri, Chandran V. PhD; Absalom, Anthony R. MBChB, FRCA, MD*; Vereecke, Hugo E. M. MD, PhD; Struys, Michel M. R. F. MD, PhD

Free Access
Article Outline
Collapse Box

Author Information

From the *Department of Anesthesiology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands; Demed, Temse, Belgium; Respiratory Monitoring Solutions, Covidien, Boulder, Colorado; and §Department of Anesthesia, University of Ghent, Gent, Belgium.

Accepted for publication March 14, 2014.

Published ahead of print June 3, 2014.

Funding: This study was partially supported by institutional and departmental grants from the Department of Anesthesiology, University of Groningen, University Medical Center Groningen, The Netherlands. The department of Anesthesiology, University of Groningen, University Medical Center Groningen, The Netherlands has received non-restrictive research support from Covidien, Boulder, CO.

Conflicts of Interest: See Disclosures at the end of the article.

Reprints will not be available from the authors.

Address correspondence to Michel M. R. F. Struys, MD, PhD, Department of Anesthesiology, University of Groningen, University Medical Center Groningen, Groningen, Hanzeplein 1, 9713 GZ Groningen, The Netherlands. Address e-mail to M.M.R.F.Struys@umcg.nl.

Collapse Box

Abstract

BACKGROUND: The Composite Variability Index (CVI), derived from the electroencephalogram, was developed to assess the antinociception–nociception balance, whereas the Bispectral Index (BIS) was developed to assess the hypnotic state during anesthesia. We studied the relationships between these indices, level of hypnosis (BIS level), and antinociception (predicted remifentanil effect-site concentrations, CeREMI) before and after stimulation. Also, we measured their association with movement in response to a noxious stimulus.

METHODS: We randomized 120 patients to one of 12 groups targeting different hypnotic levels (BIS 70, 50, and 30) and various CeREMI (0, 2, 4, or 6 ng/mL). At pseudo-steady state, baseline values were observed, and a series of stimuli were applied. Changes in BIS, CVI, heart rate (HR), and mean arterial blood pressure (MAP) between baseline and response period were analyzed in relation to level of hypnosis, antinociception, and somatic response to the stimuli.

RESULTS: CVI and BIS more accurately correlate with somatic response to an Observer Assessment of Alertness and Sedation-noxious stimulation than HR, MAP, CeREMI, and propofol effect-site concentration (Tukey post hoc tests P < 0.01). Change in CVI is more adequate to monitor response to stimulation than changes in BIS, HR, or MAP (as described by the Mathews Correlation Coefficient with significance level set at P < 0.001). In contrast, none of the candidate analgesic state indices was uniquely related to a specific opioid concentration and is extensively influenced by the hypnotic state as measured by BIS.

CONCLUSIONS: CVI appears to correlate with somatic responses to noxious stimuli. However, unstimulated CVI depends more on hypnotic drug effect than on opioid concentration.

The cerebral effect of hypnotic drugs can be monitored by means of validated monitors that analyze and condense information derived from the electroencephalogram (EEG) into a single value to simplify interpretation.1 Less progress has been made with development of objective indices of the analgesic state. The development and validation of such indices are important since optimal analgesic titration during anesthesia might reduce total drug consumption and ultimately improve clinical outcomes.2–4

The analgesic state is the net physiological result of the simultaneously opposing effects of nociceptive stimulation (e.g., surgery) and antinociceptive (analgesic) medication.5 An imbalance between nociception and antinociception may result in somatic (movement), autonomic (cardiovascular), and/or arousal (an apparent decrease in the hypnotic component of anesthesia) reactions. Nociception-induced arousal responses during anesthesia result from the activation of brainstem arousal systems6 by ascending sensory signals, which lead to cortical activation. These arousal reactions can be blunted by administering antinociceptive medication such as opioids that attenuate ascending sensory traffic and inhibit activation of subcortical arousal systems.7 Very high concentrations of hypnotics can also blunt arousal reactions, even in the absence of analgesics; in the case of propofol, this is predominantly achieved by suppression of cortical activity and responsiveness.8 Interestingly, in the absence of hypnotics, even high doses of opioids are unable to fully attenuate these arousal responses. These phenomena illustrate the complex interaction between hypnotics and analgesics.9

Recently, an EEG-derived index of the analgesic state has been introduced, described as the Composite Variability Index (CVI).10 CVI was developed as an index of the combined variability of the Bispectral Index (BIS) and electromyographic power (EMG) measured by frontal electroencephalographic electrodes, and increases in CVI have been associated with intraoperative somatic responses.10 Recently, Ellerkmann et al.11 studied the influence of remifentanil on CVI and tested the ability of CVI to predict patient movement, following a noxious stimulus. In all patients, propofol effect-site concentration (CePROP) was titrated to maintain a BIS between 35 and 65. However, they did not study the effect of the hypnotic level on CVI. They concluded that changes in CVI following a noxious stimulus were dependent on the remifentanil effect-site concentration (CeREMI) and change in CVI correlated with movement significantly better than changes in heart rate (HR) and noninvasively measured mean arterial blood pressure (MAP). However, since some indices of the analgesic state such as movement and autonomic reactions might be influenced by the level of hypnosis, the performance and use of CVI may not be equal at all levels of hypnotic drug effect (as titrated using BIS) nor at all levels of antinociception (using CeREMI as a surrogate).

The aim of this study was to investigate the relationship among CVI, BIS, and CeREMI across a wide range of hypnotic and antinociceptive levels, both with and without nociceptive stimulation. We also assessed the correlation of different thresholds for each measure with the probability of responsiveness and tolerance to stimulation. We considered HR and MAP as measures because they are still clinically used to titrate hypnotic and analgesic administration.

Back to Top | Article Outline

METHODS

After institutional ethics committee (University Medical Hospital Groningen, The Netherlands) approval and registration at ClinicalTrials.gov (NCT01053611), written informed consent was obtained from 147 ASA physical status I and II patients, aged 18 to 65 years, scheduled for surgery requiring general anesthesia. Complete group allocation is reported in Figure 1. Exclusion criteria were: weight <70% or >130% of ideal body weight, use of loco-regional anesthetic techniques, neurological disorder, or any condition or treatment that could potentially interfere with cardiovascular status or level of consciousness during anesthetic induction. After exclusion, 120 patients were randomly allocated to 1 of 12 study groups (3 × 4 groups of 10 patients) to receive propofol with or without remifentanil infusions. Cohorts of 10 patients per group were used as a convenience sample based on previous drug interaction studies.12–14

Figure 1
Figure 1
Image Tools

On arrival in the operating room, an IV cannula was inserted in the nondominant hand or forearm. All patients received an IV infusion of crystalloid solution, at a rate of 500 cc/h, to deliver the required drugs and fluids during the study period. The study was performed before any surgical intervention in a quiet operating room. Standard vital signs monitoring was used, including EEG, hemodynamic, and respiratory monitoring. Frontal EEG activity was recorded using a Vista BIS monitor (Covidien, Boulder, CO) with 4 electrodes (BISTM Quatro Sensor, Covidien, Boulder, CO). The cerebral drug effect was monitored using the BIS (Bispectral Index [BIS™], Covidien, Boulder, CO), derived from the raw frontal EEG. BIS electrodes were placed as recommended by the manufacturer. HR, 3-lead electrocardiogram, capnography, and pulse oximetry were recorded continuously, and noninvasive MAP was recorded every minute using a Philips IntelliVue MP50 (Philips, Eindhoven, The Netherlands) monitor. Numerical and waveform data were recorded electronically using Rugloop II © software (Demed, Temse, Belgium). The raw EEG was recorded at a rate of 128 Hz and stored for post hoc analysis.

Patients were allocated to one of the 12 groups by stratified randomization. Three different hypnotic levels, “light,” “intermediate,” and “deep,” were allocated to BIS targets of 70, 50, and 30, respectively. Within each of the hypnotic levels, patients were randomized into 4 groups receiving predicted CeREMI of 0, 2, 4, or 6 ng/mL. To ensure that the intended hypnotic level was achieved, a previously published and validated BIS-guided closed-loop system15 was used for propofol administration. Propofol and remifentanil predicted concentrations were calculated using the Schnider and Minto model, respectively.16–19 The specific remifentanil concentrations were selected to cover a clinically relevant spectrum of antinociception.

The timeline of the study is shown in Figure 2. All event times are reported in seconds relative to the start of the Modified Observer Assessment of Alertness and Sedation (OAA/S) evaluation. For each patient, a target-controlled infusion system for remifentanil infusion was started at the beginning of each study, targeting the allocated predicted remifentanil effect-compartment concentration. The start of the remifentanil infusion was defined as −1080 seconds. Three minutes later (900 seconds before the start of the OAA/S scale evaluation), the closed-loop controlled propofol infusion was started, targeting the allocated BIS level. Immediately after loss of consciousness, a laryngeal mask airway was inserted to maintain a patent airway (in patients who remained responsive to verbal commands, a tight-fitting facemask was used to administer oxygen). If required, manually assisted ventilation was performed to maintain oxygenation and normocapnia. At 30 seconds before the start of the OAA/S scale evaluation, the CePROP and CeREMI were assumed to be sufficiently close to steady-state equilibrium. At this time, the propofol closed-loop infusion was switched to an open-loop target-controlled infusion administration, and the effect-site target concentration was fixed at the “steady-state” CePROP last achieved by the closed-loop infusion system. This approach ensured that CePROP remained stable during the response period (Fig. 2) and that arousal in the BIS signal, if present, could be attributed to the stimulus. At timepoint “0 seconds,” we applied the sequence of stepwise, graded stimuli of the OAA/S) scale20 (Table 1). Stimuli for OAA/S values between 5 and 3 focus on the hypnotic component of anesthesia while those for 2 or lower contain a nociceptive component. Sixty seconds after the start of the OAA/S scale evaluation, defined as “60 seconds”, patients received a 30-second long (100 Hz, 60 mA) electrical stimulus.21,22 The stimulation electrodes were placed over the ulnar nerve on the volar side of the wrist. The presence or absence of purposeful movement in response to tetanic stimulation was noted by the observer. For ethical reasons, patients who scored an OAA/S of 5 did not receive a tetanic stimulus and were considered responders. All observations were continued until 180 seconds after the start of the OAA/S scale evaluation.

Figure 2
Figure 2
Image Tools
Table 1
Table 1
Image Tools

We used CeREMI as a measure of the level of antinociception, a standardized stimulus as nociception and the observed BIS as the measure of the degree of hypnosis. Individuals with an OAA/S score of 0 and no movement in response to tetanic stimulation were considered tolerant (=nonresponders), while individuals with an OAA/S greater than 0 and/or purposeful movement in response to tetanic stimulus were considered responsive (=responders). The systematic application of OAA/S and tetanic stimulation as performed in this study was treated as 1 stimulation sequence.

Offline analysis of the recorded raw electroencephalogram was later performed to calculate CVI (v2.1). CVI, which has previously been described,10 is a weighted combination of BIS, sBIS, and sEMG, where sBIS is the standard deviation (SD) of the BIS signal over the previous 3 minutes, and sEMG is the SD of the EMG over the previous 3 minutes. BIS is derived from the electroencephalogram power in the frequency band from 0 to 47 Hz using a proprietary algorithm based on the fixed, weighted combination of several subparameters, including β-ratio, SynchFastSlow, QUAZI, and burst suppression ratio. EMG is a measure of the power of the signal in the frequency band between 70 and 110 Hz recorded using the frontal electrodes of the BIS monitor. The fixed-weighted combination of BIS, sBIS, and sEMG results in CVI, which is a dimensionless index that ranges between 0 and 10 with lower values indicating an ideal balance between nociception and antinociception. The applied weights for BIS, sBIS, and sEMG into the CVI calculation and specific artifact filters are proprietary.

Back to Top | Article Outline

STATISTICAL ANALYSIS

The measures considered during the analysis were BIS, CePROP, CeREMI, CVI, HR, and MAP.

We defined 2 analysis periods relative to timing of the application of OAA/S: (1) the baseline period, from −30 to 0 seconds and (2) the response period, from 45 to 180 seconds after the start of the OAA/S scale evaluation. The tetanic stimulus was applied during the response period (always given at 60 seconds until 90 seconds after the start of the OAA/S scale evaluation). The value of a measure for an individual was defined as its median value over the relevant time period. For example, the baseline BIS for an individual was the median BIS observation over the period 30 seconds before until 0 second, the start of the OAA/S scale evaluation.

In addition, we calculated for each individual patient (1) the change in measure as a result of stimulation as the difference between baseline and response value, defined as below:

Equation (Uncited)
Equation (Uncited)
Image Tools

Median values are used to reduce the possibility of imbalanced noise and outlier observations unduly influencing the estimate of the typical value of a signal during the period.

To explore relationships between the measures and Baseline BIS or CeREMI, we performed a linear regression analysis between each measure (CVI, HR, and MAP) and Baseline BIS or CeREMI and determined the coefficient of determination (R2) for each pair. In addition, a Pearson product-moment correlation coefficient (ρ) between CVI, HR, and MAP versus Baseline BIS or CeREMI was calculated. Significance of this correlation, defined as a 95% confidence interval excluding 0, was determined.

For all the measures, descriptive statistics were investigated for each group. Two-way analysis of variance with Tukey post hoc test was done to analyze differences between various CeREMI levels or various BIS levels. Significance level was set at 5%.

The ability of the measures (BIS, CVI, CePROP, CeREMI, HR, and MAP) to correlate with the observed tolerance or responsiveness to the combined OAA/S-noxious stimulus was evaluated using the prediction probability (PK), which compares the performance of the measures having different units, as developed by Smith et al.23,24 and described previously in anesthetic research.25,26 Briefly, consider a variable such as BIS, CVI, etc… and a “gold standard” such as the observed tolerance or response to the combined OAA/S-noxious stimulus. Then, a PK of 1 for the measure would mean that measure always shows higher values when the patient responded to the combined OAA/S-noxious stimulus. Alternatively, a PK value of 0.5 would mean that the specific variable is useless for measuring the clinical response. For CePROP and CeREMI, this relationship was inverted. The jack-knife method was used to compute the standard error (SE) of the estimate, based on the assumption that all assessments were independent. Prediction probability was calculated using a custom spreadsheet macro, PKMACRO, developed by Smith et al.23,24

For the clinician, it is important to know the cutoff (threshold) values at which the measure can maximally discriminate responders and nonresponders, and also the sensitivity and specificity of this value.27 Practically, this is the answer to the question: “which number should I put as alarm limit for BIS, CVI, …on my vital signs monitor to detect responsiveness and how accurate is this value?” As PK calculations only offer a general impression of accuracy, this question should be answered by using sensitivity/specificity analysis. As such, the number of false positives (FP), false negatives (FN), true positives (TP), and true negatives (TN) were determined at each threshold. From these values, we calculated the positive and negative predictive values (PPV = TP/(TP + FP) and NPV = TN/(TN + FN)). PPV is the proportion of individuals with a measure value higher than the threshold (claiming somatic responsiveness) that are observed as responders. NPV is the proportion of individuals with a measure value lower than the threshold (claiming somatic tolerance) that is observed as nonresponders. We considered an appropriate range of threshold values for each measure and determined whether the observed value for each individual was greater (or less) than the threshold. For CePROP and CeREMI, this relationship was inverted. Good classifiers have some threshold where both PPV and NPV are close to 1.

To determine which threshold value offers the best discrimination and accuracy, we determined an optimal threshold by the maximum of the Matthews Correlation Coefficient (MCC) with respect to the threshold, defined as below:

Equation (Uncited)
Equation (Uncited)
Image Tools

The MCC is a useful indicator of binary classifier performance, unaffected by differences in the sizes of the classes. It returns a value between −1 and +1, where +1 indicates perfect prediction, 0 no better than random prediction, and −1 total disagreement between prediction and observation.28 Significance level was determined by using the (2-sided) t test for correlation coefficients. We calculated the MCC value corresponding to a P value of 0.001.

Comparison between BIS, CVI, CePROP, CeREMI, HR, and MAP for responders versus nonresponders was done using a t test.

Back to Top | Article Outline

RESULTS

All included patients completed the study. As described in Figure 1, data from 120 patients were analyzed. Characteristics of analyzed patients were similar between groups (data not shown). One individual from the BIS 70 REMI 2 group was not analyzed due to unstable predicted plasma propofol concentrations during the response period due to a user error when opening the closed-loop system. For another individual from the BIS 70 REMI 2 group, the OAA/S response was not recorded. Table 1 shows the OAA/S scores registered during the study and the number of patients classified as responders to the OAA/S and/or tetanic stimulation.

For all groups combined, the individual measures (BIS, CePROP, CeREMI, CVI, HR, and MAP) at baseline and during the response period are plotted for tolerant and responsive patients separately in Figure 3. It can be observed that the closed-loop control system for BIS performed well, showing peaks in density around the BIS targets of 30, 50, and 70 at baseline. As BIS closed-loop control was stopped before baseline measure, changes in BIS were observed during the stimulation period. Per protocol, both CePROP and CeREMI were stable at baseline and during the response period. De visu differences (especially in the inter-individual variability range) can be observed between tolerant and responsive patients for most of the measures.

Figure 3
Figure 3
Image Tools

Figure 4 displays the relationship between the individual median baseline, response CVI, and ΔCVI versus Baseline BIS and CeREMI using 2-dimensional regressions. A significant correlation is shown between Baseline CVI and Response CVI versus Baseline BIS. More scatter is seen in the Response CVI versus Baseline BIS plot, resulting in a lower R2. Although significant, ΔCVI shows a weak correlation and a lot of scatter versus Baseline BIS. No significant correlation is seen between CVI and CeREMI.

Figure 4
Figure 4
Image Tools

We also detailed the mean (SD) CVI values for all groups separately. Table 2 shows the average of individual median CVI values during baseline and response period for all groups separately. Table 3 shows the data for ΔCVI. When the BIS target was 50, CVI values during both the baseline and response periods were higher when no remifentanil was given. For the BIS 70 and BIS 30 target levels, similar CVI values were found between various CeREMI. As stated before, for a given CeREMI, lower CVI values are seen at lower BIS targets for both baseline and response periods. Table 3 reveals very few significant differences in ΔCVI versus various BIS and CeREMI levels. As higher ΔCVI values were observed for all groups at targeted BIS of 70 and for the group BIS 50—CeREMI 0 versus all more extensive drug levels, it seems there is a clear differentiation between these 2 cohorts of groups.

Table 2
Table 2
Image Tools
Table 3
Table 3
Image Tools

Figure 5 shows the relationship between the individual median baseline, response HR and ΔHR versus baseline BIS and CeREMI using 2-dimensional regressions, where only a weak correlation between ΔHR and Baseline BIS is observed. The relationship between Baseline and Response HR versus CeREMI is moderate, however, not resulting in a significant relation between ΔHR and CeREMI. For Baseline and Response HR, Table 2 shows a significant decrease when remifentanil is given at BIS levels 50 and 70. No overall differences in ΔHR are observed (Table 3). Figure 6 shows the relationship between the individual median baseline, response MAP and ΔMAP versus BIS and CeREMI. Similar to HR, MAP only reveals a significant relationship with CeREMI at Baseline and Response level, not resulting in significance for ΔMAP. No relationship with Baseline BIS is observed. These finding are confirmed in the detailed values shown in Tables 2 and 3.

Figure 5
Figure 5
Image Tools
Figure 6
Figure 6
Image Tools

For CVI, HR, and MAP, the individual data for subjects who were tolerant (nonresponders) versus responsive (responders) to stimulation are plotted in Figures 4–6. For CVI, de visu observation of Figure 4 shows a clear differentiation of responders and nonresponders in both baseline and response conditions. For HR and MAP, this is less clear. Table 4 shows the significantly different average values (with SD) of each metric between tolerant and responsive subjects. As differences in the mean are not sensitive enough to observe the accuracy of differentiation, we tested the ability of the measures to differentiate responders versus nonresponders using the PK assessment, and the results are shown in Table 4. PK values for BIS and CVI at baseline showed a similar and acceptable prediction probability. Other measures performed poorly. During the response period, CVI and BIS showed a good association with the response to the combined OAA/S-noxious stimulus. Other measures performed significantly less accurately. ΔCVI is a significantly better measure to monitor the response to stimulation than ΔBIS, ΔHR, or ΔMAP, as depicted by a significantly higher PK value.

Table 4
Table 4
Image Tools

Figures 7–9 show the PPV and NPV for a range of thresholds (being specific measure values) for each measure to classify individuals as tolerant or responsive to the applied stimuli for the baseline, stimulation period, and the change or “Δ” between baseline and response, respectively. In addition, the MCC is shown for each specific threshold, to indicate the binary classifier performance. Only the MCC values outside the shaded range are statistically significant. The highest (or lowest) MCC value indicates the best threshold value for discrimination. During baseline (Fig. 7), similar performance was seen for BIS and CVI. The overall MCC values are only showing moderate performance (all values around 0.5). The highest MCC for BIS and CVI is found around a BIS of 60, and a CVI of 1.8. For the other measures, no meaningful thresholds can be defined indicating overall low performance. During the stimulation period (Fig. 8), CVI performed the best, followed by BIS. Maximum MCC values for BIS and CVI were 0.55 and 0.8, at BIS and CVI thresholds of 66 and 2.5, respectively. Thus, CVI shows better performance than BIS at their most accurate threshold values. Other measures showed poor performance without meaningful threshold values.

Figure 7
Figure 7
Image Tools
Figure 8
Figure 8
Image Tools
Figure 9
Figure 9
Image Tools

Figure 9 shows the accuracy of the change in values between baseline and stimulation period for each measure in relation to tolerance or response to stimulation. For ΔCVI, the highest combined PPV and NPV for a range of thresholds to detect tolerance/responsiveness to stimulation (being the value where the lines cross) is seen at a difference in value of 0.8 between tolerant and responsive subjects with an accuracy, defined by the MCC, of around 0.9. In contrast, ΔBIS was was not useful to classify tolerance versus response, achieving an MCC of 0.4 at the value of highest combined PPV and NPV.

Back to Top | Article Outline

DISCUSSION

In a patient undergoing anesthesia, adequate suppression of consciousness and an optimal nociception–antinociception balance should ensure somatic tolerance to subsequent painful stimuli. We found that values of 2 indices, BIS and CVI, derived from the EEG detected this optimal state more accurately than hemodynamic measures. However, the interpretation of these indices in clinical practice may be challenging because the cerebrally derived information of analgesic state is strongly influenced by the hypnotic drug effect and does not differentiate well between opioid levels in the unstimulated condition. The performance seems to be improved in conditions where noxious stimulation is present.

No significant relation was observed between CVI and CeREMI in the absence of a noxious stimulus, except between no remifentanil and a very high dose of remifentanil (6 ng/mL) at a target BIS of 50 (decrease in CVI). During stimulation at a target BIS level of 50, a significantly lower CVI was found when remifentanil was given, although without a concentration–effect relation. As a result, ΔCVI was not different at various CeREMI levels. Our findings are partially in agreement with those from Ellerkman et al.,11 who also did not find a significant decrease in CVI at increasing CeREMI without stimulation, but found significant changes during stimulations, resulting in a better relationship between ΔCVI and CeREMI. Their study only included a lower range of CeREMI concentrations (0–3 ng/mL).

Relating analgesic state indices to specific opioid concentrations in the absence of a noxious stimulus remains challenging as published previously for other measures derived from the autonomic nervous system.5,22,29–32 In the presence of a noxious stimulus, we and others found an influence of remifentanil on the performance of the surgical pleth index, an analgesic state index derived from the plethysmographic wave. However, as in the current study, only differences between no- and low-dose remifentanil were revealed, and no relationship was seen with further increases in CeREMI.22,31 Although Rantanen et al.21 demonstrated that a 30-second tetanic stimulus emulates the noxious stimulus generated at skin incision, use of other noxious stimuli might result in more significant relations between CVI and CeREMI. This has to be clarified in further studies.

Some authors claim that an ideal index of the analgesic state should be independent from the hypnotic level of anesthesia. Using this reasoning, indices of the analgesic state should correlate directly with the level of antinociception (e.g., an opioid concentration) in the absence of a noxious stimulus. Without noxious stimulus, these indices should ideally not be influenced by the level of hypnosis.5,22,29–31 In our study, CVI values both before and during the noxious stimulation were influenced by the hypnotic level. Lower BIS levels were associated with lower CVI levels, both in the absence and presence of a noxious stimulus. As seen in Figure 4, at very deep levels of hypnosis, the influence of remifentanil on CVI is overwhelmed by the high concentration of propofol. Previous studies11 did not include a determination of the influence of various hypnotic levels on CVI. As such, our study demonstrates that one should examine various hypnotic levels before making conclusions on the accuracy of analgesic state monitors in the presence or absence of a noxious stimulus. The primary site of action for analgesic drugs such as opioids is in the spinal cord and brainstem. Due to the challenges and limitation of currently available technology to adequately monitor these areas noninvasively, all current analgesic state monitors assess some physiological signal at more accessible sites to try to calculate the modulation on the observed signal by the noxious stimuli.33

The results of this study confirm that CVI, an index of the variability of BIS and frontal EMG activity, is not independent of the hypnotic state and is not purely related to CeREMI. We also realize that BIS and CVI are not measuring the effect on the spinal cord but reveal secondary information from the noxious stimulus when reaching the cerebral cortex via ascending pathways. As such, we would rather define CVI as an arousability index and not as a measure of the balance between nociception and antinociception, similar to the suggestion by Glass in his mental model stating that “opioids attenuate the noxious stimulus that activates the neural response circuitry, whereas hypnotics directly suppress the neural response circuitry.”9,34 Interestingly, we also observed (Table 3) that ΔCVI depends on both hypnotic and antinociceptive levels, and see this as concordant with the abovementioned suggestion of Glass. In all BIS 70 groups and the BIS 50 CeREMI 0 group, higher ΔCVI values occurred in response to nociception. At other combinations, no ΔCVIs were observed. As such, ΔCVI as a result of a noxious stimulus could be an interesting indicator of overall arousal caused by inadequate hypnosis, inadequate antinociception, or the combination of both.

We successfully controlled the hypnotic component of anesthesia by titrating CePROP to a specified BIS target using a previously validated closed-loop system.15 Our aim was to standardize cerebral drug effect to minimize inter-individual variability and to automatically consider the pharmacodynamic synergy between propofol and remifentanil at a specific hypnotic level. Although BIS is a surrogate measure of the hypnotic component of anesthesia and has its limitations,1 we contend that standardization of cerebral effect would not have been possible with a fixed CePROP and CeREMI approach.

Because HR and MAP are still used in clinical practice to titrate administration of both hypnotics and analgesics, we investigated their relationship with CeREMI and BIS. Similar to previous publications,25,26 we found neither useful relationships for both HR and MAP nor for ΔHR and ΔMAP. Only the known hemodynamic side effects (bradycardia and hypotension) were identifiable when remifentanil was given.

Somatic response such as movement in response to a noxious stimulus can also be considered a mixed clinical phenomenon. Immobility is recognized as a cardinal feature of general anesthesia and can be achieved by a high concentration of hypnotics or opioids or by optimal combinations of different drugs.35,36 As such, movement after a tetanic stimulus can be caused by inadequate hypnosis or a mismatch in the analgesic state or a combination of both. The ability to identify individuals likely to respond to stimulation is useful because it gives the anesthesiologist the opportunity to increase 1 or both of the components of anesthesia before stimulation/response occurs. BIS, CVI, HR, and MAP were significantly different between responders and nonresponders to an OAA/S-noxious stimulus, both at baseline and during the stimulation period. The changes in these measures were also significantly larger in the response group. This is in agreement with that of previous studies.10,11

Because differences in the mean values do not have to result in high sensitivity and specificity for a specific measure per se, we also searched for an objective accuracy evaluation (using PK) and an identifiable threshold significantly discriminating the presence or absence of responses to an OAA/S-noxious stimulation. We first assessed the ability of the baseline measures to accurately predict movement responses. Using the PK and the MCC, we found that BIS and CVI were the most accurate predictors of somatic responses among those tested. CePROP, CeREMI, HR, and MAP were not found to be reliable predictors. As movement in response to a noxious stimulus is a spinal cord and not brain-based function, we could question whether CVI and BIS can predict movement at all. The spinal–cerebral relationship might be too complex to call cerebrally derived indices “predictors of movement.” Whereas there are some anesthetics that are classified as immobilizers (can achieve minimum alveolar concentration [MAC]), there are others that, while providing unconsciousness, will not immobilize the subject at any concentration (nonimmobilizers). Movement or an assessment of MAC still has a spinal cord origin. The concentration–effect relation for hypnotics versus MAC is well-defined for the volatile anesthetics that is fairly steep.37 Because we used a combination of propofol and opioids, the concentration–effect relationship between interaction drugs and immobility becomes rather complex. Although clinically applicable, we should realize we are not measuring the spinal cord, but we are just revealing information from an epiphenomenon. Although the definition of PK could be interpreted in a predictive way, we would rather be careful and agree with the conclusion of Mathews et al.10 stating that “CVI might detect periods of somatic response.” This was certainly revealed by the more accurate performance during the stimulus period. Here, we clearly found that BIS and CVI correlated with the probability of somatic responses. The other measures provided no clear predictions or indications of somatic response.

The detailed sensitivity/specificity analysis showed that only BIS and CVI are meaningful measures to detect somatic responsiveness. At baseline, a threshold value for BIS and CVI indicating that an individual would be responsive to an upcoming OAA/S-noxious stimulus was revealed, however, with only moderate accuracy (MCC values approximately 0.55). During stimulation, CVI showed a higher MCC than BIS, indicating a better performance at the best discriminating threshold value. During periods of stimulation, a CVI value higher than 2.5 can be associated with periods of somatic responsiveness. For BIS and CVI, our findings are in agreement with a recent publication by Mathews et al.10 who found that increases in EEG and EMG variability were associated with an increased incidence of intraoperative somatic responses. Different findings are observed for the changes in the indices (“Δ”) between baseline and stimulation. The ΔBIS is not accurate for monitoring response to stimulation compared to BIS during the Baseline and Response period. Although derived partially from BIS, ΔCVI is a much better measure than BIS for this purpose. This suggests that changes in frontal EMG activity (a component of CVI)10,38 provide useful information. Although the frontal muscles are rather resistant to neuromuscular blockade, the performance of CVI during profound neuromuscular blockade might be compromised. This has to be investigated in future studies. All other measures performed poorly as binary classifiers for tolerance or response to noxious stimulus.

One could question whether CVI offers more clinical benefits than BIS in detecting response to a noxious stimulus. CVI offers a value of arousability that is derived from changes in BIS and frontal EMG over time. We are convinced that clinicians cannot focus enough on the fast temporary changes in BIS and EMG measures to detect arousability online, so a derived measure might be of clinical value. In addition, our group has proven that static views of small changes in displayed vital signs information are not always accurately diagnosed by the clinician.39

In conclusion, CVI and BIS more accurately detect somatic responses to an OAA/S-noxious stimulation than HR, MAP, or CeREMI and CePROP. ΔCVI is a better measure to monitor the time course of the response to stimulation than ΔBIS, ΔHR, or ΔMAP. We cannot conclude that CVI is an independent index of the balance between nociception and antinociception; however, we can state that CVI and ΔCVI offer information on arousability. None of the candidate analgesic state measures was uniquely related to a specific opioid concentration and is extensively influenced by the hypnotic state as measured by BIS. More research is required to investigate if combined drug concentration information from both hypnotic and analgesic drugs resulting in accurate clinical anesthesia would be better correlated to measures of the balance between nociception and antinociception.

Back to Top | Article Outline

DISCLOSURES

Name: Marco M. Sahinovic, MD.

Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.

Attestation: Marco M. Sahinovic has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Douglas J. Eleveld, PhD.

Contribution: This author helped analyze the data and write the manuscript.

Attestation: Douglas J. Eleveld has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Alain F. Kalmar, MD, PhD.

Contribution: This author helped design and conduct the study and write the manuscript.

Attestation: Alain F. Kalmar has seen the original study data and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Eleonora H. Heeremans, MD.

Contribution: This author helped design and conduct the study.

Attestation: Eleonora H. Heeremans has seen the original study data and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Tom De Smet, Ir, MSc.

Contribution: This author helped design the study and developed the closed-loop system applied in this study.

Attestation: Tom De Smet approved the final manuscript.

Conflicts of Interest: Tom De Smet is the coinventor of the applied closed-loop system. In the past, he received honorarium as a consultant for Covidien (Boulder, CO).

Name: Chandran V. Seshagiri, PhD.

Contribution: This author helped analyze the data and write the manuscript.

Attestation: Chandran V. Seshagiri has seen the original study data and approved the final manuscript.

Conflicts of Interest: Chandran V. Seshagiri was a full time employee of Covidien, Boulder, CO.

Name: Anthony R. Absalom, MBChB, FRCA, MD.

Contribution: This author helped design and conduct the study and wrote the manuscript.

Attestation: Anthony R. Absalom has seen the original study data and approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Hugo E. M. Vereecke, MD, PhD.

Contribution: This author helped design and conduct the study and write the manuscript.

Attestation: Hugo E. M. Vereecke approved the final manuscript.

Conflicts of Interest: The author has no conflicts of interest to declare.

Name: Michel M. R. F. Struys, MD, PhD.

Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.

Attestation: Michel M. R. F. Struys helped design the study, has seen the original study data, reviewed the analysis of the data, approved the final manuscript, coauthored the writing of the manuscript, and is the author responsible for archiving the study files.

Conflicts of Interest: During the last 5 years, Michel M. R. F. Struys served twice as a member of an advisory panel organized by Covidien (Boulder, CO). He received speakers honoraria from Covidien. He is also a coinventor of the applied closed-loop system in the study.

This manuscript was handled by: Tony Gin, MD, FRCA, FANZCA.

Back to Top | Article Outline

REFERENCES

1. Bruhn J, Myles PS, Sneyd R, Struys MM. Depth of anaesthesia monitoring: what’s available, what’s validated and what’s next? Br J Anaesth. 2006;97:85–94

2. Parker SD, Breslow MJ, Frank SM, Rosenfeld BA, Norris EJ, Christopherson R, Rock P, Gottlieb SO, Raff H, Perler BA. Catecholamine and cortisol responses to lower extremity revascularization: correlation with outcome variables. Perioperative Ischemia Randomized Anesthesia Trial Study Group. Crit Care Med. 1995;23:1954–61

3. Riles TS, Fisher FS, Schaefer S, Pasternack PF, Baumann FG. Plasma catecholamine concentrations during abdominal aortic aneurysm surgery: the link to perioperative myocardial ischemia. Ann Vasc Surg. 1993;7:213–9

4. Monk TG, Ding Y, White PF. Total intravenous anesthesia: effects of opioid versus hypnotic supplementation on autonomic responses and recovery. Anesth Analg. 1992;75:798–804

5. Rantanen M, Yli-Hankala A, van Gils M, Yppärilä-Wolters H, Takala P, Huiku M, Kymäläinen M, Seitsonen E, Korhonen I. Novel multiparameter approach for measurement of nociception at skin incision during general anaesthesia. Br J Anaesth. 2006;96:367–76

6. Lydic R, Baghdoyan HA. Sleep, anesthesiology, and the neurobiology of arousal state control. Anesthesiology. 2005;103:1268–95

7. Guignard B, Menigaux C, Dupont X, Fletcher D, Chauvin M. The effect of remifentanil on the bispectral index change and hemodynamic responses after orotracheal intubation. Anesth Analg. 2000;90:161–7

8. Franks NP. General anaesthesia: from molecular targets to neuronal pathways of sleep and arousal. Nat Rev Neurosci. 2008;9:370–86

9. Shafer SL. All models are wrong. Anesthesiology. 2012;116:240–1

10. Mathews DM, Clark L, Johansen J, Matute E, Seshagiri CV. Increases in electroencephalogram and electromyogram variability are associated with an increased incidence of intraoperative somatic response. Anesth Analg. 2012;114:759–70

11. Ellerkmann RK, Grass A, Hoeft A, Soehle M. Response of the Composite Variability Index to a standardized noxious stimulus during propofol remifentanil anesthesia. Anesth Analg. 2013;116:580–8

12. Short TG, Ho TY, Minto CF, Schnider TW, Shafer SL. Efficient trial design for eliciting a pharmacokinetic-pharmacodynamic model-based response surface describing the interaction between two intravenous anesthetic drugs. Anesthesiology. 2002;96:400–8

13. Bouillon T, Bruhn J, Radu-Radulescu L, Bertaccini E, Park S, Shafer S. Non-steady state analysis of the pharmacokinetic interaction between propofol and remifentanil. Anesthesiology. 2002;97:1350–62

14. Bouillon TW, Bruhn J, Radulescu L, Andresen C, Shafer TJ, Cohane C, Shafer SL. Pharmacodynamic interaction between propofol and remifentanil regarding hypnosis, tolerance of laryngoscopy, bispectral index, and electroencephalographic approximate entropy. Anesthesiology. 2004;100:1353–72

15. De Smet T, Struys MM, Neckebroek MM, Van den Hauwe K, Bonte S, Mortier EP. The accuracy and clinical feasibility of a new bayesian-based closed-loop control system for propofol administration using the bispectral index as a controlled variable. Anesth Analg. 2008;107:1200–10

16. Minto CF, Schnider TW, Egan TD, Youngs E, Lemmens HJ, Gambus PL, Billard V, Hoke JF, Moore KH, Hermann DJ, Muir KT, Mandema JW, Shafer SL. Influence of age and gender on the pharmacokinetics and pharmacodynamics of remifentanil. I. Model development. Anesthesiology. 1997;86:10–23

17. Minto CF, Schnider TW, Shafer SL. Pharmacokinetics and pharmacodynamics of remifentanil. II. Model application. Anesthesiology. 1997;86:24–33

18. Schnider TW, Minto CF, Gambus PL, Andresen C, Goodale DB, Shafer SL, Youngs EJ. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology. 1998;88:1170–82

19. Schnider TW, Minto CF, Shafer SL, Gambus PL, Andresen C, Goodale DB, Youngs EJ. The influence of age on propofol pharmacodynamics. Anesthesiology. 1999;90:1502–16

20. Chernik DA, Gillings D, Laine H, Hendler J, Silver JM, Davidson AB, Schwam EM, Siegel JL. Validity and reliability of the Observer’s Assessment of Alertness/Sedation Scale: study with intravenous midazolam. J Clin Psychopharmacol. 1990;10:244–51

21. Rantanen M, Yppärilä-Wolters H, Yli-Hankala A, Kymalainen M, Korhonen I. Tetanic stimulus of the ulnar nerve as a predictor of heart rate response to skin incision in propofol-remifentanil anaesthesia. Br J Anaesth. 2007;99:509–13

22. Struys MM, Vanpeteghem C, Huiku M, Uutela K, Blyaert NB, Mortier EP. Changes in a surgical stress index in response to standardized pain stimuli during propofol-remifentanil infusion. Br J Anaesth. 2007;99:359–67

23. Smith WD, Dutton RC, Smith NT. A measure of association for assessing prediction accuracy that is a generalization of non-parametric ROC area. Stat Med. 1996;15:1199–215

24. Smith WD, Dutton RC, Smith NT. Measuring the performance of anesthetic depth indicators. Anesthesiology. 1996;84:38–51

25. Struys MM, Jensen EW, Smith W, Smith NT, Rampil I, Dumortier FJ, Mestach C, Mortier EP. Performance of the ARX-derived auditory evoked potential index as an indicator of anesthetic depth: a comparison with bispectral index and hemodynamic measures during propofol administration. Anesthesiology. 2002;96:803–16

26. Struys MM, Vereecke H, Moerman A, Jensen EW, Verhaeghen D, De Neve N, Dumortier FJ, Mortier EP. Ability of the bispectral index, autoregressive modelling with exogenous input-derived auditory evoked potentials, and predicted propofol concentrations to measure patient responsiveness during anesthesia with propofol and remifentanil. Anesthesiology. 2003;99:802–12

27. Kalkman CJ, Drummond JC. Monitors of depth of anesthesia, quo vadis? Anesthesiology. 2002;96:784–7

28. Matthews BW. Comparison of the predicted and observed secondary structure of T4 phage lysozyme. Biochim Biophys Acta. 1975;405:442–51

29. Luginbühl M, Yppärilä-Wolters H, Rüfenacht M, Petersen-Felix S, Korhonen I. Heart rate variability does not discriminate between different levels of haemodynamic responsiveness during surgical anaesthesia. Br J Anaesth. 2007;98:728–36

30. von Dincklage F, Hackbarth M, Mager R, Rehberg B, Baars JH. Monitoring of the responsiveness to noxious stimuli during anaesthesia with propofol and remifentanil by using RIII reflex threshold and bispectral index. Br J Anaesth. 2010;104:201–8

31. Gruenewald M, Meybohm P, Ilies C, Höcker J, Hanss R, Scholz J, Bein B. Influence of different remifentanil concentrations on the performance of the surgical stress index to detect a standardized painful stimulus during sevoflurane anaesthesia. Br J Anaesth. 2009;103:586–93

32. Bergmann I, Göhner A, Crozier TA, Hesjedal B, Wiese CH, Popov AF, Bauer M, Hinz JM. Surgical pleth index-guided remifentanil administration reduces remifentanil and propofol consumption and shortens recovery times in outpatient anaesthesia. Br J Anaesth. 2013;110:622–8

33. Mashour GA. Neurophysiology and intraoperative nociception: new potentials? Anesthesiology. 2013;118:239–40

34. Glass PS. Anesthetic drug interactions: an insight into general anesthesia–its mechanism and dosing strategies. Anesthesiology. 1998;88:5–6

35. Eger EI 2nd, Raines DE, Shafer SL, Hemmings HC Jr, Sonner JM. Is a new paradigm needed to explain how inhaled anesthetics produce immobility? Anesth Analg. 2008;107:832–48

36. Hendrickx JF, Eger EI 2nd, Sonner JM, Shafer SL. Is synergy the rule? A review of anesthetic interactions producing hypnosis and immobility. Anesth Analg. 2008;107:494–506

37. Zbinden AM, Maggiorini M, Petersen-Felix S, Lauber R, Thomson DA, Minder CE. Anesthetic depth defined using multiple noxious stimuli during isoflurane/oxygen anesthesia. I. Motor reactions. Anesthesiology. 1994;80:253–60

38. Yli-Hankala A, Edmonds HL Jr, Heine MF, Strickland T Jr, Tsueda K. Auditory steady-state response, upper facial EMG, EEG and heart rate as predictors of movement during isoflurane-nitrous oxide anaesthesia. Br J Anaesth. 1994;73:174–9

39. van Amsterdam K, Cnossen F, Ballast A, Struys MM. Visual metaphors on anaesthesia monitors do not improve anaesthetists’ performance in the operating theatre. Br J Anaesth. 2013;110:816–22

© 2014 International Anesthesia Research Society

Login

Become a Society Member

Article Level Metrics