Most critically ill patients receive sedative and analgesic drugs to attenuate discomfort and pain. The excessive use of sedatives and analgesics prolongs ventilation time and intensive care unit (ICU) stay and increases costs (1,2), whereas strategies to reduce the use of sedatives and analgesics may improve the outcome (3,4). Monitoring of the depth of sedation is difficult and is based on clinical assessments of patients’ behavior, tolerance of therapy, and physiological variables (1). More structured assessment can be achieved with the introduction of sedation scores, which evaluate the response to a defined external stimulus (5). Drawbacks of these tools are their subjectivity, lack of standardization, and the risk of unnecessary deep sedation because of problems in assessing deeply sedated or pharmacologically paralyzed patients. Although stopping sedation daily helps to avoid gross oversedation, this is not always possible as a result of the unstable condition of the patient. Also, accumulation of sedatives and analgesics may occur rapidly, especially in patients with multiorgan dysfunction.
To overcome subjectivity, several methods based on the electroencephalogram (EEG) have been tested, but the results have been disappointing (6–8). This might be explained by the fact that EEG is a passive measure, whereas sedation scores measure the presence or absence of a reaction to an active stimulus.
Evaluation of the EEG response to standardized external stimuli offers an alternative approach. Cortical function and responsiveness can be assessed using the long latency evoked potentials (EP), also referred to as event-related potentials (ERP). These potentials reflect both exogenous and endogenous components of the cortical response. Exogenous components appear after detection of a sensory stimulus, and endogenous components are related to cognitive processes, such as attention and memory (9).
The N100 component appears approximately 100 ms after the auditory stimulus and follows the detection of any change in acoustic surroundings (10). The first stimulus after a long silence elicits a large N100 component, which reflects mainly the activation of an orienting reaction (OR). Repetition of the stimulus inhibits OR, called “habituation,” and therefore the stimuli that follow elicit smaller N100 components (10) (Fig. 1).
Another ERP component, mismatch negativity (MMN), reflects the brain’s automatic auditory change detection mechanism, which depends on the integrity of the auditory sensory memory. MMN is elicited by infrequently presented stimuli that differ in some physical dimension from the standard stimuli (11).
We hypothesized that recording the long latency EP to standard external stimuli could provide an objective and reproducible assessment of the level of sedation. We therefore studied the EP response to auditory stimuli in healthy volunteers during administration of sedative (propofol) and analgesic (remifentanil) drugs.
The study was approved by the Ethics Committee of the Canton of Bern and written informed consent was obtained from each participant. We studied 10 healthy male volunteers (30 ± 4 yrs; range, 23–39 yrs) in 3 sessions (infusion of propofol, remifentanil, and a combination of both drugs) separated by a 1-wk washout period. Before entry in the study, volunteers underwent a full clinical examination and laboratory screening (white blood cell count, glucose, creatinine, alanine aminotransferase, γ-glutamyl transferase) to assess normal organ function. In addition, a drug screening was performed before each session. The study took place in a fully equipped ICU environment with a dedicated intensivist responsible solely for volunteer safety.
Venous and arterial lines were inserted, Ag/AgCl EEG electrodes were placed on the scalp according to the international 10–20 system (standard method to describe the location of the scalp electrodes), and the electrode location of Cz (crossing of lines nasion-inion and ear-ear) was used with the right mastoid as reference electrode. Electrode-skin impedances were kept below 5 kOhm. Electrocardiogram, invasive and noninvasive arterial blood pressure, oxygen saturation, and end-tidal carbon dioxide were monitored by a standard patient monitor (S/5; Datex-Ohmeda, Instrumentarium Corporation, Helsinki, Finland) and recorded by data recording software (Wincollect; Datex-Ohmeda). All volunteers received supplemental oxygen via nasal probe.
After baseline hemodynamic measurements, the EPs were measured using the habituation and oddball paradigms. Briefly, the habituation paradigm consisted of a train-of-four standard tones (800 Hz; duration, 84 ms) delivered by earphones, with an interstimulus interval of 1 s and an interval of 12 s between the trains. A total of 40 trains were recorded. The oddball paradigm consisted of 85% standard tones and 15% deviant tones (560 Hz) in random order, with an interstimulus interval of 1 s (see Fig. 7). Six-hundred stimuli were delivered in each recording. The signals were amplified and digitized continuously at 279 Hz with EMMA (an ERP measuring machine developed and custom-made in the Department of Clinical Neurophysiology, Kuopio University Hospital, Kuopio, Finland).
Afterwards, infusion of either propofol, remifentanil, or a combination of both drugs was started. The order of the drug used in the first two sessions (either propofol or remifentanil) was randomized. Propofol (Disoprivan® 1%; AstraZeneca, Zug, Switzerland) and remifentanil (Ultiva®; GlaxoSmithKline, Schönbühl, Switzerland) were administered by a computer-controlled pump (Harvard-pump 22; Harvard apparatus, Kent, UK) driven by the STANPUMP program (S. Shafer, Stanford, CA) with the pharmacokinetic variable set of Schnider et al. (12) for propofol and that of Minto et al. (13) for remifentanil. In the propofol sessions, propofol infusion was titrated to reach sedation levels corresponding to the Modified Ramsay Sedation Scores 2, 3, and 4 (Table 1) (14,15) in increasing order (dosage range Ramsay Score 4, 1.0–2.1 mg · kg−1 · h−1). After 15 min of stabilization at each clinical level of sedation, EPs were measured. Remifentanil infusion was targeted to reach fixed plasma concentration levels of 1, 2, and 3 ng/mL. In the third session, remifentanil was infused to a target plasma concentration of 2 ng/mL (dosage range, 0.07–0.08 μg · kg−1 · min−1). In addition, propofol was infused as needed to achieve the same Ramsay Scores as in the propofol session. In an attempt to increase the accuracy of the sedation level assessment, Ramsay Scores before and after the EP recordings were determined by three examiners who were always present and who agreed on the level of sedation. When discrepancies between the initial and the final determination at each level of sedation appeared (usually an increase of the Ramsay Score during the last sedation level), the Ramsay Score before EP measurements was used. Each infusion period at the respective Ramsay Scores lasted 1 h. Reasons for discontinuation of the drug infusion included excessively deep sedation (Ramsay score 5 or 6), hypoventilation (respiratory rate < 8/min), and airway obstruction. At the beginning and end of each sedation level, arterial blood samples were drawn for later analysis of propofol and remifentanil plasma concentrations.
EP data from each stimulus were transformed off-line to epochs of −100 ms to 900 ms relative to the stimulus onset. After artifacts were removed (potentials with amplitudes larger than ±75 μV at baseline and those larger than ± 100 μV during sedation, mainly as a result of motion; <2% of all sweeps), the EPs were separately averaged for standard and deviant stimuli (oddball paradigm) and for the first, second, third, and fourth stimulus (habituation paradigm). To cleave the influence of corticothalamic oscillations (approximate range, 10–14 Hz) induced by propofol (16), the raw data were bandpass filtered digitally with cut-off frequencies at 1 Hz and 8 Hz before averaging (Fig. 2). The amplitudes of the N100 components were measured visually from the screen relative to the prestimulus baseline, defined as the 100-ms epoch before stimulation. If no responses to the stimulus were visible (which occurred during the deeper sedation levels) the magnitude of the amplitude was considered to be 0 μV. Habituation was defined as ratio first/second N100 amplitude of the habituation trial. The MMN waveform was obtained by subtracting the response to the standard stimulus from that to the deviant stimulus, and the mean area (between 100–230 ms) corresponding to MMN was calculated.
Blood samples for determination of drug concentrations were collected in lithium-heparin tubes (Sarstad, Nümbrecht, Germany). To prevent cleavage by endogenous esterases and by chemical hydrolysis, remifentanil samples were processed immediately by adding 20 μL of 50% citric acid per 1 mL blood. Remifentanil and centrifuged propofol samples were stored at −30°C for later analysis. Propofol was measured by high-pressure liquid chromatography with ultraviolet detection (Varian Pro Star; Varian Inc., Darmstadt, Germany), and remifentanil by liquid chromatography–tandem mass spectrography (Advantage Finnigan, Thermo Finnigan GmbH, Egelsbach, Germany). The analyses were performed at the laboratory of the University Hospital Basel, Basel, Switzerland. The quantification range was 0.1–10 μg/mL for propofol determination and 0.5–10 ng/mL for remifentanil. The accuracy (precision) of the methods was 2.8% (run-to-run) and 9.2% (day-to-day) for propofol and 9.1% (run-to-run) and 10.2% (day-to-day) for remifentanil.
Results are expressed as mean ± sd, and a two-factorial (medication and sedation level) analysis of variance design was used as a statistical test, computed by the General Linear Model-Repeated Measurements Module of the SPSS for Windows 11.0 statistical program (SPSS Inc., Chicago, IL). The comparisons of the plasma concentrations of the medication within each sedation level were done with Student’s t-test for related measurements.
One of the 10 volunteers declined further participation after the first session with remifentanil. One remifentanil session could not be included in the analysis because of improper functioning of the infusion pump. In propofol sessions the sedation level of 6 volunteers increased from Ramsay Score 4 before EP recording to Ramsay Score 5 afterwards. In the text and the figures, these sedation levels were handled as Ramsay Score 4. Five combined sessions had to be discontinued before the last measurement because one or several of the criteria for discontinuation of drug infusion were reached. Therefore, only baseline and Ramsay Scores 2 and 3 are included in the statistical analysis. No volunteer needed airway support; the only complaints were severe headaches and nausea after pure remifentanil administration. For the final analysis we compared 9 propofol, 8 remifentanil, and 9 combined sessions.
The amplitudes of the N100 components of the first tone of the habituation paradigm decreased as the level of sedation increased, independent of whether sedation was induced by propofol alone (baseline: −10.4 ± 3.5 μV; Ramsay Score 3: −4.8 ± 2.1 μV, P < 0.01) or by propofol in combination with remifentanil (baseline: −9.0 ± 4.1 μV; Ramsay Score 3: −2.9 ± 2.3 μV, P < 0.01; Table 2, within-subjects effect test of the medication: P = 0.55). The individual responses demonstrate that the N100 amplitudes decreased uniformly when the deeper levels of sedation (Ramsay Score 3) were achieved and disappeared in all individuals at Ramsay Score 4 (Fig. 3, Table 2). Similar results but with poorer discrimination were observed with the deviant tone and with the standard tone of the oddball paradigm (Table 2). Neither habituation nor MMN paradigm could be used to discriminate the sedation levels (Table 2). The latencies of the N100 responses did not differ significantly from baseline (Table 2). The individual N100 amplitudes at baseline of each session did not differ within the paradigm used (Fig. 4). In the remifentanil sessions, the volunteers never reached Ramsay Score 3 or deeper sedation levels and the measured EPs did not differ at the selected target concentrations.
At each sedation level, smaller plasma propofol concentrations were needed to achieve the corresponding sedation level in the combined sessions as compared with propofol infusion alone (3.4 ± 0.5 μg/mL propofol session RS 4 versus 1.9 ± 0.3 μg/mL combined session, P < 0.01). Drug plasma concentrations remained fairly constant during each sedation level (Fig. 5).
With standard filtering, the amplitudes of the N100 components of the first tone of the habituation paradigm still decreased as the level of sedation increased, also independent of whether sedation was induced by propofol alone (baseline: −12.5 ± 6.4 μV; Ramsay Score 3: −2.3 ± 3.7 μV, P = 0.01) or by propofol in combination with remifentanil (baseline: −11.5 ± 5.8 μV; Ramsay Score 3: −2.9 ± 2.7 μV, P = 0.01; Table 3), but the individual responses demonstrated that 2 of 3 volunteers who reached RS 4 in the combined session had N100 responses different from 0 (Fig. 6, Table 3). The test of the within-subjects effect of the medication is still not significant (P = 0.14).
Heart rate and end-tidal CO2 remained stable throughout the different sessions, whereas there was a significant decrease in mean arterial blood pressure from approximately 90 mm Hg to 70 mm Hg in both the propofol and the propofol/remifentanil sessions (P < 0.01 for each).
The main finding of our study was an inverse relationship between the tested levels of sedation and N100 amplitudes. This relationship was uniform among individuals and independent of whether propofol was used alone or in combination with the opioid remifentanil. Because the N100 amplitudes either disappeared or approached zero at Ramsay Score 4, this particular component cannot be used to evaluate the deepest sedation levels (Ramsay scores 5 and 6). However, Ramsay score 3-4 reflects the desirable target level of sedation in most clinical scenarios. Hence, despite the fact that individual amplitudes vary over a wide range at baseline and at Ramsay 2, amplitudes approaching zero seem to indicate sufficient sedation in most individuals while avoiding oversedation.
Although providing sedation and analgesia to reduce pain and the stress imposed on critically ill patients is clearly beneficial, oversedation is harmful. Despite the widespread use of various clinical sedation scores to assess the sedation level, oversedation remains a major problem in clinical practice. Use of EEG for the measurement of depth of sedation has been studied extensively in recent years, but the results have been disappointing. EEG power, though related to drug plasma concentrations, is also dependent on the specific drug used. Further EEG-derived variables, such as bispectral index (BIS), spectral edge frequency, or entropy, have gained widespread use in the operating room, whereas their use for monitoring less deep levels of sedation in the ICU patient has been problematic. The BIS values overlap widely between various sedation levels and are also dependent on the sedative used (6,8,17), and compared with the N100 responses, there is no threshold value where the subjects are in the asleep sedation levels. Mid-latency acoustic evoked potentials (MLAEPs) have shown graded changes with increasing doses of hypnotics (18–21), but unfortunately the large interindividual variation of the Nb (second negative peak) latencies in the desired sedation range of ICU patients makes them unsuitable (22,23).
Whereas MLAEPs measure mainly the activation of the auditory cortex, EPs assess higher cortical functions, such as sensory memory and cognition. They are influenced by drugs and disorders (21). If a stimulus is novel and significant, a “nonspecific” component at 100 ms is presumed to reflect activation of a relatively widespread neural network related to a general arousal response of the brain (10). This activation is part of an OR, which may also occur during sleep. In healthy subjects, the N100 component is decreased in different circumstances of reduced attention and wakefulness, including sedation (24). Similarly, N100 decreases in various degrees in anesthetized patients independent of the drug used (25–27). These previous studies may have faced the problem of overlap of evoked responses with thalamocortical oscillations at 12-14 Hz, which have been filtered in our study (Fig. 2). This filtering improved the performance of the method by establishing a clear-cut threshold the volunteer exceeds when reaching the asleep sedation levels, reducing the variability of the individual responses, and improving the independence of the results regardless of the medication used. Also, in contrast to other studies, in the present study the dose of the infused drug was targeted to clinically relevant sedation levels and not to general anesthesia. A decrease in N100 amplitudes has also been found during induction of anesthesia with propofol in patients undergoing elective surgery (27). Our results demonstrate that N100 amplitudes also reflect the level of sedation when propofol is infused to reach a clinical state of sedation without loss of consciousness.
Our concept of changes in N100 amplitude reflecting primarily the level of sedation is supported by the lack of changes during infusion of remifentanil alone. Remifentanil alone produces little sedation if any. In intensive care, sedative and analgesic drugs are almost invariably used in combination. Hence, any method for assessing sedation should be independent of concomitantly administered analgesics. The suppression of N100 amplitudes in our study was consistent at a given clinical sedation level, independent of whether propofol was used alone or in combination with remifentanil. Whether these findings can be extrapolated to other sedatives and opiates must be studied separately.
Limitations of this study are the small number of volunteers and a wide range of individual N100 amplitudes, which reflects physiologic variability. We tested the hypothesis in young, healthy volunteers who did not have impaired auditory capacity. We do not know how the method performs in patients with organ failure, particularly those with encephalopathy or traumatic brain injury. It is conceivable that a loud and busy ICU environment can have a major impact. For practical reasons we also applied the sedation in stepwise increasing rather than in random order. This might theoretically influence the results. With the paradigm used, a rather long recording time is necessary to achieve enough sweeps for averaging (roughly 10 minutes). Improved signal detection methods and shorter interstimulus intervals may help to shorten recording times in the future, and modules integrated in standard monitoring systems can help to easily titrate medication to achieve lighter levels of sedation. Whether this translates into a further reduction of the use of sedatives and a reduction in adverse events associated with oversedation must be tested in clinical trials. Meanwhile, we recommend frequent clinical assessment of sedation and careful titration of the sedative drug (1,15).
The authors would like to thank Klaus Maier, RN, and Patrick Munch, RN, for their professional care of the volunteers during the study, André Scholer, MD, for support to analyze the drug concentration, and Thomas Schnider, MD, for his assistance with the infusion pumps.
1. Jacobi J, Fraser GL, Coursin DB, et al. Clinical practice guidelines for the sustained use of sedatives and analgesics in the critically ill adult. Crit Care Med 2002;30:119-41.
2. Shafer A. Complications of sedation with midazolam in the intensive care unit and a comparison with other sedative regimens. Crit Care Med 1998;26:947-56.
3. Kollef MH, Levy NT, Ahrens TS, et al. The use of continuous i.v. sedation is associated with prolongation of mechanical ventilation. Chest 1998;114:541-8.
4. Kress JP, Pohlman AS, O’Connor MF, Hall JB. Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation. N Engl J Med 2000;342:1471-7.
5. De Jonghe B, Cook D, Appere-De-Vecchi C, et al. Using and understanding sedation scoring systems: a systematic review. Intensive Care Med 2000;26:275-85.
6. Sleigh JW, Donovan J. Comparison of bispectral index, 95% spectral edge frequency and approximate entropy of the EEG, with changes in heart rate variability during induction of general anaesthesia. Br J Anaesth 1999;82:666-71.
7. Riker RR, Fraser GL, Simmons LE, Wilkins ML. Validating the sedation-agitation scale with the bispectral index and visual analog scale in adult ICU patients after cardiac surgery. Intensive Care Med 2001;27:853-8.
8. Ely EW, Truman B, Shintani A, et al. Monitoring sedation status over time in ICU patients: reliability and validity of the Richmond Agitation-Sedation Scale (RASS). JAMA 2003;289:2983-91.
9. Picton TW, Bentin S, Berg P, et al. Guidelines for using human event-related potentials to study cognition: recording standards and publication criteria. Psychophysiology 2000;37:127-52.
10. Naatanen R, Picton T. The N1 wave of the human electric and magnetic response to sound: a review and an analysis of the component structure. Psychophysiology 1987;24:375-425.
11. Naatanen R. The mismatch negativity: a powerful tool for cognitive neuroscience. Ear Hear 1995;16:6-18.
12. Schnider TW, Minto CF, Gambus PL, et al. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology 1998;88:1170-82.
13. Minto CF, Schnider TW, Shafer SL. Pharmacokinetics and pharmacodynamics of remifentanil. II. Model application. Anesthesiology 1997;86:24-33.
14. Ramsay MA, Savege TM, Simpson BR, Goodwin R. Controlled sedation with alphaxalone-alphadolone. BMJ 1974;2:656-9.
15. Barr J, Egan TD, Sandoval NF, et al. Propofol dosing regimens for ICU sedation based upon an integrated pharmacokinetic-pharmacodynamic model. Anesthesiology 2001;95:324-33.
16. Feshchenko VA, Veselis RA, Reinsel RA. Comparison of the EEG effects of midazolam, thiopental, and propofol: the role of underlying oscillatory systems. Neuropsychobiology 1997;35:211-20.
17. Simmons LE, Riker RR, Prato BS, Fraser GL. Assessing sedation during intensive care unit mechanical ventilation with the Bispectral Index and the Sedation-Agitation Scale. Crit Care Med 1999;27:1499-504.
18. Iselin-Chaves IA, El Moalem HE, Gan TJ, et al. Changes in the auditory evoked potentials and the bispectral index followingpropofol or propofol and alfentanil. Anesthesiology 2000;92:1300-10.
19. Dutton RC, Smith WD, Rampil IJ, et al. Forty-hertz midlatency auditory evoked potential activity predicts wakeful response during desflurane and propofol anesthesia in volunteers. Anesthesiology 1999;91:1209-20.
20. Struys MM, Jensen EW, Smith W, et al. 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.
21. Sneyd JR, Samra SK, Davidson B, et al. Electrophysiologic effects of propofol sedation. Anesth Analg 1994;79:1151-8.
22. Schulte-Tamburen AM, Scheier J, Briegel J, et al. Comparison of five sedation scoring systems by means of auditory evoked potentials. Intensive Care Med 1999;25:377-82.
23. Rundshagen I, Schnabel K, Pothmann W, et al. Cortical arousal in critically ill patients: an evoked response study. Intensive Care Med 2000;26:1312-8.
24. Ogilvie RD, Simons IA, Kuderian RH, et al. Behavioral, event-related potential, and EEG/FFT changes at sleep onset. Psychophysiology 1991;28:54-64.
25. Nielsen-Bohlman L, Knight RT, Woods DL, Woodward K. Differential auditory processing continues during sleep. Electroencephalogr Clin Neurophysiol 1991;79:281-90.
26. Plourde G, Picton TW. Long-latency auditory evoked potentials during general anesthesia: N1 and P3 components. Anesth Analg 1991;72:342-50.
© 2004 International Anesthesia Research Society
27. Simpson TP, Manara AR, Kane NM, et al. Effect of propofol anaesthesia on the event-related potential mismatch negativity and the auditory-evoked potential N1. Br J Anaesth 2002;89:382-8.