During mechanical ventilation in the pediatric intensive care unit (PICU), sedative and analgesics are routinely administered to provide comfort, anxiolysis and relief of pain, to facilitate patient-ventilator synchrony, and to prevent inadvertent removal of invasive devices. These drugs are titrated to the clinical needs of the patient, usually assessed via numeric sedation and analgesia scores that are based on various clinical and physiologic variables.
Although most patients are effectively managed in this way, neuromuscular blockade is occasionally required. Consequently, physical assessment is lost and the practitioner must rely on physiologic variables to assess comfort. However, because these patients tend to be the most critically ill, changes in physiologic variables may be more reflective of the patient’s state of health than level of comfort. Concern has been expressed that, to avoid undetected awareness and discomfort, larger doses of sedative drugs may be administered to these patients, which may increase morbidity, particularly the incidence of drug withdrawal (1).
The bispectral index (BIS) monitor processes a modified electroencephalogram (EEG) to assess the hypnotic effects of sedatives and anesthetics, replacing the reliance on physiologic variables for determining the depth of anesthesia. A number is assigned between 0 (isoelectric) and 100 (fully awake), making interpretation simple and available to any bedside caregiver. Intraoperative studies have demonstrated that adequate anesthesia, as assessed by movement at incision, correlates well with BIS values <60 (2) and that BIS values of 40–60 are typical during maintenance of general anesthesia (3,4). Of perhaps more relevance to the critical care practitioner, similar studies suggest that amnesia reliably occurs at BIS values <64–80 (5,6). Although interest in applying the BIS technology in the critical care setting is growing, studies to date have only evaluated short-term use of the BIS (7,8) and have not included PICU patients. In the current study, we evaluated longer-term use of the BIS monitor in sedated, mechanically ventilated PICU patients and assessed its correlation with clinical sedation scores. We hypothesized that the BIS monitor would allow differentiation among clinically relevant levels of sedation and, therefore, be of potential value when the clinical examination is unavailable.
This study was approved by the IRB of the University of Missouri. Written, informed consent was obtained from a parent. All children admitted to the PICU and receiving IV sedation with or without analgesia during mechanical ventilation were eligible for enrollment. Patients were excluded if they were receiving neuromuscular blocking drugs. Choices of sedation and/or analgesic regimens were made at the discretion of the attending physician, independent of the BIS.
Clinical assessment of sedation was determined by simultaneous recording of three sedation scales (Table 1). The PICU scale was developed for and is routinely used in our PICU and is familiar to our PICU nursing staff. The modified Ramsay scale is commonly used, has been previously validated, and assesses sedation by using similar criteria to the PICU scale (9). The tracheal suctioning scale (TSS) more specifically assesses response to a noxious stimulus (endotracheal tube suctioning) and also effectively evaluates sedation in children (10). The TSS was recorded only when suctioning was performed. The BIS was obtained with the BIS monitor (model A-2000; Aspect Medical Systems, Newton, MA). After electrode placement above the bridge of the nose, over the temple area, and between the corner of the eye and the hairline, the monitor undergoes automatic impedence testing to ensure acceptable signal reception. When an inadequate signal is sensed, the display changes color pattern, allowing easy differentiation between true and spurious values. Electrodes were repositioned or replaced if impedances increased sufficiently as to impair EEG evaluation.
Simultaneous recording of the clinical sedation scores, sedative drug(s), and the BIS value was performed by the PICU bedside nurse every 2–4 h. Although nursing staff were not formally blinded to the BIS, the sedation assessment was performed before recording the BIS value to avoid bias. In addition, although staff were trained to interpret the quality of the BIS signal, they received no instruction on the significance of the actual BIS value. Data were collected until extubation or, to avoid sampling bias, for a maximum of 5 days.
The median and interquartile range of the BIS values and sedation scores were calculated. To determine whether the correlation between the BIS and the sedation score was affected by the sedative used, we compared the BIS values and sedation scores for patients sedated with midazolam and fentanyl with those patients sedated with propofol. Because the ultimate goal of sedation assessment is to determine the adequacy of sedation, we evaluated the ability of the BIS to distinguish among three groupings of each sedation scale, designed to differentiate inadequate, adequate, or excessive sedation. Because acceptable depths of sedation can vary among patients, two groupings were made based on the authors’ consensus as to what would constitute acceptable levels of activity or responsiveness. When lighter levels of sedation were appropriate, under-sedation was defined as PICU, Ramsay, and TSS scores of 1, adequate sedation as PICU scores of 2 or 3, Ramsay scores of 2, 3, or 4, and TSS scores of 2, 3, or 4, and over-sedation as a PICU score of 4, Ramsay scores of 5 or 6, and TSS score of 5. When deeper levels of sedation were required, under-sedation was defined as PICU or TSS scores of 1 or 2, and Ramsay scores of 1, 2, or 3; adequate sedation as a PICU score of 3, Ramsay scores of 4 or 5, and TSS scores of 3 or 4; and over-sedation as a PICU score of 4, Ramsay score of 6, and TSS score of 5.
Variables were described using the median and interquartile range. Correlation between BIS values and sedation scores was determined by linear regression analysis. Between-group analysis was performed using the Kruskal-Wallis one-way analysis of variance. A P value < 0.05 was considered significant. The sensitivity and positive predictive value of various BIS values were calculated to determine the ability of the BIS to differentiate among adequate, inadequate, and excessive levels of sedation.
Four hundred twenty-eight sample sets were recorded from 24 patients aged 1 mo to 20 yr (5.7 ± 6.1). Underlying illnesses included upper airway obstruction (n = 7), sepsis (n = 5), acute respiratory distress syndrome (n = 4), cardiac surgery (n = 3), pneumonia (n = 2), asthma (n = 1), hypertensive encephalopathy (n = 1), and closed head injury (n = 1). Patients were studied for 45.6 ± 35.8 h (range 10–150 h) and 18 ± 14 sample sets were recorded for each patient. Sedation was primarily provided with infusions (±boluses) of midazolam and fentanyl (n = 16), propofol (n = 6), lorazepam (n = 1), or fentanyl (n = 1). Three patients received additional sedation with pentobarbital or chloral hydrate. In three patients, data collection was temporarily suspended when neuromuscular blockade became clinically necessary.
No significant adverse effects were associated with prolonged placement of the BIS electrode. In several patients, the lead required replacement when patient movement caused it to become dislodged or when increased impedence consistently prevented recording of a true BIS value. In one patient, the same lead was used for the entire 5-day monitoring period. Although electrode removal revealed erythematous indentations in the skin, there was no skin breakdown and the lesions healed rapidly.
The correlation between sedation score and BIS values is shown in Table 2. For each sedation scale, the BIS value correlated with increasing depth of sedation (P < 0.001). Regression analysis comparing the BIS value with clinical sedation score revealed r2 values of 0.21, 0.12, and 0.08 for the PICU, Ramsay, and TSS scores, respectively.
To evaluate whether the BIS could distinguish among more clinically relevant levels of sedation, each sedation scale was divided into three groups (under-sedated, adequately sedated, and over-sedated) as described above. For both divisions, there was a significant decrease in BIS value with increasing depth of sedation (P < 0.0001). Within each sedation category, BIS values did not vary between sedation scales.
Because the BIS value may vary with the sedative used, we compared the BIS values obtained during sedation with either a midazolam/fentanyl combination or propofol. Clinically, patients were more deeply sedated during propofol compared with midazolam/fentanyl sedation regimens (P < 0.001, Table 3). At individual sedation scores, there were no differences in BIS values between the two groups (Table 3).
To determine the potential utility of the BIS when clinical examination may be unavailable, we calculated the sensitivity and predictive value of various BIS values to differentiate among inadequate, adequate, and excessive levels of sedation using the divisions defined for when a deeper level of sedation would be desired. Values are reported in the order of PICU, Ramsay, and TSS scale. To differentiate adequate from inadequate levels of sedation, a BIS value of 70 had a high sensitivity (0.89, 0.89, 0.87) but lower positive predictive value (0.84, 0.68, 0.82). Using a BIS value of ≤60, the sensitivity decreased slightly (0.78, 0.81, 0.75) whereas the positive predictive value increased (0.87, 0.73, 0.85). To differentiate adequate from excessive levels of sedation, a BIS value of ≤40 had both a low sensitivity (0.48, 0.55, 0.50) and positive predictive value (0.55, 0.38, 0.08). Increasing this value to 50 improved the sensitivity (0.75, 0.70, 0.67) but did not alter the positive predictive value (0.52, 0.35, 0.07).
The need for appropriate sedation during pediatric mechanical ventilation is well recognized. In addition to the relief of fear, anxiety, pain, and suffering, such measures facilitate patient-ventilator synchrony, prevent inadvertent removal of invasive devices, and may positively impact outcome (11). However, there is still uncertainty regarding optimal definitions for, and assessment of, appropriate sedation. This distinction is important, because over-sedation may also be associated with increased morbidity (1,12,13). Within the PICU, sedation is usually assessed via numeric sedation scales based on patient appearances, responses to stimulation, and physiologic variables. However, these scores are semi-objective and become ineffective during neuromuscular blockade. Applications of the EEG, such as the BIS, may allow more objective assessment of sedation. The BIS accurately predicts intraoperative levels of anesthesia in both adults (14,15) and children (3), and preliminary studies suggest that it may be useful in the adult ICU (7,8).
We report, for the first time, the use of the BIS in critically ill children and demonstrate its reliable use for prolonged periods of time. Similar to previous reports (5–8,14,15), the BIS correlated with the clinically assessed level of sedation and did not vary with the drug used for sedation. The BIS reliably differentiated between inadequate and adequate sedation, but was relatively insensitive for differentiating between adequate and over-sedation.
Although sedation scales use a varying number of actual scores, their most clinically relevant function is differentiation among inadequate, optimal, or excessive sedation. Therefore, comparison of the BIS with these scales is dependent on the criteria used, a priori, to define these categories. No such definitions have been published for use in the PICU, probably because sedation requirements vary greatly within this patient population. In general, older and/or more stable patients do well with lighter levels of sedation, whereas younger and/or sicker patients require deeper sedation. Therefore, we analyzed the BIS against two divisions of the sedation scales, designed to simulate both of these scenarios. In both divisions, the BIS value decreased with increasing depth of sedation, although differences among individual groups were not always significant, likely because of the large range of values in each category. The reasons for this variability are unclear, but the finding is consistent with reports evaluating the BIS in critically ill adults (7,8).
Most of the high BIS/deep sedation data points occurred during night-time hours and it has been suggested that the reduction in the BIS during natural sleep may not be as great as that seen during pharmacologic sedation (16). This seems to contrast with another report that, in healthy volunteers, changes in the BIS during natural sleep stage progression mimic those seen during general anesthesia (17). However, the background EEG pattern in critically ill children may be altered (18) and the effects of these alterations on the BIS, including during sleep, remain uninvestigated. In addition, to encourage retention of a day-night cycle, attempts are made to decrease patient stimulation during the night shift so stimulation during these assessments may have been less vigorous, causing overestimation of the clinical sedation score. In support of this, BIS values for the three “unresponsive” sedation scores were much lower for the TSS compared with the PICU or Ramsay scales. Because this scale uses a more potent noxious stimulus (endotracheal tube suctioning) than the other two, it should be more effective at verifying true patient unresponsiveness.
Conversely, individual scores within a scale are not always mutually exclusive. For example, a patient may appear awake and undistressed, suggesting a Ramsay or TSS score of 2, but be relatively unresponsive to physical stimulation, suggesting a higher score. This may be particularly true when the sedation regimen includes opioids, which should decrease the response to physical stimulus but are less sedating than benzodiazepines or propofol.
One limitation of this study is that the sedation assessments were performed by many individuals and it is possible that interobserver variability accounts for some of the variation noted. However, similar variability was reported in studies using a single observer evaluating the BIS in critically ill adults (7,8). This suggests that the variability we found should not all be attributed to intraobserver variability but may be, at least in part, a function of critical illness.
An additional limitation of this study is the fact that the nurse performing the sedation assessment was not formally blinded to the BIS, introducing the potential for a recording bias. Therefore, we introduced a number of measures to limit this possibility. First, we required that the sedation assessment be performed before the BIS value was recorded, removing the possibility that the sedation assessment would be biased by the BIS value. Second, although the nurses were oriented to the BIS before starting the study, this orientation was limited to the technical aspects of the machine, including differentiation between true and spurious BIS signals. There was no instruction regarding the interpretation of the actual BIS value itself, decreasing the possibility that inadvertent recording of the BIS before the sedation assessment would have affected this assessment. Finally, we refrained from defining target sedation scores or BIS values for the duration of the study, ensuring that clinical decision making and sedation assessments were made independently of stated protocol goals. Although not foolproof, we believe that these measures should have successfully removed the majority of recording bias.
A primary goal of this study was to determine whether the BIS might reliably differentiate adequate from inadequate or excessive levels of sedation, and thereby be useful when the clinical examination is lost, such as during neuromuscular blockade. Because anxiety is likely to be increased if the ability to respond and interact with one’s surroundings is lost (19), and because neuromuscular blockade tends to be reserved for the sickest patients, our present practice is to aim for deep sedation during neuromuscular blockade. By using the definitions outlined above for deep sedation, our data suggest that most (80% or 85%) patients are adequately sedated when BIS values are maintained at less than 70 or 60, respectively. Although our definitions are somewhat arbitrary, they were developed with the goal of differentiating consciousness from unconsciousness as previously defined by Ramsay et al. (9). Also, although it is possible that amnesia and adequate comfort may still occur without complete loss of consciousness, one must weigh this against the likelihood of excessive anxiety and recall.
To avoid under-sedation during neuromuscular blockade, practitioners may choose to err on the side of caution, resulting in over-sedation and the potential for increased morbidity (1). Therefore, we also evaluated the ability of the BIS to detect clinically excessive sedation. Using a BIS value of <40, said to indicate a deep hypnotic state, the BIS was a poor predictor of over-sedation. Less than half of the sedation scores recorded at these values correlated with clinically excessive sedation and almost half of the excessively sedated patients had BIS values >40. Although using a BIS value of ≤50 significantly increased the sensitivity to detect over-sedation, most patients with BIS values <50 actually had clinically appropriate levels of sedation. These data suggest that maintaining BIS values between 50 and 70 should provide adequate sedation for the majority of PICU patients and limit the likelihood of inadequate or excessive sedation. However, the lack of data correlating clinical sedation scores to awareness, anxiety, and amnesia remains concerning, as is the lack of indicators of optimal sedation during neuromuscular blockade. Improved information should allow better targeting of BIS values during critical illness.
The differential effects of anesthetics, sedatives, and hypnotics on the EEG are well known. Therefore, it is important to know whether the choice of sedative has an effect on the correlation between BIS value and level of sedation. Twenty-two of the 24 patients in our study were sedated with either a combination of midazolam and fentanyl or with propofol. For both drugs, BIS values correlated equally well with sedation level and this correlation was independent of sedative regimen, consistent with previous reports performed during anesthesia (6,20). However, questions remain regarding the effect on the BIS of adding opioids. Whereas Sebel et al. (2) found the BIS to be a good predictor of movement at incision when anesthesia was primarily maintained with hypnotic drugs, the correlation became less significant when opioids were added. Similarly, Iselin-Chaves et al. (14) reported that the addition of alfentanil to propofol anesthesia had no effect on the baseline BIS value, but blunted the increase in BIS that occurred upon painful stimulus, a finding attributed to the fact that opioids provide little sedation or amnesia when used at analgesic concentrations. Glass et al. (6) also reported minimal reductions in BIS values with large-dose alfentanil infusions. In contrast, Strachan and Edwards (21) reported a dose-dependent reduction on the BIS with remifentanil. One patient in our study was sedated solely with fentanyl. His BIS values were low (38) and he was deeply sedated clinically. Additionally, in some patients sedated with midazolam/fentanyl combinations, changes in both sedation scores and BIS values seemed to be associated with increases in the fentanyl infusion, suggesting that, in critically ill children, opioids do provide sedation and may alter the BIS. Further study evaluating opioid-specific effects on the BIS and comparing these effects with sedative/hypnotic-based sedation would be valuable.
Finally, it is interesting that the BIS values we recorded seem lower than those reported for similar levels of sedation in anesthesia-based studies. These findings are similar to those of De Deyne et al. (7) who reported average BIS values of 34 in 18 critically ill adults deeply sedated with morphine and midazolam. Therefore, one must consider that critical illness itself may alter the BIS and that target BIS values may differ between anesthesia and critical care-based applications.
In conclusion, we have demonstrated that the BIS monitor may be effectively used for prolonged periods in sedated, mechanically ventilated, PICU patients and that the BIS correlates with clinical sedation scores. The BIS monitor effectively differentiates clinically adequate from inadequate sedation but is less sensitive to predicting over-sedation. Although further study is necessary, the BIS may be a useful adjunct to monitor and guide sedative use when the clinical examination is unavailable.
1. Cammarano WB, Pittet JF, Weitz S, et al. Acute withdrawal syndrome related to the administration of analgesic and sedative medications in adult intensive care unit patients. Crit Care Med 1998; 26: 676–84.
2. Sebel PS, Lang E, Rampil IJ, et al. A multicenter study of bispectral electroencephalogram analysis for monitoring anesthetic effect. Anesth Analg 1997; 84: 891–9.
3. Denman WT, Swanson EL, Rosow D, et al. Pediatric evaluation of the bispectral index (BIS) monitor and correlation of BIS with end-tidal sevoflurane concentration in infants and children. Anesth Analg 2000; 90: 872–8.
4. Wuesten R, Van Aken H, Glass PS, Buerkle H. Assessment of depth of anesthesia and postoperative respiratory recovery after remifentanil- versus alfentanil-based total intravenous anesthesia in patients undergoing ear-nose-throat surgery. Anesthesiology 2001; 94: 211–7.
5. Liu J, Singh H, While PF. Electroencephalographic bispectral index correlates with intraoperative recall and depth of propofol-induced sedation. Anesth Analg 1997; 84: 185–9.
6. Glass PS, Bloom M, Kearse L, et al. Bispectral analysis measures sedation and memory effects of propofol, midazolam, isoflurane, and alfentanil in healthy volunteers. Anesthesiology 1997; 86: 836–47.
7. De Deyne C, Struys M, Decruyenaere J, et al. Use of continuous bispectral EEG monitoring to assess the depth of sedation in ICU patients. Intensive Care Med 1998; 24: 1294–8.
8. 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: 1499–1504.
9. Ramsay M, Savege T, Simpson BRJ, et al. Controlled sedation with alphaxalone/alphadolone. Br Med J 1974; 2: 656–9.
10. Parkinson L, Hughes J, Gill A, et al. A randomized controlled trial of sedation in the critically ill. Paediatr Anaesth 1997; 7: 405–10.
11. Anand KJS, Hickey PR. Halothane-morphine compared with high-dose sufentanil for anesthesia and post-operative analgesia in neonatal cardiac surgery. N Engl J Med 1992; 326: 1–9.
12. Prielipp RC, Coursin DB, Wood KE, Murray MJ. Complications associated with sedative and neuromuscular blocking drugs in critically ill patients. Crit Care Clin 1995; 11: 983–1003.
13. 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.
14. Iselin-Chaves IA, Flaishon R, Sebel PS, et al. The effect of the interaction of propofol and alfentanil on recall, loss of consciousness, and the bispectral index. Anesth Analg 1998; 87: 949–55.
15. Vernon JM, Lang E, Sebel PS, Manberg P. Prediction of movement using bispectral electroencephalographic analysis during propofol/alfentanil or isoflurane/alfentanil anesthesia. Anesth Analg 1995; 80: 780–5.
16. Venn R, Cusack RJ, Rhodes A, Grounds RM. Monitoring of the depth of sedation in the intensive care unit. Clin Intensive Care 1999; 10: 81–90.
17. Sleigh JW, Andrzejowski J, Steyn-Ross A, Steyn-Ross M. The bispectral index: a measure of depth of sleep? Anesth Analg 1999; 88: 659–61.
18. Si Y, Gotman J, Pasupathy A, et al. An expert system for EEG monitoring in the pediatric intensive care unit. Electroencephalogr Clin Neurophysiol 1998; 106: 488–500.
19. Parker MM, Schubert W, Shelhamer JH, Parillo JE. Perceptions of a critically ill patient experiencing therapeutic paralysis in an ICU. Crit Care Med 1984; 12: 69–71.
20. Flaishon R, Windsor A, Sigl J, Sebel PS. Recovery of consciousness after thiopental or propofol. Anesthesiology 1997; 86: 613–9.
21. Strachan AN, Edwards ND. Randomized placebo-controlled trial to assess the effect of remifentanil and propofol on bispectral index and sedation. Br J Anaesth 2000; 84: 489–90.