A failure of M-Entropy[TM] to correctly detect burst suppression leading to sevoflurane overdosage.
A 37-year-old woman with grade 4 placenta praevia presented for elective caesarean section under general anaesthesia. Anaesthesia was induced with propofol 200 mg intravenously, followed by suxamethonium 100 mg intravenously. The patient coughed as cricoid pressure was applied and made semi-purposeful movements. A further 100 mg of propofol was given intravenously. The patient's trachea was intubated and her lungs ventilated with sevoflurane, nitrous oxide and oxygen. After intubation she was hypertensive (175/110 mmHg), tachycardic (145 bpm), lacrimating profusely and had widely dilated reactive pupils. An M-Entropy[TM] sensor was applied using the recommended frontal montage. The response entropy value was 98. Initially only the numerical value was displayed on the monitor screen. In response to these clinical signs and high response entropy, the inspired sevoflurane concentration was increased to 6% and the patient given midazolam 3 mg intravenously. The patient subsequently became hypotensive and two boluses of phenylephrine 100 [micro]g were administered intravenously to restore normal haemodynamic parameters (heart rate 100 bpm and blood pressure 90/50 mmHg). Fifteen minutes after induction the monitor screen was altered to display the processed EEG waveform, which showed a flat line indicating profound burst suppression. At this time the numerical value of the response entropy was 98 but shortly thereafter dropped abruptly to 2, with no apparent change in waveform or further administration of drugs. Surgery commenced and the baby was born five minutes later, apnoeic and hypotonic. Prompt intubation by the consultant anaesthetist resulted in successful resuscitation. The neonate was taken to the newborn unit where it was extubated and observed for a few hours before being returned to the parents with no apparent evidence of ongoing sequelae. The patient was carefully interviewed immediately after anaesthesia and the day after surgery. She had no explicit recall of any intraoperative events.
This case highlights the importance of being able to interpret the raw EEG waveform in the context of the clinical picture (4,5). The patient initially showed clinical signs of being under-anaesthetised with a corresponding high response entropy value. The delivery of sevoflurane was increased appropriately. After a period of time in which the partial pressure of sevoflurane in the brain would have equilibrated with the alveolar partial pressure (6,7), the patient showed clinical signs of adequate anaesthesia and had an end-tidal sevoflurane concentration of more than 1 minimum alveolar concentration, but the entropy monitor still indicated an awake state. At this point the correct response by the anaesthetist would have been to assess the raw EEG waveform and probably disregard the numerical value of the response entropy.
The response entropy varies between 0 (flat line EEG) and 100 (the awake state) (8). The state entropy is derived from the frequency band 0.8 to 32 Hz and includes mainly EEG with a small amount of frontal electromyogram activity. The shortest time epoch for state entropy analysis is 15 seconds. The response entropy is derived from the frequency band 0.8 to 47 Hz and includes both EEG and frontal electromyogram activity. The shortest time epoch for response entropy analysis is 1.92 seconds (over frequency range 32 to 47 Hz) (9). When burst suppression occurs, the entropy module treats the part of the signal that contains the suppressed EEG as a regular signal with entropy of zero, with the entropy of the bursts being calculated in the usual way. This is done over a one-minute time interval. The algorithm to detect the presence of suppression is as follows: the baseline fluctuations are eliminated by subtracting a local average from each signal sample (the local average is obtained by computing a weighted average in a pre-determined window, using a sliding one second window) (10). The signal is then divided into two frequency bands by filters: a high pass filter (>75 Hz) to detect artefact and a low pass filter (<25 Hz) to detect burst suppression. Signal sampling is at 200 Hz. A non-linear energy operator is applied to estimate signal power in both bands for each 0.05 second epoch (9). Suppression is detected if: 1) the signal power is below a fixed threshold for greater than 0.05 seconds and 2) artefacts are not present. A burst suppression ratio is then calculated as the percentage of 0.05 second epochs that were considered suppressed in the last 60 seconds. We note that the non-linear energy operator method used by the M-Entropy[TM] module has been shown to classify EEG segments correctly only 94% of the time during propofol anaesthesia and 92.8% of the time with thiopentone (10). In our case the M-entropy module failed to correctly detect suppressed epochs. To use the monitor safely we need to understand why it failed. From the discussion above, there are at least two possibilities.
[FIGURE 1 OMITTED]
1. The power of the suppressed (flat line) part of 1. 'EEG signal' was greater than the threshold. This might arise from either electrical interference or a variety of true physiological reasons. Clearly the threshold in this monitor is set too low for some patients and situations.
2. Artefacts are detected as being present, which 2. prevents the algorithm recognising the burst suppression. Is the M-Entropy algorithm too sensitive to artefacts?
We did not have a recording of the raw and processed EEG from this case, but we have noted this phenomenon in quite a number of other cases. A similar example that was able to be recorded is shown in Figure 1.
As shown in Figure 1, our calculation of the non-linear energy operator did show a progressive decrease in its value during the period of EEG burst suppression. Presumably, at the point when it dropped below a threshold value (of about 22 [micro]V) it triggered the burst suppression recognition in the monitor. This is a plausible but unproven explanation for our original case.
The one minute processing time used to detect burst suppression results in a short delay in the entropy numerical value reading, but does not explain the approximate eight minute period of burst suppression on EEG, seen in Figure 1, that was undetected by the M-entropy algorithm.
In summary, this case highlights the limitations of the M-Entropy[TM] depth of anaesthesia monitoring system in the detection of the burst suppression pattern. The overall outcome of this case was good, with mother and baby doing very well. However, it serves to illustrate the pitfalls of being guided by a single monitor to gauge depth of anaesthesia. As with all patient monitors, the information provided by an M-entropy[TM] monitor must be interpreted within the clinical context and questioned when it steers the anaesthetist in a direction that would not be their usual practice.
Accepted for publication on May 20, 2009.
(1.) Myles PS, Leslie K, Forbes A, Chan MTV. Bispectral index monitoring to prevent awareness during anaesthesia: the B-Aware randomised controlled trial. lancet 2004; 363:1757-1763.
(2.) Avidan MS, Zhang L, Burnside BA, Finkel KJ, Searleman AC, Selvidge JA et al. Anaesthesia Awareness and the Bispectral Index. N Engl J Med 2008; 358:1097-1108.
(3.) Ekman A, Linholm M-L, lennmarken C, Sandin R. Reduction in the incidence of awareness using BIS monitoring. Acta Anaesthesiol Scand 2004; 48:20-26.
(4.) Barnard JP, Bennett C, Voss LJ, Sleigh JW. Can anaesthetists be taught to interpret the effects of general anaesthesia on the electroencephalogram? Comparison of performance with the BIS and spectral entropy. Br J Anaesth 2007; 99:532-537.
(5.) Bennett C, Voss LJ, Barnard JPM, Sleigh JW. Practical use of the raw electroencephalogram waveform during general anaesthesia: the art and the science. Anesth Analg 2009; 109:539-550.
(6.) Olofsen E, Sleigh JW, Dahan A. Permutation entropy of the electroencephalogram: a measure of anaesthetic drug effect. Br J Anaesth 2008; 101:810-821.
(7.) Soehle M, Ellerkmann RK, Grube M, Kuech M, Wirz S, Hoeft A et al. Comparison between bispectral index and patient state index as measures of the electroencephalographic effects of sevoflurane. Anesthesiology 2008; 109:799-805.
(8.) Vakkuri A, Yli-Hankala A, Talja P, Mustola S, Tolvanen-Laakso H, Sampson T et al. Time frequency balanced spectral entropy as a measure of anaesthetic drug effect in central nervous system during sevoflurane, propofol, and thiopental anaesthesia. Acta Anaesthesiol Scand 2004; 48:145-153.
(9.) Viertio-Oja H, Maja V, Sarkela M, Talja P, Tenkanen N, Tolvanen-laakso H et al. Description of the Entropy[TM] algorithm as applied in the Datex-Ohmeda S/5[TM] Entropy module. Acta Anaesthesiol Scand 2004; 48:154-161.
(10.) Sarkela M, Mustola S, Seppanen T, Koskinen M, Lepola P, Suominen K et al. Automatic analysis and monitoring of burst suppression in anesthesia. J Clin Monit Comput 2002; 17:125-134.
S. M. HART *, C. R. BUCHANNAN [[dagger]], J. W. SLEIGH [[double dagger]]
Department of Anaesthetics, Waikato Hospital, Hamilton, Waikato, New Zealand
* M.B., Ch.B., Anaesthetic Registrar.
[[dagger]] M.B.,Ch.B., F.A.N.Z.C.A, Anaesthetic Consultant.
[[double dagger]] M.B., Ch.B., M.D., F.A.N.Z.C.A, Anaesthetic Consultant.
Address for correspondence: Dr S. Hart, Waikato Hospital, Pembroke Street, Private Bag 3200, Hamilton 3240, New Zealand. Email: firstname.lastname@example.org
|Printer friendly Cite/link Email Feedback|
|Author:||Hart, S.M.; Buchannan, C.R.; Sleigh, J.W.|
|Publication:||Anaesthesia and Intensive Care|
|Date:||Nov 1, 2009|
|Previous Article:||Audit of performance of size 1.5 ProSeal laryngeal mask airways in infants less than six months undergoing inguinal herniotomy.|
|Next Article:||Thyroid storm presenting as intra-abdominal sepsis with multi-organ failure requiring intensive care.|