Nitrous oxide (N2O) has been used clinically for more than a century as an analgesic, sedative, and hypnotic agent. Whereas the majority of IV and volatile anesthetics induce a range of stereotypical dose-dependent changes in the human electroencephalogram (EEG),1 which have been exploited to monitor their clinical action, N2O, in contrast, is generally thought to produce atypical changes in the EEG.
Most processed EEG methods fail to correctly capture the dose-dependent effects of N2O.1–7 Despite the absence of clear empirical evidence, the basis for this failure is generally thought to be attributable to N2O paradoxically triggering high-frequency EEG activity indicative of arousal. For example, Rampil et al.2 reported increased power in the high β (40–50 Hz) and low-EMG (70–110 Hz) spectral bands in response to the sole administration of N2O in healthy participants. Similarly, Yamamura et al.8 reported that the sole administration of 50% to 70% N2O was associated with increased activity in the high β band (peaked at 34 Hz) but reduced α band (8–13 Hz) activity. Thus, the use of adjuvant N2O is believed to impair the utility of EEG indices of depth of anesthesia (DOA) by falsely enhancing EEG measures of awareness.9 However, confirming these effects while controlling for experimental confounds such as electromyographic interference remain largely uninvestigated, particularly in healthy cohorts.
To clarify the paradoxical effects of N2O on resting EEG activity, the present study quantified changes in EEG spectral power during sole N2O inhalation in healthy participants during multichannel EEG recordings in a noise-minimized laboratory setting.
Healthy male participants were recruited for participation in this study, which was approved by the Swinburne University of Technology Human Research Ethics Committee. Female participants were excluded because of an increased likelihood of nausea and emesis, along with difficulties in controlling for the effects of menstrual cycle and contraceptive pill on resting EEG.10,11 Study inclusion required passing a general medical examination, performed by a registered physician, and the absence of any history of psychiatric or neurological illness. All participants provided informed, voluntary, written consent and were free to withdraw from the study at any point.
EEG recordings were performed using a 62-channel SynAmps™ EEG system (NeuroScan; Compumedics Ltd., Melbourne, Australia). The EEG montage was positioned according to the extended international 10:20 system, with a linked mastoids reference. Electrode impedance was maintained below 10 kΩ for all recordings, which were acquired using Scan 4.3™ (NeuroScan) with a sample rate of 500 Hz, bounded by a digital bandpass filter (0.1–70 Hz) in a noise-minimized recording laboratory.
N2O and O2 were administered through a closed 1.5-m Bain coaxial nonrebreathing circuit. Oxygen saturation and heart rate were obtained using a standard clinical pulse oximeter finger clip. Expired gas was analyzed online using a Normocap 200™ (Datex-Ohmeda, Helsinki, Finland) infrared gas analyzer to determine the end-tidal concentrations of N2O, O2, and CO2. The outputs from both the pulse oximeter and gas analyzer were manually logged once every minute during the gas condition by the experimenter.
To monitor behavioral state and minimize the interparticipant variability of spontaneous resting EEG and cognitive state, participants were asked to perform a low-effort auditory continuous performance task during gas inhalation. Auditory-based measures of vigilance are particularly well suited to studies of sedation/anesthesia in healthy volunteers,12 with their use consistent with the methods of valid pharmaco-EEG profiling.13 Participants had to respond (left or right button) to 2 auditory tones of differing frequency (1 or 2 kHz, respectively), but of fixed stereo amplitude (70 dB) occurring every 2.5 seconds during baseline and gas recordings. Targets and responses (correctness and latency) were automatically logged for each trial.
Participants were subject to only a single N2O condition. Before the recording session, each participant was randomly allocated to 1 of 3 conditions: 20%, 40%, or 60% inspired N2O/O2, respectively. All testing sessions commenced at 9 am and continued for approximately 2 hours. Participants were asked to fast for 8 hours before participation.
Baseline involved a 5-minute eyes-closed recording of spontaneous EEG during the auditory task. Recordings of EEG during N2O inhalation were 20 minutes in duration. This period comprised a 5-minute equilibration phase after which a marker was placed in the EEG trace to signify the beginning of a 10-minute period of continuous equilibrated gas flow. After this 10-minute period had elapsed (15 minutes total), N2O was discontinued with pure O2 (100%) administered for a further 5 minutes (20 minutes total).
All EEG recordings were subject to postprocessing using MATLAB™ (The MathWorks, Natick, MA) to remove noise and any recording artifacts. Artifact-rejected raw 62-channel EEG was subsequently bandpass filtered between 1 and 40 Hz. Any remaining artifact such as eye blinks and electromyogram were successfully removed using independent-component analysis as implemented in the EEGLAB toolbox.14 For this study, a subset of electrodes was inspected taking the Fpz, Cz, and Oz electrodes from the central-longitudinal axis (referred to here as the “midline-montage”).
Quantitative EEG comparisons between rest and gas were made using the 5-minute baseline condition (rest) and the final 5 minutes (i.e., approximately 600–900 seconds) of the respective gas condition (gas) as the latter would correspond to maximal gas effect. Additionally, comparisons were made with the washout period (end) after the cessation of N2O (last 3 minutes of recording, after removing 30 seconds from data end to avoid transients induced at termination of the EEG recording). For estimating spectral values, Welch's method was used by calculating spectral power via the fast Fourier transform for short 2-second Hamming windows with 50% overlap and taking the average of these estimates over the time window of interest.
All omnibus tests were performed using a mixed-effects analysis of variance (ANOVA) in which the fixed effect was the inspired level of N2O and the random effect was the time of observation (rest, peak gas, gas washout). To the best of our knowledge, there is no nonparametric alternative to a mixed-effects ANOVA. Therefore, departures from the underlying parametric assumptions of normality, homoscedasticity, and sphericity were dealt with using standard approaches. In particular, although transformations could be used to minimize data skewness and to improve normality and homoscedasticity, such transformations are not typically applied to quantitative EEG data. Further skewness (<3) and departures from homoscedasticity for all data were found to be mild. ANOVA methods are robust to such mild departures.15 Violations of sphericity (compound symmetry) were dealt with using Greenhouse-Geisser corrections. As required, post hoc t tests were performed, modified for multiple comparisons. Because of discontinuation of the 60% condition, statistical analysis was restricted to the 20% and 40% conditions. All statistical analyses were performed using SPSS for Windows version 16 (SPSS, Inc., Chicago, IL). A value of P < 0.05 was considered statistically significant.
During the study, a number of participants had adverse responses to the 60% N2O condition, which included nausea and emesis both during and after recording. There was also 1 case of emesis in the 40% condition. Because of the high incidence of adverse effects (3 of 5 tested participants), the 60% condition was discontinued for the remainder of the project as required by the local institutional research ethics guidelines regarding healthy volunteers. Ten participants were successfully tested for each of the 20% (mean [SD]: 26 years [3 years]; 71.3 kg [8.7 kg]; 177.4 cm [8.1 cm]) and 40% (25.7 years [5 years]; 73 kg [13.3 kg]; 174.1 cm [6.8 cm]) N2O conditions. Participants with incomplete or erroneous data were excluded from analyses where appropriate. In most cases, the remaining data from the 60% condition (n = 2; 23.7 years [3.1 years]; 72.7 kg [24.42 kg]; 177.3 cm [2.1 cm]) are presented for qualitative comparison only.
Mixed-effect ANOVAs revealed that 20% and 40% N2O significantly reduced auditory vigilance task-performance accuracy (F[1,14] = 5.455, P = 0.035) and response latency (F[1,14] = 5.301, P = 0.037). Mean accuracy decreased and mean response latency increased for both 20% (accuracy: ↓2.2%; latency: ↑0.13 seconds) and 40% (accuracy: ↓12.2%; latency: ↑0.38 seconds) N2O, respectively. Although there were insufficient numbers for the 60% condition to be included in the statistical analysis, the changes in task performance attending this condition were on average consistent with the mixed ANOVA results (accuracy: ↓55%; latency: ↑1 second).
Mixed-effect ANOVAs revealed that end-tidal CO2 levels (F[2,13] = 3.388, P = 0.076), O2 saturation (F[2,13] = 0.594, P = 0.486), and heart rate (F[2,13] = 2.750, P = 0.081) did not significantly differ across N2O conditions or time of observation (rest, gas, and end), indicating that any significant changes in EEG activity were attributable to N2O and not respiratory (hypercapnia or hypoxia) or autonomic effects. Given that all N2O treatment groups were matched for age, height, and weight and were associated with no discernible difference on a range of physiological variables, observed changes in EEG activity can be assumed to have occurred in pharmacokinetically equivalent participants.
Median Power Frequency, Spectral Edge Frequency, and Total Power
Estimates of spectral edge frequency 95% (SEF95), median power frequency (MPF), and total power are standard indicators of bulk spectral change and are often used in DOA research.16 These respective estimates define the value below which 50% (MPF) or 95% (SEF95) of the total spectral power decreases. Table 1 shows the means and standard deviations for MPF, SEF95, and total power across the midline montage.
Mixed-effect ANOVAs revealed that N2O exposure caused significant changes across states (rest, gas, end) in MPF at both midline frontal (Fpz) (F[2,13] = 4.23, P = 0.043) and central (Cz) (F[2,13] = 8.75, P = 0.001) electrodes. Post hoc t tests revealed that only the MPF in electrode Cz increased significantly (P = 0.002; mean increase [95% confidence interval, CI]: 0.68 Hz [0.26–1.12 Hz]) after N2O exposure. N2O had no effect on MPF of the occipital electrode (P = 0.52). As shown in Table 1, MPF generally increased from rest to gas state and decreased from gas to washout (end). Such changes across exposure condition were not significantly different between 20% and 40% N2O at Fpz (P = 0.44), Cz (P = 0.053), or Oz (P = 0.43).
Significant changes in SEF95 values across states were only observed at Fpz (F[2,13] = 5.74, P = 0.020), which was not significantly different between 20% and 40% N2O (P = 0.33). However, SEF95 values for the 60% condition, although only representing 2 cases, did show marked differences as seen in Table 1. Typically, the trend was again only minor increases in SEF95 from rest to gas condition and decreases from gas to washout. However, post hoc t tests revealed that only reductions in SEF95 at washout were significant (P = 0.003; mean decrease [95% CI]: 2.29 Hz [0.77–3.81 Hz]). The overall stability of SEF estimates (Table 1) suggested that generally the shape of the EEG spectrum was unchanged by N2O. Figure 1 shows representative time courses for MPF and SEF95 at Cz for example 20%, 40%, and 60% N2O participants, respectively.
Interestingly, mean total power was observed to consistently decrease between rest and gas, but rebounded during washout (Table 1). The level-dependent effect of N2O inhalation on total power is strongest frontally (Fpz) and diminishes progressively in midline central (Cz) and occipital (Oz) channels. Mixed-effect ANOVAs revealed that N2O caused significant changes in total power at Fpz (F[2,13] = 7.52, P = 0.007), Cz (F[2,13] = 4.72, P = 0.041), and Oz (F[2,13] = 4.94, P = 0.028) electrodes. These effects were not significantly different between 20% or 40% N2O at Fpz (P = 0.11), Cz (P = 0.74), or Oz (P = 0.33) electrodes. Post hoc tests revealed that only Fpz showed both a significant reduction in total power between rest and gas (P = 0.001; mean reduction [95% CI]: 41.90 μV2 [18.19–65.61 μV2]) and an increase in total power between gas and washout (P = 0.039; mean increase [95% CI]: 45.73 μV2 [2.21–89.25 μV2]).
Figure 2 shows total power values for Fpz, Cz, and Oz over the full 20-minute recording for 3 example cases from the respective gas conditions. The most striking feature of these plots is the large increase in total power soon after the cessation of N2O (≈900 seconds). Indeed, N2O-induced total power reductions are less noticeable when the total power is scaled to include this rebound response. Figure 3 also suggests that even when the participant's end-tidal N2O levels are undetectable (≈1200 seconds), EEG effects are still present.
To further explore this response over all participants, changes in total power were examined during the cessation of N2O for each recording. Figure 3 shows EEG total power for Cz and the mean EEG spectra for Fpz, Cz, and Oz during N2O withdrawal for each gas condition, clearly presenting an increase in power during decreasing levels of N2O over the majority of participants.
N2O reduced the power of the EEG (total power) while in large part preserving its spectral distribution (MPF/95); however, this influence was clearly reversed after the discontinuation of N2O. More qualitatively, total power during N2O washout was enhanced by higher proceeding N2O concentrations. In light of these identified spectral changes, the power in individual frequency bands was investigated.
To investigate the quantitative changes in band-limited power, relative and absolute band powers were calculated and compared for rest, peak gas, and washout recordings over the midline montage. The spectral bands were δ (1–4 Hz), θ (4–8 Hz), α (8–15 Hz), β (15–30 Hz), and γ (30–40 Hz). Figure 4 shows the changes (gas − rest) in relative band powers at Fpz, Cz, and Oz, respectively, for both 20% and 40% N2O. Mixed-effects ANOVAs revealed changes in relative band power across recordings (rest, gas, end) for only δ (F[2,13] = 16.8, P < 0.001) and θ (F[2,13] = 5.93, P = 0.007) bands at Fpz and only the δ band (F[2,13] = 28.14, P < 0.001) for Cz, with no significant change in relative band power observed at Oz. As can be seen in Figure 4, changes in mean relative power were more dominant frontally than occipitally. In particular, relative δ power was reduced whereas all other bands were either marginally enhanced or unchanged (minor increase or decrease <5%). Post hoc t tests revealed pronounced decreases in relative δ power for Fpz for both peak gas (P < 0.001; mean decrease [95% CI]: 12.53% [6.28%–18.79%]) and washout periods (P = 0.007; mean decrease [95% CI]: 11.43% [3.03%–19.81%]) compared with the resting condition. Given that resting relative δ power at Fpz is of the order of 40%, these represent substantial fractional decreases. By contrast, relative θ power for Fpz at washout was significantly increased over resting values (P = 0.033; mean increase [95% CI]: 7.71% [0.58%–14.85%]). Reductions in relative δ power at Cz at peak gas (P < 0.001; mean decrease [95% CI]: 7.19% [4.12%–10.27%]) and washout periods (P < 0.001; mean decrease [95% CI]: 5.69% [2.74%–8.65%]) were somewhat less pronounced.
Changes in relative band power may be the result of variations in the absolute power of a band, or of changes in the absolute power of other bands. For this reason, inspection of absolute band power values was expected to provide additional information regarding the EEG effects of N2O. Figure 5 shows the mean absolute power change (gas − rest) values for Fpz, Cz, and Oz, respectively, for both 20% and 40% N2O. Mean values of absolute power change show different findings when compared with Figure 4, with most bands being suppressed or unchanged. However, statistical analyses showed similar findings to relative power values, with significant changes in absolute band power across recordings (rest, gas, end) observed for only δ (F[2,13] = 15.88, P < 0.001) and θ (F[2,13] = 4.89, P = 0.038) bands at Fpz and only the δ band (F[2,13] = 7.50, P = 0.011) for Cz, with no absolute band power change at Oz. Similar to the previous results for relative band powers, post hoc t tests revealed pronounced reductions in absolute δ power at peak gas (P < 0.001; mean decrease [95% CI]: 31.35 μV2 [16.89–45.80 μV2]) compared with the resting condition. In contrast, reductions in absolute δ power during the washout period compared with the resting period were at the boundaries of significance (P = 0.046; mean decrease [95% CI]: 17.43 μV2 [0.24–34.62 μV2]) in part because of the increase in absolute δ activity at washout compared with peak gas (P = 0.041; mean increase [95% CI]: 13.92 μV2 [0.51–27.33 μV2]). Reductions in absolute δ at Cz at peak gas compared with rest (P < 0.001; mean decrease [95% CI]: 24.95 μV2 [12.07–37.88 μV2]) were similar in magnitude to those at Fpz.
During the inhalation of the putative N-methyl-d-aspartate (NMDA) receptor antagonist N2O,17 it was found that the EEG recorded in healthy male volunteers underwent reductions in total power (amplitude). This reduction was principally the result of selective decreases in δ (1–4 Hz) and θ (4–8 Hz) band power. As a consequence of this reduction, along with the general preservation of the peak α frequency, estimates of relative band power were enhanced for higher frequencies. These observed spectral changes were disproportionately reversed with the withdrawal of N2O, whereby EEG total power increased, primarily through increases in δ and θ band power beyond resting values. These spectral changes were most pronounced at frontal and vertex locations.
Previously reported findings of clear EEG acceleration in response to N2O exposure8 were not replicated by the present study. Therefore, the ostensibly accelerating effects of N2O on the EEG may have less to do with increases in specific high-frequency (>15 Hz) components of the EEG, and more to do with preservation of waking/resting spectra (α) during sedation. Such findings are consistent with reports of N2O's apparently paradoxical influence on the EEG and its confounding effect on the quantitative monitoring of the brain state using the EEG. For example, Anderson and Jakobsson5 were able to detect consistent changes in the EEG (reduced spectral entropy) for hypnotic doses of propofol, but were unable to detect any consistent change for hypnotically comparable levels of N2O. On the basis of this and other studies,2–4,7,18,19 it has been speculated that most methods for EEG-based DOA monitoring are insensitive to N2O because they anticipate loss of “waking” desynchronized EEG and the emergence of “slow wave” somnographic-type EEG, whereas the converse seems to be true for N2O.6 Indeed, our own results add further support to such a conclusion. It should be noted, however, that the admixture of N2O during standard general anesthesia may produce EEG arousal and acceleration by promoting the restoration of desynchronized EEG patterns. Such adjuvant effects are of particular importance because they more closely reflect the standard clinical situation in which N2O is used. More generally, N2O-induced changes in the EEG may confuse estimates of patient state if focus is placed on monitor indices alone rather than in conjunction with the raw waveform.1
Most descriptions in the literature of the effects of N2O on the EEG typically make no mention of withdrawal responses. However, in light of the present findings, a number of earlier studies may be interpreted as providing some anecdotal evidence for such an effect. Perhaps the earliest report of a N2O-induced EEG withdrawal response was from Henrie et al.,20 who reported “overswing” in the EEG seen as high-voltage θ oscillations (4–7 Hz) soon after the removal of the breathing apparatus in some participants (30% N2O). Williams et al.21 also showed in a subset of participants that N2O inhalation produced reductions in EEG amplitude, which were subsequently increased above baseline upon the cessation of gas inhalation. Rampil et al.2 made similar observations in 2 participants who showed a strong N2O withdrawal response that was defined by increased δ power and reduced Bispectral Index (BIS) scores. Such anecdotal evidence was observed to replicate across participants in the present study.
Evidence for cortical excitation after N2O withdrawal is found in animal studies in which rebound hyperexcitability is observed.22,23 Stevens et al.23 showed that seizures were more easily induced via electrical stimulation directly after N2O withdrawal than at rest, even though N2O was shown to be an effective anticonvulsant. Potentially similar to the effects of ethanol, N2O withdrawal excitability seems to be in part a consequence of acute tolerance, such that increasing levels of effective concentrations decrease the seizure threshold upon withdrawal.24–27 N2O's ability to produce acute tolerance suggests it may directly alter the gating kinetics of NMDA receptor action. Chronic exposure of NMDA antagonists produces a strong rebound response not only in neural excitability but also in synaptogenesis in animal models.28 An effect of smaller magnitude proportional to acute exposure may account for the withdrawal response presented herein. A candidate for this effect would be the voltage gating of magnesium ions, potentially being expelled from the channel pore at lower membrane voltages because of enhanced NMDA synaptic activity during N2O withdrawal.
In developing a more standard pharmaco-EEG approach to the study of N2O action, our findings, along with previous research, help to explain the confounding effects of N2O on a variety of EEG-based anesthesia monitoring approaches. DOA methods that rely on quantifying some feature of the EEG power spectrum generally assume that increases in relative and/or absolute high-frequency activity implies increased arousal, while often accounting for disinhibition (biphasic EEG arousal with induction/ emergence). Therefore, the observed reductions in δ and θ band activity during N2O inhalation would anomalously imply increased arousal. This explains the observed return of BIS scores toward fully alert values of 100 with increases in inspired N2O level.2 Second, the withdrawal of N2O produces a large low-frequency response (δ/θ waves), which is enhanced by both the level and duration of the preceding N2O concentration and possibly the rapidity of its removal. The paradoxical decreases in BIS scores during N2O reduction in the Rampil et al.2 investigation may chiefly be accounted for by the “overswing” or “withdrawal” phenomena. This peculiar response of suppression and excitation was presciently suggested2 to produce the paradoxical changes in BIS values, an interpretation supported by the present study. Surprisingly it would seem, based on the present results that a measure as simple as total power or EEG amplitude would behave more consistently as an index of patient state during sole N2O sedation. Finally, it is of note that the frontal dominance of these effects is of particular interest to anesthesia monitoring given the common use of frontal montages for EEG recording during anesthesia.
The failure to acquire sufficient data for the 60% condition, and to perform a repeated-measures design, needs to be addressed in future investigations of N2O action. Along with inclusion of female participants, attempts should be made to study the effects of higher concentrations of N2O if achievable without participant withdrawal in healthy controls. The latter might be achieved with the coadministration of a psychoactively and EEG-neutral antiemetic agent. At these higher concentrations, the psychoactive effects of N2O are likely to further confound attempts to estimate sedation level, in a manner possibly similar to the actions of ketamine.29 Consequently, we believe that improvements in understanding the combined effects of multiple drugs on the EEG is best predicated on clear empirical knowledge of their sole effects.
The ability of our study to suggest the potential reasons for the response of existing DOA monitoring approaches to N2O highlights the importance of characterizing the basic EEG phenomenology of anesthetics. Although the number of studies focusing on the processed EEG during anesthesia has rapidly increased, typically few attempt to clarify the raw EEG responses under investigation. Dissociative agents such as N2O and ketamine provide the perfect case example of the need to focus on controlled replication and to develop a range of consistent and nonproprietary measures to sensitively monitor physiological brain state. Almost all of the commercially available DOA monitors fail to capture the effects of N2O, yet there has been little systematic empirical study of why this is, or what its origin might be. Although N2O may be declining in its clinical use, basic scientific investigation beyond typical clinical conditions and cohorts is necessary for ongoing improvements in intraoperative monitoring and the quantitative monitoring of psychoactive drug effects.
Name: Brett L. Foster, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Brett L. Foster 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: Brett L. Foster consulted for Cortical Dynamics Pty. Ltd.
Name: David T. J. Liley, MBChB, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: David T. J. Liley has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Conflicts of Interest: David T. J. Liley consulted for Cortical Dynamics Pty. Ltd., has equity interest in Cortical Dynamics Pty. Ltd., and received research funding from Cortical Dynamics Pty. Ltd.
This manuscript was handled by: Dwayne R. Westenskow, PhD.
1. Bennett C, Voss LJ, Barnard JP, Sleigh JW. Practical use of the raw electroencephalogram waveform during general anesthesia: the art and science. Anesth Analg 2009;109:539–50
2. Rampil IJ, Jin-Soo K, Lenhardt R, Negishi C, Sessler DI. Bispectral EEG index during nitrous oxide administration. Anesthesiology 1998;89:671–7
3. Barr G, Jakobsson JG, Owall A, Anderson RE. Nitrous oxide does not alter bispectral index: study with nitrous oxide as sole agent and as an adjunct to i.v. anesthesia. Br J Anaesth 1999;82:827–30
4. Barr G, Anderson R, Jakobsson J. The effects of nitrous oxide on the auditory evoked potential index during sevoflurane anesthesia. Anesthesia 2002;57:736–9
5. Anderson R, Jakobsson J. Entropy of EEG during anesthetic induction: a comparative study with propofol or nitrous oxide as sole agent. Br J Anaesth 2004;92:167–70
6. Sleigh JW, Barnard JPM. Entropy is blind to nitrous oxide: can we see why? Br J Anaesth 2004;92:150–61
7. Anderson RE, Barr G, Jakobsson JG. Cerebral state index during anesthetic induction: a comparative study with propofol or nitrous oxide. Acta Anesthesiol Scand 2005;49:750–3
8. Yamamura T, Fukuda M, Takeya H, Goto Y, Furukawa K. Fast oscillatory EEG activity by analgesic concentrations of nitrous oxide in man. Anesth Analg 1981;60:283–8
9. Prabhakar H, Ali Z, Bithal PK, Singh GP, Laithangbam PK, Dash HH. EEG entropy values during isoflurane, sevoflurane and halothane anesthesia with and without nitrous oxide. J Neurosurg Anesthesiol 2009;21:108–11
10. Solis-Ortiz S, Arce JRC, Guevara MA, Corsi-Cabrera M. EEG oscillations during menstrual cycle. Int J Neurosci 1994;76:279–92
11. Kaneda Y, Ikuta T, Nakayama H, Kagawa K, Furuta N. Visual evoked potential and electroencephalogram of healthy females during the menstrual cycle. J Med Invest 1997;44:41–6
12. Purdon PL, Pierce ET, Bonmassar G, Walsh J, Harrell PG, Kwo J, Deschler D, Barlow M, Merhar RC, Lamus C, Mullaly CM, Sullivan M, Maginnis S, Skoniecki D, Higgins HA, Brown EN. Simultaneous electroencephalography and functional magnetic resonance imaging of general anesthesia. Ann NY Acad Sci 2009;1157:61–70
13. Knott V. Quantitative EEG methods and measures in human psychopharmacological research. Hum Psychopharmacol 2000;15:479–98
14. Delorme A, Makeig S. EEGlab: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 2004;134:9–21
15. Schmider E, Ziegler M, Danay E, Beyer L, Buhner M. Is it really robust? Reinvestigating the robustness of ANOVA against violations of the normal distribution assumption. Methodology 2010;6:147–51
16. Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology 1998;89:980–1002
17. Jevtovi-Todorovi V, Todorov SM, Mennerick S, Powell S, Dikranian K, Benshoff N, Zorumski CF, Olney JW. Nitrous oxide (laughing gas) is an NMDA antagonist, neuroprotectant and neurotoxin. Nat Med 1998;4:460–3
18. Bein B. Entropy. Best Pract Res Clin Anesthesiol 2006;20:101–9
19. Liley DTJ, Leslie K, Sinclair NC, Feckie M. Dissociating the effects of nitrous oxide on brain electrical activity using fixed order time series modeling. Comput Biol Med 2008;38:1121–30
20. Henrie JR, Parkhouse J, Bickford RG. Alteration of human consciousness by nitrous oxide as assessed by electroencephalography and psychological tests. Anesthesiology 1961;22:247–59
21. Williams DJ, Morgan RJ, Sebel PS, Maynard DE. The effects of nitrous oxide on cerebral electrical activity. Anesthesia 1984;39:422–5
22. Harper M, Winter P, Johnson B, Koblin D, Eger E. Withdrawal convulsions in mice following nitrous oxide. Anesth Analg 1980;59:19–21
23. Stevens JE, Oshima E, Mori K. Effects of nitrous oxide on the epileptogenic property of enflurane in cats. Br J Anaesth 1983;55:145–54
24. Belknap JK, Laursen SE, Crabbe JC. Ethanol and nitrous oxide produce withdrawal-induced convulsions by similar mechanisms in mice. Life Sci 1987;41:2033–40
25. Vaughn LK, Pruhs RJ. Strain-dependent variability in nitrous oxide withdrawal seizure frequency. Life Sci 1995;57:1125–30
26. Brooks WJ, Weeks AC, Leboutillier JC, Petit TL. Altered NMDA sensitivity and learning following chronic developmental NMDA antagonism. Physiol Behav 1997;62:955–62
27. Yang L, Long C, Faingold CL. Audiogenic seizure susceptibility is induced by termination of continuous infusion of gamma-aminobutyric acid or an N-methyl-D-aspartic acid antagonist into the inferior colliculus. Exp Neurol 2001;171:147–52
28. Brooks WJ, Petit TL, LeBoutillier JC. Effect of chronic administration of NMDA antagonists on synaptic development. Synapse 1997;26:104–13
29. Hirota K. Special cases: ketamine, nitrous oxide and xenon. Best Pract Res Clin Anesthesiol 2006;20:69–79