Distinct Cortical Signatures Associated with Sedation and Respiratory Rate Depression by Morphine in a Pediatric Population

ANALGESIA and sedation are essential components of perioperative care in children and adults.¹  Opioid drugs are used as analgesic agents but present potent sedating properties²  characterized by reduction of motor control, loss of attention, and sleep architecture changes.³  Most opioid drugs exert their physiologic and neurologic effects by binding to opioid receptors at multiple targets in the brain. Action at these targets alters cortical activity and impacts on brain arousal functions. Opioids when combined with anesthetics decrease β power (14 to 30 Hz) recorded in the electroencephalogram,⁴  electroencephalogram coherence (i.e., synchronization of activity between electroencephalogram channels) in the β frequency range that may relate to the loss of stability of sustained attention by opioid drugs.⁵  Similarly, anesthetics produce a powerful suppression of β powers and increase α power and coherence.^6,7  These changes suggest that opioids impact on arousal states. Indeed, behaviorally, opioids reduce time spent in non–rapid eye movement (non-REM) N1 sleep and increase time spent in N2 sleep.⁸  Overall, opioids reduce states of heightened arousal associated with high-frequency electroencephalogram activity,⁹  but the neural circuit-level mechanisms of opioid sedation are unknown. Hence, comparing the changes in electroencephalogram dynamics induced by opioids with those observed at wakefulness and sleep provides insights into the neural circuit mechanisms through which opioids induce sedation.

Although opioids are the mainstay of pain management, they can be associated with potentially hazardous respiratory depression.^8,10  Opioid-related misuse, overdose, and mortality^11,12  are serious health issues, with more than 15,000 deaths/yr due to prescription of opioid analgesics in the United States alone.¹³  In postoperative care, opioids can lead to sleep-disordered breathing¹⁴  and reduced respiratory rate.^10,15  Importantly, in rodent models, respiratory sensitivity to opioids is more pronounced in states of reduced arousal such as sleep or anesthesia than in wakefulness.¹⁶  Consistent with these data, opioid-related respiratory depression may be potentiated by sedation in the perioperative setting.¹⁷  In addition, sedation and respiratory depression are prevalent in young and naive patients,¹⁸  due to their high sensitivity to opioids.^19,20  The relationship between the cortical signatures associated with sedation and respiratory depression by opioids has never been assessed in young patients during postoperative care.

Analysis of the scalp electroencephalogram, a readily accessible measure of the average cortical activity, combined with polysomnography, can provide insights into the mechanisms of sedation and respiratory changes associated with opioids. Here, we performed an observational study to identify the key cortical signatures of sedation by opioids and to determine how these signatures are associated with respiratory changes. We propose that opioids elicit distinct electroencephalogram changes associated with sedation and that these signatures are linked to changes in respiratory activity. We determined the impact of morphine on cortical activity, as measured by electroencephalogram spectral content, topography, and coherence, and how it affects respiratory activity when given to young patients for pain relief after elective surgery. We aimed to answer the following questions: Are there specific cortical changes induced by morphine explaining differences in clinical response and behavior? Can these cortical changes with morphine be detected in the electroencephalogram and are they associated with respiratory changes?

This study was approved by the Research and Ethics Board at the Hospital for Sick Children (Toronto, Ontario, Canada) and written informed consent and assent were obtained.

All suitable patients meeting eligibility criteria were approached in the Ear, Nose, and Throat (ENT) clinic at the Hospital for Sick Children, and the study was discussed with them by an ENT surgeon. Potential participants were identified and contacted by phone by a research coordinator to provide further details of the study. When potential participants and their guardians were in verbal agreement to proceed with this research study, they met with the research coordinator on the morning of the planned surgery to provide written consent and assent. The participant flow chart is shown in figure 1. Inclusion criteria are (1) age 4 to 18 yr, (2) no previous polysomnography, (3) no known sleep-disordered breathing, (4) no known airway problems, and (5) elective ENT day surgery. An exclusion criterion is chronic opioid therapy. On the day of surgery, polysomnography was performed for one night starting 1 to 2 h before morphine was administered for pain relief. The research team had no influence on the type, timing, or dose of medication prescribed to the patient at any stage of this study (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B309). Patients underwent surgery and polysomnography between February 2013 and September 2014.

All parents completed a detailed sleep questionnaire specific for the sleep laboratory at the Hospital for Sick Children to collect demographic and clinical data. Data obtained for each patient included age, gender, height, weight, body mass index, type of surgery performed, and current regular medications. During the overnight polysomnography, the medications and their route of administration were recorded.

All patients received standard anesthetic, surgical, and perioperative care. Several hours (between 3 and 10 h) after being released from the postanesthesia care unit, the patients receiving morphine as analgesics were categorized as the morphine group. The patients who did not receive morphine were categorized as the control group. To determine the effects of morphine on electroencephalogram and breathing, we compared electroencephalogram (fig. 2A) powers and respiratory activity between baseline/premorphine condition (when the patient was awake) and postmorphine condition when the patient was sedated. We analyzed control patients using a similar timeframe as the morphine patients (see table, Supplemental Digital Content 2, https://links.lww.com/ALN/B309). We compared electroencephalogram and respiratory activities between baseline condition (wakefulness between 10:00 pm and 10:30 pm) and postcondition (non-REM sleep between 11:30 pm and 12:00 am). Using this experimental design, comparisons were made between baseline (wakefulness) and sedation/morphine for the morphine group and baseline (wakefulness) and non-REM sleep for the control group.

After receiving standard anesthetic, surgical, and perioperative care (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B309), patients were monitored for one night at the Hospital for Sick Children. They were administered analgesics as indicated for pain relief as per current standard recommended by the acute pain service of the Hospital for Sick Children. Full polysomnography, including (electroencephalogram) and cardiorespiratory recordings, was performed at their inpatient bed during the first night of their hospital stay after surgery using a Natus SleepWorks system (XLTEC, Canada) according to standard guidelines.²¹  Polysomnography was started between 8:00 pm and 10:00 pm, and morphine was administered between 10:00 pm and 2:00 am (see table, Supplemental Digital Content 2, https://links.lww.com/ALN/B309). Polysomnography was recorded for at least 1 h before morphine was administered. Electroencephalogram and electrooculogram were performed according to the 10:20 system²²  (see fig. 2B for electrode placements). The 10:20 system is an internationally recognized method to ensure standard placements of electrodes and reproducibility.²¹  Physiologic measurements included nasal air pressure, end-tidal pressure of carbon dioxide, and oxygen saturation (fig. 2C). Data were digitized at a sampling rate of 512 Hz.

Using Natus SleepWorks software, polysomnography data of each patient were exported into European Data Format (edf) files and then imported into MATLAB (MathWorks Inc., USA) for further analysis. A combination of custom-made scripts using the MATLAB Signal Processing Toolbox were used to process data from the polysomnography recordings. We first applied a 60-Hz notch filter to all channels. The following derivations (differences between specific channels) were processed according to the American Academy of Sleep Medicine²¹ : frontal (F4-M1 and F3-M2), central (C4-M1 and C3-M2), and occipital (O2-M1 and O1-M2). We also filtered electroencephalogram derivations with a high-pass filter (more than 1 Hz) to remove movement artifacts. For each 30-s epoch, we then computed the power spectral density (frequency range between 0 and 30 Hz), also referred as the power spectrum, to determine the frequency distribution within the electroencephalogram signals (fig. 2D). We computed the power spectral density estimate using a discrete Fourier transform (function periodogram, MATLAB). The periodogram returns the power spectral density estimate for a specific frequency range using fast Fourier transform. The average powers in the following frequency ranges or bands were calculated: δ₂ (1 to 2 Hz), δ₁ (2 to 4 Hz), θ (4 to 7.5 Hz), α (7.5 to 13.5 Hz), β₁ (13.5 to 20 Hz), and β₂ (20 to 30 Hz) for the following electroencephalogram derivations: frontal (F4-M1), central (C4-M1), and occipital (O2-M1) as recommended by the American Academy of Sleep Medicine²¹  (fig. 2E). Frequency band powers were estimated for each subject in baseline and morphine conditions or non-REM sleep. Spectrograms, which are time-varying plots of the electroencephalogram spectra, were computed for frequencies between 0 and 50 Hz using short-time Fourier transform (function spectrogram, MATLAB) and a Hamming window of 1 s.

Topographic power maps display the intensity of electroencephalogram activity for each frequency band over the surface of the scalp. Topographic maps were calculated for each frequency band using the EEGLAB toolbox.²³  Briefly, polysomnography data were imported into the EEGLAB toolbox, and an independent component analysis was performed and topographic maps were then created. For each frequency band, this approach quantified the topographic changes in band powers induced by morphine. We compared morphine-induced changes in band powers between frontal (F4-M1) and occipital (O2-M1) electroencephalogram signals.

Finally, we determined the coherence between the frontal and occipital electroencephalogram signals for each frequency band to quantify the degree of correlation between the two signals. The frontal-occipital coherence was estimated as the magnitude squared coherence using the Welch averaged modified periodogram method²⁴  between the frontal (F4-M1) and occipital (O2-M1) derivations in baseline and morphine conditions using the following formula:

where C is the magnitude squared coherence estimate between frontal electroencephalogram signal F = (F4-M1) and occipital signal O = (O2-M1) at each frequency f, P_FO is the cross power spectral density of F and O, P_FF is the power spectral density of F, and P_OO is the power spectral density of O. Magnitude squared coherence was computed using the function mscohere (MATLAB). For each frequency, this function provides an index (C_FO) ranging from 0 to 1, where 0 means the two signals are not correlated and 1 the two signals are perfectly correlated.

Data were analyzed for each subject 60 min before morphine administration (baseline condition) for a 30-min period and starting 30 min after morphine (postmorphine condition) also for a 30-min period (fig. 2A). The impact of morphine on respiratory and electroencephalogram variables was quantified by comparing values in baseline (premorphine) versus postmorphine conditions. Changes were defined as percentage of the baseline mean value. As examples, 100% indicates no change due to morphine, 60% indicates a reduction of 40%, and 200% a twofold increase compared to baseline (premorphine). This repeated-measure design allows us to detect changes directly induced by morphine. We also used percentage of baseline powers to quantify the changes induced by morphine compared to baseline.

Pain intensity was evaluated using the numerical rating scale for pain intensity and unpleasantness as previously used and tested in a pediatric population of postoperative patients.²⁵  Children verbally rated the intensity of pain on a scale from “0” (no pain) to “10” (worst pain possible; most hurt possible).

Continuous variables were expressed as mean ± 95% CI. Normality (Shapiro–Wilk) and equality of variances tests were performed. Paired Student’s t tests, or Wilcoxon signed-rank tests (if Shapiro–Wilk tests were significant), were used to compare respiratory rate, carbon dioxide, and electroencephalogram powers between baseline and postmorphine conditions. The relationships between the percentage changes in electroencephalogram powers and respiratory rate depression by morphine were examined by linear regressions. Interactions between subject groups (morphine vs. control) and experimental conditions (baseline vs. post) were tested using two-way repeated-measure ANOVAs. ANOVAs were followed by Holm–Sidak post hoc tests. To compare ordinal pain ratings before and after opioid administration, Wilcoxon signed-rank test was used, and medians were indicated. Statistical tests were considered significant when P < 0.05. For a repeated-measure design, our previous experience showed that to detect significant differences, 10 patients were required. Analyses were performed with Sigma Plot 11 (Systat Software, USA).

The study participant flow chart is shown in figure 1. Initially, 25 patients were included in the study. Of the 25 patients who underwent a polysomnography, 7 did not complete the study because they felt uncomfortable before receiving morphine and refused to complete the polysomnography. Of the 18 patients who completed the study (table 1), 10 patients received morphine analgesia (morphine group) as a result of elective otologic surgery and 8 patients did not receive morphine (control group). The medications given to control and morphine patients are listed in table, Supplemental Digital Content 1, https://links.lww.com/ALN/B309. There were four females and six males in the morphine group, the mean ± SD body weight was 51.1 ± 15.5 kg, and the mean ± SD morphine dose was 0.157 ± 0.044 mg/kg. In the control group, there were three males and five females, and the mean ± SD body weight was 58.2 ± 13.0 kg.

To evaluate pain levels in patients, we used a self-report numerical pain rating scale.²⁵  In control patients, the pain rating median did not significantly change between the baseline and post conditions (Wilcoxon signed-rank test, P = 0.578, n = 8, table 2). In morphine patients, administration of morphine decreased significantly the pain scale (P = 0.020, n = 10).

We first identified the electroencephalogram spectral changes in the central lead (C4-M1) by quantifying the power changes between the baseline and postmorphine conditions according to the experimental design described in figure 2. Spectrograms of a representative patient show a reduction in β₁ and β₂ electroencephalogram powers with morphine compared to baseline premorphine (fig. 3, A and B). Morphine significantly decreased β₁ (P = 0.003, n = 10), β₂ (P = 0.020, n = 10), and α powers (P = 0.039, n = 10; fig. 3C), but did not change δ₂, δ₁, θ, and α powers (all P > 0.331, n = 10).

To determine the spatial changes in electroencephalogram powers induced by morphine, we therefore compared the topographic distribution of electroencephalogram powers in baseline and postmorphine. In a representative patient (fig. 4A) and in the analyzed group data (fig. 4, B and C), morphine reduced frontal, but not occipital β₁ and β₂ electroencephalogram powers. In the frontal region, morphine significantly reduced β₁ (P = 0.005, n = 9) and β₂(P = 0.008, n = 9) powers, but did not change power in the lower frequency bands (i.e., δ₂, δ₁, θ, and α; fig. 4, B and C). In contrast, in the occipital region, there were no significant changes induced by morphine (P > 0.121, n = 9) over the entire 0- to 30-Hz frequency range (fig. 4C). In this analysis, one patient was not considered because his frontal and occipital channels were not recorded properly and did not contain valid data.

To put these results into a physiologic perspective, we then compared the changes in spectral band powers induced by morphine with the changes induced by non-REM sleep in a separate control group. Data from REM sleep were not considered because, at the time of recordings (evenings), there were not enough REM episodes to statistically compare band powers. We then plotted topographic electroencephalogram maps from representative patients of the control and morphine groups (fig. 5, A and B) and also analyzed group data band powers for central electroencephalogram channels (fig. 5C). Two-way repeated-measure ANOVA showed that there were significant interactions between conditions and groups for the following band powers: α, β₁, and β₂ powers (P = 0.034, P = 0.042, and P = 0.027, respectively), but not for δ₂, δ₁, and θ bands. Post hoc tests showed that there were significant differences between postmorphine and non-REM sleep for β₁ and β₂ powers (P = 0.004 and P = 0.002, respectively; fig. 5C). Interestingly, when analyzed separately, δ₂ and δ₁ powers were increased by non-REM sleep in control patients (P = 0.019 and P = 0.039, respectively, n = 8) consistent with an increased slow-wave activity in non-REM sleep.

The results in figure 4 showed that morphine reduced the higher (β) frequency powers in the frontal but not in the occipital lobe (fig. 5). To directly test whether morphine selectively alters powers in the frontal or occipital regions, we measured the coherence between the frontal and occipital electroencephalogram signals as the degree of correlation between them at a given frequency band. As an example, the coherence estimate was computed for two identical electroencephalogram signals (fig. 6A), and this (as expected) resulted in coherence estimates of 1.0 at all frequencies (gray line in fig. 6B). We then quantified coherences between frontal (F4-M1) and occipital (O2-M1) signals in the baseline and postmorphine conditions (fig. 6, B and C). At frequencies above 20 Hz, we identified that frontal-occipital coherence decreased in the morphine compared to the baseline condition (fig. 6D). Coherence was significantly decreased by morphine in the β₂ frequency band (P = 0.043, n = 9; fig. 6D). In this analysis, one patient was not considered because his frontal and occipital channels were not recorded properly and did not contain valid data. There was no significant differences found between baseline and postmorphine conditions at the lower electroencephalogram frequencies (P > 0.142, n = 9; fig. 6D). For δ₂, δ₁, θ, α, and β₁ bands, the statistical power of the performed tests were less than 80%. To detect a physiologically relevant difference of 10% between baseline and morphine conditions, it would require more than 30 patients.

To determine changes in the degree of respiratory rate depression induced by morphine, we measured respiratory rate using the continuous recordings of expired carbon dioxide partial pressure and oral and nasal airflow (fig. 7, A and B). Although the doses of morphine administered in the study were relatively low and given according to clinical judgment, morphine decreased respiratory rate by an average of 8.3 ± 9.5% (mean ± SD, P = 0.021, n = 10; fig. 7C). Morphine did not significantly change arterial oxygen saturation (P = 0.678, n = 10; fig. 7D), but significantly reduced heart rate by 4.1 beats/min (P = 0.023, n = 10; fig. 7E). Compared to the control group, morphine significantly decreased respiratory rate in the morphine group (two-way ANOVA, groups × conditions, P = 0.009; fig. 7F). There was no significant relationship between the severity of respiratory rate depression and either the morphine dose administered (R = 0.04, P = 0.924, n = 10; fig. 7G) or age (R = 0.30, P = 0.396, respectively, n = 10; fig. 7H). End-tidal partial pressure of carbon dioxide was not significantly changed by morphine (P = 0.062, n = 10).

Since reductions in β₁ and β₂ powers were the major electroencephalogram signatures associated with the sedative properties of morphine, we then performed regression analyses between electroencephalogram powers and respiratory activity in control (fig. 8A) and morphine patients (fig. 8B). There were no significant correlations between changes in δ₂, δ₁, and θ electroencephalogram frequency activity and respiratory rate depression with morphine (fig. 8C). There was no significant correlations between changes in α and respiratory rate depression (R = 0.611, P = 0.061, n = 10; fig. 8C), whereas there was a statistically significant positive correlation between the reductions in β₁ power and respiratory rate depression (R = 0.715, P = 0.020, n = 10; fig. 8C). There was no significant correlation between β₂ power and rate changes (R = 0.568, P = 0.087, n = 10; fig. 8C) and respiratory rate depression with morphine. In control patients, there was no relationship between changes in power bands and changes in respiratory rate, therefore suggesting that the relationships observed were specific to morphine. We then looked more closely at the correlation between changes in β₁ power and the severity of respiratory rate depression. In a patient with initially relatively high β₁ power (fig. 9A), morphine reduced β₁ power, and this reduction was accompanied by respiratory rate depression of 27%. In another patient with initially relatively low β₁ power (fig. 9B), morphine did not reduce power and respiratory rate. This relationship between β₁ power and respiratory rate depression was further confirmed by the positive correlation between the severity of respiratory rate depression and reduction of β₁ power found in 10 patients (fig. 9C).

We previously showed that morphine reduced frontal-occipital coherence in the high-frequency (β₂) band. We then determined whether these changes in coherence are also associated with morphine-induced respiratory rate depression (fig. 10). We identified a significant positive correlation between the change in coherence in the β₂ band and the degree of respiratory rate depression due to morphine (fig. 10; R = 0.745, P = 0.021, n = 9). In this analysis, one patient was not considered because his frontal and occipital channels were not recorded properly and did not contain valid data. There were no significant relationships between respiratory rate suppression with morphine and the changes in coherence for the δ₂, δ₁ θ, α, and β₁ frequency bands.

Here, we identify a significant impact of morphine on electroencephalogram spectral content and respiratory activity when given to young patients for pain relief after elective surgery. The main findings were (1) morphine reduced high-frequency (β₁ and β₂ band powers) cortical activity in the central and frontal lobes, without changing low-frequency activity. Low- and high-frequency activities in the occipital lobe were not affected by morphine. These results were characteristic of morphine-induced sedation and distinct from the changes induced by non-REM sleep. (2) Morphine decreased frontal-occipital coherence in the β₂ electroencephalogram frequency band. (3) Importantly, the degree of respiratory rate changes by morphine was significantly related to the reduction in β₁ frequency cortical activity and frontal-occipital coherence. Overall, these data show that analgesic doses of morphine are associated with reductions in β power and accompanied by respiratory rate depression, indicating a powerful effect of cortical arousal state per se in respiratory rate depression by morphine.

States of heightened arousal in wakefulness are characterized by high-frequency electroencephalogram activity compared to states of reduced arousal such as sleep or anesthesia.⁹  The changes in electroencephalogram spectral content induced by sedatives and anesthetics have been widely documented. Anesthetics produce a powerful suppression of β powers and also increase α power and coherence.^6,7  Here, we identified that morphine significantly decreased β₁ and β₂ powers in the frontal region as well as decreased frontal-occipital coherence in the β₂ range. We showed that central electroencephalogram α, β₁, and β₂ powers were decreased by morphine. Similarly, a combination of general anesthetics and the opioid analgesic remifentanil decreased electroencephalogram β power.⁴  Remifentanil or alfentanil alone similarly decreased β powers.²⁶  Overall, the acute effects of morphine on electroencephalogram spectral content are in accord with the impact of various opioid drugs on electroencephalogram activity. Although reductions in high frequency above 20 Hz may be due to the decrease of electromyographic activity by morphine,²⁷  the changes observed in α and β₁ powers (less than 20 Hz) are likely due to changes in electroencephalogram activity. In addition, we identified that morphine decreased frontal-occipital coherence, reflecting a decrease in functional connectivity between frontal and occipital regions. Electroencephalogram coherence provides an index of functional interactions between neural systems operating in a specific frequency band. Similarly, the short-acting opioid remifentanil decreases coherence in healthy adults in the β frequency range that may relate to the loss of stability of sustained attention by opioid drugs.⁵  In this study, morphine was administered to decrease pain in postoperative patients. Pain levels, however, directly influence electroencephalogram activity. In fact, chronic pain increased δ and θ powers,²⁸  which is not consistent with the changes in β power associated with morphine in our study. Above all, these changes in β electroencephalogram activity fit well with the concept that cortical arousal is reduced by opioid drugs.

After the finding that morphine reduced β₁ and β₂ electroencephalogram powers and frontal-occipital coherence, we then associated these electroencephalogram changes with changes in respiratory activity. Although this study was not specifically designed to examine correlations and have a small sample size, it suggests that patients exhibiting reductions in frontal β electroencephalogram power by morphine displayed respiratory rate depression. Interestingly, it was only in those patients who initially had increased high-frequency β electroencephalogram activity that morphine reduced both β power and respiratory rate. In contrast, when β power was already low, patients did not display respiratory rate depression. These results support two important conclusions: (1) When patients exhibit a high level of cortical arousal before morphine administration, then morphine produced electroencephalogram changes consistent with sedation, and this was accompanied by respiratory rate depression. (2) In conditions of lesser cortical arousal characterized by reduced frontal β electroencephalogram activity, the same doses of morphine were not powerful enough to reduce breathing. These findings indicate a powerful effect of cortical arousal state per se in the respiratory rate depression produced by morphine.

Identification of the electroencephalogram spectral changes elicited by morphine implicates certain putative networks that may mediate reduced arousal by morphine and its impact on breathing. Our data showing that morphine reduced electroencephalogram signatures of cortical arousal suggest that opioid drugs inhibit arousal circuits involved in wakefulness. The diffuse reticular activating system, for instance, contributes to the ascending arousal system^29,30  and is thought to constitute one of the components underlying the wakefulness drive to breathe.^31–33  The diffuse arousal system is active during wakefulness, and its activation increases high-frequency coherence of cortical signals.³⁴  A reduction of frontal-occipital coherence produced by morphine, along with a reduction in β spectral content, is consistent with a decreased activity of the reticular activating system. Also, the periaqueductal gray modulates pain and arousal, expresses μ-opioid receptors,³⁵  is inhibited by fentanyl,³⁶  modulates respiratory activity in wakefulness,³⁷  and has reciprocal connections with the pre-Bötzinger complex,^38,39  a small population of cells in the medulla contributing to opioid-induced respiratory rate depression in rodents.^16,40  The periaqueductal gray may, therefore, be a key component mediating reduced arousal and respiratory rate depression by opioid drugs. In summary, the current results are consistent with the proposition that morphine reduces the tonic excitatory wakefulness drive to breathe by reducing the activity of cortical arousal systems, which then promote respiratory rate depression via withdrawal of this tonic drive to the respiratory network.

The conduct of research in young patients receiving postoperative care carries the obligation of not exposing children to risks beyond the routine care after surgery. Randomized studies are, therefore, not ethical in this population. Here, we used a quasi-experimental study design where patients received morphine when needed after elective surgery. This study design cannot eliminate the possibility of confounding bias that may limit the ability to draw causal inferences. For example, the patients receiving morphine initially experienced pain, and this may have been responsible for increasing their β activity, although we note that baseline respiratory rates, heart rates, and pain levels were similar. Also, patients were administered anesthetics, analgesics, and other drugs before the study as part of standard surgical procedures, but the last opioid analgesic regimen was given to patients at least 5 h before the study. To minimize these and other potential confounding factors, we used a repeated-measures design of premedication versus postmedication and compared control versus morphine groups that isolated the effects of treatment. Despite these limitations, this article generates new findings and knowledge regarding the sedative properties of morphine and its effects on cardiorespiratory function, by identifying distinct cortical signatures associated with respiratory rate depression by morphine.

Supported by a Parker B. Francis Fellowship (Kansas City, Missouri; to Dr. Montandon); the Ontario Thoracic Society Grant-in-Aid (Toronto, Ontario, Canada; to Drs. Narang, Horner, and Montandon); National Sanitarium Association Innovative Research Program (Toronto, Ontario, Canada; to Dr. Horner); and Tier 1 Canada Research Chair in Sleep and Respiratory Neurobiology (Ottawa, Ontario, Canada; to Dr. Horner).

The authors declare no competing interests.