Research and clinical implications of emerging evidence regarding patterns of postoperative opioid-induced respiratory depression

The wider availability of continuous respiratory monitors and advanced data abstraction techniques has led to a substantial increase in understanding of postoperative opioid-induced respiratory depression (OIRD), particularly regarding its incidence, presentation, temporal distribution, and risk factors. Self-limited episodes of OIRD are relatively common, typically presenting as repetitive apneas beginning in the postoperative period and continuing through the first night after surgery. In contrast, life-threatening episodes of OIRD are rare and usually occur on the day of surgery. Traditional monitoring of patient vital signs may be insensitive in detecting OIRD, while healthcare staff may be more adept at recognizing the concurrent development of somnolence. Although obstructive sleep apnea is a known risk factor for OIRD, a more comprehensive phenotype is emerging-elderly patients with debility and substantial comorbidity. These advances have significant implications for managing postoperative OIRD. This review will focus on how these new insights into OIRD have highlighted knowledge gaps and created opportunities for future research and practice initiatives.

Determination of the effectiveness of any of these interventions in preventing critical OIRD events remains elusive because traditional concepts regarding OIRD may be outdated or incorrect.In this review, we will consider the latest evidence regarding the incidence, phenotypic and temporal presentation and risk factors for OIRD; and the implications of how these new understandings have on guiding future research and shaping current practices.These concepts are summarized in Table 1.
Incidence of postoperative OIRD: a common event albeit severe events are rare

Continuous monitoring and OIRD incidence
Over the past fifteen years, a variety of technologies with the capacity to continuously monitor and record data related to respiratory function in the postoperative period have been studied.These studies have consistently found very high rates of postoperative respiratory depression events.The first large cohort reported was conducted by Sun et al. (11) which utilized pulse oximetry blinded to healthcare providers on 833 postoperative patients on general care wards.They found that 37% of subjects had clinically undetected hypoxemia (oxyhemoglobin saturation [SpO2] < 90% for ≥ 1 hour).The PRediction of Opioid-induced respiratory Depression In patients monitored by capnoGraphY (PRODIGY) trial was a large multinational study of 1,335 subjects that used both capnography and pulse oximetry blinded to healthcare providers on patients who were receiving opioid analgesics on general care wards and found that 46% of patients had at least one OIRD episode (5).Measurement of respiratory effort via noninvasive respiratory volume monitors utilizing changes in chest wall bioimpedance, indicated that approximately one-third of patients on general care wards have OIRD (6).Finally, the use of acoustic respiratory rate monitors revealed that bradypnea occurs in 48% of postoperative patients on general care wards (7).
However, the clinical relevance of these OIRD episodes remains uncertain.Khanna et al. (12) did find that PRODIGY patients who had OIRD episodes had longer hospital stays, and higher hospital costs compared to those who did not have OIRD.Yet, among the 614 patients in PRODIGY who had an episode of OIRD, only 7 had respiratory depression recognized by healthcare staff (13).There is a possibility that a sizeable proportion of these postoperative ORID episodes may be a component of underlying sleep breathing disorders, such as undiagnosed obstructive sleep apnea (OSA).OSA is a common but often undiagnosed comorbidity, with an estimated 20% of adult surgical patients not being preoperatively diagnosed with OSA (14).The implication of the commonality of OSA can be illustrated by considering a postoperative patient with known mild sleep apnea diagnosed with polysomnography, who has 10 apnea-hypopnea index (AHI) events per hour.Using the alarm settings from the PRODIGY trial (5), this patient would experience 80 respiratory events during 8 hours of uninterrupted sleep.While these events may have long-term health consequences, such as an increased risk of developing hypertension, they are unlikely to significantly impact postoperative outcomes in the acute period.

Incidence of severe OIRD
Quantifying the incidence of life-threatening episodes of OIRD is more challenging because these events are rare.Acute respiratory failure requiring endotracheal intubation and initiation of mechanical ventilation may result from opioid toxicity but may arise from other etiologies such as severe neurologic disorders (e.g., stroke, encephalitis), hypoxemia secondary to ventilation-perfusion mismatch (e.g., pneumonia, pulmonary embolism), or other conditions (sepsis, multiorgan failure) (15).A less sensitive but more specific marker of life-threatening OIRD is the administration of opioid reversal drug naloxone (16).A recent review reported that the incidence of postoperative naloxone administration on the general care wards ranges from 3.7 to 53.3 per 10,000 patients ( 16).An aggregate of these studies suggests an incidence of 1 emergency naloxone administration per 1,000 postoperative patients admitted to the general care wards.The rarity of severe respiratory events has important implications for ORID research which could lead to establishing more precise clinical practice guidelines.

Implications for clinical research
Prospective randomized controlled trials are conducted primarily to assess the efficacy of an intervention, often with the goal of obtaining regulatory approval for a drug or device.
However, a variety of methodological shortcomings usually limit the ability of such studies to quantify the risk of adverse events, especially the low rate of life-threatening OIRD events (17,18).Furthermore, strict protocols designed to minimize the risk of adverse events and inconsistencies in reporting adverse events represent further limitations in these trials (17,18).Meta-analyses of such studies inherit these methodological limitations and thus may underestimate the risk of ORID (18).

Determining safety of analgesics
Perioperative gabapentinoids were advocated by practice guidelines as a component of non-opioid multimodal analgesic protocols (19).Because these drugs were considered to be 'opioid-sparing' they, in theory, should reduce the risk of serious OIRD events.Many of the gabapentinoids studies were designed to minimize the risk of OIRD.For example, Hah et al. (20) conducted a prospective study on perioperative gabapentin use to assess postoperative pain resolution and opioid cessation.Exclusion criteria included patients with OSA, a known risk factor for postoperative OIRD (16).In addition, the study reported only 410 subjects which is insufficient to reliably assess the OIRD risk.Larger retrospective studies demonstrated a positive association between perioperative gabapentinoid use and serious episodes of OIRD (21)(22)(23), leading to a 'black box warning' by the US Food and Drug Administration (FDA) regarding the risk of serious respiratory depression when gabapentinoids are used with other sedating medications (24).
More recently, there has been interest in perioperative methadone for analgesic therapy (25).A recent report suggested that methadone use is not associated with an increased need for perioperative naloxone compared to conventional opioid analgesics (26).In that study, 337 patients who were administered methadone were matched to 674 controls administered conventional opioids.The rate of postoperative naloxone administration was 2 (0.6%) in the methadone group and 1 (0.1%) in the conventional opioid group.The sample size in this study (26) is underpowered to assess the safety of methadone use in the perioperative period.It is important to note that the majority of FDA safety notices are derived after the regulatory process is completed (27).Thus, it is incumbent upon perioperative physicians and healthcare systems to continually audit their practices to review serious OIRD events and assess if there are changing trends in incidence or risk factors associated with novel drug practices.

Assessing efficacy of respiratory stimulants and novel opioids on OIRD incidence
There are several drugs that act as respiratory stimulants without reversing the analgesic effects of opioids unlike the μ-antagonist naloxone (10).For example, ENA001 (formerly known as GAL-021, a drug related to almitrine), a selective antagonist of large-conductance big potassium channels of the carotid body chemoreceptors which enhances hypoxic respiratory drive and has been shown to reverse alfentanil-induced OIRD without affecting its analgesia (28), (refer to van der Schrier et al. (10) for a comprehensive review of novel respiratory stimulants).Oliceridine is a unique opioid that induces μ-opioid receptor G-protein signaling, which confers analgesia and has minimal activation of the μ-receptor β-arrestin signaling, which induces respiratory depression and gastrointestinal adverse effects (29).Compared to morphine, oliceridine has been found to have a better therapeutic index regarding offering analgesia with less respiratory and gastrointestinal adverse effects (30,31).However, designing a study sufficiently powered to determine if respiratory stimulants or novel opioids could reduce the incidence of severe OIRD in clinical settings would be prohibitive.For example, a study powered to detect a 50% reduction in naloxone administration would require approximately 50,000 subjects.Thus, alternative metrics to establish safety are required, but such metrics have not yet been established, a concept we will explore in the next section.

Postoperative monitoring strategies
To mitigate the risk of serious or fatal OIRD, there have been calls for universal monitoring of hospitalized postsurgical patients (32).Why not, use an oximeter on all patients so the hospital staff can be alerted if a patient's oxygen saturation is dropping?Pulse oximetry is widely available, inexpensive, and well-understood by both laypeople and healthcare providers.
A challenge to such an approach is defining clinically relevant thresholds for alarm activation.Sun et al. (11), using SpO2 ≤ 90% as a definition of hypoxemia, found that 37% of patients met this criterion.However, many of these patients were not in acute danger of respiratory arrest.
Such overly sensitive settings lead to alarm fatigue, rendering the monitors less effective.This dilemma was addressed by McGrath et al. (33) with the introduction of the concept of "condition" and "surveillance" alarm thresholds."Condition" specific alarm thresholds are those used in high acuity settings, such as during general anesthesia where vital signs are closely monitored, and small changes in patient physiology necessitate constant adjustments of anesthetic variables.In contrast, "surveillance" specific alarm thresholds are more useful in a lower-acuity setting.In one large academic hospital all postoperative patients were continuously monitored with pulse oximetry using "surveillance" specific alarm settings (SpO2 below 80% for 30 seconds), and when this threshold was reached the automated system alerted the nurse via pager (4).Using this algorithm, the nurses received on average 4 pages per patient per day.
Questions about the clinical safety of these surveillance alarm settings were addressed by Taenzer et al. (4).The primary outcomes of Taenzer's (4) study were rescue events and transfers to the intensive care unit before and after implementation of monitoring changes.They found that the number of rescue events decreased from 3.4 to 1.2 per 1,000 patient discharges, and intensive care unit transfers from 5.6 to 2.9 per 1,000 patient days.The observed death rate after implementation was two, compared to four in the previous time frame.Furthermore, McGrath et al. (33) retrospectively analyzed 6,130 patients who received 3 days of continuously recorded SpO2 data from patients hospitalized on general care wards.They examined the effect of decreasing the alarm threshold from 88% to 80% and changing from an instant alarm to a 15second delay.Using "surveillance" settings there would be 2.5 pulse oximeter alarms per patient day, compared to 148 alarms per patient day with traditional "condition" settings (33).
Cutoffs for safe thresholds for continuous pulse oximetry monitoring to detect respiratory failure on general care wards have not been well defined in the literature.The unanswered question is whether "surveillance" settings should use universal thresholds or if a tailored approach based on specific patient comorbidities would be more appropriate.The latter approach would introduce an additional layer of clinical complexity.Furthermore, ideal thresholds for more advanced monitoring technologies such as capnography, have not been established.For example, the threshold of apnea in the PRODIGY trial was based on definitions used for polysomnography (5).However, as mentioned previously, a patient with underlying sleep apnea might frequently trigger the capnography alarm.The pattern of apneas leading to clinically relevant OIRD (e.g.need for re/intubation, etc) is not well understood.Is it the frequency of apneas, the duration of apneas, accompanying changes in pulse oximetry and heart rate, or a combination of these factors?Does it differ between patients with or without OSA?Given these unknowns, it remains unclear how this technology should be best utilized on hospital care wards.

Presentation with continuous monitoring
Classic teaching about the effect of opioids on ventilation suggests that these drugs slow the respiratory rate and decrease the sensitivity to arterial carbon dioxide and oxygen concentrations (34).The PRODIGY trial (5) found that 97% of detected OIRDs presented as apnea and 58% as bradypnea.The typical pattern, shown in Figure 1a, involves intermittent apnea episodes interspersed with arousal periods of normal respiratory effort.For patients on room air, the apneas are accompanied by episodic decreases in SpO2 and increases in heart rate.
In contrast, for patients receiving supplemental oxygen, the intermittent apneas are often not associated with changes in SpO2 and heart rate (Figure 1b).Isolated bradypnea was encountered in less than 3% of cases (Figure 1c).These observations are more consistent with opioidsinducing pauses of ventilation through a combination of central apneas as well as obstructive apneas due to reduced hypoglossal nerve tone to the upper airway causing obstructions during sleep.Another pattern of respiratory instability would be isolated hypoxemia without signs of decreased respiratory drive (Figure 1d), which would be more compatible with a pathological lung condition (e.g., pneumonia, atelectasis) resulting in a ventilation-perfusion mismatch.

Healthcare providers perception of OIRD
Analysis of how healthcare providers evaluate patients experiencing OIRD has yielded unexpected results.For example, Lee et al. (1) studied the American Society of Anesthesiology Close Claim Database of 92 malpractice claims related to adverse outcomes from postoperative respiratory depression.Their study revealed that somnolence was recorded by nurses in 62% of cases prior to the adverse respiratory event.Additionally, 'heavy snoring' was noted in 15% of cases, and somnolence accompanied snoring in 93% of those patients.Studies examining postoperative naloxone administration on general care wards also found that nurses frequently recorded somnolence or decreased levels of consciousness preceding severe respiratory events.
Weingarten et al. (35) examined postoperative naloxone administration on 134 patients on general care wards within the first 48 hours of surgery and found that the indication for naloxone administration was excessive sedation in 43% of cases and respiratory depression in 52% of cases.Similarly, Deljou et al. (21) examined postoperative naloxone administration on 128 patients on general care wards within the first 48 hours of surgery and found that the indication for naloxone administration was excessive sedation in 44% of cases and respiratory depression in 56% of cases.Valencia Morales et al. (36) examined 162 medical and postsurgical patients on general care wards who were administered naloxone and found that nurses appreciated neurologic signs (e.g., somnolence, decreased level of consciousness) in 80% of cases and pulmonary symptoms in 50% of cases.There is a possibility that some cases of naloxone administration 'to treat somnolence' were unrelated to OIRD, but it is also possible that patients with somnolence had unrecognized concurrent respiratory depression.This was illustrated in PRODIGY trial, where a patient was administered naloxone for a 'decreased level of consciousness' rather than respiratory depression, yet capnography identified eight apneic episodes immediately prior to the naloxone administration (13).Examining respiratory drive requires a higher level of expertise and prolonged patient observation to differentiate between apneic spells and isolated bradypnea.In contrast, assessing the level of consciousness can be done quickly.This is especially true if the respiratory assessment relies solely on pulse oximetry when the patient is on supplemental oxygen, which can mask decreases in oxyhemoglobin levels due to hypoventilation (37).Further, if nurses awaken patients during the assessment, transitioning from an asleep to an awake state should terminate apnea spells.A recent simulation study (38) assessed how nurses perform postoperative vital checks and found that in 96% of assessments, the nurses woke patients prior to assessments resulting in the missed detection of 85% of apnea spells occurring while the patient was asleep.

Implications for clinical research
Life-threatening cases of postoperative OIRD are rare (16), thus using this endpoint presents a challenge to construct adequately powered prospective studies regarding the safety of new analgesics/sedatives, the effectiveness of respiratory stimulants, or practice initiatives design to mitigate the effects of OIRD associated oversedation.The introduction of continuous monitoring technologies has provided plenty of possibilities to assess respiratory status including oxyhemoglobin saturation levels, respiratory rate, apnea spells, and minute ventilation.Each of these measurements has its advantages and limitations.

Pulse oximetry
Pulse oximetry provides data on oxyhemoglobin saturation and pulse rate.This technology can be utilized in various ways to assess interventions aimed at mitigating OIRD.
Simplistically, one could calculate the duration patients spend below a defined oxyhemoglobin saturation threshold.Another approach is to use pulse oximetry to calculate the oxygen desaturation index, which measures the number of desaturation events (defined as a 3% decrease in SpO2 for ≥ 10 seconds within a 120-second period) per hour.However, there are several limitations to either approach.Importantly, supplemental oxygen, commonly administered to postoperative patients, mitigates oxyhemoglobin desaturation in response to hypoventilation (Figure 1b) (37).Thus, patients on supplemental oxygen may be experiencing profound episodes of respiratory drive depression due to opioids and other sedatives yet appear 'normal' when assessed by pulse oximetry alone.The other major limitation is that common postoperative lung conditions, such as atelectasis, can lower oxyhemoglobin levels, while preserving or even enhancing respiratory drive (hypoxemic drive).In this circumstance, an intervention designed to prevent respiratory depression may seem ineffective because oxyhemoglobin levels would be low despite adequate respiratory drive.

Advanced monitors
Other technologies, such as capnography, thoracic bioimpedance, and acoustic respiratory monitoring systems, measure respiratory drive by providing data on tidal volume, exhaled CO2, and respiratory rate.These technologies are often used alongside pulse oximetry to provide a holistic view of pulmonary function.However, they face ergonomic challenges.Capnography devices, typically incorporated into nasal cannulas, may interfere with eating, drinking, and speaking due to the sampling device extending from the nasal cannula to the mouth.While this may be manageable in a sedation setting (e.g., endoscopy suite), it is less likely to be tolerated on general care wards by awake patients.Bioimpedance monitors can be dislodged by patient movement and their adhesive backing can be undone by patient perspiration or chest hair.Lastly, acoustic respiratory monitors may be less accurate with facial hair and are prone to artifacts from patient movement and activities such as eating and speaking.Thus, the output from these monitors is complex and often requires time-intensive expert interpretation (author's observations).Furthermore, clinically relevant thresholds for respiratory depression have not been established, though definitions from sleep studies are often used (e.g., apneic episode is a period of apnea that lasts ≥ 30 seconds) (5).These monitors can capture respiratory depressive episodes when supplemental oxygen is used, even when pulse oximetry does not identify oxyhemoglobin desaturation, or when the pulse oximetry indicates desaturation, but respiratory drive is preserved.Yet, how to quantify the effect of an intervention using such technologies has not been determined.Potential approaches include measuring the number of respiratory depressive events over time (akin to the apnea-hypopnea index), or the total duration of respiratory drive instability.However, such definitions require further research and development of consensus.

Changing healthcare workers assessment of OIRD
The typical bedside assessment of respiratory status in hospitalized patients is to count breaths to establish the respiratory rate, measure oxyhemoglobin saturation with pulse oximetry, auscultation of the chest, and ask the patient about symptoms of dyspnea (38).We have observed that nurses usually awaken patients prior to these assessments (38).This approach has notable flaws.Awakening a patient extinguishes the repetitive apneic breathing pattern typically associated with OIRD.By the same token, if patients with ORID are awakened, their breathing may normalize temporarily and be misleadingly recorded as normal.This practice can lead to problems when using continuous respiratory monitors.Specifically, in a sleeping patient with OIRD (or underlying sleep breathing disorders), continuous monitoring may trigger frequent alarms.However, each time the nurse assesses the patient by waking them, their respiratory status may seem reassuring.Thus, trust in the monitor will erode, resulting in alarm fatigue.
Adoption of 'surveillance' monitoring strategies with less sensitive thresholds might decrease the frequency of alarms and only alert the nurses to more serious episodes of respiratory depression (4).However, whether this approach may represent a safe clinical practice is not known.
To reiterate the above-discussed important concept, The American Society for Pain Management Nursing Guidelines clearly recommends that nurses quietly observe a sleeping patient's breathing pattern for at least one minute for signs of airway obstruction or apneas (39).This practice would increase the likelihood of detecting OIRD by nursing staff.However, this approach is not widely practiced (38) and needs to be emphasized in nursing training and clinical practice by nurse managers.Even so, busy healthcare providers may not have adequate time to observe a patient sleeping for several minutes to conclude if apneic episodes are occurring.
Current practices seem to be adept at identifying decreased levels of consciousness, which often precede critical respiratory events (1), therefore nursing guidelines should emphasize that such patients should be considered high-risk and should be more frequent assessments of respiratory status.

Temporal distribution of OIRD: dead at dinnertime
The data from the capnography and pulse oximetry patterns in the PRODIGY trial (5) provided a unique opportunity to determine the timing of postoperative OIRD.In the initial PRODIGY study patients were categorized into two groups, those with ORID and those without.This approach focused only on the first adjudicated OIRD episode due to the overwhelming number of detected episodes.However, in Driver et al. (40) all capnography and pulse oximetry data from two PRODIGY sites were closely analyzed.The authors found that among patients who experienced OIRD 87.7% had multiple episodes.They were also able to construct detailed maps of when the initial event occurred (Figure 2a) and the concentration of all events (Figure 2b).The initial OIRD events were detected in the late afternoon and early evening (Figure 2a), though it is important to note that patients did not have capnography or pulse oximetry data during their postoperative care unit (PACU) stay and transport from the PACU to the ward.
Additionally, there was a bilobed concentration of all events, one peak in the late afternoon/early evening, and a larger, significant peak during the early morning hours (Figure 2b).These early morning events have been postulated to be more clinically important (41) because patients are less closely monitored, and thus respiratory instability is more likely to progress to a critical level.However, data from the Mayo Clinic indicate that the most frequent time of naloxone administration on the surgical wards is in the late afternoon/early evening of the day of surgery (Figure 2c) (16).
Interestingly, Mayo Clinic naloxone data has also indicated that patients who have PACU episodes of respiratory depression are five times more likely to receive naloxone on the general care wards (21,35).Another study that used bioimpedance monitors on patients in the PACU and then for 12 hours on the wards found that those patients who experienced respiratory depression in the last 30 minutes of the PACU admission continued to have episodes of respiratory depression for hours on the wards compared to their controls (42).These observations suggest that residual sedation from anesthesia and opioids during surgery and PACU admission may place vulnerable patients at increased risk for critical events at the beginning of their general care ward admission.

Implications for clinical research
Clinical studies examining the effects of novel respiratory stimulants (10) and opioids with less effects on respiratory drive (29)(30)(31) need to consider these temporal observations when determining the duration of postoperative monitoring.Ideally, administration of respiratory stimulants would be initiated in the PACU and continue through the postoperative period until the following morning.Likewise, continuous monitoring for episodes of OIRD should extend beyond the PACU admission and continue onto the general care wards through the night until the patient awakens the next day.Similarly, studies examining mitigation practices should follow a similar timeline.

Implications for clinical practice
Appreciating the temporal patterns of postoperative OIRD has several practical implications that can be easily integrated into the daily routines of healthcare staff.First, the PACU stay of any patient experiencing an OIRD episode should be extended.For example, the Mayo Clinic extends the PACU stay by at least one hour for any patient who is witnessed to have OIRD during anesthesia recovery (3).Second, the PACU staff should alert the accepting ward if a patient has an episode of OIRD so that heightened vigilance for this complication can be provided.Also, the ward staff should be educated that patients are prone to decompensation early on the ward admission and apply continuous monitors upon admission, rather than waiting until the patients are about to sleep.

Risks for OIRD: beyond OSA and opioids
OSA is the most appreciated risk factor for postoperative respiratory complications and is the focus of several prominent perioperative management guidelines (2,19,43).However, when considering the risk for OIRD it is important to look beyond OSA and have a more holistic approach to assessing patients.The PRODIGY trial found that advancing age is associated with OIRD risk, with an odds ratio of 2.2 for ages 60 to 70 years; 3.4 for ages 70 to 80 years, and 4.8 for patients over 80 years old (5).Furthermore, studies that relied on naloxone administration as a surrogate for severe OIRD found a four-fold increase in risk among patients with underlying neurodegenerative conditions (35) and frail patients (21).These two studies found a two-fold increased risk for OIRD among patients with OSA (21,35).Moreover, cardiopulmonary diseases, other than sleep breathing disorders, have been also found to be associated with increased OIRD risk (5,35).
Not surprising, more extensive surgeries and more sedating anesthetic regimens are associated with OIRD.As expected, intraoperative opioids, especially longer acting and higher doses are associated with an increased risk of OIRD (44)(45)(46).The intrathecal administration of hydrophilic opioids (morphine, hydromorphone) is also known to increase risk for postoperative OIRD, which may have a delayed onset necessitating the need for prolonged monitoring (47,48).
Lastly, the PACU course can yield important information regarding OIRD risk on the wards, specifically patients who have episodes of respiratory depression in the PACU are at increased risk for naloxone administration after PACU discharge (see next section) (21,35).

Implications for clinical research
The PRODIGY study developed a PRODIGY score which can be used to categorize patients into low, intermediate, and high risk for postoperative OIRD as detected by capnography and/or pulse oximetry (5).However, as mentioned earlier, almost half the PRODIGY cohort developed postoperative OIRD.The PRODIGY score has never been validated for severe OIRD episodes (e.g., patients requiring naloxone therapy) but given the commonality of self-limited OIRD and the rarity of life-threatening OIRD, the PRODIGY score may be limited by being overly sensitive with a lack specificity to identify patients at risk for serious OIRD.Electronic medical records now serve as vast repositories of clinical data and offer powerful data extraction capabilities.Statistical software can leverage this information with the goal of developing improved prediction tools for rare clinical events such as serious postoperative OIRD.Further, these data repositories should be interrogated to assess the safety of anesthetic management and emerging trends, such as the perioperative administration of methadone for non-cardiac surgeries (25).
Results from retrospective studies have found strong associations with various aspects of anesthetic care and postoperative OIRD (21,46), and that changes in anesthetic management were associated with a reduction of the incidence of postoperative OIRD (49).These findings support the need for robust prospective trials to assess if various anesthetic techniques can be used to decrease the incidence of OIRD.For example, ultrashort-acting remifentanil and remimazolam are attractive agents to investigate if they, as components of anesthetic management, reduce OIRD risk.

Implications for clinical practice
Efforts by medical societies to prioritize OSA in mitigating the risk of OIRD are laudable (2,19,43).However, clinicians should have a holistic view of surgical patients and appreciate the contributions that advancing age, neurodegenerative conditions, debility, and other disease processes have on OIRD risk.Patients undergoing therapy for OSA should continue this therapy in the postoperative period.Further, the anesthetic regimens for high-risk patients should be modified to promote more rapid anesthetic recovery by utilizing less soluble anesthetics and minimizing sedating analgesic medications (49).Regional anesthetic techniques for patients at high risk for OIRD should be utilized if feasible (46).Because the onset of OIRD can be delayed following intrathecal administration of hydrophilic opioids, patients should be monitored with continuous pulse oximetry for the first 24 postoperative hours (53).Recent evidence has cast doubt on the analgesic efficacy of perioperative gabapentinoids (54), which in conjunction for their propensity to contribute to OIRD (21)(22)(23), suggests that these medications should no longer be used during the perioperative period.Clinicians should carefully assess patients experiencing respiratory depression during anesthesia recovery.The anesthesiologist should consider the entire perioperative course, from preoperative patient assessment to the PACU course to determine the optimal postoperative care plan prior to PACU discharge (the proposed pathway is summarized in Figure 3) (55).

CONCLUSIONS
Postoperative OIRD is a common complication typically presenting as repetitive apneic episodes which first develop during anesthesia recovery and may continue through the first postoperative night.Rarely but importantly, self-limited episodes of OIRD can deteriorate into life-threatening respiratory failure, which can result in serious morbidity and even mortality.
When OIRD is used as a research endpoint, investigators need to appreciate this continuum of OIRD severity to set appropriate definitions and ensure studies are adequately powered.
Likewise, clinical practices need to appreciate these nuances of postoperative OIRD to develop the best perioperative care and postoperative monitoring strategies.Lastly, clinical outcomes need to be continuously reviewed to identify emerging trends of changes in OIRD incidence.

Incidence of postoperative OIRD
• Self-limited episodes of OIRD occur in 1/3 -½ of postoperative patients on general care wards

Figure 1 .
Figure 1.Capnography and pulse oximetry patterns encountered during opioid-induced respiratory depression.The top panels show measured exhaled CO2 (blue lines) over time in mmHg.The bottom panels show pulse oximetry oxyhemoglobin concentration (red line) and heart rate (green line).

Figure 1a .
Figure 1a.The panel depicts the typical pattern of postoperative opioid-induced respiratory depression.The top panel shows pauses in respiratory effort with the dashed blue lines indicating when apneic episodes last ≥ 30 seconds.In this panel, apneic episodes are accompanied by episodic decreases in oxyhemoglobin concentrations and increases in heart rate (40).Reprinted with permission from Anesthesia & Analgesia and Wolters Kluwer Health, Inc.

Figure 1b .
Figure 1b.The panel depicts typical postoperative opioid-induced respiratory depression in a patient with supplemental oxygen.The top panel exhibits intermittent apneic episodes, but in this case, changes in the oxyhemoglobin concentration and heart rate are mitigated.Courtesy of the author's work, data from the PRODIGY trial (5).

Figure 1c .
Figure 1c.The panel depicts a classical bradypnea pattern of postoperative opioid-induced respiratory depression.The top panel exhibits a slow and widened, but stable exhaled CO2

Figure 1d .
Figure 1d.The panel depicts a patient with normal respiratory effort but a decreased oxyhemoglobin concentration level.This pattern can emerge in patients with conditions such as atelectasis.Courtesy of author's work.

Figure 2 .
Figure 2. Timing of opioid-induced respiratory depression episodes depicted on 24-hour clocks.The magnitude of each spoke is the total number of events between the previous spoke time and the current spoke time (eg, at 02:00, the number of episodes that occurred between 00:00 and 02:00).The red line indicates the end of the surgery, and the blue line indicates respiratory depressive episodes.The scale of episodes is different between the three plots(40).Reprinted with permission from Anesthesia & Analgesia and Wolters Kluwer Health, Inc.

Figure 2a .
Figure 2a.Shows the time of day for the first detected episode of respiratory depression among a subset of patients from the PRODIGY trial.(40) The majority of initial episodes occurred in the afternoon and evening following surgery.Reprinted with permission from Anesthesia & Analgesia and Wolters Kluwer Health, Inc.

Figure 2b .
Figure 2b.Shows the concentration of all respiratory depressive episodes among a subset of patients from the PRODIGY trial (40).There is a bilobed distribution with many episodes occur in the afternoon and evening, but the majority of episodes occurring in the early morning hours.Reprinted with permission from Anesthesia & Analgesia and Wolters Kluwer Health, Inc.

Figure 2c .
Figure 2c.Shows the time of day naloxone was administered on general care wards following surgery from two studies of postoperative naloxone administration from Mayo Clinic (16).The majority of naloxone administrations occurred in the afternoon and evening following surgery.Reprinted with permission from International Anesthesiology Clinics and Wolters Kluwer Health, Inc.

Figure 3 .
Figure 3. Proposed clinical pathway for patients with postoperative opioid-induced respiratory depression.Clinical decisions on the postoperative level of care are complex and unique for each patient.Preoperatively, patients should have a risk assessment for respiratory depression.The surgical and anesthetic management should be tailored to this risk.During anesthesia recovery, patients' respiratory status should be monitored for various signs of respiratory depression.Postoperative management decisions regarding the level of monitoring and care should be guided by preoperative status, intraoperative status, and the anesthesia recovery course.Home therapies for sleep-disordered breathing should be continued into the postoperative period.PACU, postanesthesia care unit; OIRD, opioid-induced respiratory depression; SDB, sleep-disordered breathing; PAP, positive airway pressure (55).Reprinted with permission from Anesthesiology and Wolters Kluwer Health, Inc.
Often biased to mitigate risk for adverse events o Inadequately powered to detect severe OIRD events o Require quantification of self-limited OIRD episodes • Retrospective observational studies o Large cohorts required to assess risks potentially associated with severe OIRD • Optimal alarm thresholds for continuous monitors have not been established Clinical implications • Universal continuous monitoring postoperative patients o Alarm fatigue with high rate of self-limited OIRD episodes o Optimal alarm thresholds for monitors may reduce alarm fatigue Presentation of OIRD • Intermitted apneic episodes are the most dominant presentation of OIRD • Isolated bradypnea is very rare • Supplemental oxygen mitigates desaturation events secondary to hypoventilation • Healthcare providers often awaken sleeping patients prior to vital sign assessments o Can temporarily extinguish OIRD breathing patterns • Healthcare providers often do not recognize OIRD but do recognize somnolence Research implications • Best modalities (capnography, pulse oximetry) to quantify OIRD episodes for prospective trials have not been established • Parameters of OIRD episodes have not been established Clinical implications • Healthcare providers should be trained to assess respiratory efforts while patients are sleeping • Healthcare providers should be trained to recognize somnolence can precede severe OIRD events Temporal distribution of OIRD • Self-limited OIRD episodes begin in the PACU and continue onto the ward