Perioperative sleep apnea: a real problem or did we invent a new disease?

Depending on the subpopulation, obstructive sleep apnea (OSA) can affect more than 75% of surgical patients. An increasing body of evidence supports the association between OSA and perioperative complications, but some data indicate important perioperative outcomes do not differ between patients with and without OSA. In this review we will provide an overview of the pathophysiology of sleep apnea and the risk factors for perioperative complications related to sleep apnea. We also discuss a clinical algorithm for the identification and management of OSA patients facing surgery.

Perioperative sleep apnea is becoming a major concern for anesthesiologists and intensivists 1 . However, similar to another problem in the perioperative field (i.e. residual paralysis) 2 , the negative impact of sleep apnea on postsurgical management and patient outcomes remains unclear. In this article, we will (1) summarize currently available data on the prevalence and pathophysiology of sleep apnea in the perioperative context, (2) discuss the bidirectional effect of anesthesia and surgery on sleep apnea, and (3) suggest a clinical pathway for the perioperative identification and management of sleep apnea patients that is being used by many physicians at the Massachusetts General Hospital in Boston, MA, USA based on in-house discussions and input we received from Dr Shiroh Isono, Chiba, Japan during his visits in our institution.

Obstructive sleep apnea and why it is important in the perioperative setting Definition and Epidemiology
Obstructive sleep apnea (OSA) is characterized by recurrent episodes of reduction or cessation of airflow despite continued or increased respiratory effort. Hypopneas are shallow breaths resulting from partial obstruction and reduced intraluminal diameter of the upper airway (UA). Apneas are characterized by the absence of airflow due to complete airway collapse. These respiratory events are associated with oxyhemoglobin desaturations, neuronal arousal, disrupted sleep, and impaired daytime functioning 3 .
Based on daytime symptoms, the incidence of OSA in the general population ranges from 0.3% to 5% 4-6 . However, several studies investigating the prevalence of OSA without daytime symptoms using polysomnography (PSG) found much higher rates 7-9 (Table 1A) with undiagnosed OSA in up to 80% of patients 10 . Furthermore, with obesity representing a major risk factor for OSA, one can expect a higher prevalence of sleep apnea as rates of obesity continue to climb 11-13 .
In the perioperative population, the prevalence of OSA varies widely among different subgroups (Table 1B). Bariatric surgery patients, the subpopulation most extensively studied, have been shown to have rates of up to 77.5% 14 . Many of these patients are asymptomatic despite severe sleep apnea 15 . Other surgical populations, such as orthopedic surgery patients, have been shown to have rates of OSA that are only slightly higher than the general population 16 . This broad range of prevalence rates may be related to the diverse distribution of risk factors for OSA, such as obesity based on a high body mass index (BMI), age, and/or comorbidities (e.g. stroke and myocardial infarction). Further research is needed to evaluate if the surgical population has a higher risk of OSA independent of these factors.
Perioperative complications occur more often in patients with OSA compared to controls 26-29 . These include delirium 30,31 , reintubation, pneumonia 32-35 , atrial arrhythmias, myocardial infarction, and pulmonary embolism 36 (Table 2). Delirium in the postoperative period is associated with increased morbidity and mortality 37,38 , as well as long-term cognitive and functional decline 39,40 . These complications increase utilization of intensive care and length of stay as described in a recent retrospective study in 1,058,710 patients undergoing elective surgeries 35 . Our group also found that, independent of OSA, reintubation and unplanned intensive care unit (ICU) admission may result in a substantial increase in in-hospital mortality 41,42 .
Yet some studies suggest a decreased risk of postoperative mortality in patients with a known diagnosis of sleep apnea 34,35,43 . Ischemic preconditioning was hypothesized to be involved in this protective effect of OSA, despite higher rates of cardiovascular comorbidities 44,45 . Ischemic preconditioning is an experimental strategy during which exposure to short, non-lethal episodes of ischemia results in attenuated tissue injury from ischemia and reperfusion 46 . The underlying mechanisms may include increased blood vessel collaterality 47 and reduced oxidative stress 48 . Recent studies found patients with OSA to have less severe cardiac injury after acute non-fatal myocardial infarction 49 . Protective preconditioning from OSA may not be limited to the heart muscle, but may also have beneficial effects in the kidney and the brain 50-52 .
It is important to note that published studies investigating the effect of OSA on postoperative mortality are based on retrospective chart review. These retrospective analyses used diagnostic coding of OSA as an independent variable. These studies did not control for intraoperative predictors of postoperative complications, such as blood loss, anesthetics used, and mode of mechanical ventilation 53,54 . Therefore, one can assume that the true impact of OSA on postoperative outcomes remains unclear.
Additionally, one could argue that patients already diagnosed with OSA might receive more careful postoperative management. Longer time to extubation 55,56 and increased utilization of noninvasive ventilation have been reported in OSA patients 33 . Furthermore, it is a challenge to isolate the effect of OSA from the known adverse perioperative outcome of other comorbidities seen in patients with OSA, such as hypertension, diabetes, dyslipidemia, and obesity 57-59 . Although randomized controlled trials are important, such trials are unlikely to be feasible. Given that postoperative pulmonary complications are rare and multifactorial, and that the phenotypes of OSA differ by patient, it is difficult to undertake a trial that can capture all the nuances of this question. Observational studies reflect the real world practice of anesthesiologists and allow for the large sample size that is needed to be able to make inferences on the ideal settings for specific patient populations. Despite the limited data on OSA as an independent perioperative risk factor 60 , it is intuitive to conclude that OSA patients are at risk of developing  severe perioperative complications. Therefore, identification and optimal perioperative management of OSA patients is mandatory.

Why does OSA occur and how can the perioperative setting affect OSA?
Upper airway physiology and pathogenesis of obstructive sleep apnea The respiratory pump consists of an anatomically diverse group of muscles including thoracic wall muscles, the diaphragm, and other muscles of the trunk 61 . The contraction of these muscles increases the thoracic volume, and the lung generates negative intra-thoracic pressure. That negative pressure translates to a negative intraluminal UA pressure and thereby results in inspiratory airflow. If the negative UA pressure drops below a critical value (Pcrit), the UA collapses 62,63 . In contrast to healthy controls, Pcrit is positive (>0 cmH 2 O) in OSA patients when paralyzed 64 or sedated, and UA dilator muscle activity is required to maintain airway patency 65 .
The activity of UA dilating muscles depends on neuronal innervation from subcortical and cortical brain regions that are modulated by physiological feedback and feed forward mechanisms. Subcortical regions of the brainstem and midbrain receive inputs for peripheral and central chemoreceptors that are sensitive to partial pressures of oxygen and carbon dioxide levels. Breathing and UA patency therefore respond to changes in gas exchange 66  The underlying mechanisms of pharmacological agents on breathing are diverse 106 . Dose-dependent increases in collapsibility of the UA through depressed respiratory drive, direct inhibition of UA dilator muscle activity (e.g. propofol) 104 , and reduced responsiveness of UA dilator muscles to negative pressure (e.g. isoflurane) 105 have been shown for all GABAergic drugs, but N-methyl-Daspartate (NMDA) antagonistic drugs such as ketamine and nitrous oxide may have respiratory protective effects, at least in the lowdose range. Nishino and colleagues investigated the differential effects of anesthetics and found greater dampening of hypoglossal nerve input relative to the phrenic nerve 115 . Since this results in greater impairment of UA dilator muscles compared to respiratory pump muscles, the effects can lead to increased risk for UA collapse. In contrast, ketamine has been found to reduce neural input to the UA dilator muscles and equally to respiratory pump muscles. Ketamine's effect on the UA dilator muscles was less when compared to GABAergic anesthetics 115 . Studies in rats have demonstrated dissociation between loss of consciousness and UA dilator muscle function under ketamine anesthesia 112 . Taken together, these studies suggest that patients with OSA, who have preoperative UA instability, may be at a heightened risk of UA collapse when under the influence of some, but not all, anesthetics. The unique effects associated with ketamine suggest that some anesthetic agents may be a safer choice for patients with OSA. However, prospective clinical studies in surgical patients with OSA are still required to confirm this preclinical finding and investigate the resulting effects on postoperative outcome.
In addition to reducing arousal and inducing loss of consciousness during surgery with medication, the anesthetist needs to apply neuromuscular blocking agents (NMBAs) carefully to cause muscle relaxation and optimize surgical conditions. The effects of NMBAs may outlast the duration of the surgical procedure. Postoperative residual neuromuscular blockade (rNMB) can affect postoperative respiratory outcome 116,117 and has been reported to occur frequently after surgery 118-120 . rNMB as well as neostigmine reversal may also be associated with an even higher risk of complications in OSA patients. UA and respiratory pump muscles differ in their sensitivity to the effects of NMBAs 121,122 . These differences may lead to imbalanced activation of pump and dilator muscles, thereby generating excessive negative intraluminal pressure. Weakened UA dilator muscles would be unable to compensate for the excessive negative pressure. Even at levels that produce minimal blockade (as measured by train-of-four ratio 0.5 to 1), NMBAs increased UA collapsibility and impaired the compensatory genioglossus response to negative pharyngeal pressure challenges 117  Following surgery, opioids are commonly used for the management of postoperative pain. The use of opioids has been increasingly identified as a contributor to postoperative exacerbation of sleepdisordered breathing 94, 125,126 . OSA patients show a lower pain threshold [127][128][129][130] and increased sensitivity to the respiratory depressant effects of opioids 130 , both of which are of particular relevance in the postoperative setting. Increased UA resistance has been described in cats after opioid application 131 and may be mediated by direct inhibition of hypoglossal motor neurons with suppressed genioglossus activity 132 . Therefore, the use of opioids during and immediately after surgery can be an important perioperative factor to consider in patients with OSA. Some data suggest that some interventions such as elevated upper body position 94 or continuous positive airway pressure (CPAP) 133 can ameliorate the respiratory depressant effects of opioids.
Finally, following surgery, patients commonly experience disrupted, reduced, and poor-quality sleep. Sleep fragmentation can reduce the rapid eye movement (REM) sleep stage 134,135 . Following sleep deprivation and fragmentation, a rebound effect with increased amount of REM sleep can be seen a few nights after surgery 134,136 . REM sleep is primetime for sleep apnea, since it is associated with muscle atonia and impaired respiratory arousal 137 . Sleep deprivation may also contribute to the development of delirium, further disrupting sleep patterns and cortical arousals 138 .

How to manage patients with OSA perioperatively
Despite the high prevalence of OSA in surgical populations, standardized guidelines for the safe handling of this patient population are limited 1 . The imperative is to provide the highest level of quality care while scheduling surgery in a timely manner and minding the expanding cost of providing care. Healthcare resources need to be optimally allocated to improve patients' safety without undue economic impact. Given these restrictions, it is probably not feasible to conduct a sleep study on each patient scheduled for surgery, and there is so far no data indicating that a preoperative sleep medicine consultation improves patient safety. However, a stepwise approach for the detection of patients at risk of sleep apnea may help guide the need for diagnostics and treatment.
Patients should be tested for the risk of sleep apnea, and there are several validated scores available for preoperative testing such as the STOP BANG questionnaire 139,140 . We have recently developed an OSA screening instrument that is supposed to be applied to patients who have not been scheduled to see an anesthesiologist prior to the day of surgery. The Score for Preoperative prediction of OSA (SPOSA) can be used based on data available preoperatively in the electronic medical record without the need to take a physical exam 141 . Once patients are identified to be at a high risk for sleep apnea in the perioperative setting, they demand special attention and care during the perioperative period and anesthesia. To some extent, the need for additional testing may depend on the perioperative risk of the scheduled surgical intervention. Data on the optimal intraoperative and perioperative management of sleep apnea is still limited, but OSA patients undergoing surgery and anesthesia with high risk of morbidity should receive specialized treatment during the perioperative period based on the best available local evidence and experience level.
The use of a standardized algorithm, such as the one developed by physicians at the Massachusetts General Hospital in Boston (Figure 2), may help identify and manage OSA patients facing surgery. In patients with known OSA who have been prescribed CPAP, the use of CPAP is continued during the postoperative period (e.g. in the recovery room). A respiratory therapist may visit the patient preoperatively or postoperatively in the recovery room to make sure the device and its interface function properly. During the postoperative period, when the patient is under lingering effects of anesthetics, CPAP treatment under close guidance of respiratory therapists reduces the number of respiratory events 142 and improves breathing early after surgery 126 . Furthermore, CPAP may even improve postoperative hemodynamics (i.e. blood pressure variance) in patients who are not hypovolemic 142 . CPAP treatment under close guidance of respiratory therapists likely reduces postoperative complications in OSA patients 143 . However, successful perioperative CPAP treatment needs a close collaboration among patients, surgeons, anesthesiologists, and respiratory therapists, and sleep physicians may need to be consulted in selected patients.
For patients not previously diagnosed with OSA, clinical management should be performed based on risk stratification. Patients undergoing surgical procedures with moderate to high risk of perioperative complications should be stratified based on OSA risk. Risk for occult or undiagnosed OSA is based on the clinical assessment (e.g. symptoms and/or comorbidities of sleep apnea; see Table 3) and the use of standardized validated questionnaires (e.g. Berlin questionnaire 144 , STOP questionnaire 140 , P-SAP score 139 , or the SPOSA 141 ). Note that these questionnaires have been validated for the identification of OSA, but not other forms of sleepdisordered breathing (e.g. obesity hypoventilation or central sleep apnea) 145 . Additional testing, such as arterial blood sampling, is required to detect hypercapnia (increased blood carbon dioxide). However, blood gas analysis is not typically included in the standard preoperative workup for surgical patients and might not be available in some settings. In these cases, venous serum bicarbonate concentration, as available on most biochemistry profiles, might be a helpful screening tool for an occult, chronic, respiratory acidosis. A serum bicarbonate level greater than 27 mmol/l has been shown to be highly sensitive (92%) for an elevated arterial partial pressure of carbon dioxide. An elevated bicarbonate level accompanied by mild hypoxemia (peripheral oxygen saturation of 94%) may also indicate high risk of obesity-related alveolar hypoventilation 145,146 .
Patients deemed at high risk for OSA and/or obesity hypoventilation syndrome based on these screening tools may require sleep medicine consultation prior to or following anesthesia. The sleep specialist can help determine the role of a sleep study (home vs. laboratory based), start therapy with positive airway pressure therapy (e.g. auto-titrating CPAP), and develop strategies to "desensitize" patients to the mask and pressures prior to or following an elective surgical procedure.
Throughout the perioperative period, special attention should be paid to patients with confirmed OSA or high-risk patients under special circumstances. In these patients, specific methods should be used during intubation, intraoperatively, during and early after extubation, and during post-anesthesia care unit (PACU) treatment including fluid management, patient positioning, neuromuscular blockade, protective ventilation 53,147 , pain management, and choice of anesthesia type and anesthetic (Table 4). However, the currently available data on the choice of anesthetic for the OSA patient facing surgery are limited, but the choice of a sedative agent to be Opioids are commonly used for the control of surgical pain during and after anesthesia. This is of special importance in OSA patients who have been found to require higher doses of opioids for adequate pain control 162 . While pain management may improve the use of the respiratory pump, these analgesics induce a dose-dependent  132,163 . Recent data from our group and others indicate that treatment of OSA patients with CPAP early after surgery improves sleep apnea and mitigates negative effects of opioid application. An alternative, such as nonsteroidal anti-inflammatory drugs (NSAIDs) or regional anesthesia with local anesthetics, should also be considered.
Given these perioperative factors, the transfer of an OSA patient from the recovery room or ICU to an unmonitored floor should be carefully considered. Patients should not be moved until their vital signs have recovered to values similar to pre-anesthesia baseline and after passing a room air challenge test. Furthermore, adequate treatment of nausea and pain should be accomplished, preferably by NSAIDs, prior to transfer.

Conclusion
When caring for OSA patients facing surgery, the therapeutic team needs to be aware of the increased risk for post-anesthesia respiratory complications. While these complications are not associated with increased mortality risk, the morbidity of preventable complications may lead to undesired expenses, jeopardize available resources, and may lead to an increased hospital readmission rate.
Since the currently available literature on perioperative management of OSA patients is still limited, additional clinical and basic research in this area is needed to improve the safety of OSA patients undergoing anesthesia.
We hypothesize that early recognition and treatment of sleep apnea reduces perioperative complications. Further research is needed to confirm this clinical suspicion and support the use of diagnostic or therapeutic algorithms for these patients.
Pending that research, an institution-specific plan (based on setting, personnel, equipment, medications, and resources) needs to be established for the identification, testing, monitoring, and care of the surgical population. The plan should include (1) stepwise preoperative screening procedures for OSA, (2) an optimized anesthesia regimen and sedation protocol for this high-risk group which eliminates drug-induced respiratory depressant effects at the end of the case, (3) intraoperative neuromuscular monitoring with goal-directed reversal of rNMB, (4) a protocol for the use of CPAP therapy in the recovery room, (5) optimal opioid titration for postoperative pain control, and (6) specific discharge criteria for transfer to the unmonitored ward. Competing interests Sebastian Zaremba declares that he has no disclosures.

CPAP
James E. Mojica declares that he has no disclosures.

Grant information
Matthias Eikermann has received research funding from MERCK and ResMed Foundation, and received grants from Jeff and Judy Buzen.

Open Peer Review Current Referee Status:
Editorial Note on the Review Process are commissioned from members of the prestigious and are edited as a F1000 Faculty Reviews F1000 Faculty service to readers. In order to make these reviews as comprehensive and accessible as possible, the referees provide input before publication and only the final, revised version is published. The referees who approved the final version are listed with their names and affiliations but without their reports on earlier versions (any comments will already have been addressed in the published version).