Upper airway obstruction—partial and complete—during anesthesia and sleep is a consequence of the combination of a narrow, anatomically predisposed airway and the permissive effect of state-related upper airway muscle relaxation, the degree of which varies with anesthetic depth or sleep stage (being most profound in REM sleep). Predisposing features acting to narrow the airway include: skeletal factors that constrict the skeletal confines, such as retrognathia; increased soft tissue around the pharynx, as in obesity and macroglossia; pharyngeal muscle weakness, as seen in many neuromuscular disorders; and as tonsillar and adenoidal hypertrophy. Familial issues also play a part: Some have inherently narrow pharynges. Further, it appears that upper airway neuromotor responses are impaired in OSA patients relative to controls; variability in collapsibility of the upper airway between individuals cannot be explained by structural loads alone [23, 24]. Certain postures also predispose to upper airway obstruction including recumbency, supine position and neck flexion. Alcohol, sedatives and opioids also exacerbate narrowing of the pharyngeal airway through their effects on pharyngeal muscle tone.

There are two main sites for obstruction in the pharynx: the velopharynx (that portion behind the soft palate) and the retrolingual hypopharynx. Indeed, the velopharynx—usually the narrowest segment of the pharynx—appears to be the primary site of obstruction in 80 % of events in both sleep and anesthesia. Increased collapsibility of the upper airway occurs quite abruptly with the encephalographic alpha–theta transition at sleep onset and with loss of consciousness/rousability during anesthetic induction [25, 26]. This emphasizes the importance of conscious state as a determinant of muscle activation.

Predicting those with narrow “difficult” upper airways is a time honored pursuit in anesthesia, mainly in relation to predicting difficult tracheal intubation. “Can’t intubate, can’t ventilate,” where the patient cannot be tracheally intubated or ventilated by bag and mask after induction of anesthesia is a dreaded scenario and identifying patients at risk allows such problems to be circumvented (e.g., by awake intubation) or avoided (e.g., by use of regional anesthetic techniques rather than general anesthesia). Clinical predictors of difficult intubation include oropharyngeal crowding (the Mallampati score) and a short thyromental distance (signifying micro- or retrognathia) [27]. These have also been found to be of value in predicting those with obstructive sleep apnea (OSA). Indeed, there is a close relationship between difficult intubation and the presence of OSA [28]. Furthermore, the tendencies to upper airway obstruction during anesthesia and during sleep are related [7]. Pharyngeal collapsibility under anesthesia is substantially higher among OSA patients than those without OSA [29]. An airway that proves “difficult” in one state tends to also be difficult during the other.

It follows that where such problems are noted in sleep or anesthesia, then implications for behavior in the other must be considered. For example, difficulty with intubation should prompt the anesthesiologist to consider whether the patient has OSA, as should increased propensity to upper airway obstruction during anesthesia or recovery. Similarly a history of OSA should alert the anesthesiologist of increased difficulty with airway management during anesthesia and increased risk of obstruction in the postoperative period. This risk is compounded by compromise of protective arousal responses by anesthetic, analgesic and sedative drugs during this time.

These issues emphasize the importance of identifying at-risk patients preoperatively. There is a growing literature documenting increased risk of perioperative cardiorespiratory complications in patients with OSA [30–32]. Particular risk appears to attend major surgery where there is a need for postoperative opioids and sedatives, presumably because of the depressant effects these can have on protective arousal responses. There is significant variability between individuals in their sensitivity to these depressant effects. Recently, it has been shown that in volunteers at high risk of OSA, nocturnal hypoxemia is associated with greater sensitivity to opioids, suggesting that increased potency of opioid analgesia could be a marker of its propensity to depress respiration during sleep [33].

Identifying OSA and predicting patients those most at risk of perioperative complications (“risk stratification”) is a challenge for anesthesiologists. Important factors determining risk are likely to include degree of collapsibility of the relaxed upper airway, susceptibility to the depressant effects of analgesic and sedative drugs, and sensitivity of arousal responses in the case of an obstructive event. There is a general relationship between degree of airway collapsibility and apnea–hypopnea index (AHI), making this (as in sleep medicine) an important metric of severity [34]. However, the safety issues center on capacity to arouse from an obstructive event. The propensity to arousal failure might be better reflected by other metrics such as length of apneas and hypopneas and the degree of associated hypoxemia. Those individuals with lengthy apneas accompanied by deep desaturations on preoperative sleep study are of concern, as these characteristics suggest little tolerance for further depression of arousal [35]. It is notable that (in contrast to multiday surgery) outpatient surgery does not appear to be associated with demonstrable increased risk of postoperative adverse events in OSA patients [36]. This may reflect the fact that such patients, having emerged from anesthesia, are not exposed to the risk of re-suppression of arousal responses through postoperative administration of opioids and sedatives [37].

Obesity is an important OSA consideration in several ways. Besides being a risk factor for OSA, it predisposes to hypoventilation and to disordered gas exchange during sleep and anesthesia. Obesity-related reduction in functional residual capacity (FRC) when recumbent and asleep or anesthetized promotes atelectasis in the dependent regions of the lungs with consequent shunt. The reduction in FRC reduces oxygen stores in the lungs. Both these effects magnify the degree of desaturation observed during an obstructive event of a given duration [38].

Upper airway surgery presents additional risk for patients with OSA, as postoperative edema can temporarily further narrow the airway. Somewhat paradoxically, such problems can be seen in the early postoperative course following palatal surgery to treat snoring and OSA [39]. Children undergoing upper airway surgery for OSA present particular risks based on their small airway dimensions and low lung volumes relative to body size [40].

Because of its potential for perioperative problems and because its presence helps alert the anesthesiologist to the possibility of difficulties with airway maintenance intraoperatively, identifying OSA preoperatively is an important task. A popular tool is the acronymic Stop-BANG questionnaire, which asks 8 questions relating to OSA risk: the presence of loud Snoring, daytime Tiredness and lethargy, Observed apneas, history of elevated blood Pressure, a Body mass index of greater than 35 kg/m², Age greater than 50 years, a Neck circumference more than 40 cm and male Gender [41, 42]. A positive response to 5 or more of these questions indicates high risk of OSA, while a score of 3–4 indicates intermediate risk. A problem with the questionnaire is that while highly sensitive for the presence of OSA, it is relatively non-specific. This together with a high community prevalence of OSA means that it designates many patients as at risk, with a high rate of false positives. This leaves anesthesiologists with a dilemma: which patients justify extra resources perioperatively? How do they risk stratify? The answers are not yet clear, but factors such as severity of OSA, inherent arousal thresholds, postoperative opioid and sedative requirements, sensitivity to the depressant effects of these, and presence of comorbidities including obesity and vascular disease are all likely to be influential.

Current guidelines for the perioperative management of OSA are understandably cautious [43]. Some general principles apply (Table 10.1). Circumventing the difficulties associated with general anesthesia and postoperative sedative and analgesic by use of regional anesthetic/analgesic techniques is an attractive option if appropriate for the surgical procedure [44]. Where general anesthesia is required, then use of anesthetic drugs that readily reverse at the end of the procedure and, where possible, use of non-opioid analgesics following it are likely to help. Close observation is required until the patient is sentient and not likely to be exposed to further treatments which have potential to compromise arousal. Availability and use of positive airway pressure (PAP) therapies postoperatively are further tenet. These treatments are much more easily implemented if the patient already has some familiarity with them. This provides a strong rationale for preoperative diagnosis and initiation of treatment where practicable.

Besides OSA, another major category of sleep-disordered breathing is (non-obstructive) sleep hypoventilation. The sleep-associated decrease in ventilatory drive, a consequence of the combination of loss of the stimulatory effects of wakefulness and a specific state-related down-regulation of hypercapnic and hypoxic ventilatory responses, ensures that any disorder with the potential for hypoventilation is exacerbated by sleep, particularly REM sleep where ventilatory drive is at its nadir. Such disorders have in common an imbalance between the load on respiratory muscles and their capacity to cope with these loads. Common examples of conditions that increase the load on muscles include obesity and advanced lung diseases such as chronic obstructive pulmonary disease. Conditions that decrease their capacity to cope include a wide variety of neuromuscular disorders affecting the respiratory (and upper airway) muscles, high spinal injuries, diaphragmatic palsy and the inefficiencies associated with the low flat diaphragm of emphysema and use of respiratory depressant drugs. Emphysema is an example of a condition that both loads the muscles (by increasing airway resistance) and decreases their capacity to cope (through the hyperinflation-associated decrease in inspiratory muscle mechanical advantage). Often hypoventilation coexists with upper airway obstruction, as is the case in obesity hypoventilation syndrome and in neuromuscular disorders that involve both upper airway and respiratory pump muscles. Obesity hypoventilation is common in obese patients with OSA: a recent meta-analysis of more than 4000 such patients demonstrated that 19 % of them had daytime hypercapnia, fulfilling the criteria for obesity hypoventilation syndrome [45].

Where there is a gross imbalance between load on respiratory muscles and their capacity to cope, a state of type 2 (hypercapnic) ventilatory failure exists and continuous mechanical ventilatory support is required. In conditions associated with a slow evolution in this imbalance (such as progressive neuromuscular problems or obesity), sleep hypoventilation may precede daytime respiratory failure by months or years [46]. Indeed sleep hypoventilation is both a harbinger of wakeful type 2 respiratory failure and a contributor to its evolution. The hypercapnia and hypoxemia that occur with significant sleep hypoventilation cause a loss of sensitivity to these ventilatory stimulants, through buffering of hypercapnia-associated decreases in extracellular pH levels and roll-off of hypoxic responsiveness. Intervention along this evolutionary pathway with bi-level ventilatory assistance, delivered noninvasively during sleep, can prevent or reverse daytime respiratory failure where the imbalance between load and capacity on muscles is of moderate degree [46].

Not surprisingly, given their similar effects on ventilatory drive, the same problems predispose to hypoventilation during anesthesia where spontaneous ventilation is preserved. Despite this, sleep hypoventilation has not yet been seriously examined as a risk factor for postoperative respiratory complications, including respiratory failure. However, a recent retrospective study has shown that most patients with obesity hypoventilation syndrome are unrecognized at the time of elective surgery and that they are at substantially greater risk of respiratory failure after elective surgery than patients with OSA alone (44.4 vs. 2.6 %) [30]. Where postoperative respiratory failure occurs, the possibility of sleep-related hypoventilation should be considered, particularly where predispositions exist and where there is not an acute event (such a pneumonia or pulmonary edema) to explain it. Such patients often require bi-level ventilatory assistance rather than CPAP. An empiric inspiratory PAP of 16–18 cm H₂O and expiratory PAP of 9–10 cm H₂O can be used to initiate therapy to overcome upper airway obstruction and improve ventilation with subsequent titration conducted according to therapeutic response [42].

Besides OSA and sleep hypoventilation, a further form of sleep-disordered breathing is periodic breathing. While this can also be present during wakefulness, it is always worse during, and often confined to, sleep because wakefulness provides its own stimulatory, dampening effect on periodicity [47]. It can take the form of Cheyne–Stokes respiration where there is regular waxing and waning of ventilation, as seen in heart failure and cerebrovascular disease, or ataxic (Biot’s) breathing where the waxing/waning pattern is irregular, as seen with chronic opioid use. The combination of periodic breathing and OSA is a form of “complex” sleep apnea. It is notable that untreated OSA aggravates heart failure and so the problems of OSA and periodic breathing may not only coexist, but also be inter-dependent.