Perioperative incidence of airway obstructive and hypoxemic events in patients with confirmed or suspected sleep apnea - a prospective, randomized pilot study comparing propofol/remifentanil and sevoflurane/remifentanil anesthesia

Obstructive sleep apnea (OSA) is a common sleep related breathing disorder [1, 2] and it is characterized by recurrent nocturnal episodes of hypopnea and apnea evoking decreases in peripheral arterial oxygen saturation (SpO₂) and other sequelae [3, 4]. The prevalence of OSA among the general population is high [5] and it is estimated that 11.4% of men and 4.7% of women suffer from moderate to severe OSA [6‐8]. In surgical patients, its prevalence is even greater with up to 22% [9].

Several studies and meta-analyses have shown that OSA is an independent risk factor for a variety of perioperative pulmonary complications such as aspiration pneumonia, acute respiratory distress syndrome, pulmonary emboli, and reintubation / mechanical ventilation [10, 11]. Unfortunately, the vast majority (i.e., 70–90%) of OSA patients presents undiagnosed before surgery [2, 12]. While the American Society of Anesthesiologists has published guidelines for the perioperative management of patients with OSA, recommendations for specific anesthetic regimen do not exist [13]. In fact, randomized studies addressing specific anesthetic regimen for the management of OSA patients are missing. Furthermore, it is unclear whether OSA patients need special monitoring or surveillance following anesthesia/surgery, which is particular relevant in an ambulatory setting [14].

Accordingly, in a cohort of patients with established or strongly suspected OSA, we prospectively compared in a randomized fashion the effects on nocturnal apnea-hypopnea-index (AHI) and SpO₂ of the volatile anesthetic sevoflurane and the intravenous anesthetic propofol when combined with the short acting opioid remifentanil for surgery of moderate duration and severity. To determine whether anesthetic regimens altered the incidence of obstructive and hypoxemic events variables were assessed both during the pre- and postoperative nights. Specifically, we tested the null hypotheses that nocturnal AHI and minimum SpO₂ did not differ between 1) anesthetic regimens, and 2) preoperative vs. postoperative nights.

After IRB approval and study registration (German Clinical Trials Register number DRKS00005824) we screened and enrolled after informed consent adult elective surgery patients with an established or suspected diagnosis of OSA, as defined by a positive prior overnight polysomnography and/or ≥2 positive STOP-criteria [15]. We enrolled patients undergoing surgery of moderate duration and severity which might be performed also in an ambulatory setting. Briefly, the STOP-questionnaire contains four questions (Do you Snore loudly? Do you often feel Tired, fatigued, or sleepy during daytime? Has anyone Observed you stop breathing during your sleep? Do you have or are you being treated for high blood Pressure?), and positive answers to two or more questions classify patients at high risk for OSA.

OSA patients already on nocturnal continuous positive airway pressure (CPAP) therapy were excluded to avoid withdrawal of an established therapy solely for the purpose of this study. Further exclusion criteria were planned airway surgery, any procedure likely requiring postoperative mechanical ventilation, patients undergoing regional anesthesia, and a patient’s refusal to study participation.

The main types of surgery in our patients were eye surgery (including eye enucleation) (16 in the propofol group, 12 in the sevoflurane group), musculoskeletal surgery (7 and 6, respectively), and gynecology/urology surgery (3 and 6, respectively). The types of surgery, including the positioning of the patients, were not different between the groups.

The study flow chart is depicted in Fig. 1. Seventy-seven patients gave consent to participate and were instrumented for the preoperative nocturnal measurements. Of those patients, 23 had to be excluded after study enrolment because of loss of data acquisition (involuntary or voluntary removal of sensors) or secondary withdrawal of consent, leaving 54 patients for analysis. Twenty-one (38%) patients reported a sleep-laboratory established diagnosis of OSA (but did not receive CPAP-therapy) and 33 (62%) patients fulfilled two or more positive STOP-criteria.

Twenty-seven patients received propofol/remifentanil anesthesia and another 27 patients sevoflurane/remifentanil anesthesia. Anesthesia duration averaged 97 min ± 43 (SD) and included breast cancer surgery, intervertebral disc surgery, eye/ENT surgery, and urologic surgery. The patients’ demographic data are shown in Table 1.

This study, to our knowledge, is the first that prospectively and randomly compares the effects of two anesthetic regimens generally deemed suitable in OSA patients. Our data show, that 1) there was no difference in variables between propofol/remifentanil or sevoflurane/remifentanil anesthesia, and 2) postoperative nocturnal AHI and minimum SpO₂ were unchanged when compared to the patients preoperative night. Thus, both anesthesia regimens did not worsen nocturnal airway obstruction or increase hypoxemic episodes, at least following surgeries of moderate duration and severity that could have been performed also in an ambulatory setting.

While OSA clearly is a risk factor for postoperative complications (for review see [14]) the contribution of specific anesthetic regimen is unknown and randomized studies are missing. Our finding, that neither anesthetic regimen increased postoperative nocturnal obstructive and hypoxemic episodes is consistent with other studies in OSA patients following general anesthesia [19, 20]. In fact, in a study of 31 obese OSA and non-OSA patients undergoing desflurane/remifentanil anesthesia no postoperative alterations of nocturnal postoperative oxygen saturation were found compared to their preoperative baseline [19]. Likewise, in 58 patients undergoing either regional or general anesthesia with various agents (sevoflurane, desflurane, fentanyl, morphine, and/or hydromorphone) and receiving polysomnography for up to 7 postoperative nights, AHI and minimum SpO₂ also did not significantly change during the first postoperative night [20]. These findings are also consistent with a retrospective analysis of 234 ambulatory surgical patients with polysomnography confirmed OSA which revealed no greater incidence of unanticipated hospital admission or other adverse events than in non-OSA case controls [21]. In contrast, Chung et al. found a small increase in AHI in patients with mild and moderate OSA but surprisingly not in patients with severe OSA [22]. Accordingly, while OSA patients are at greater postoperative risk than non-OSA patients it is not evident whether specific anesthesia regimens alter nocturnal AHI and oxygenation. Thus, our study shows that the two anesthetic regimens based on the inhalational anesthetic sevoflurane and the intravenous anesthetic propofol appear safe, at least for the type of surgeries and patient cohorts studied, and do not increase nocturnal obstructive events.

In our study, anesthesia regimens were strictly standardized, rather short acting drugs were used, and substantial residual neuromuscular blockade was ruled out before extubation. Most prior observational studies in OSA patients do not have such a high degree of standardization. In particular, if longer acting neuromuscular drugs are used in the absence of neuromuscular monitoring greater interference with nocturnal upper airway patency and oxygenation irrespective of the specific anesthetic drugs may not be excluded [23].

While the literature and guidelines provide little help in defining an appropriate duration of postoperative respiratory monitoring in patients with OSA, the first postoperative night is generally believed to constitute a particular risk for OSA patients [24] and, therefore, monitoring is usually applied during the first postoperative night only in OSA patients. The question remains whether measurements of AHI and minimum oxygenation during the first postoperative night in OSA patients are representative for their entire postoperative course. In fact, in patients undergoing polysomnography for up to 7 postoperative nights [20, 22] the frequency of sleep disordered breathing appeared to increase on the third postoperative night, raising the hypothesis that sleep disordered breathing, if anything, may not deteriorate until the third postoperative night. While this question has to be addressed, the latter data are consistent with the provocative speculation that OSA patients undergoing surgery/anesthesia of moderate duration do not deteriorate the night following anesthesia, but may require special monitoring or surveillance later.

Several limitations of our pilot study should be mentioned. First, the majority of our patients did not suffer from the most severe degree of OSA, due to the exclusion of OSA patients already on nocturnal home CPAP therapy. Even some patients with a preexisting sleep laboratory derived OSA diagnosis had a low AHI in our study. Intra-individual variability in OSA severity is well recognized and may be partly explained by day-to-day changes in evening leg fluid volume and overnight rostral fluid shift [25]. However, even in our patients with severe OSA AHI and SpO₂ were not altered by either anesthetic regimen. Second, not all of our patients had a polysomnography established OSA diagnosis and the STOP-Questionnaire has a lesser sensitivity and specificity in identifying such patients [15]. However, this applies to both of our subcohorts and even patients with a high preoperative nocturnal AHI revealed an unchanged AHI the night following anesthesia. Third, we did not assess the doses of analgesics that our patients received after discharge from the PACU, what might be a potential confounder if these doses were significantly different between the groups. However, as the doses of opioids received in the PACU and the types and length of surgery did not differ between the groups, we are confident that unmeasured opioid intake did not affect results. Fourth, another potential confounder would be chronic alcohol consumption, as past and/or current alcohol consumption influences the prevalence and severity of obstructive sleep apnea [26, 27]. However, none of our patients revealed such a history on preanesthetic evaluation. Finally, the number of patients was small and different results may be observed in patients with more extensive surgeries, in particular abdominal or intrathoracic surgery. These issues must be specifically addressed in different, also very standardized studies.

Neither propofol/remifentanil nor sevoflurane/remifentanil anesthesia worsened the incidence of postoperative nocturnal obstructive or hypoxemic events following medium duration surgery in a cohort of patients with established or strongly suspected OSA when compared to the preoperative night. Furthermore, there was no difference between these anesthetic regimens. Thus, while our data suggest that both anesthetic regimen are safe in patients with confirmed or suspected OSA undergoing anesthesia of moderate duration, they also support the view that these patients do not deteriorate below their preoperative baseline the night following anesthesia.