This study represents the largest multi-surgeon series in the literature of patients undergoing UPPP with or without additional procedures as both outpatient and inpatient surgery individually and combined. Ninety-seven patients (28.1%) were discharged on the day of the surgery. Only three patients suffered serious complications, all occurring within two hours of surgery, that could be directly associated to the underlying OSA.
Fairbanks et al . describe 12 anecdotal cases of post-operative mortality secondary to airway obstruction. However, no information is available regarding patient’s preoperative health, OSA severity, or postoperative course in that study. Haavisto & Juonpaa  describe one case of death immediately after extubation. Kezirian et al . describe 7 deaths in 3130 patients following UPPP, but the causes of death are not described. No other studies report post-operative death secondary to airway complications. Recent studies show no postoperative mortality [5, 7, 8, 13–15]. This is likely due to increased awareness of risk factors, decreased narcotic use, and perioperative steroid use.
The incidence of all respiratory complications (including laryngospasm and pulmonary edema) in our series was 13.3%; when examining only uncomplicated oxygen desaturations, the incidence was lower at 12.8%. Respiratory complication rates in the literature range from 1.4%  to 15.3% . Some authors only include serious respiratory complications such as pneumonia, tracheostomy, or pulmonary edema, and do not consider a simple oxygen desaturation to be a complication. In addition, many centres in previous studies provide their OSA population oxygen supplementation until discharge from hospital as routine care. These factors may explain the variability in the respiratory complication rates stated in previous studies. In our study, all respiratory complications, with the exception of one case of laryngospasm, were that of decreased oxyhemoglobin saturation and were treated with oxygen supplementation. It is questionable however if simple desaturations should actually be considered complications in this population. Patients with OSA have generally been desaturating nightly for years prior to surgery as part of their OSA, and thus it may be invalid to consider post-operative desaturations as causally related to the UPPP as opposed to the underlying disease status. It is also questionable whether routine oxygen administration should be given after OSA surgery to treat desaturations. These topics are significantly understudied in the OSA literature, and merit further investigation.
Comparison of inpatient and outpatient groups showed that outpatients were significantly younger and had a significantly lower mean AHI. This likely represents a selection bias as surgeons may have preferentially selected younger patients with mild OSA for outpatient UPPP. The incidence of total complications and oxyhemoglobin desaturations in the PACU alone were not significantly different between inpatient and outpatient groups. This result suggests that complications in the PACU may not have influenced surgeon’s decisions to discharge patients.
Our results suggest that the most significant risk factor for the development of post-PACU respiratory complications is the presence of single or multiple co-morbidities. Previous studies describe AHI [3, 14–16], preoperative LSAT (lowest oxyhemoglobin saturation) [13, 14, 16], BMI , or no factors [6, 8] related to the incidence of postoperative respiratory complications. This discrepancy likely represents the inaccuracy in determining the rate of respiratory complications due to oxygen supplementation post-operatively. To correct for this factor, we compared patients undergoing UPPP with or without tonsillectomy that permanently stopped oxygen supplementation in the PACU while maintaining an oxyhemoglobin saturation > 88% with those requiring oxygen in the ward. We found that both BMI and AHI were significantly higher in patients requiring supplemental oxygen in the ward. Odds ratio calculations revealed that obese and morbidly obese patients (BMI ≥ 30) who had an AHI ≥ 22 were at significant risk of requiring oxygen supplementation on the ward and may warrant an overnight stay. Our results, therefore, suggests that these AHI and BMI values can be useful for identifying patients who can undergo safe outpatient UPPP surgery. It should be noted that it is likely that many patients received oxygen supplementation on the ward unnecessarily as a trial of oxygen discontinuation was not attempted in the PACU in all patients. This may bias our results.
The decision to discharge postoperative patients varies across surgical centres. Chung et al . devised the post-anesthesia discharge scoring system (PADS) for providing a uniform assessment of all ambulatory surgical patients based on vital signs, pain, ambulation, nausea / vomiting, and bleeding. Specifically, vital signs within 20% of preoperative baseline are generally considered as adequate prior to discharge. All previous studies looking at postoperative sleep studies show no significant change in post-operative AHI, LSAT (lowest oxyhemoglobin saturation), or both in the immediate post-operative setting compared to preoperative levels [8, 18, 19]. In our study, 91.3% of all inpatients who were successfully taken off oxygen supplementation in the PACU did not require further oxygen treatment overnight. This data suggests that the risk of postoperative oxyhemoglobin desaturation in OSA patients is not significantly elevated compared to their preoperative baseline.
There are a few limitations to our study. Being a retrospective study, there is potential for a selection bias. However, we strictly adhered to our inclusion and exclusion criteria. As in previous studies, our respiratory complication rates may have been skewed by postoperative oxygen supplementation. We tried to account for this by recording oxygen use postoperatively and utilizing this data in our analysis. A key variable of lowest oxygen saturation level could not be identified in many of the records, therefore limiting the conclusions that can be drawn from this review. In addition, many patients did not have available preoperative AHI values. It is unclear whether this may have introduced a surgical selection bias. The level of CPAP adherence prior to surgery was not available for many records, especially the remote ones; CPAP adherence pre-operatively can in some cases affect respiratory functioning after surgery, hence this variable could not be accounted for in our study. However, it is known that the overall rate of CPAP adherence is low especially in a study population of patients undergoing surgery, hence we do not feel the lack of this variable will significantly affect the overall findings of this study. Finally, the decision to perform outpatient UPPP patients was generally based on surgeon’s preference. Although patients were required to maintain oxyhemoglobin saturation > 88% with pain and nausea controlled on oral medication, some surgeons may have elected to keep patients overnight despite patient factors. This may bias comparisons of outpatient and inpatient UPPP.