Department of Anesthesia and Perioperative Medicine, St Joseph’s Health Care, Western University, London Ontario
Department of Medicine, Division of Critical Care, Western University, London, Ontario
Department of Otolaryngology – Head and Neck Surgery, Western University, London, Ontario
Published Date: 13/07/2021.
*Corresponding author: Raymond Kao, Department of Medicine, Division of Critical Care, Western University, 800 Commissioners’ Rd, London, Ontario, Canada
Purpose: Practice guidelines recommend monitoring post-operative obstructive sleep apnea (OSA) patients longer than non-OSA patients. Literature suggests patients with OSA have a higher risk for postoperative pulmonary complications, cardiac events and ICU admission. The objective of this study is to determine the optimal duration of post-operative monitoring.
Methods: This is a retrospective review of two-hundred and three patient charts, age 18-83 years old (69.46% male) who underwent surgery from June 2011- March 2014. OSA patients were identified either by AHI diagnosis or by preoperative questionnaire. Patients were monitored in the Post Anesthetic Care Unit (PACU) for 4 hours seeking desaturation and/or apnea events equals ‘fail’ to guide overnight ward monitoring (5-24 hours) or no event equals ‘pass’ monitoring.
Results: Seventy-five patients (N=38 with AHI and N=37 at risk, 36.95%) had desaturation events in PACU. The mean time to desaturation was 55 ±43.33min. Females had more desaturation than males, 50.00% vs. 31.21%, p=0.01 with OR=1.91, 95%CI 0.90-4.05; Female age ≥ 50 with BMI >35 had more desaturation that male age ≥ 50 with BMI>35, 77.42% vs. 63.64%, but not statistically significant, p=0.51 with OR=3.95, 95% CI 2.00-7.80. No significant difference in mean AHI between passed or failed post-op monitoring (51.04±23.75 vs. 43.54±25.23, p=0.17), but AHI>40 and male (OR=3.18, 95%CI 0.38-26.62) has higher risk of desaturation compared to AHI=0-20. The logistic regression model has good discriminating ability between passed and failed monitored patients, C-statistic 0.7756.
Conclusion: All PACU desaturations occurred less than 3 hours after surgery, with no other adverse events recorded. Females, patients with OSA and BMI >35 had significantly higher rates of oxygen desaturation. It may be possible to limit postoperative monitoring for OSA positive and at risk patients up to 3 hours.
Keywords: Post-operative; Obstructive sleep apnea (OSA); Complications; Monitoring
Patients with obstructive sleep apnea (OSA) experience periods of partial or complete upper airway obstruction lasting >10 seconds during sleep which can lead to frequent arousals, hypoxia, hypercarbia and/or cardiovascular dysfunction . Approximately 24% of patients presenting for elective surgeries are found to be at high risk for OSA based on preoperative screening . These patients are thought to be more sensitive to the respiratory depressant effects of opioids and sedatives. There is also evidence of increased apnea-hypopnea index (AHI) in postoperative OSA and non-OSA patients . These post-operative patients are at higher risk of pulmonary complications including respiratory failure with desaturation and reintubation; cardiac events such as myocardial infarction and arrhythmia; and ICU admission [4,5]. However, other data suggests that neither diagnosis of OSA nor positive risk screen for OSA was associated with increased 30-day or 1-year postoperative mortality .
There is limited evidence in guiding postoperative monitoring and management of OSA patients or patients at risk for OSA. The 2006 American Society of Anesthesiologists practice guideline, recommended monitoring of postoperative patients with OSA for at least 3 hours longer than non-OSA patients [1,7]. At St. Josephs Health Care (SJHC) London, protocol for postoperative monitoring of OSA patients were developed, based on the 2006 ASA guideline. In some centers such monitoring can mean admission to the intensive care unit postoperatively. The ASA guidelines recognize that there is insufficient evidence to determine either a) the appropriate duration of PACU monitoring or b) whether it is safe to discharge the patient to a regular ward or home versus overnight oximetry monitoring. At SJHC in London, Ontario, 80% are outpatients’ day surgeries. The main surgical specialties are hand and upper limb, urology, ophthalmology, outpatient Otolaryngology – Head and Neck Surgery, general surgery and gynecology. Surgeries are low to intermediate intensity and patients are mainly ASA class I-III. Patients with known or suspected OSA are monitored for in the PACU for apneas or desaturations, as per the ASA guidelines. The 4 hours duration includes time to meet usual PACU discharge criteria plus 3 hours additional monitoring. Patients were deemed to ‘pass’ the monitoring if no events were detected, and to have ‘fail’ if desaturation or apnea occurred. Physicians and clinical nursing staff working in the PACU developed the clinical impression that patients would pass or fail the monitoring protocol significantly before completion of the 4 hours monitoring. It has been suggested that the monitoring could be shortened to 2 hours, potentially economizing hospital resources and allowing patients to be discharged to unmonitored ward or home sooner. Our primary study objective is to determine if postoperative monitoring of OSA patients and patients at risk for OSA can be shorter than 4 hours. Our hypothesis is that postoperative monitoring for OSA patients and patients at risk for OSA can be limited to 2 hours in the PACU. We also aimed to determine if this monitoring would allow us to predict and/or prevent postoperative complications in these patients.
A retrospective chart review between June 2011 and March 2014 of all patients undergoing extended postoperative monitoring at SJHC was completed. This study was approved by the Research Ethics Board at Western University, approval number 105093. We reviewed preoperative questionnaires, polysomnography (PSG) records, anesthesia records, and PACU records. From these sources we recorded documented apnea-hypopnea index (AHI), demographics, comorbidities, surgical procedure, and perioperative medications including inhalation agents, muscle relaxant and narcotics, length of procedure, as well as postoperative oxyhemoglobin saturation, PACU administration of narcotics, time from admission to PACU to desaturation or apnea. We also reviewed the inpatient chart if the patient was admitted prior to their surgery, and the extended electronic medical record for evidence of adverse outcomes. We searched for postoperative visits to the Emergency Room, clinics, and readmissions for up to ten days postoperatively. A total of 237 patients were reviewed and after exclusion of patients for reasons outlined in Figure 1, a total of 203 patients were entered into the study.
Our institutional policy identifies patients for extended monitoring if they have an existing PSG diagnosis of moderate (AHI 21-40) or severe (AHI >40) OSA, or are classified as probably having obstructive sleep apnea by the tool provided in the 2006 ASA Practice Guideline.1 Patients who met the criteria for definite OSA were extended monitored with oximetry, Table 1A. Eligible patients were monitored in the PACU for 4 hours, assessing for apnea of greater than 10 seconds in duration or desaturation with oxyhemoglobin saturation less than 90% while the patient was asleep or awake. Patients were given Continuous Positive Airway Pressure (CPAP) therapy if routinely used at home. Disposition after PACU was determined by results of monitoring. The options were either if there was no respiratory event during 4 hours monitoring the patient was discharged home or to the regular ward at the discretion of the surgeon or if there were apneas or desaturations during monitoring, the patient was admitted for overnight oximetry monitoring on the ward. The outcomes measured were failure of 4 hours monitoring as defined by desaturation or apnea event in PACU and incidence of postoperative complication occurring on the postoperative ward or following discharge.
Table 1A: Baseline characteristics comparing patients identified with or at risk for obstructive sleep apnea for extended Post Anesthesia Care Unit (PACU) monitoring: passed (no desaturation) or failed (desaturation < 90%).
Figure 1: 237 patients identified in SJHC Registry extended monitoring. After excluding patients who failed to meet the inclusion criteria, a total of 203 patients were entered into the study.
For continuous variables, data are expressed as mean ± SD and comparisons conducted using the student’s t-test. For categorical variables, data are reported as proportions and comparison made using Pearson’s Chi-square test. A 95% confidence interval computed for all continuous variables. All tests presented are two-sided, a p-value ≥ 0.05 and the range of the 95% confidence interval contains the value of no effect then considered statistically insignificant. For dichotomous outcomes, logistic regression was used to evaluate clinical factors predicting passed or failed at 4 hours post-operative monitoring. Independent variables included in the regression models were age, sex, body mass index (BMI), AHI, post-operative narcotic use and duration of anesthesia for the surgical procedure. All statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).
Table 1B: Baseline characteristics of patients failed (desaturation<90%) based on gender.
Table 2: Multivariable logistic regression analysis of post-operative monitoring of patients with or at risk for obstructive sleep apnea, N=203 at Saint Joseph Health Centre, London Ontario, Canada.
Table 3: Hosmer and Lemeshow and Goodness-of-fit test for the multivariable logistic regression model.
Figure 2: Receiver operating curve (ROC) of post-operative monitoring of patients with or at risk for obstructive sleep apnea (OSA), N=203 at St. Joseph Health Centre, London Ontario, Canada. C-Statistic=0.7756.
From June 2011 to March 2014, 237 patients underwent extended postoperative monitoring for obstructive sleep apnea. After administrative cleaning, 203 patients were included in the study, Figure 1. Twelve patients had repeat surgeries, so the second admission was excluded. One patient could not be identified due to typographic error. Nine patients did not complete the full monitoring period due to insufficient PACU staffing and 12 patients had incomplete data in their medical records.
In Table 1A, no apnea events were recorded and 75 patients (36.95%) had a desaturation in the PACU. Females, N=31/62 (50.00%) were more likely to have desaturation than males, N=44/141 (31.24%), p=0.01. The mean time from PACU admission to desaturation was 87.55±43.33 minutes. No patients had desaturation between 0-30minutes, likely because all patients received oxygen by mask from PACU arrival until fully awake; 19/75(25.33%) patients had desaturation between 31-60minutes; 39/75(52.00%) had desaturation between 61-90minutes; 11/75(14.67%) had desaturation between 91-120minutes and 6/75(8.00%) had desaturation between 121-180minutes. There was no significant difference in desaturation based on mean age, p=0.17, but for age > 50, N=55 (73.33%) had a higher desaturation rate than age ≤ 50, N=20 (26.67%). Patients with high mean BMI more likely to have desaturation then lower mean BMI, (39.15±8.23 vs. 33.09±7.46, p<0.0001); BMI>35, N=52 (69.33%) were more likely to have desaturation compared to patients with BMI<35, N=23 (30.67%). Patients identified by screening questionnaire for OSA, N=37/121 (30.58%) were less likely to fail desaturation than patients with AHI diagnosis of OSA, N=38/82 (46.34%). In the desaturation group, patients with ASA 3, N=44 (58.67%) compared to ASA 1, 2 and 4 and those with chronic disease hypertension, N=50 (66.67%) and diabetes, N=26 (34.67%) compared to other chronic diseases had the highest failed monitoring at PACU. There was no statistically significant difference between the failed and passed monitoring groups at PACU for the adjusted post-operation narcotic use.
In Table 1B, desaturation by gender revealed male N=44 (58.67%) and female N=31(41.33%). Although the mean age between sexes with desaturation was not statistically significant, p=0.05, but for both gender age > 50 years old had higher proportion of desaturation [male N=32/44(72.73%) vs. female N=23/31(74.19%)] than age ≤ 50 years old [male N=12/44(27.27%) vs. female N=8/31(25.81%)]. The mean BMI between gender was not statistically significant, p=0.51, but for both gender BMI > 35 had higher proportion of desaturation [male, N=28/44(63.64%) vs. female, N=24/31(77.42%)] than BMI ≤ 35 [male, N=16/44(36.36%) vs. female, N=7/31(22.58%)]. For those patients with AHI, this was not statistically significant between genders. But male with severe AHI, N =13/23(56.52%) and female with moderate AHI, N=8/16 (50.00%) had the highest proportion of desaturation. The following chronic diseases have the highest proportion of desaturation in both genders, hypertension [male, N=30/44 (68.18%) vs. female, N=20/31 (64.52%)], diabetes [male, N=14/44 (31.82%) vs. female, N=20/31 (64.52%)] and asthma [male, N=8/44 (18.18%) vs. female, N=7/31 (22.58%)]. There is no statistical significance between genders with respect to number of pack-year smoking, the mean time to desaturation and adjusted post-operation opiates utilization.
Fourteen patients who passed PACU monitoring and 2 patients who failed PACU monitoring presented to the emergency room greater than 24 hours after discharge. The ER complaints found were pain (N=3), bleeding (N=4), urinary retention (N=2), adverse reaction to medication (N=1) and prescription refill/musculoskeletal injury/dressing change (N=6). There were no complications related to OSA, and no respiratory or cardiac events found in patients discharged home.
In Table 2, we performed multiple logistic regression analysis adjusting for age, sex, BMI, AHI, anesthesia duration and adjusted post-op narcotic dose. The model containing the explanatory variables compared to that with the intercept only, significantly impacted the predictive ability of the model with the likelihood ratio, χ2 =35.86, p<0.0001. The overall effect of each of the covariates on desaturation at PACU revealed only BMI (χ2 =35.86, p<0.0001) and anesthesia duration (χ2 =11.10, p=0.0009) had a significant independent effect on desaturation at PACU. Categorically, the odds of desaturation during PACU monitoring are higher for patient age ≥ 50 is 1.52 (95% CI 0.74, 3.13) and female is 1.91 (95% CI 0.91, 4.02); patients with BMI≥35 is 3.95 (95% CI 2.0, 7.80); patients with AHI 21-40 (moderate) is 1.07 (95% CI 0.12, 9.23) and AHI >40 (severe)is 3.10 (95% CI 0.38, 25.67) as compared to AHI 6-20 (mild); anesthesia duration between 2-4 hours 3.55 (95% CI 1.68, 7.47) compared to anesthesia duration 1-2 hours and adjusted post-op narcotic dose 5.1-25 mg 1.48 (95% CI 0.65, 3.36) compared to dose 0-5 mg. The discriminatory performance of the training model revealed the AUC was 0.776, Figure 2. This indicates that the model has good ability to distinguish between patients with diagnosis or at risk for OSA and those without OSA for post-operative desaturation. As a measure of calibration from the model in Table 3, the Hosmer and Lemeshow goodness-of-fit statistics revealed χ2=1.62 (p>0.99) indicating strong agreement between observed and expected post-operative desaturation in patient diagnosis or at risk for OSA.
A large number of patients have ambulatory surgery . The perioperative period is particularly high risk for patients with or at high risk for OSA due to the effect of anesthesia, narcotics and sedatives. Several studies found patients with OSA undergoing non-cardiac surgery have higher incidence of post-operative hypoxia, respiratory failure, cardiac events and ICU transfers compared to those without OSA [4,9]. An ability to predict patients at greater risk for postoperative complications would help hospitals target effective resource utilization while complying with various guidelines for managing OSA patients postoperatively. The 2006 American Society of Anesthesiologists (ASA) guideline for perioperative management of patients with OSA, which the SJHC protocol was based upon, recommends postoperative monitoring in the PACU for at least 4 hours, and for 7 hours after an episode of airway obstruction or hypoxemia.1 Subsequent publications have recommended different approaches to postoperative monitoring. Despite these recommendations, many institutions do not have a policy for perioperative management of patients with OSA , due to a lack of evidence clarifying duration of monitoring and or prevention of clinically significant adverse events. The implementation of the postoperative monitoring of OSA patients can have significant resource implications for hospitals. Our institution adopted a 4 hours PACU monitoring policy on the basis of the 2006 ASA guidelines. The protocol was chosen as it identifies only higher risk patients, which allowed us to restrict the number of patients monitored. This retrospective chart review found the mean time for patients who had desaturations in the PACU occurred at 1.46 hours of admission; 69/75 (92.00%) with desaturation between 0-2 hours and 6/75 (8.00%) with desaturation between 2-3 hours. Females with BMI≥35 with or at risk for OSA were more likely to have desaturation. These findings are similar to that of a study of perioperative risk assessment in patients with a propensity for OSA .
In general sleep physiology changes with age and gender [12-14]. Female have better objective sleep quality with shorter sleep onset latency and better sleep efficiency than male [12,15]. Factors that can impact female more than male sleep efficiency including anxiety and depression and hormonal changes specifically luteal phase of the menstrual cycle [16-18]. However, OSA is more common in men than women in the general population with a male to female ratio ranges 3:1 to 5:1 [19-21]. Obesity a well-known risk for OSA, and higher body mass index (BMI) is associated with greater severity of OSA for both sexes.22 But for the same AHI women tends to be more obese than men [23,24], this is likely due to differences in fat distribution between the sexes . In our study, women with moderate AHI had similar proportion of desaturation as compared to male with sever AHI which can be explained by the fat distribution differences between genders which has physiological and mechanical effect in patients with OSA . Female who are obese and with OSA have shown to have significantly increased hypercapnic and hypoxic response, which is not the case in male . Furthermore, women with OSA are less likely to be evaluated and diagnosed for OSA . In our study, the limitation for gender difference specifically female is the small numbers (N=62) with BMI >35 (N=24) and AHI moderate-severe (N=12/16)) are more likely to have post-operative desaturation.
Procedural related risk factors such as surgical site, surgical (anesthesia) duration, anesthetic techniques and emergency surgery have been shown to predict peri-operative complications . In particular pulmonary complications for non-cardiac-thoracic surgeries requiring greater than 2-hours anesthesia and mechanical ventilation would need for prolonged oxygen therapy and atelectasis which required ICU admission may complicated by increasing post-operative-mortality and increasing ICU/hospital length of stay . Russell KM et al.  reported patients after ophthalmologic surgery found that the expected duration of PACU recovery depends on the operations that requires general anesthesia. Patients having orbitotomy and strabismus procedures had a median anesthesia recovery time 2 to 3 times longer than for other procedures and more likely to have prolonged recovery with other procedures. In addition, their post-hoc analysis found a higher proportion of patients with respiratory depression during anesthesia recovery also had OSA [31. Our study indicated that in our ambulatory surgery cohort 56/75 (78.67%) of the patients with desaturation had longer anesthesia duration between 181-200 minutes than those with shorter anesthesia duration.
A major determinant for discharge after ambulatory surgery is the quality of post-operative pain control. Opioid consumption in the PACU can be used as an earlier surrogate for poor global quality of recovery after surgery. An inverse relationship noted between opioid consumption in the PACU and global quality of recovery at 24-hours after the surgical procedure . It is also known obese patients with or without OSA experiences frequent oxygen desaturation episodes post-operatively after total anesthesia followed by patient-controlled intravenous analgesia with morphine . However, in our study the adjusted opioid dose is much less as compared to other published studies [32,34], which had a much higher opioid dose for the higher pain patients than the lower pain patients. This may well be because our anesthesiologists focus on opiate sparing techniques for OSA patients. Multimodal analgesia and regional analgesia techniques are used routinely.
Certain co-morbidities predispose adult patients for high risk of post-operative OSA includes obesity, hypertension, diabetes, male sex, alcohol use and large neck size , coupled with chronic opioid use . In our study there is a high proportion of patients with desaturation had hypertension and diabetes. Therefore a robust screening questionnaire would be very useful to select that patient at risk such as the STOP-BANG questionnaire . Studies have shown that the STOP-BANG questionnaire can identify 93% of those patients as being at risk for OSA [35,38]. However, that sensitivity would lead to the inclusion of substantially larger numbers of patients, which would have taxed hospital resources significantly.
In an aging population the frequency of OSA increases and plateau after 65 years . In our study group, patients > 50 years old had a higher number of desaturations as compared those patient ≤ 50 years old. There are multiple risk factors associated with the older population desaturation post ambulatory surgery such as preferential deposition of fat around the pharynx which reduces airway patency ; overnight fluid shift to the neck affect the surface tension of the upper airway [41,42] and restriction of chest wall movement due to decrease lung volume . The greatest exacerbation of OSA and sleep-disordered breathing has been shown to occur on postoperative day three , at which point all of our patients had been discharged home. We did not identify evidence of OSA related complications following discharge from hospital. It is not clear whether increased AHI and increase in desaturations after general anesthesia are associated with clinically significant events.
There are several limitations with our study. Due to a small study population and exclusion of higher risk patients from this ambulatory surgical hospital we did not find any clinically significant adverse postoperative events in our study patients. Patients admitted to the inpatient ward for postoperative oximetry can be awakened by the oximeter alarm, and commence breathing prior to developing any further complication. Another limitation of this study was the lack of dedicated apnea monitoring, both in PACU and on the ward. Apneas presumable occurred but the events were only recognized when there was a resulting desaturation and oximeter alarm. Our protocol did not include assessment for sedation analgesia mismatch, but patients in the cohort did not exhibit this finding: there was a low rate of opiate requirement, and minimal sedation.
The absence of clinically significant adverse effects is notable. We cannot know whether the overnight oximetry monitoring, and application of oxygen to those with desaturations, prevented any complications. However, the data show that the majority of patients in our cohort of select lower acuity patients could safely be discharged home on the day of surgery. These findings echo that of a recent study of patients undergoing surgery to relieve obstructive sleep apnea symptoms .
In conclusion, this retrospective chart review found that all desaturation events in PACU occurred within the first 3 hours of arrival to PACU. Patients with a PSG diagnosis of OSA appear to be at higher risk than those identified by preoperative screening. Females and those with a BMI >35 were more likely to have desaturations. We did not capture any clinically significant adverse events in our study patients. A larger study may be required to determine the utility and optimal duration of postoperative monitoring.
John Fuller contributed substantially to all aspects of the manuscript, including conception and design; acquisition, analysis and interpretation of data, and drafting the article. Raymond Kao contributed substantially to analysis and interpretation of data, and drafting the article. Caitlin Gallagher contributed substantially to data acquisition, interpretation of data, and drafting the article. Brian Rotenberg contributed substantially to interpretation of data, conception and design, and drafting the article.