Secondary Logo

Journal Logo

Preoperatively Screened Obstructive Sleep Apnea Is Associated With Worse Postoperative Outcomes Than Previously Diagnosed Obstructive Sleep Apnea

Fernandez-Bustamante, Ana MD, PhD*; Bartels, Karsten MD*; Clavijo, Claudia MD*; Scott, Benjamin K. MD*; Kacmar, Rachel MD*; Bullard, Kenneth BS*; Moss, Angela F. D. MS; Henderson, William PhD; Juarez-Colunga, Elizabeth PhD; Jameson, Leslie MD*

doi: 10.1213/ANE.0000000000002241
Respiration and Sleep Medicine: Original Clinical Research Report

BACKGROUND: Obstructive sleep apnea (OSA) affects up to 26% of US adults, is often undiagnosed, and increases perioperative morbidity. We hypothesized that patients screened on the day of surgery as moderate/high risk for OSA (S-OSA) present similar perioperative respiratory complications, hospital use, and mortality than patients with previously diagnosed OSA (D-OSA). Second, we hypothesized that both OSA groups have more respiratory complications than No-OSA patients.

METHODS: The electronic medical database from 1 academic and 2 community hospitals was retrospectively queried to identify adults undergoing nonemergent inpatient surgery (January 1, 2012, to December 31, 2014). Based on the day-of-surgery preoperative assessment and STOP-BANG (Snoring, Tiredness, Observed apnea during sleep, high blood Pressure, Body mass index >35, Age >50 years, thick Neck, Gender male) score, they were classified as D-OSA, S-OSA, or No-OSA. Perioperative respiratory events and interventions, hospital use, and mortality were measured. The primary outcome composite (adverse respiratory events [AREs]) included perioperative hypoxemic events and difficult airway management. Hypoxemic event was defined as peripheral saturation of oxygen (Spo2) <90% by continuous pulse oximetry for ≥3 minutes, or if validated and/or manually entered into the medical chart. Hypoxemia was classified as mild (lowest Spo2 86%–89%) or moderate/severe (lowest Spo2 ≤85%). Secondary outcomes included postoperative respiratory interventions, intensive care unit admission, hospital length of stay, and 30-day and 1-year all-cause mortality. Outcomes were compared using linear and logistic regression analyses.

RESULTS: A total of 28,912 patients were assessed: 3432 (11.9%) D-OSA; 1546 (5.3%) S-OSA; and 23,934 (82.8%) No-OSA patients. At least 1 ARE was present in 68.0% of S-OSA; 71.0% of D-OSA; and 52.1% of No-OSA patients (unadjusted P < .001), primarily ≥1 moderate/severe hypoxemic event after discharge from the postanesthesia care unit (PACU; 39.9% in S-OSA; 39.5% in D-OSA; and 27.1% in No-OSA patients). S-OSA patients compared to D-OSA patients presented lower rates of moderate/severe hypoxemia in the PACU but similar intraoperatively and postoperatively, higher difficult mask ventilation rates, and similar difficult intubation reports. After adjusting for demographic, health, and surgical differences and hospital type, the likelihood of ≥1 ARE was not different in S-OSA and D-OSA patients (adjusted odds ratio 0.90 [99% confidence interval, 0.75–1.09]; P = .15). S-OSA patients compared to D-OSA patients had significantly increased postoperative reintubation, mechanical ventilation, direct intensive care unit admission after surgery, hospital length of stay, and 30-day all-cause mortality.

CONCLUSIONS: Patients classified as S-OSA have similar rates of AREs to D-OSA patients, but increased postoperative respiratory interventions, hospital use, and 30-day all-cause mortality. These worse postoperative outcomes in S-OSA patients than D-OSA patients could reflect the lack of awareness and appropriate management of this bedside S-OSA diagnosis after PACU discharge. Multidisciplinary interventions are needed for these high-risk patients.

From the *Department of Anesthesiology and Adult and Child Center for Health Outcomes and Delivery Science, University of Colorado Denver, Denver, Colorado.

Accepted for publication April 24, 2017.

Funding: This study was supported by departmental sources. K.B. was supported by the National Institutes of Health/National Institute on Drug Abuse, Award Number K23DA040923.

The authors declare no conflicts of interest.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website.

Reprints will not be available from the authors.

Address correspondence to Ana Fernandez-Bustamante, MD, PhD, Department of Anesthesiology, University of Colorado School of Medicine, 12631 E 17th Ave, AO-1, R2012, MS 8202, Aurora, CO 80045. Address e-mail to

Obstructive sleep apnea (OSA) affects up to 1 in every 4, or 25 million, adults in the United States, and is increasing in parallel with obesity.1,2 About 80% to 90% of patients with OSA present for surgery undiagnosed.3,4 Characteristic OSA features include brief episodes of upper airway obstruction during sleep, leading to periods of apnea (no airflow) or hypopnea (decreased airflow), and repeated hypoxemic events.5 Intermittent hypoxemia triggers sympathetic activation and systemic inflammation, and contributes to respiratory and cardiovascular conditions, hyperglycemia, and overall increased mortality in OSA patients.4,6–13 Anesthesiologists routinely screen surgical patients for OSA preoperatively14–16 because it predisposes patients to cardiopulmonary complications and intensive care unit (ICU) admissions.6,8,17,18

Perioperative hypoxemia is multifactorial (eg, sedatives and opioids, residual muscle weakness, atelectasis, pain, and surgical trauma) and underrecognized.19 Most perioperative hypoxemic episodes are brief, easily reversed with oxygen supplementation, and considered irrelevant. However, the sympathetic consequences of brief hypoxemia, increased risk for perioperative respiratory depression, and greater incidence of difficult airway management in OSA patients20 support the importance of hypoxemic events in surgical patients at high risk for OSA. A recent multicenter prospective study analyzed postoperative pulmonary complications in 1202 American Society of Anesthesiologists physical status III patients after general anesthesia for noncardiothoracic surgery lasting 2 hours or more. Although this study did not specifically target OSA patients, the presence of inadequate oxygenation requiring prolonged (>1 day) postoperative oxygen therapy in this population was associated with longer hospital length of stay (LOS) and ICU admission.21 These data strengthen the need to understand the incidence, severity, duration, contributing factors, and impact of transient hypoxemia after surgery.

It is unknown whether patients with a day-of-surgery screened OSA diagnosis (S-OSA) have the same frequency and severity of hypoxemia and other postoperative adverse outcomes as patients with a previously established OSA diagnosis (D-OSA). The S-OSA classification is often based on the STOP-BANG (Snoring, Tiredness, Observed apnea during sleep, high blood Pressure, Body mass index >35, Age >50 years, thick Neck, Gender male) score14–16 or other OSA screening tools, but their reliability to predict outcomes compared to polysomnography has not been well established.18 Further, several STOP-BANG features are not routinely documented in the general medical record, making the S-OSA diagnosis often not available by chart review. The importance of providing the bedside S-OSA classification would be validated if S-OSA and D-OSA patients had similar adverse outcomes.

In a health system database study including an academic and 2 community practices, we tested the primary hypothesis that patients classified as S-OSA by anesthesiology personnel on the day-of-surgery preoperative evaluation would experience the same risk of perioperative adverse respiratory events (AREs), including hypoxemic events and/or difficult airway management, and other secondary postoperative outcomes related to respiratory interventions, hospital resource utilization and all-cause mortality after surgery, compared to patients arriving for surgery with a D-OSA diagnosis. Secondarily, we hypothesized that both OSA groups would have more postoperative adverse outcomes compared to a group without S-OSA or D-OSA.

Back to Top | Article Outline


We performed a retrospective database study after approval by the Colorado Multiple Institutional Review Board (COMIRB #09-0674).

Back to Top | Article Outline

Study Design



This retrospective study was designed using the “Strengthening the Reporting of Observational Studies in Epidemiology” guideline. Clinical anesthesia and surgical care were not investigated and it was decided by individual patient’s care teams not following a specific care protocol. The primary and secondary study outcomes were defined and established a priori at initiation of the study design. We queried the Epic Clarity Electronic Medical Record (EMR) database at the University of Colorado Health (UCHealth) system for all adults undergoing inpatient anesthesia care in one of its hospitals between January 1, 2012 and December 31, 2014. During this period, UCHealth was composed of a large academic institution and 2 community hospitals. Patients were included if they fulfilled the following criteria: ≥18 years and receiving anesthesia care for any surgical procedure. Only the first eligible procedure per patient was included. Subjects with missing demographic data (weight, surgical details) were also excluded. Figure 1 shows the study schematic flowchart.

Back to Top | Article Outline

Data Collected

We extracted demographic information, surgical service, OSA classification and STOP-BANG screening criteria,15,16 comorbid conditions, hypoxemic events in the operating room, postanesthesia care unit (PACU) and postoperative care units, airway management, and presence and duration of respiratory interventions (supplemental oxygen therapy, positive airway pressure [PAP, including continuous, automatic, or bilevel PAP] and reintubation with mechanical ventilation). Hypoxemia was defined intraoperatively as a value of peripheral saturation of oxygen (Spo2) <90% by continuous pulse oximetry for a minimum of 3 minutes. For episodes occurring in the PACU or hospital floor, hypoxemia was defined as a Spo2< 90% value if validated and/or manually entered into the medical chart by standard nursing protocol. Mild hypoxemia was defined as Spo2 86% to 89% while Spo2 ≤85% was considered moderate/severe hypoxemia.

Back to Top | Article Outline

Patient Classification

Patients were classified as “previously diagnosed OSA” (D-OSA) if an existing diagnosis of OSA or OSA therapy was pre-existing in the EMR or if self-reported by the patient on the date of surgery, “preoperative suspected OSA” (S-OSA) if classified as OSA by anesthesia providers in their day-of-surgery preoperative note in absence of an D-OSA diagnosis and had 3 or more STOP-BANG criteria15,16 reported, and “No OSA” (No-OSA) if none of the previous criteria was present and had less than 3 STOP-BANG criteria reported. The lack of a positively reported STOP-BANG criterion was presumed as negative. The requirement of 3 or more STOP-BANG criteria for S-OSA classification could include any of the 8 available criteria that were positively reported in the EMR. The STOP-BANG scoring criteria are available in the day-of-surgery preoperative note used by anesthesia providers. Methods utilized for the determination of the D-OSA diagnosis (polysomnography or other) were not investigated.

Back to Top | Article Outline

End Points

The primary outcome was a composite requiring at least 1 perioperative ARE: 1 or more perioperative hypoxemic events (intraoperative and/or postoperative) and/or reported difficult airway management incident (mask ventilation or difficult intubation). Secondary outcomes analyzed were as follows: postoperative (from PACU discharge to hospital discharge) respiratory interventions (oxygen therapy, PAP and/or reintubation, and invasive mechanical ventilation), any ICU admission, hospital LOS, and all-cause mortality reported to UCHealth within 30 days and/or 1 year after the date of surgery. ICU admission was considered “direct” if patients were transferred directly from the operating room to the ICU and “unplanned” if there was a postoperative floor stay period between the operating room and the ICU admission.

Back to Top | Article Outline

Statistical Analysis

Data were inspected for outliers, missing data, and data entry errors using univariable analyses. Missing data and outlier data values were individually revised for completion, corrected or left as missing data points. Cases with insufficient data for body mass index (BMI) calculation, BMI values <12 or >80, and surgery durations >24 hours or <15 minutes were excluded. Unrecorded subjective STOP-BANG criteria (eg, snoring and tiredness) were presumed as negative findings. Values of Spo2 <50% were assumed inaccurate and considered missing and values >100% were coded as 100%.

A bivariable analysis was performed to compare patients’ characteristics, comorbidities, and outcomes in the 3 OSA groups using χ2 tests of association for categorical variables and analysis of variance or Kruskal–Wallis tests for continuous variables. Pairwise comparisons of the OSA groups were performed using χ2 test or the Fisher exact (categorical variables) and t tests or Wilcoxon rank sum (continuous variables) tests.

The primary outcome, ARE, and its components were first described by bivariable analysis of the OSA group with χ2 tests. Then multiple logistic regression was used to adjust the association between the OSA group and ARE for all baseline variables, where OSA group was the primary explanatory variable and ARE was the dependent variable. Due to the smaller frequencies, American Society of Anesthesiologists (ASA) physical status I and II patients were combined. Age and BMI were dichotomized using the STOP-BANG tool thresholds. Collinearity between variables was assessed using the Spearman correlation coefficient. The c statistic was used to assess the discriminatory ability of the model, and the Hosmer-Lemeshow test was used to assess model calibration. Because the Hosmer-Lemeshow test is very sensitive to large sample sizes and can be statistically significant even though the differences from the proposed model are small, expected and observed events were plotted by deciles of risk to further examine the magnitude of the differences.

To examine the association between the OSA groups and the secondary outcomes, first Kruskal–Wallis, Wilcoxon, and χ2 tests were used to describe the unadjusted association and then multiple logistic or linear regression were used to describe the adjusted association. Nonparsimonious models were fit for each secondary outcome adjusting for all baseline variables. Non-normally distributed continuous outcome variables were log-transformed before linear regression was performed and diagnostic plots were inspected. The discriminatory ability of the models was assessed using the c statistic (logistic models), and the model fit was assessed with adjusted r2 values (linear regression).

Assuming an overall event rate of 55% for ARE in the sample, with a sample size of 28,912, this study had >95% power to detect an odds ratio of <0.9 or >1.1 for the OSA variable in the ARE model in Table 3, assuming a multiple correlation of ≤0.90 between OSA and the other variables in the model.22

Table 1.

Table 1.

Table 2.

Table 2.

Table 3.

Table 3.

All analyses were performed using SAS version 9.4. A Bonferroni adjustment for the multiple comparisons (between the 3 OSA groups) of our primary analysis would result in an adjusted significance level of 0.017 (0.05/3). Due to the large sample size, we selected a level of significance of 0.01 for all statistical tests, which is even more conservative than the adjusted α of .017.

Back to Top | Article Outline



A total of 28,912 patients were included in the study (Figure) and distributed as D-OSA (11.9%), S-OSA (5.3%), or No-OSA (82.8%). For the S-OSA classification, the objective STOP-BANG criteria (“P-BANG,” including high blood Pressure, Body mass index >35, Age >50 years, thick Neck, and Gender male) were present in all patients, but subjective criteria (“STO,” including Snoring, Tiredness, and Observed apnea during sleep) were missing in 674 (43.6%; snoring), 1280 (82.8%; tiredness), and 1272 (82.3%; observed apnea) of the patients, respectively. The lack of a positive reported criterion was presumed as negative. Table 1 details the comparison between D-OSA, S-OSA, and No-OSA patients. Patient characteristics per medical center can be found in Supplemental Digital Content 1, Table 1 ( The number of patients, especially the S-OSA patients, was significantly greater from the academic center than from the 2 community centers. A total of 1947 (56.7%) patients with a D-OSA diagnosis had self-reported compliance with noninvasive ventilation therapy. Patients with S-OSA were primarily males (68.2%) with an average age and BMI similar to those of the D-OSA group. S-OSA patients showed lower incidence of respiratory disease, diabetes, and neurological disease than D-OSA patients (Table 1). Compared to No-OSA patients, S-OSA had a greater number of comorbidities with the exceptions of similar smoking, asthma, neurological, and liver disease. Surgical duration was significantly longer in S-OSA than D-OSA or No-OSA patients.

Back to Top | Article Outline

Perioperative AREs, Unadjusted

The frequency of any perioperative AREs (ARE composite) in the S-OSA patients (68.0%) was similar to the D-OSA patients (71.0%) but greater than in the No-OSA group (52.1%; Table 2). The frequency of individual perioperative AREs in the D-OSA, S-OSA, and No-OSA patients is shown in Table 2. In all groups, the lowest frequency of moderate/severe hypoxemic events (Spo2 ≤85%) was observed in the PACU and the highest frequency occurred in a postoperative unit. The highest frequency of difficult mask ventilation was observed in S-OSA patients.

Back to Top | Article Outline

OSA Diagnosis Type and Perioperative AREs

The association between presenting at least 1 perioperative ARE and the OSA diagnosis type (S-OSA versus D-OSA) was not significantly different (P = .15) after adjusting for all baseline covariates: age >50 years, obesity (BMI > 35 kg/m2), comorbidities, ASA physical status, surgery duration (hours), surgical service, and hospital type (community, academic; Table 3). This model showed fair discriminatory ability (c index = 0.71). The Hosmer-Lemeshow test result indicated poor goodness of fit (P = .0006). This may be an artifact of the large sample size. Further inspection of the plot of the expected and observed event rates by deciles of risk showed very small differences indicating reasonable fit of the model to the data (see Supplemental Digital Content 2, Figure 1,

Back to Top | Article Outline

Secondary Outcomes

Table 4.

Table 4.

Table 4 shows the logistic or linear regression analyses results for each secondary outcome in D-OSA, S-OSA, and No-OSA patients. The S-OSA patients, compared to those with a D-OSA diagnosis, had similar frequency and duration of oxygen therapy, a significantly lower frequency of postoperative PAP therapy but higher rates of mechanical ventilation and reintubation. S-OSA classified patients were also significantly more often directly transferred to the ICU than D-OSA patients. The hospital LOS and postoperative mortality within 30 days were significantly greater in S-OSA than in D-OSA patients.

Back to Top | Article Outline

Subanalysis Comparison of Characteristics and Outcomes of D-OSA Patients With, Versus Without, Self-Reported Compliance to Prescribed OSA Therapy

Patients reportedly compliant to OSA therapy presented a greater frequency of obesity but reduced rates of smoking, COPD, and neurological disease (see full details in Supplemental Digital Content 3, Table 2, The frequency of AREs was not significantly different in D-OSA patients compliant and noncompliant with any PAP therapy. The postoperative PAP use was higher in PAP-compliant D-OSA patients (69.4%) than in PAP noncompliant D-OSA patients (26.1%) (unadjusted P < .001). No other differences were observed in secondary outcomes between PAP-compliant and noncompliant D-OSA patients (see Supplemental Digital Content 3, Table 2,

Back to Top | Article Outline


This study examines the adverse perioperative outcomes in surgical patients with a preoperative suspected OSA diagnosis (S-OSA) compared to patients with previously diagnosed OSA (D-OSA) in 1 academic and 2 community practices. S-OSA and D-OSA groups had distinct demographics, comorbidities, and surgical categories. S-OSA patients presented similar rates of postoperative moderate/severe hypoxemia and difficult intubation, but increased difficult mask ventilation, compared to D-OSA patients. After adjusting for all demographic, health, and surgical differences, S-OSA patients showed increased postoperative reintubation, ventilation, direct ICU admission after surgery, prolonged hospital LOS, and all-cause 30-day mortality after surgery, compared to D-OSA patients. No-OSA patients had fewer comorbidities and a decreased overall frequency of perioperative adverse respiratory outcome compared to the S-OSA group.

OSA is a significant public health concern because of its frequency, health risks, and associated costs.1,2,13,23–26 OSA is common, affecting up to 26% of all adults in the general population (43% men and 27% women 50 years of age and older)2,27 and greater than 70% of certain subpopulations (ie, bariatric surgery).28,29 In our study, approximately 17% of patients were classified as OSA, with about one-third identified on the day of surgery (S-OSA, 5.3%). The lower-than-average rates of detected OSA in our cohort may have been influenced by the incomplete STOP-BANG screening criteria (eg, tiredness, observed apnea) and the lower BMI in the Colorado population compared to other US populations.30 Our S-OSA classification was based on a risk score tool performed during the physician bedside evaluation, which may have affected the sensitivity of our OSA diagnosis (Figure). Nonetheless, the S-OSA diagnosis reflects the reality of perioperative medicine, where diagnosis and treatment must be performed within a very limited time period.

The STOP-BANG tool had the best predictive value for OSA screening in recent comparisons of 4 screening tools: STOP-BANG, Berlin, ASA Check List, and SA-SDQ (Sleep Apnea scale of the Sleep Disorders Questionnaire) score.16,31 Recent meta-analyses18,32 have suggested that the STOP-BANG-based OSA diagnosis is an acceptable, although imperfect, alternative to oximetry or polysomnography-based diagnoses. We selected a STOP-BANG criteria ≥3 because a previous study showed that this threshold presents an odds ratio for any degree of OSA (apnea-hypopnea index >5) of 3.01 (1.92–4.70) compared with patients with STOP-BANG scores ≤2.33 In our system, the 8 STOP-BANG criteria are available on the perioperative evaluation tool but criteria selection is not required for anesthesia providers to select the S-OSA classification. Reporting of individual subjective criteria (eg, snoring, daytime sleepiness) are often omitted, which may have contributed to underestimate the number of S-OSA patients.

The academic center, compared to the community centers, contributed about two-thirds of all patients classified as S-OSA, and slightly more than half of patients classified as D-OSA and No-OSA (Table 1). It is impossible to know the reasons from our study, but greater patient volume with diverse health and primary care backgrounds, higher surgical complexity and increased awareness by anesthesia care teams could be among contributing factors.

Patients classified as S-OSA had an overall lower comorbidity burden than D-OSA patients. Despite this, S-OSA patients had similar (postoperative hypoxemia, duration of oxygen therapy) or worse secondary outcomes (postoperative reintubation, ventilation, ICU admission, and 30-day all-cause mortality) than D-OSA patients after adjusting for potential confounders (Tables 3 and 4). The impact of preoperative treatment of OSA on surgical outcomes is unclear.34 However, our results show that this S-OSA population is a high-risk population that may benefit from increased medical attention and focused care. Compared to the No-OSA group, D-OSA patients had significantly increased comorbidities, perioperative hypoxemia and difficult airway management, postoperative oxygen therapy and PAP use, but similar reintubation and ventilation rates, unplanned ICU admission, hospital LOS, and all-cause mortality (Table 4). Interestingly, STOP-BANG criteria scores were not associated with postoperative hypoxemia in the subanalysis of the vascular events in noncardiac surgery patIents cohort evaluation (VISION) study, a prospective international cohort analysis of plasma troponin levels from 15,133 patients after noncardiac surgery and their impact on 30-day mortality.35 The different patient populations and design make comparisons difficult.

No significant differences in outcomes were observed between D-OSA patients with, versus without, self-reported compliance to OSA therapy, except the expected higher postoperative PAP use in PAP compliant D-OSA patients (see Supplemental Digital Content 3, Table 2, Of note, approximately 3 every 4 D-OSA PAP-compliant patients received PAP after surgery, but our study cannot determine if postoperative PAP was established as a routine nocturnal therapy or rescue for respiratory failure. Further investigation is needed to analyze postoperative compliance with OSA therapy and its impact on surgical outcomes.

Due to multifactorial barriers, the inability to obtain diagnostic testing for OSA preoperatively contributes to the high proportion of patients at moderate/high risk for OSA presenting for surgery without a formal diagnosis.3,4 Our results suggest that anesthesiologists can reliably detect patients with S-OSA who are at high risk for perioperative AREs. This risk is greatest beyond the immediate postoperative period (after PACU discharge). The value of a S-OSA diagnosis and subsequent interventions to mitigate perioperative events has not been specifically evaluated. Several authors have proposed initiating OSA therapy based on clinical diagnosis only, without polysomnography confirmation15,36–38 both in the hospital and community settings. Our results support testing if the postoperative standard OSA precautions (ie, increased respiratory monitoring) and therapy (ie, nocturnal PAP) in the S-OSA group could improve patient health and reduce health care costs.

The accurate frequency, duration, or impact of hypoxemia in the perioperative period are challenging to achieve and often reasonably questioned. Most perioperative hypoxemic episodes are brief, easily reversed with oxygen supplementation, acutely influenced by medications (eg, opioids) or conditions (eg, incisional pain), and not associated with immediate harm to patients. However, intermittent hypoxemia is intrinsically linked to the OSA physiology and its cardiovascular and metabolic negative consequences12,13 through activation of the sympathetic system and stress responses. Even if inaccurate, it is likely underdetected and under-reported, based on a prospective blinded assessment of the incidence and duration of hypoxemia after noncardiothoracic surgery by Sun et al.19 Further, hypoxemia, prolonged oxygen therapy, atelectasis, and other mild postoperative pulmonary outcomes may have a significant impact on the postoperative course (including mortality) and hospital resource use (ICU admission, hospital/ICU LOS). Specifically, in a recent multicenter prospective analysis of postoperative pulmonary complications in 1202 ASA III patients with ≥2 hours general anesthesia for noncardiothoracic surgery, the need for prolonged (>1 day) oxygen therapy after surgery was significantly associated with ICU admission as well as hospital and ICU LOS21 but not with 7-day mortality. This study did not specifically focus on OSA, but its overall results suggest an opportunity to improve perioperative outcomes by investigating postoperative hypoxemia as early subclinical signs of suboptimum respiratory function.

There are several limitations to our study, primarily related to the use of retrospective data extracted from an electronic database. Electronic database extracts depend on charted data, and its reliability is best for selected predetermined entries but ignores unformatted text information. Obstacles for a clinical diagnosis of OSA using a screening tool are present in the primary care office and the surgical preoperative area. The high frequency of missing subjective STOP-BANG criteria in our database, particularly of “tiredness” and “observed apnea,” may have underestimated the number of patients classified as S-OSA. Anesthesiologists have the advantage of direct observations of medical events (ie, respiratory depression, airway obstruction, difficulty of airway management) that reinforce the suspected OSA diagnosis. While OSA screening in asymptomatic patients in primary care is still controversial,39,40 recent guidelines from the Society of Anesthesia and Sleep Medicine support this screening preoperatively as an effective method to improve care.34 It is also possible that some D-OSA patients received the OSA diagnosis via screening without the current gold standard polysomnography confirmation. Thus, the specificity of D-OSA and S-OSA diagnosis requires further confirmation. The incidence of hypoxemic events, the occurrence of respiratory complications, and not protocolized oxygen therapy, particularly in less intense monitoring environments like surgical wards, may be underestimated. This underestimation has been previously reported.19 The clinical impact of perioperative hypoxemic events present in the EMR, particularly if isolated brief ones, is still unclear. Characterizing this relationship would require continuous pulse oximetry, respiratory rate monitoring, and detailed cardiovascular and biochemical follow-up in prospective observational studies. Other limitations of our study include use of a composite primary outcome variable that can be driven by the component with the largest incidence, and use of mortality data based on deaths of patients as reported to our hospitals which might be under-reported. Further prospective studies will be needed to confirm the relationship between the type of OSA diagnosis and the observed outcomes. The effect of long-term treatment of S-OSA patients needs evaluation for improved health, outcomes, and reduction of future hospital costs.

In conclusion, a preoperative suspected OSA diagnosis is associated with worse postoperative respiratory outcomes, increased ICU admission, prolonged hospital LOS, and greater all-cause 30-day mortality after surgery, compared to an established OSA diagnosis. Our findings expose a large high-risk surgical population that could benefit from increased awareness and focused interventions. Future research should aim at developing multidisciplinary perioperative care interventions to improve clinical outcomes of all preoperatively identified OSA patients.

Back to Top | Article Outline


Name: Ana Fernandez-Bustamante, MD, PhD.

Contribution: This author helped with study design, conducting of the study, data collection, data analysis, interpretation of the data, and writing the manuscript.

Name: Karsten Bartels, MD.

Contribution: This author helped with interpretation of the data and critically revising the manuscript.

Name: Claudia Clavijo, MD.

Contribution: This author helped with interpretation of the data and critically revising the manuscript.

Name: Benjamin K. Scott, MD.

Contribution: This author helped with interpretation of the data and critically revising the manuscript.

Name: Rachel Kacmar, MD.

Contribution: This author helped with interpretation of the data and critically revising the manuscript.

Name: Kenneth Bullard, BS.

Contribution: This author helped with acquisition of the data and critically revising the manuscript.

Name: Angela F.D. Moss, MS.

Contribution: This author helped with study design, data analysis, and critically revising the manuscript.

Name: William Henderson, PhD.

Contribution: This author helped with study design, data analysis, and critically revising the manuscript.

Name: Elizabeth Juarez-Colunga, PhD.

Contribution: This author helped with study design, data analysis, and critically revising the manuscript.

Name: Leslie Jameson, MD.

Contribution: This author helped with study design, conducting the study, data collection, data analysis, interpretation of the data, and critically revising the manuscript.

This manuscript was handled by: David Hillman, MD.

Back to Top | Article Outline


1. American Academy of Sleep Medicine. Rising prevalence of sleep apnea in U.S. threatens public health. 2014. Available at: Accessed January 16, 2016.
2. Peppard PE, Young T, Barnet JH, Palta M, Hagen EW, Hla KM. Increased prevalence of sleep-disordered breathing in adults. Am J Epidemiol. 2013;177:1006–1014.
3. Singh M, Liao P, Kobah S, Wijeysundera DN, Shapiro C, Chung F. Proportion of surgical patients with undiagnosed obstructive sleep apnoea. Br J Anaesth. 2013;110:629–636.
4. Memtsoudis SG, Besculides MC, Mazumdar M. A rude awakening–the perioperative sleep apnea epidemic. N Engl J Med. 2013;368:2352–2353.
5. Berry RB, Budhiraja R, Gottlieb DJ, et alAmerican Academy of Sleep Medicine. Rules for scoring respiratory events in sleep: update of the 2007 AASM Manual for the Scoring of Sleep and Associated Events. Deliberations of the Sleep Apnea Definitions Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med. 2012;8:597–619.
6. Fouladpour N, Jesudoss R, Bolden N, Shaman Z, Auckley D. Perioperative complications in obstructive sleep apnea patients undergoing surgery: a review of the legal literature. Anesth Analg. 2016;122:145–151.
7. Memtsoudis SG, Stundner O, Rasul R, et al.The impact of sleep apnea on postoperative utilization of resources and adverse outcomes. Anesth Analg. 2014;118:407–418.
8. Kaw R, Pasupuleti V, Walker E, Ramaswamy A, Foldvary-Schafer N. Postoperative complications in patients with obstructive sleep apnea. Chest. 2012;141:436–441.
9. Marin JM, Agusti A, Villar I, et al.Association between treated and untreated obstructive sleep apnea and risk of hypertension. JAMA. 2012;307:2169–2176.
10. Martinez-Ceron E, Barquiel B, Bezos AM, et alEffect of CPAP on glycemic control in patients with obstructive sleep apnea and type 2 diabetes. A randomized clinical trial. 2016;194:476–485.
11. Kendzerska T, Gershon AS, Hawker G, Tomlinson G, Leung RS. Obstructive sleep apnea and incident diabetes. A historical cohort study. Am J Respir Crit Care Med. 2014;190:218–225.
12. Lurie A. Hemodynamic and autonomic changes in adults with obstructive sleep apnea. Adv Cardiol. 2011;46:171–195.
13. Drager LF, Jun JC, Polotsky VY. Metabolic consequences of intermittent hypoxia: relevance to obstructive sleep apnea. Best Pract Res Clin Endocrinol Metab. 2010;24:843–851.
14. Chung F, Abdullah HR, Liao P. STOP-Bang Questionnaire: a practical approach to screen for obstructive sleep apnea. Chest. 2016;149:631–638.
15. Chung F, Yang Y, Brown R, Liao P. Alternative scoring models of STOP-bang questionnaire improve specificity to detect undiagnosed obstructive sleep apnea. J Clin Sleep Med. 2014;10:951–958.
16. Abrishami A, Khajehdehi A, Chung F. A systematic review of screening questionnaires for obstructive sleep apnea. Can J Anaesth. 2010;57:423–438.
17. Mutter TC, Chateau D, Moffatt M, Ramsey C, Roos LL, Kryger M. A matched cohort study of postoperative outcomes in obstructive sleep apnea: could preoperative diagnosis and treatment prevent complications? Anesthesiology. 2014;121:707–718.
18. Kaw R, Chung F, Pasupuleti V, Mehta J, Gay PC, Hernandez AV. Meta-analysis of the association between obstructive sleep apnoea and postoperative outcome. Br J Anaesth. 2012;109:897–906.
19. Sun Z, Sessler DI, Dalton JE, et al.Postoperative hypoxemia is common and persistent: a prospective blinded observational study. Anesth Analg. 2015;121:709–715.
20. Kheterpal S, Healy D, Aziz MF, et alMulticenter Perioperative Outcomes Group (MPOG) Perioperative Clinical Research Committee. Incidence, predictors, and outcome of difficult mask ventilation combined with difficult laryngoscopy: a report from the multicenter perioperative outcomes group. Anesthesiology. 2013;119:1360–1369.
21. Fernandez-Bustamante A, Frendl G, Sprung J, et al.Postoperative pulmonary complications, early mortality, and hospital stay following noncardiothoracic surgery: a multicenter study by the Perioperative Research Network investigators. JAMA Surg. 2017;152:157–166.
22. Hsieh FY. Sample size tables for logistic regression. Stat Med. 1989;8:795–802.
23. Drager LF, Lopes HF, Maki-Nunes C, et al.The impact of obstructive sleep apnea on metabolic and inflammatory markers in consecutive patients with metabolic syndrome. PLoS One. 2010;5:e12065.
24. Strohl KP, Brown DB, Collop N, et alATS Ad Hoc Committee on Sleep Apnea, Sleepiness, and Driving Risk in Noncommercial Drivers. An official American Thoracic Society Clinical Practice Guideline: sleep apnea, sleepiness, and driving risk in noncommercial drivers. An update of a 1994 Statement. Am J Respir Crit Care Med. 2013;187:1259–1266.
25. Franco CM, Lima AM, Ataíde L Jr, et al.Obstructive sleep apnea severity correlates with cellular and plasma oxidative stress parameters and affective symptoms. J Mol Neurosci. 2012;47:300–310.
26. Lopez-Jimenez F, Sert Kuniyoshi FH, Gami A, Somers VK. Obstructive sleep apnea: implications for cardiac and vascular disease. Chest. 2008;133:793–804.
27. Heinzer R, Vat S, Marques-Vidal P, et al.Prevalence of sleep-disordered breathing in the general population: the HypnoLaus study. Lancet Respir Med. 2015;3:310–318.
28. Hallowell PT, Stellato TA, Schuster M, et al.Potentially life-threatening sleep apnea is unrecognized without aggressive evaluation. Am J Surg. 2007;193:364–367.
29. Frey WC, Pilcher J. Obstructive sleep-related breathing disorders in patients evaluated for bariatric surgery. Obes Surg. 2003;13:676–683.
30. CDC/NCHS. Prevalence of Self-Reported Obesity Among U.S. Adults by State and Territory, BRFSS, 2014. Available at: Accessed February 14, 2016.
31. Ramachandran SK, Josephs LA. A meta-analysis of clinical screening tests for obstructive sleep apnea. Anesthesiology. 2009;110:928–939.
32. Nagappa M, Liao P, Wong J, et al.Validation of the STOP-Bang Questionnaire as a screening tool for obstructive sleep apnea among different populations: a systematic review and meta-analysis. PLoS One. 2015;10:e0143697.
33. Chung F, Subramanyam R, Liao P, Sasaki E, Shapiro C, Sun Y. High STOP-Bang score indicates a high probability of obstructive sleep apnoea. Br J Anaesth. 2012;108:768–775.
34. Chung F, Memtsoudis SG, Ramachandran SK, et al.Society of anesthesia and sleep medicine guidelines on preoperative screening and assessment of adult patients with obstructive sleep apnea. Anesth Analg. 2016;123:452–473.
35. Khanna AK, Sessler DI, Sun Z, et al.Using the STOP-BANG questionnaire to predict hypoxaemia in patients recovering from noncardiac surgery: a prospective cohort analysis. Br J Anaesth. 2016;116:632–640.
36. Nigro CA, Dibur E, Aragone MR, Borsini E, Ernst G, Nogueria F. Can CPAP be indicated in adult patients with suspected obstructive sleep apnea only on the basis of clinical data? Sleep Breath. 2016;20:175–182.
37. Mulgrew AT, Fox N, Ayas NT, Ryan CF. Diagnosis and initial management of obstructive sleep apnea without polysomnography: a randomized validation study. Ann Intern Med. 2007;146:157–166.
38. Senn O, Brack T, Russi EW, Bloch KE. A continuous positive airway pressure trial as a novel approach to the diagnosis of the obstructive sleep apnea syndrome. Chest. 2006;129:67–75.
39. Redline S. Screening for obstructive sleep apnea: implications for the sleep health of the population. JAMA. 2017;317:368–370.
40. Bibbins-Domingo K, Grossman DC, Curry SJ, et alUS Preventive Services Task Force. Screening for obstructive sleep apnea in adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2017;317:407–414.

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2017 International Anesthesia Research Society