Introduction
Obstructive sleep apnea (OSA) is a condition caused by repetitive episodes of partial or complete airway collapse during sleep. The prevalence rates of OSA have increased substantially over the last decades, ranging from 14 to 55% depending on the age and gender of the patient population [
1]. The significant relationship with cardiovascular and metabolic diseases forms the major health burden associated with OSA which leads to substantial morbidity and mortality [
2‐
4].
Symptoms of OSA include increased daytime sleepiness, fatigue, irritability, inattention and a decrease in cognitive function resulting in a highly heterogeneous disease with multiple phenotypes [
5‐
7]. Although OSA is a common problem, the fact that the respiratory events, (apnea and hypopnea) occur during sleep, results in an unawareness of- and underdiagnosed disease [
8]. Various questionnaires including the Epworth Sleepiness scale (ESS), [
9]; Berlin questionnaire [
10]; the snoring, tiredness, observed apnea, high blood pressure, body mass index (BMI), age, neck circumference, and male gender (STOP-BANG) questionnaire [
11] and scores Lausanne NoSAS score [
12] and the Multivariable Apnea Prediction (MVAP) score [
13] exist to aid in identifying patients with OSA.
Flexible bronchoscopy (FB) is a generally safe minimally invasive procedure used to assess, diagnose and treat patients with respiratory disease [
14]. Transient hypoxemia due to upper airway obstruction is known to occur in patients undergoing flexible bronchoscopy [
15‐
18]. However, the association between transient hypoxemia during FB and the presence of sleep apnea remains unexplored.
We hypothesize that transient hypoxemia during flexible bronchoscopy under conscious sedation might be associated with the apnea–hypopnea index (AHI) assessed by polygraphy and therefore could inform about the presence of sleep apnea.
Discussion
Transient hypoxemia is known to occur in patients undergoing FB, however, the association between transient hypoxemia during FB and OSA is unknown. We found that patients who experienced hypoxemia during FB under conscious sedation had increased AHI in the polygraphy. The association between oxygen desaturation < 90% during bronchoscopy and AHI remained significant after adjustment for duration of the procedure and propofol dose. Accordingly, patients presenting oxygen desaturation < 90% during bronchoscopy had a 9.1/h higher AHI than those not presenting oxygen desaturation during the procedure.
Hypoxia is the most commonly cited adverse event in patients undergoing bronchoscopy [
16,
17,
28]. The occurrence of hypoxemia was highly prevalent during FB in our study, with an oxygen saturation of < 90% for ≥ 5 s measured in 132 patients (91%). This reflects the severity and multi-morbidity of the patient population included in the study with almost 50% of the participants suffering from COPD and 74% classified as ASA Class III or higher. In a recent study by Cho et al. [
18] they had an incidence of hypoxemia of 35% during moderate sedation bronchoscopy. Their population consisted of 35% ASA III patients and 65% ASA II patients. Although ASA classification was introduced as a subjectively determined marker of general health used to evaluate perioperative morbidity [
29], it was also reported as a predictor for occurrence of adverse events, including hypoxemia, during endoscopic procedures [
30‐
32]. Nonetheless, the occurrence of hypoxemia during flexible bronchoscopy with sedation is similar in studies with and without supplemental oxygen [
16,
33‐
35]. A recent study in Norway has shown that when patients chose to have no sedation, it significantly increased unplanned interventions during the bronchoscopy [
17].
Although the safety and efficacy of propofol used as a sedative during bronchoscopy has been explored [
21‐
23,
25], airway collapsibility is affected by propofol in a dose dependent manner making proper titration a decisive factor in avoiding overestimating the severity of airway obstruction and implicitly OSA [
36]. The mean dose of propofol required to achieve proper sedation in the present study was slightly higher than previously reported [
21,
22,
37]. However, the number of complex procedures performed during bronchoscopy has increased compared to earlier studies, thus underpinning a higher sedative requirement. In addition the intravenous continuous infusion of propofol, which is as safe as bolus administration, is known to be associated with an increased dosage during FB [
25]. Endoscopic procedures known as drug-induced sleep endoscopy (DISE) are widely recognized as diagnostic instruments for OSA [
38] due to induction of airway obstruction and collapse during sedation [
39]. Propofol is the recommended pharmacologic agent for DISE [
38]. The use of propofol to induce a sleep-like state, mimics the drop in oxygen saturation seen during polysomnography and the respiratory events are comparable to the respiratory events seen during polysomnography [
40,
41].
We found that the association between oxygen desaturation during bronchoscopy and higher AHI remained significant after adjusting for propofol dose even though patients do not enter REM sleep during the bronchoscopy. It has been shown that most of the respiratory events in sleep apnea occur during N2 sleep [
40,
42‐
44].
Sleep apnea assessed using polygraphy, was highly prevalent in our population with 90% of patients having an AHI above 5/h, even though BMI and neck circumference were normal. Similar results were found by Cho et al. [
18] who, using the STOP-BANG score ≥ 3 found that 67% of their patients were at high risk of sleep apnea even though the average neck circumference and BMI were in the normal range. The high prevalence of sleep apnea found in the present study is similar to values previously reported [
45]. Although the subjects enrolled in the other study had similar BMI (25.6 vs 25.6) and neck circumference (36.9 vs 39.7) the patients in our study were older (57 vs 66 years) [
45]. In addition, a moderate to severe sleep apnea was observed in 54% of our patients and a gender difference in prevalence was evident. These results are similar to previously published rates of sleep apnea in the general population [
45‐
47]. The difference in sleep apnea incidence between males and females may be due to differences in upper airway mechanical or neuromuscular properties, chemoreflex control of breathing or sex hormone levels [
46]. There was, however, no difference in the occurrence of oxygen desaturation during bronchoscopy when comparing males and females.
The Mallampati index is a quick instrument for assessing airway patency before intubation [
48]. There is growing data pointing out the association of a higher Mallampati index to severity of OSA [
49,
50]. Most of the patients enrolled in our study (72%) had a Mallampati index of three or more. In the study by Wang et al., they found that 85% of patients with sleep apnea had a Mallampati score of three or higher [
51]. In addition, Sharara et al. [
52] also found a Mallampati score of at least three with a prevalence of 83% in a population of snorers. Furthermore, in our study the Mallampati score was assessed in the supine position, which may have led to an increased number of patients being graded with a Mallampati score of three or four. In a study by Bindra et al. [
53] the number of patients with a Mallampati score of three or four, more than doubled when assessed in the supine position. We, however, found no association between Mallampati index and sleep apnea or no sleep apnea.
The incidence of OSAS ranged between 15 and 59% depending on the questionnaire or score used. This was much higher than the incidence reported in the literature [
54‐
56] and could be due to the population studied, as our population was older and had more severe disease.
In our study, patients who had hypoxemia during flexible bronchoscopy under conscious sedation had increased AHI as measured by polygraphy. Harvin et al. found that patients with a high risk of sleep apnea as assessed by the Berlin score had a higher rate of hypoxemia during conscious sedation for colonoscopy [
28]. It is unclear whether this association remained true after adjusting for confounding variables. Cho et al. found an association between a high risk for sleep apnea, assessed using the STOP-BANG score, and cardiopulmonary events, such as hypoxia, during bronchoscopy [
18]. As with male gender, there is also a known association between sleep apnea, age and BMI [
47]. Thus both characteristics are also included in sleep apnea screening tools like Lausanne NoSAS [
12] and the STOP-Bang questionnaire [
11]. The association between AHI and oxygen desaturation disappeared in a multivariate analysis when adjusting for age and BMI and duration of bronchoscopy. Whereas age might be a confounder, BMI is probably related to effect modification, as risks tend to increase as a certain cut-off is reached.
In our non-selected group of patients, we had an incidence of overlap syndrome (COPD + sleep apnea) in 66/145 patients (46%). This is similar to incidence rates observed by Wang et al. [
57] and Zhang et al. [
58].
In our study, the Berlin Questionnaire had a 59% sensitivity and 43% specificity to identify sleep apnea as assessed by polygraphy. Only the STOP-BANG score was associated with oxygen desaturation during flexible bronchoscopy. This corroborates the findings of Cho et al. [
18]. The sensitivity and specificity of ESS was low and no association was seen with AHI. Conversely, Johns MW [
9] found that a higher ESS score correlates with the respiratory index measured during polysomnography and with the minimum oxygen saturation measured during the night.
The key message of the present study is that oxygen desaturation during endoscopic procedures, in particular flexible bronchoscopy, should prompt the treating physician to screen for sleep apnea. This could prove to be essential especially if we consider the prevalence of sleep apnea in this study, which is higher than in other epidemiologic studies. It also underlines that foremost in patient populations with high number of comorbidities, some of the screening tools, as for example the questionnaires evaluated in our study (NoSAS, STOP-BANG and ESS), could underestimate the prevalence of sleep-disordered breathing.
Limitations of this study include the fact that the sedation was applied by a trained nurse on the basis of clinical evaluation of sedation level (i.e. to achieve ptosis). This may result in an excessive sedation. Also, this was a single-center study in a tertiary hospital with patients with high ASA scores, therefore, the results may not be applicable to the general population.
Sleep apnea was determined by polygraphy and not polysomnography. However, polygraphy is confirmed as a viable diagnostic tool for sleep apnea and is included in the clinical algorithm for implementation of clinical practice guidelines by the American Academy of Sleep Medicine [
59]. The high agreement ratio between polygraphy and polysomnography [
60] may be attributable to the development of the peripheral arterial tonometry signal (PAT) which addresses the inability of other home sleep apnea tests to record and stage sleep. The algorithm used for sleep/wake detection and categorizing sleep stages REM/NREM have been described [
60,
61]. The overall agreement in detecting REM/NREM sleep was 88.7%/88.6%, respectively when compared to polysomnography [
62]. The limitation in using WatchPAT is that it is unable to reliably differentiate sleep apnea into obstructive, central or obstructive/central.
Another possible limitation is that since the exclusion criteria were few and the possibility to participate in the study was open, prevalently patients concerned about having a sleep disorder would participate. We don’t feel that this bias existed in our population. According to the Epworth Sleepiness Scale results, which is the reference standard used to determine the symptoms experienced by the patient, only 28/145 patients in the study actually had symptoms of a sleep disorder. Most of the patients had no symptoms even though they were at high risk of developing sleep apnea as shown with the STOP-BANG score, in which 93/145 (64%) patients had a score ≥ 3. In addition, 5.6% of the screened patients refused participation.
The strength of our study is the objective assessment of sleep apnea before undergoing bronchoscopy and the clinical applicability of study results.
In conclusion, oxygen desaturations during FB are associated with more severe sleep apnea. This association remains significant after adjusting for sedative dose and duration of procedure. It appears justifiable to consider sleep apnea screening for patients with oxygen desaturation during bronchoscopy.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.