Oxygen desaturation during flexible bronchoscopy with propofol sedation is associated with sleep apnea: the PROSA-Study

This was a prospective, investigator-initiated and driven, single-centre cross-sectional study performed at the Clinic of Respiratory Medicine and Pulmonary Cell Research at the University Hospital of Basel between October 2018 and August 2019. Patients older than 18 years undergoing a diagnostic flexible bronchoscopy were sequentially recruited. Exclusion criteria were hypoxemia at rest defined as an oxygen saturation of < 90% in room air, rapid fatal disease, any disease or condition precluding the initiation of continuous positive airway pressure (CPAP) therapy within the next 6 months and a new onset of cardio-respiratory symptoms (“unstable state”) as defined by a deterioration of cardio-vital signs within the last 48 h. Patients who required intubation during the bronchoscopy were withdrawn from the study.

The Study was approved by the Ethics Committee northwest/central Switzerland, EKNZ (EK 16/13) and was carried out according to the Declaration of Helsinki and Good Clinical Practice guidelines. Due to its observational character, the study did not require registration at a clinical trial registry.

Collected data included patient demographic data, current pulmonary function test, arterial blood gas analysis, oxygen saturation during bronchoscopy and all information related to the bronchoscopy. A targeted physical examination was performed on all subjects including BMI, neck circumference and examination of the oral cavity to record the oropharyngeal Mallampati score. All patients completed the ESS, and Berlin Questionnaire. The NoSAS Score and STOP-BANG score were calculated from measurements and data collected from the patient. The ESS is a self-administered questionnaire that provides insight into the subject`s general level of daytime sleepiness. [9] The ESS is applied by rating the probability of falling asleep during eight different scenarios usually encountered in daily life. A meta-analysis found the sensitivity of the ESS questionnaire ranges between 47 and 52% depending on the severity of OSA [19]. The Berlin Questionnaire is a validated instrument for screening patients at risk for sleep apnea. In this questionnaire, snoring and observed apnea (first domain, five questions), daytime sleepiness (second domain, four questions) and BMI and hypertension (third domain, one question and information on height and weight) are assessed. The Berlin Questionnaire is positive when two of the three domains are positive. It has a sensitivity of 86% in predicting patients with a respiratory disturbance index (RDI) above 5/h [10]. The NoSAS score is a screening tool used to identify persons at risk for sleep apnea [12]. It uses information about neck circumference, BMI, snoring, age and gender to predict sleep apnea. A score of 8 points or higher is predictive of sleep apnea. The NoSAS score has a sensitivity of 79–85% [12]. The STOP-Bang questionnaire is a concise screening tool for OSA. It is self-reportable and includes four subjective features (Snoring, Tiredness, Observed apnea and presence of high blood pressure) and four demographic queries (BMI, Age, Neck circumference and Gender). Every item can be scored with zero or one point with a maximum score of eight points. Using a cutoff of ≥ 3, the STOP-BANG score has a sensitivity of 84% in detecting OSA [11].

Polygraphy, was used to screen the patients for sleep apnea (WatchPAT™200 Unified, Itamar medical) and was performed in an in-patient setting within 7 days before the bronchoscopy. The following parameters were measured via three points of contact: peripheral arterial tonometry signal, heart rate, oximetry, actigraphy, body position, snoring and chest motion. AHI, RDI and oxygen desaturation index (ODI) based on true sleep time and sleep staging were assessed. Mild sleep apnea was defined as 5 ≤ AHI ≤ 15, moderate sleep apnea as 15 < AHI ≤ 30 and severe sleep apnea as AHI > 30. Sleep apnea syndrome is defined as AHI ≥ 5/h + symptoms of daytime sleepiness [8]. We calculated sleep apnea syndrome using the ESS, Berlin score, NoSAS score and STOP-BANG score.

All patients were assessed prior to the bronchoscopy by a physician or a member of the nursing team trained in anesthesiology and graded according to the American Society of Anesthesiologists (ASA) criteria. The bronchoscopy was performed trans-nasally or trans-orally with the patients in a semi-recumbent position. Pulse oximetry and electrocardiography were recorded continuously and blood pressure was monitored automatically and non-invasively every 5 min. These readings were recorded with the patient monitor, ixTrend Express (ixitos GmbH, Berlin Germany). Supplemental oxygen at 4 l/min was supplied through a nasal cannula to all patients, and oxygen delivery was increased gradually up to 12 L/min when oxygen saturation dropped below 90%. Lidocaine 2% gel was applied as a nasal anesthesia. The bronchoscopist introduced 5-ml aliquots of 1% lidocaine over the vocal cords, the trachea and both left and right main bronchi. In certain patients, up to 4 mg hydrocodone was administered prior to the procedure [20]. Nurses trained in endoscopy performed the conscious sedation with propofol, which was applied intravenously through an infusion-pump and in addition, propofol-boluses were applied and titrated to achieve adequate sedation (onset of ptosis) for the bronchoscopy. The amount of propofol applied was further titrated during the bronchoscopy in order to maintain conscious sedation, defined as a decreased state of consciousness that minimizes discomfort [21‐25]. Hydrocodone was usually applied as a cough-suppressant. [26, 27]

The event of a relevant oxygen desaturation during conscious sedation for flexible bronchoscopy was characterized as a minimal transcutaneous oxygen saturation < 90% for ≥ 5 s. For additional explorative analyses we also analysed oxygen desaturations < 90% for ≥ 1 min, desaturations ≤ 88% for 5 s and a ≥ 4% decrease in oxygen saturation from baseline. During the bronchoscopy the following additional parameters were noted: audible snoring for ≥ 10 s, non-invasive ventilation, chin support and witnessed apnea longer than 10 s.

Bronchoscopy complications were defined as bleeding requiring intervention, need for non-invasive ventilation, need to abort the examination, transfer to the intensive care unit, pneumothorax or death. The decision to discontinue the bronchoscopy, to provide chin support or non-invasive ventilation was taken by the bronchoscopist. [24, 27]

Differences in dichotomous variables were evaluated using the Chi-squared test or Fisher exact test, as appropriate. Normally distributed parameters were analyzed using the Student t-test for equality of means. All other continuously non-normally distributed parameters were evaluated using the non-parametric Mann–Whitney U test or Kruskal–Wallis test, as appropriate. The association between oxygen desaturation and a diagnosis of sleep apnea was evaluated by linear regression using a univariate model and a multivariate model adjusting for the duration of the bronchoscopy and amount of propofol used. The Statistical Package for Social Sciences Program (SPSS Inc, version 22 for Windows) was used. All tests are two-tailed, a P-value < 0.05 was considered significant. Results are expressed as mean (SEM) or median (interquartile range) unless stated otherwise.

We assumed that patients without desaturation below 90% during flexible bronchoscopy under conscious sedation would depict an AHI of 5/h ± 18 during polygraphy whereas patients with oxygen desaturation below 90% during FB would depict an AHI of 15/h ± 18. We considered an uneven distribution of desaturation during bronchoscopy of 70–30%. A total of 131 patients (77 + 54) would need to be included in the study to achieve a level of significance of 5% and a power of 80%. Considering a lost-to-follow-up of 5%, a total of 145 patients (85 + 60) was planned for inclusion in the study.

The number of patients screened for the study totaled 178. Of these 178 patients, 33 patients were excluded and 145 patients were included in the study (Fig. 1). The included population consisted of 90 (62%) males, the mean BMI was 25.6 ± 0.4 kg/m² and a third of the patients were current smokers (n = 48; Table 1). The majority of the patients had an ASA-Score III. A Mallampati score of 3–4 was present in 105/145 (72%) of the patients. The total number of patients with observed snoring during the bronchoscopy (n = 107; 74%) was higher than the number of patients with self-reported snoring (n = 74; 51%) and lower than the number of patients recorded snoring using polygraphy (n = 121; 83%). The most prevalent comorbidity was chronic obstructive pulmonary disease (COPD) followed by renal disease and coronary artery disease (Table 1).

Twenty-eight patients (19%) had an elevated ESS whereas the STOP-BANG was suggestive of 93/145 (64%) high-risk patients (Table 2). Nineteen/28 (68%), 18/28 (64%) and 22/28 (79%) patients with an elevated ESS were concomitantly ranked as high-risk by the Berlin score, the Lausanne No-SAS and the STOP-BANG score respectively.

The mean recorded time during the polygraphy was 7.2 ± 0.12 h with an average sleep time of 5.3 ± 0.12 h. The snore volume was on average 44 ± 0.76 dB and the average AHI was 20.3 ± 1.2. Information regarding sleep parameters can be found in Additional file 1: Table 1.

A normal AHI was measured in 14/145 (10%) of the patients. We detected mild sleep apnea in 52/145 patients (36%), moderate sleep apnea was observed in 49/145 patients (34%) and severe sleep apnea was seen in 30/145 patients (20%; Fig. 2).

There was a significant association between gender and incidence of sleep apnea (p = 0.034) with a higher incidence in males (94%) than females (84%). Oxygen desaturation during bronchoscopy, however, was not associated with gender (p = 0.56). When adjusting for Mallampati score and duration of bronchoscopy, oxygen desaturation during the bronchoscopy still had a significant association with AHI (p = 0.035). Logistic regression analyzing Mallampati score and its association with sleep apnea or no sleep apnea, showed no association whether as a discrete, categorical variable (p = 0.832) or when grouped according to severity (p = 0.477).

There was no association between the prevalence of sleep apnea and snoring as measured during the polygraphy (Pearson Chi-square, p = 0.729). Conversely, there was a significant correlation between AHI and age (Spearman Rho correlation coefficient = 0.254; p = 0.002), BMI (Rho = 0.347; p < 0.001), forced vital capacity (FVC) % predicted (Rho = − 0.188; p = 0.023); and total lung capacity (TLC) %predicted (Rho = − 0.277; p < 0.001). There was no correlation between AHI and pre or post bronchodilator forced expiratory volume in 1 s (FEV₁).

Bronchoscopy was performed on average 1.5 ± 0.08 days after polygraphy and had a mean duration of 23.5 ± 1.3 min. The average dosage of propofol needed for sedation was 323 ± 16.5 mg and 105/145 patients (72%) also received hydrocodone as a cough suppressant. There was no association (p = 0.35) between the administration of hydrocodone and sleep apnea. Of the patients receiving hydrocodone, 93/105 (89%) had sleep apnea. Of the patients not administered hydrocodone, 38/40 (95%) had sleep apnea. Hydrocodone had no effect on the incidence of oxygen desaturations (Pearson Chi-square, p = 0.75). Chin support was performed on 116/145 patients (80%), and 4/145 patients (3%) required non-invasive ventilation.

During bronchoscopy, an oxygen saturation of < 90% for ≥ 5 s was measured in 132/145 patients (91%; Fig. 3), an oxygen saturation < 88% for ≥ 5 s was measured in 123/145 patients (85%) and a decrease in oxygen saturation of ≥ 4% of the baseline oxygen saturation was measured in 132/145 patients (91%).

There was a significant difference in age between the patients who remained normoxemic during the bronchoscopy and the patients who had an oxygen saturation < 90% (57.9 ± 3.5 vs. 66.6 ± 1.1 years; p = 0.017), < 88% (56.0 ± 3.1 vs. 67.5 ± 1.1 years; p = 0.001) and ≥ 4% decrease from baseline (57.9 ± 3.5 vs. 66.6 ± 1.2 years; p = 0.017). There was also a difference in BMI between the patients who remained normoxemic and the patients who had an oxygen saturation < 90% (23.5 ± 1.3 vs 25.8 ± 0.42 kg/m²; p = 0.055), and those who had a ≥ 4% decrease from baseline (23.5 ± 1.3 vs. 25.8 ± 0.4 kg/m²; p = 0.055). Remarkably, lung function parameters such as post-bronchodilator FEV₁ (2.1 ± 0.3 vs. 2.3 ± 0.5 L; p = 0.248), post-bronchodilator FVC (3.4 ± 0.4 vs. 3.0 ± 0.1 L; p = 0.262), post-bronchodilator TLC (100 ± 8.5 vs. 107 ± 2.0%; p = 0.355) and post-bronchodilator vital capacity (VC) (3.6 ± 0.35 vs. 3.3 ± 0.1 L; p = 0.408) did not differ significantly between the normoxemic and the hypoxemic patients. Patients with oxygen desaturation during the bronchoscopy had significantly less sleep time during the polygraphy (Table 3).

Patients with oxygen desaturation < 90% had significantly higher AHI values compared to patients with no oxygen desaturation (Fig. 3). Thus, there was a significant association between AHI and oxygen desaturation < 90% (β 9.082 CI 0.982–17.182, p = 0.028) suggesting that patients presenting oxygen desaturation < 90% during bronchoscopy had an AHI 9.1/h higher than patients with no oxygen desaturation during bronchoscopy. This association remained significant after adjusting for the duration of the procedure and for the administered propofol dose (β 9.813 CI 1.382–18.243 p = 0.023) and also held true when examining desaturation < 90% for ≥ 1 min (Additional file 1: Fig. 1). The association disappeared when adjusting for gender, age and BMI together (β 5.729 CI − 2.261–13.719; p = 0.159).

When using polygraphy as a reference standard, oxygen desaturation during bronchoscopy had a sensitivity of 92% and a specificity of 14% to diagnose sleep apnea. The average score for the Epworth sleepiness questionnaire and the NoSAS score according to oxygen desaturation or no oxygen desaturation is depicted in Additional file 1: Table 2. STOP-BANG had the highest sensitivity and ESS the lowest sensitivity to diagnose sleep apnea when using either AHI or oxygen desaturation during bronchoscopy as reference standards (Table 4).

The sensitivity and specificity of oxygen desaturation < 90% during bronchoscopy to diagnose obstructive sleep apnea syndrome (OSAS) using AHI > 5/h and a positive symptom score as a reference standard is depicted in Table 5.

The values changed when taking into consideration sleep apnea syndrome with a pAHI ≥ 15 (Additional file 1: Table 3).