Effect of nasopharyngeal airway on hypoxemia in overweight and obese patients undergoing sedated endoscopy: A randomized trial

Summary

Hypoxemia is common during sedated gastrointestinal endoscopy (SGIE), especially in overweight and obese patients, yet the effectiveness of nasopharyngeal airways (NAs) for prevention remains unclear. This prospective randomized trial compared NA with standard nasal cannula (NC) in 256 overweight and obese patients. Results revealed that the NA group demonstrated a statistically significant reduction of hypoxemia (20/128 [15.6%] vs. 52/128 [40.6%]; p < 0.001) and severe hypoxemia (4/128 [3.1%] vs. 13/128 [10.2%]; p = 0.024), along with fewer jaw-thrust maneuvers (9/128 [7.0%] vs. 34/128 [26.6%]; p < 0.001), compared to the NC group. Multivariable analysis confirmed that higher STOP-BANG scores independently predicted hypoxemia risk, while the use of NA was an independent protective factor (odds ratio, OR, 0.23, 95% CI, 0.12–0.44, p < 0.001). These findings support using NA to reduce hypoxemia in high-risk patients and underscore the value of STOP-BANG for preoperative screening.

Graphical abstract

Highlights

• •NA reduces hypoxemia in overweight/obesity during sedated endoscopy
• •STOP-BANG score predicts hypoxemia risk in sedated endoscopy
• •NA decreases rescue interventions and improves provider satisfaction

Human physiology; Medical procedure

Introduction

Overweight and obesity have emerged as significant global public health concerns. As of 2022, approximately 2.5 billion adults aged 18 and older were classified as overweight, with 890 million of them categorized as obese, according to the World Health Organization.¹ Notably, it is worth highlighting that China reports the highest number of individuals affected by overweight and obesity worldwide.² Given this alarming prevalence, obesity is now recognized as a chronic, relapsing, and progressive disease, and its increasing prevalence imposes a considerable burden on healthcare systems due to its strong association with multiple comorbidities and severe complications.^3,4 For instance, the incidence of gastrointestinal (GI) disorders is notably higher in overweight and obese individuals compared to those with normal body weight.^5,6 Additionally, it is also important to note that overweight status is closely linked to obstructive sleep apnea (OSA), with a cohort study reporting OSA prevalence rates of 4% in middle-aged men and 2% in middle-aged women.⁷ Both obesity and OSA are established risk factors for perioperative airway complications, particularly hypoxemia.⁸

Consequently, overweight and obesity contribute to an increased risk of adverse respiratory events during sedation or anesthesia, especially during procedures like GI endoscopy. In particular, hypoxemia is a critical issue during sedated GI endoscopy (SGIE) due to the combined effects of sedative drugs that depress the respiratory center and relax the airway muscles, in addition to the endoscopic procedure itself, which can compromise ventilation.^9,10,11 Obese patients are anatomically predisposed to airway collapse from fatty tissue in the neck and restricted breathing from heavy chest and abdominal walls.^11,12,13,14 Moreover, sedative medications then critically exacerbate these issues by relaxing the airway muscles and depressing the drive to breathe. Our preliminary investigations have highlighted a significant association between higher body mass index (BMI) and the incidence of hypoxemia during SGIE.^10,15 Given these elevated risks, exploring effective strategies to prevent hypoxemia in overweight and obese patients undergoing sedated endoscopic procedures is crucial.

One such promising approach is high-flow nasal cannula (HFNC) oxygen therapy, delivering flows of 50–70 L/min, which has proven effective in significantly reducing the incidence of hypoxemia in obese patients during SGIE. Specifically, it reduces the rates of hypoxia, subclinical respiratory depression, and severe desaturation from 21.2% to 2%, without increasing the risk of other adverse events.¹⁶ However, despite its efficacy, the widespread adoption of HFNC is often hindered by its high cost, the need for specialized equipment, and time-consuming setup procedures. Thus, these challenges highlight the urgent need for a simpler, faster, and more cost-effective alternative to prevent hypoxemia in this high-risk patient group. In addition, unaddressed intraoperative hypoxemia has been linked to elevated postoperative troponin levels, suggesting increased myocardial stress in obese patients experiencing oxygen desaturation below 90% during procedures.¹⁷ Therefore, individualized oxygen titration strategies, ideally combined with continuous capnography, are essential to mitigate the neurological and cardiovascular complications associated with intraoperative hypoxemia in high-risk patients.¹⁸ Although nasopharyngeal airways (NAs) have demonstrated favorable safety profiles in airway management during procedures such as bronchoscopy,¹⁵ their potential role in reducing hypoxemia risk during SGIE in overweight and obese populations remains largely unexplored. Thus, this study aims to assess the clinical efficacy and safety of NA during SGIE in overweight and obese Chinese patients.

Results

A total of 300 overweight and obese patients scheduled for SGIE were initially enrolled in the study (Figure 1). Of these, 34 were excluded due to incomplete data and ten declined participations. As a result, the final cohort comprised 256 patients, with a median age of 51.9 years, including 174 males (67.97%). Importantly, demographic and clinical characteristics were similar between the two groups (Table 1). Additionally, Figure 2 presents NA placement during gastroscopy in subgroups stratified by sex and BMI. Notably, NA placement enhances oxygen delivery beyond the nasopharynx, near the epiglottis and glottis, thereby optimizing upper airway oxygenation.

Figure 1.

Flowchart of the present study

Table 1.

Demographic, clinical characteristics, and treatment variables of the included participants

Variables	Participants, no. (%) (N = 256)
	Group NA	Group NC	p value
Demographics and clinical characteristics
Sex (male/%)	86 (67.2)	88 (68.8)	0.789
Age, median (IQR), y	51.9 ± 11.6	51.8 ± 12.4	0.913
Body mass index (kg/m2)	27.0 ± 2.4	27.0 ± 2.1	0.761
Wrist circumference	17.0 ± 1.2	17.0 ± 1.0	0.809
Neck circumference	39.4 ± 3.7	39.9 ± 3.3	0.318
Metabolic equivalents	6.0 ± 1.6	6.1 ± 1.5	0.405
ASA	–	–	0.866
I	41 (32)	45 (35.2)	–
II	81 (63.3)	77 (60.2)	–
III	6 (4.7)	6 (4.7)	–
STOP-BANG score	3.4 ± 1.3	3.6 ± 1.2	0.300
Breath holding test (s)	39.2 ± 15.8	37.6 ± 12.2	0.348
Type of procedure			0.985
Gastroscopy	24 (18.8)	24 (18.8)	–
Colonoscopy	20 (15.6)	21 (16.4)	–
Gastrointestinal endoscopy	84 (65.6)	83 (64.8)	–
Hypertension (Y/N)	38 (29.7)	37 (28.9)	0.891
Diabetes (Y/N)	22 (17.2)	17 (13.3)	0.385
Coronary artery disease (Y/N)	7 (5.5)	5 (3.9)	0.554
History of gastrointestinal surgery (Y/%)	19 (14.8)	27 (21.1)	0.193
Dose of anesthetics
Total remimazolam, (mg)	10.0 (8.0, 14.2)	10.0 (7.5, 13.0)	0.199
Total ciprofol, (mg)	9.6 ± 4.0	9.8 ± 3.6	0.668

Abbreviations: NA, nasopharyngeal airway; NC, nasal cannula.

Figure 2.

Schematic gastroscopic views of nasopharyngeal airway placement by sex and BMI

(A–C) Endoscopic images showing placement of size 6 nasopharyngeal airway in female patients across different BMI categories.

(D–F) Endoscopic images showing placement of size 7 nasopharyngeal airway in male patients across different BMI categories.

NA markedly reduces hypoxemia incidence and severity

Compared to the nasal cannula (NC) group, the NA group exhibited a significant reduction in the incidence of hypoxemia. For hypoxemia defined as SpO₂ < 94%, the incidence decreased from 40.6% (52/128) to 15.6% (20/128), corresponding to an absolute risk reduction (ARR) of 25.0% and a relative risk reduction (RRR) of 61.6% (p < 0.001, Table 2). Although the odds ratio (OR) value is not provided in Table 2, based on further statistical analysis, the OR for SpO₂ < 94% was 0.27 (95% confidence interval [CI], 0.15–0.49, p < 0.001), demonstrating a significant effect of the NA group in reducing the incidence of hypoxemia. When hypoxemia was defined as SpO₂ < 90%, the incidence dropped from 26.6% (34/128) in the NC group to 7.0% (9/128) in the NA group, reflecting an ARR of 19.6% and an RRR of 73.7% (p < 0.001, Table 2), with an OR of 0.21 (95% CI, 0.10–0.46, p < 0.001). For severe hypoxemia (SpO₂ < 85%), the incidence fell from 10.2% (13/128) to 3.1% (4/128), resulting in an ARR of 7.1% and an RRR of 69.6% (p = 0.024, Table 2), with an OR of 0.29 (95% CI, 0.09–0.9, p = 0.032). Moreover, hypoxemia episodes occurred significantly more often in the NC group than in the NA group (p < 0.001, Table 2). Additionally, the number of patients requiring jaw-thrust maneuvers and the frequency of such interventions were significantly higher in the NC group. Interestingly, no patients in either group required endotracheal intubation. Although one patient in the NC group required temporary procedure interruption for mask ventilation, this difference was not statistically significant.

Table 2.

Outcomes in terms of peripheral desaturation events in the two groups

Variables	Group NA	Group NC	p value
Primary outcomes
Lowest SpO2 < 94%	20 (15.6)	52 (40.6)	<0.001
Lowest SpO2 < 90%	9 (7.0)	34 (26.6)	<0.001
Lowest SpO2 < 85%	4 (3.1)	13 (10.2)	0.024
Hypoxemia (Y/N)	20 (15.6)	52 (40.6)	<0.001
Lowest SpO2, Mean ± SD	95.4 ± 3.9	92.1 ± 8.9	<0.001
Frequency of hypoxemia, Median (IQR)	0.0 (0.0, 0.0)	0.0 (0.0, 1.0)	<0.001
Jaw lifting (Y/N)	9 (7.0)	34 (26.6)	<0.001
Times of jaw lifting, Median (IQR)	0.0 (0.0, 0.0)	0.0 (0.0, 1.0)	<0.001
Secondary outcomes: complications and NRS score
hiccup, n (%)	25 (19.5)	16 (12.5)	0.125
Involuntary movement	8 (6.2)	9 (7)	0.802
Hypotension	10 (7.8)	16 (12.5)	0.214
Hypertension	11 (8.6)	8 (6.2)	0.474
Bradycardia	5 (3.9)	3 (2.3)	0.722
Tachycardia, n (%)	4 (3.1)	8 (6.2)	0.237
Nausea	2 (1.6)	1 (0.8)	1.000
Vomiting	1 (0.8)	0 (0.0)	1.000
Abdominal distension	16 (12.5)	13 (10.2)	0.554
Abdominal pain	11 (8.6)	15 (11.7)	0.408
Choke	9 (7)	12 (9.4)	0.494
Nasopharyngeal pain	0.0 (0.0, 0.0)	0.0 (0.0, 0.0)	<0.001
Anesthesiologist satisfaction score, Median (IQR)	10.0 (9.0, 10.0)	9.0 (8.0, 10.0)	<0.001
Endoscopist satisfaction Score, Median (IQR)	10.0 (9.0, 10.0)	9.0 (9.0, 10.0)	<0.001
Patient Satisfaction Score, Median (IQR)	10.0 (9.0, 10.0)	10.0 (9.0, 10.0)	0.660

Continuous variables are reported as the mean ± SD or median (range), depending on data distribution; categorical (ordinal) variables are reported as median (range). Comparisons between the two groups were conducted using an unpaired t test for normally distributed continuous variables and a Mann-Whitney U test for non-normally distributed variables. For categorical variables, Pearson’s chi-square test was performed when the assumption of expected cell counts ≥5 was satisfied; otherwise, Fisher’s exact test was used. HR, heart rate; MAP, mean arterial pressure; SpO₂, pulse oxygen saturation; Y, yes; N, no; NRS, numerical rating scale.

Comparable major adverse events and improved provider satisfaction with NA

Several major adverse events were monitored, including hiccups, involuntary movements, hypertension, hypotension, tachycardia, bradycardia, nausea, vomiting, abdominal distension, abdominal pain, choking, and nasopharyngeal discomfort. However, statistical analysis revealed no significant differences in the incidence of these events between the two groups. Following the procedure, satisfaction surveys were conducted among endoscopists, anesthesiologists, and patients. Results indicated significantly higher satisfaction scores for both endoscopists and anesthesiologists in the NA group compared to the NC group, while patient-reported satisfaction showed no significant difference (Table 2). These findings suggest that the use of an NA enhances procedural satisfaction among healthcare providers without compromising patient satisfaction.

Correlation and regression analyses of risk factors associated with minimum intraoperative oxygen saturation

Correlation analyses were conducted to explore the relationships between known hypoxemia-related risk factors and patients’ minimum intraoperative oxygen saturation (Min SpO₂) (Figures 3A–3L). Age showed a significant negative correlation with Min SpO₂ in the overall cohort (R = −0.25, p < 0.0001; Figure 3A), as well as in the NC group (R = −0.28, p = 0.0015; Figure 3B) and the NA group (R = −0.25, p = 0.0047; Figure 3C). Similarly, BMI exhibited a significant inverse relationship with Min SpO₂ in all patients (R = −0.23, p = 0.0003; Figure 3D), in the NC group (R = −0.23, p = 0.0095; Figure 3E), and more strongly in the NA group (R = −0.33, p = 0.0001; Figure 3F). Neck circumference was not significantly associated with Min SpO₂ in the overall cohort (R = −0.12, p = 0.066; Figure 3G) or in the NC group (R = −0.072, p = 0.42; Figure 3H), but exhibited a weak negative correlation in the NA group (R = −0.20, p = 0.023; Figure 3I). STOP-BANG scores were significantly negatively correlated with Min SpO₂ in all patients (R = −0.29, p < 0.0001; Figure 3J), with a stronger correlation in the NC group (R = −0.37, p < 0.0001; Figure 3K) and a moderate correlation in the NA group (R = −0.20, p = 0.024; Figure 3L).

Figure 3.

Correlation between minimum oxygen saturation (Min SpO₂) and physical parameters

(A–C) Age exhibited a significant negative correlation with Min SpO₂ in the overall cohort (A), the NC group (B), and the NA group (C).

(D–F) Body mass index (BMI) was inversely correlated with Min SpO₂ in all patients (D), as well as in the NC (E) and NA (F) subgroups.

(G–I) Neck circumference was inversely correlated with Min SpO₂ only in the NA group (G), with no significant association observed in the NC group (H) or the overall cohort (I).

(J–L) STOP-BANG scores demonstrated a negative correlation with Min SpO₂ in all patients (J), and similarly in the NC (K), and NA (L) groups. Correlations between minimum SpO₂ and clinical variables were analyzed using Pearson’s correlation, as all variables were continuous and demonstrated approximately linear relationships. For each scatterplot, the Pearson correlation coefficient (r) and the corresponding two-sided p value were computed. Linear regression lines were fitted using the least-squares method, and the shaded regions represent the 95% confidence intervals of the regression estimates. Each data point corresponds to an individual participant. Abbreviations: BMI, body mass index; Min SpO₂, minimum pulse oxygen saturation.

To assess the independent effect of the STOP-BANG score, multivariable linear regression analysis was performed (Table S1). Higher STOP-BANG scores, indicating a greater risk of OSA,¹⁹ were strongly associated with lower Min SpO₂. The unadjusted model showed a β coefficient of −1.61 (95% CI, −2.27 to −0.96; p < 0.001), and this association remained robust across all adjusted models. In the fully adjusted Model V, the STOP-BANG score remained an independent predictor of reduced Min SpO₂ (β = −0.94; 95% CI, −1.79 to −0.09; p = 0.032; Table S1). This finding indicates that the STOP-BANG score is a significant independent predictor of intraoperative hypoxemia risk, highlighting the need for comprehensive preoperative assessment in patients undergoing SGIE.

Multivariable logistic regression analyses of the impact of oxygen delivery methods on intraoperative hypoxemia

A multivariable logistic regression analysis was performed to evaluate the association between different oxygen delivery methods and the risk of intraoperative hypoxemia (Table S2). Using the NC group as the reference, the NA group consistently demonstrated a statistically significant reduction in the risk of hypoxemia across all four models. In the unadjusted model (Model I), the NA group exhibited a 73% lower risk of hypoxemia compared to the NC group (OR, 0.27, 95% CI, 0.15–0.49, p < 0.001). This association remained significant after adjusting for age and sex in Model II (OR, 0.25, 95% CI, 0.14–0.47, p < 0.001). Further adjustment for American Society of Anesthesiologists physical status, BMI, and metabolic equivalent in Model III slightly strengthened the association (OR, 0.22, 95% CI, 0.12–0.42, p < 0.001). In the fully adjusted model (Model IV), which also controlled for neck circumference, STOP-BANG score, and breath-holding test results, the use of NA remained significantly protective against hypoxemia, reducing the risk by 77% (OR, 0.23, 95% CI, 0.12–0.44, p < 0.001; Table S2). These findings suggest that the use of NA significantly reduces the likelihood of hypoxemia during sedated GI endoscopy, independent of patient demographics, cardiopulmonary fitness, and upper airway risk factors.

Discussion

In overweight and obese patients undergoing SGIE, the use of NA oxygen delivery demonstrated a statistically significant reduction in the incidence of hypoxemia compared to standard low-flow NC oxygen therapy. These findings clearly highlight the clinical value of NA as an effective strategy for perioperative oxygenation management in high-risk populations, further supporting the need for broader clinical adoption of this approach.

We acknowledge that choosing low-flow NC as the control, although consistent with current common practice,^11,12,15 inherently introduced a known and expected high risk of hypoxemia, as reflected by the 40.6% incidence rate observed in the NC group. This is a limitation of our trial design and suggests that a more appropriate comparator could have been more beneficial in minimizing this risk and bias. A more appropriate comparator, such as HFNC,¹¹ would have provided a more rigorous and clinically relevant benchmark, potentially reducing the observed treatment effect size and minimizing potential performance bias. Despite this limitation, the significant risk reduction observed with NA highlights its potential clinical utility as a simple and cost-effective intervention in high-risk populations.

The potential clinical implications of these results are substantial. Specifically, these findings may lead to guidelines demoting HFNC from a first-line choice to a rescue therapy for high-risk patients only, thereby streamlining workflows and reducing equipment costs.^11,12 In contrast, NA presents as a practical alternative because it is significantly less expensive, requires minimal equipment, and the study demonstrates that it provides comparable clinical efficacy for preventing hypoxemia in the majority of patients undergoing procedural sedation. As a result, NA emerges as a highly scalable and cost-effective solution for routine use.

The physiological rationale for using NA in this context is well grounded. With the growing demand for comfortable healthcare, there has been a significant increase in SGIE procedures in China, underscoring the need for both patient comfort and safety. A meta-analysis reported that ciprofol was associated with a lower incidence of respiratory depression than propofol and with markedly less injection pain.²⁰ In addition, our prior work also demonstrated that combining ciprofol with remimazolam resulted in a further reduction in hypoxemia and significantly shorter induction and recovery times compared with ciprofol alone.¹⁰ However, despite these advantages, clinical observations indicate that hypoxemia remains common in overweight and obese patients. To address this issue, we implemented an NA to evaluate whether it could reduce hypoxemia and streamline the workflow of sedated endoscopy in these high-risk cases. Overweight and obesity exacerbate the risk of upper airway obstruction through two primary mechanisms: (1) excess soft tissue deposition within the maxillomandibular region causes pharyngeal crowding and reduced airway diameter, predisposing the airway to collapse and (2) obesity, particularly central obesity, leads to an increase in visceral fat volume, which in turn reduces lung volume. This reduction in lung volume may increase pharyngeal wall collapsibility, possibly due to decreased longitudinal tracheal traction.¹⁴ The NA effectively maintains upper airway patency by bypassing glossoptosis-induced obstruction during sedation. As shown in Figure 2, placing the NA 0.5–1 cm above the glottis facilitates direct supraglottic oxygen delivery, helping prevent airway obstruction in deeply sedated patients. This benefit has been previously validated in plastic surgery as an alternative to the laryngeal mask airway.²¹ Our findings confirm that the use of NA significantly reduces both the incidence and frequency of hypoxemia in overweight and obese patients, with the rate of SpO₂ ≤ 94% reduced to 15.6% (20/128), compared to 40.6% (52/128) in the standard NC oxygen group.

Recent international guidelines, including those from the British Society of Gastroenterology,²² Japan,²³ and the German DGVS,²⁴ consistently recommend moderate sedation as the standard of care for GI endoscopy, reserving deep sedation for selected cases because of its higher risk profile, including the potential loss of airway protective reflexes. We acknowledge that our study’s use of deep sedation for all enrolled patients represents a deviation from these guideline recommendations. The guideline-concordant approach for the majority of diagnostic endoscopic procedures in this population would have been moderate sedation. However, previous studies have also demonstrated that deep sedation can be safely used for gastroscopy and colonoscopy in obese patients, particularly when performed under the supervision of an anesthesiologist.^11,12 In this context, our study specifically enrolled overweight and obese individuals, who are at increased risk of airway collapse and oxygen desaturation. An anesthesiologist was present throughout every procedure to ensure the safety of deep sedation. Notably, the incidence of hypoxemia in the NA group was only 15.6%, and all cases were promptly managed without adverse outcomes. By investigating deep sedation, this study evaluated the performance of NA under the most clinically demanding conditions, in which its benefit is likely to be most pronounced. This demonstrates its efficacy in a high-risk population. Notwithstanding this rationale, the generalizability of our findings to settings utilizing moderate sedation may be limited. Future studies should directly evaluate the efficacy of NA under moderate sedation, which represents the standard of care for most patients.

Finally, our results reinforce the value of preoperative risk stratification. The STOP-BANG questionnaire consists of eight items, with each “Yes” response scoring one point and each “No” response scoring zero, resulting in a total score ranging from zero to eight.^25,26 It is a widely used screening tool for assessing the risk of OSA and has also been investigated in relation to hypoxemia, particularly in surgical and perioperative settings. While the STOP-BANG score is effective in identifying individuals at high risk for OSA, its ability to predict sedation-related hypoxemia during gastroscopy remains uncertain. Some studies have suggested a positive correlation between higher STOP-BANG scores and increased risk of intraoperative hypoxemia.²⁵ Conversely, other research has indicated that its predictive value for postoperative hypoxemia may be limited, potentially influenced by factors such as the type and dosage of anesthetic agents administered.²⁷ In our study, a significant negative correlation was observed between STOP-BANG scores and intraoperative oxygen saturation levels, indicating that higher scores were associated with greater severity of hypoxemia. These findings suggest that patients with elevated STOP-BANG scores may require enhanced intraoperative respiratory monitoring and preventive strategies to mitigate the risk of severe hypoxemia during SGIE.

This study highlights the clinical value of NA use in overweight and obese patients undergoing SGIE. In particular, NA significantly reduces the incidence of hypoxemia, and the STOP-BANG questionnaire proves to be a useful tool for preoperative risk stratification. Taken together, these findings provide strong support for the broader clinical adoption of NA in high-risk populations and underscore the importance of individualized airway management strategies during procedural sedation.

Limitations of the study

This study has several limitations that should be acknowledged. Firstly, its single-center design may limit the generalizability of the findings, as patient demographics and local clinical practices could introduce bias. As such, future multicenter studies are warranted to validate our results across diverse settings. In addition, the assessment was confined to in-hospital outcomes, leaving the long-term safety profile unexplored. Therefore, subsequent studies with extended follow-up are crucial to evaluate delayed complications, functional recovery, and other pertinent long-term clinical endpoints. Moreover, the efficacy of NA and HFNC in preventing hypoxia during sedated GI endoscopy among obese patients requires further investigation, which will be the focus of our subsequent research. Lastly, a key methodological limitation is the inability to blind the clinicians performing the procedure, which was unavoidable because of the obvious differences in appearance and placement between the NA and the NC.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Ru-Yi Luo (luoruyi@csu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data reported in this paper will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This trial was funded by the Scientific Project of Hunan Health Commission (grant no. B202304116531 to R.-Y.L.), the Natural Science Foundation of Hunan (grant no. 2024JJ5503 to R.-Y.L.), and the Natural Science Foundation of Changsha (grant no. kq2403082 to R.-Y.L.).

Author contributions

R.-Y.L. and H.-D.Z. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design, R.-Y.L. and H.-D.Z.; acquisition, analysis, or interpretation of data, R.-Y.L. and H.-D.Z.; drafting of the manuscript, R.-Y.L. and H.-D.Z.; critical review of the manuscript for important intellectual content, all authors; statistical analysis, H.-D.Z. and R.-Y.L.; administrative, technical, or material support, Q.L., J.Z., Y.-X.H., Z.-L.H., C.L., and P.Z.; supervision, R.-Y.L.

Declaration of interests

The investigators of this trial have no financial or competing interests to declare.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Software and algorithms
PASS 15.0 software	PASS software	https://www.ncss.com/software/pass/
R studio version 4.3.3	R software	https://cran.rstudio.com/
Free Statistics software version 2.1	Free Statistics software	https://www.clinicalscientists.cn/freestatistics/

Experimental model and study participant details

Ethics approval

This study was approved by the Second Xiangya Hospital of Central South University Review Board (approval no. LYEC2025-0144).

Study population

This study enrolled 300 patients undergoing SGIE at the Endoscopy Center of the Second Xiangya Hospital, Central South University, between June and July 2025. The inclusion criteria were: American Society of Anesthesiologists (ASA) physical status classification I–III, age between 18 and 85 years, metabolic equivalents (METs) ≥ 3, BMI ≥ 24 kg/m², and no history of nasal abnormalities or recent nasal surgery. On the other hand, exclusion criteria were: severe obstructive ventilatory dysfunction, baseline pre-anesthesia SpO₂ < 94%, known nasal or nasopharyngeal disorders, significant hepatic or renal insufficiency, pregnancy or lactation, known allergies, or contraindications to benzodiazepines or ciprofol.

Method details

Randomization and blinding

Overweight was defined as 24.0 ≤ BMI < 28.0 kg/m², while obesity was defined as BMI ≥ 28.0 kg/m², in accordance with the Chinese adult BMI classification criteria.²⁸ Participants classified as overweight or obese were randomized in a 1:1 ratio using a random number table generated by R software (version 4.3.3). Participants with a remainder of 0 when the random number was divided by 2 were assigned to the nasal cannula oxygen therapy (NC) group, whereas those with a remainder of 1 were assigned to the NA group. The NAs used in this study (manufactured by Tianjin Meidisi Company) were available in four sizes: 4, 5, 6, and 7, with sizes 6 and 7 intended for adults. The NA, made of medical-grade polyvinyl chloride (PVC), was selected according to patient anatomy and was lubricated prior to insertion. Following placement, it was secured by an adjustable positioning ring and nasal cavity, and tube position was confirmed through clinical assessment and physiologic monitoring such as capnography. The allocation of participants to groups was concealed from patients, endoscopists, postoperative care providers, outcome assessors, and statisticians. Due to the visible differences between NA and NC, blinding of anesthesiologists performing the procedure was not feasible.

Anesthesia protocol

The anesthesia protocol followed our previously published methodology.¹⁰ All patients adhered to standard preoperative fasting guidelines, abstaining from solid food for at least 8 hours and clear liquids for 2 hours prior to the procedure. Upon arrival, intravenous (IV) access was established. In the operating room, patients were positioned in the left lateral decubitus position and continuously monitored using electrocardiography (ECG), non-invasive blood pressure (NIBP), peripheral oxygen saturation (SpO₂), and heart rate (HR). To begin with, all patients initially received oxygen via nasal cannula at a flow rate of 5 L/min.

In the experimental group, pre-oxygenation was administered via nasal cannula prior to sedation. Once the Modified Observer’s Assessment of Alertness/Sedation (MOAA/S) score reached ≤ 1, the nasal cannula was replaced with an NA to improve patient comfort during insertion, after which endoscopy was initiated. In contrast, the control group received nasal cannula only (5 L/min) during the SGIE procedure.

All patients received an initial IV bolus of remimazolam (0.1 mg/kg), followed by ciprofol (0.35–0.7 mg/kg) to achieve a target MOAA/S score of ≤ 1. Sedation was maintained using additional bolus doses of remimazolam (2–3 mg) or ciprofol (2.5–7.5 mg) as needed, with preference given to remimazolam during the later stages of the procedure. At the conclusion of the procedure, flumazenil (0.3 mg IV) was routinely administered for sedation reversal. Finally, all patients were monitored in the post-anesthesia care unit (PACU) before discharge.

Outcome measurements

Baseline data were collected for all participants, including demographic and clinical characteristics such as medical history, age, height, weight, METs, BMI, ASA classification, STOP-BANG score, and neck and wrist circumferences. Results from the preoperative breath-holding test were also recorded. Vital signs, including mean arterial pressure (MAP), HR, and SpO₂, were continuously monitored.

Intraoperative outcomes were recorded, including the incidence and frequency of hypoxemia, minimum SpO₂ values, and the use of airway interventions (e.g., jaw thrust, mask ventilation, or endotracheal intubation). Additional outcomes included the occurrence of hypotension requiring vasopressor administration, total anesthetic drug consumption, and recovery times—defined as the time to reach a MOAA/S score ≥ 4, and the time to discharge from the PACU with a MOAA/S score of 5. Postoperative satisfaction was evaluated using a 10-point Numeric Rating Scale (NRS) by the attending anesthesiologist, the endoscopist, and the patient.

Primary and secondary outcomes

The primary outcome was the incidence of hypoxemia (defined as SpO₂ < 94%) in overweight and obese patients undergoing SGIE, comparing those who received NA placement with those in the control group who received low-flow nasal cannula oxygen therapy, in line with current clinical practice guidelines.

Secondary outcomes included intraoperative adverse events such as hypotension, hypertension, tachycardia, bradycardia, and coughing, as well as satisfaction ratings from the endoscopist, anesthesiologist, and patient. Additionally, risk factors associated with hypoxemia, including demographic characteristics, physical parameters (e.g., BMI, neck circumference), and STOP-BANG scores, were analyzed.

Sample size estimation

Sample size calculation was performed using PASS 15.0 software for two independent proportions, based on the estimated difference in hypoxemia incidence between the two groups. Previous studies comparing various oxygenation devices with standard oxygen therapy reported differences in hypoxemia incidence ranging from 6% to 25%.^12,29,30 In the preliminary clinical trial, 63 patients were enrolled in each group. Hypoxemia occurred in 11 patients in the NA group and 21 patients in the nasal cannula group, corresponding to incidence rates of 17.46% and 33.33%, respectively. Assuming a 1:1 allocation ratio between groups, PASS 15.0 software was used to calculate the required sample size, with a one-tailed significance level of 0.025 and a power of 80%. The estimated sample size was 230 patients (115 per group). To account for a projected 10% dropout rate, a total of 256 patients (128 per group) were ultimately enrolled in the study.

Quantification and statistical analysis

Statistical analyses were conducted using R software (version 4.3.3) and Free Statistics software (version 2.1, Beijing, China).³¹ Continuous variables with a normal distribution, as determined by the Shapiro–Wilk test, were presented as means ± standard deviations (SD) and compared using independent samples t-tests. Non-normally distributed continuous variables were expressed as medians with interquartile ranges (IQRs) and compared using the Mann–Whitney U test. Categorical variables were presented as counts and percentages and compared using Pearson’s χ² test or Fisher’s exact test when any expected cell count was < 5. Correlations between minimum SpO₂ and clinical variables were analyzed using Pearson’s correlation, as all variables were continuous and demonstrated approximately linear relationships. For each scatter plot, the Pearson correlation coefficient (r) and the corresponding two-sided P value were computed. Linear regression lines were fitted using the least-squares method, and the shaded regions represent the 95% confidence intervals of the regression estimates. Each data point corresponds to an individual participant. Multivariate linear regression was performed to evaluate the association between the STOP-BANG score and minimum oxygen saturation, adjusting for relevant covariates. Odds ratios (ORs) and their 95% confidence intervals for the risk of hypoxemia were calculated using univariate binary logistic regression models, with oxygen delivery method (NA vs. NC) as the sole independent variable and hypoxemia status as the binary outcome. Separate logistic regression models were fitted for each hypoxemia definition (SpO₂ < 94%, SpO₂ < 90%, and SpO₂ < 85%). P values were obtained from Wald tests of the regression coefficients. Multivariable logistic regression was performed to evaluate the relationship between oxygen delivery methods and the occurrence of hypoxemia. All tests were two-tailed, with a P-value < 0.05 considered statistically significant. Exact P values are reported throughout the manuscript.

Additional resources

The trial was registered at https://www.chictr.org.cn/showproj.html?proj=274359 (registration no. ChiCTR2500103631, principal investigator: Hai-Ding Zou, date of registration: June 3, 2025).

Published: December 6, 2025