Risk of respiratory complications following gastrointestinal endoscopy under monitored anesthesia care: A prospective cohort study

Abstract

BACKGROUND

Compared to immediate cardiopulmonary complications during gastrointestinal endoscopy (GIE), post-procedural respiratory symptoms are often overlooked. Furthermore, an increased incidence of respiratory complications has been reported following GIE under moderate sedation (MS), administered by endoscopists or endoscopist-directed nurse sedation. However, monitored anesthesia care (MAC), which primarily involves deep sedation/general anesthesia (DS/GA), has not been evaluated yet regarding post-endoscopic respiratory complications.

AIM

To investigate the incidence of post-procedural respiratory complications and explore their links to different sedation strategies.

METHODS

In this single-center prospective cohort study, outpatients undergoing GIE under MAC were stratified into DS/GA (exposed) and MS (non-exposed) groups. Telephone follow-up post-procedure was used to assess symptoms such as cough, fever, and shortness of breath. Respiratory complications were defined as the presence of ≥ 2 respiratory symptoms, with further follow-up conducted to determine treatment and outcomes. Multivariate logistic regression was used to analyze the association between sedation depth and respiratory complications. Sensitivity analyses were conducted to verify reliability.

RESULTS

Seventy-seven of the 7230 patients experienced respiratory complications. Respiratory complications occurred 8.0 (4.0, 12.0) hours post-operatively, with a median duration of 1.5 (1.0, 2.5) days. Additional medical intervention was required in 23 cases. One patient (1/7230) was diagnosed with aspiration pneumonia and hospitalized for one week. All patients recovered without any severe sequelae. Compared with that in the MS group, the adjusted odds ratio for respiratory complications in the DS/GA group was 2.75 (95%CI: 1.32-5.71; P = 0.007). Subgroup analyses revealed similar patterns (all P values for interactions > 0.05). The association between DS/GA and respiratory complications remained significant in the sensitivity analysis after multiple imputation for 293 lost-to-follow-up patients (odds ratio = 2.86, 95%CI: 1.37-5.98).

CONCLUSION

Although DS/GA is associated with an increased risk of GIE-induced respiratory complications, MAC with sedation strategies tailored to patients’ risk factors may not significantly increase the incidence of respiratory complications.

Key Words: Gastrointestinal endoscopy; Monitored anesthesia care; Respiratory complications; Moderate sedation; Deep sedation; General anesthesia

Core Tip: This prospective cohort study investigated delayed respiratory complications after gastrointestinal endoscopy under monitored anesthesia care. Deep sedation or general anesthesia increased complication risk, while risk-stratified monitored anesthesia care strategies reduced adverse events in high-risk patients, providing key evidence for clinical sedation selection.

INTRODUCTION

Existing studies on sedation-related complications in patients with gastrointestinal endoscopy (GIE) have focused primarily on adverse acute cardiopulmonary events during the procedure[1-4]. However, delayed postprocedural respiratory complications remain understudied. Clinically significant respiratory symptoms often manifest after the procedure, leading to frequent misdiagnosis. Furthermore, many patients who undergo GIE have comorbid gastroesophageal reflux disease (GERD). Moreover, compared with other medical interventions, endoscopic procedures that require air insufflation and water irrigation might increase the risk of reflux and aspiration[5-7]. Additionally, 5.3% of patients developed ≥ 1 respiratory symptoms following GIE with endoscopist-directed sedation[6]. A retrospective analysis revealed that 1.9% of patients who underwent esophagogastroduodenoscopy (EGD) experienced aspiration pneumonia, which was associated with significantly increased mortality and prolonged hospitalization[8].

Monitored anesthesia care (MAC) describes a specific anesthesia service performed by a qualified anesthesia provider for a diagnostic or therapeutic procedure. In contrast to endoscopist-directed moderate sedation (MS) protocols, MAC represents an anesthesiologist-administered continuum from MS to general anesthesia (GA). While the literature documents differential complication rates between these approaches[2-4,7,9], GIE-associated respiratory complications under MAC remain poorly characterized. Our prospective cohort study aimed to investigate the incidence of GIE-associated respiratory complications under MAC and evaluate the association between sedation levels and respiratory complications.

MATERIALS AND METHODS

Study design and population

We conducted this prospective, single-center study from December 2024 to February 2025 at the Affiliated Hospital of Qingdao University. The study was conducted in accordance with the principles of the Declaration of Helsinki. It was approved by the Ethics Committee of Affiliated Hospital of Qingdao University (No. QYFYEC2024-218) and was registered on clinicaltrials.gov (NCT06705894). All patients provided informed consent before the procedure.

Inclusion criteria: (1) Aged ≥ 18 years; (2) Adequate bowel preparation (Boston score ≥ 6); and (3) Scheduled for colonoscopy, EGD, or double examination under MAC.

Exclusion criteria: (1) Pregnant individuals; (2) Patients with pre-existing respiratory symptoms, including fever, cough, and shortness of breath (SOB); (3) Patients scheduled for endotracheal intubation under GA; and (4) Those who did not provide informed consent.

Data collection

Data were prospectively collected using a standardized Microsoft Excel template, which included the following parameters: (1) Age; (2) Sex; (3) Body mass index (BMI); (4) American Society of Anesthesiologists (ASA) classification; (5) Smoking and alcohol consumption history; (6) Comorbidities (asthma, chronic obstructive pulmonary disease, hypertension, heart disease, cerebrovascular diseases, depression, diabetes mellitus, GERD, hiatus hernia, and gastrointestinal ulcer); (7) Previous surgery (upper gastrointestinal, colorectal, pulmonary, oral-pharyngeal); (8) Intraoperative adverse events (severe cough, regurgitation aspiration, hypoxemia); (9) Type of procedures (EGD, colonoscopy, or combined); (10) Procedure duration; and (11) Sedation depth. The standardized Excel sheet required input for the study parameters, preventing finalization of the endoscopy report without data on sedation.

Standardized sedation/anesthesia protocol

Our sedation strategy comprises MS and deep sedation (DS)/GA, which are all determined and administered by anesthesiologists. Using balanced propofol sedation (combining sufentanil, remimazolam, and low-dose propofol)[10], MS was administered to patients with ASA class IV or higher and those with elevated aspiration risk. DS/GA comprising propofol alone or with sufentanil was administered to other patients. During the procedure, the depth of sedation was assessed every 5 minutes by anesthesiologists, and additional sedative agents were administered as needed according to the assessment results to maintain the target sedation level. The distinction between DS and GA is difficult to determine when high-dose propofol is used alone or in combination with opioids; therefore, in clinical practice, we do not strictly differentiate between the two states[11,12]. DS/GA and MS levels were indicated by a MOAA/S score of ≤ 1 and a score of 2-3, respectively[7].

Subsequently, oxygen (5 L/minutes) was delivered via a nasal cannula, and continuous electrocardiogram, oxygen saturation, and noninvasive blood pressure monitoring were performed. All procedures were performed under standardized monitoring. Intraoperative hypoxemia was managed stepwise: (1) The fraction of inspired oxygen was increased, jaw thrust was performed, and an oropharyngeal/nasopharyngeal airway was inserted if needed; and (2) If ineffective, bag-mask ventilation or endotracheal intubation was performed. In the case of regurgitation/aspiration, the procedure should be stopped immediately, the patient placed in the Trendelenburg position, the airway suctioned, and intubation performed if indicated. We recorded adverse respiratory events that occurred during the examination. Severe coughing was defined as persistent, forceful coughing occurring at a frequency exceeding ten episodes per minute or any coughing that resulted in interruption of the procedure. Regurgitation and aspiration were defined as definitive evidence of gastric contents moving into the oropharyngeal cavity or airway. An oxygen saturation ≤ 90% for > 30 seconds was defined as hypoxemia. The patients were transferred to the post-anesthesia care unit after the examination. Discharge criteria included a modified Aldrete score ≥ 9[13].

GIE

All patients fasted for ≥ 8 hours and underwent standard bowel preparation for colonoscopy. Intravenous access was established with a continuous normal saline infusion on the procedure day. Subsequently, in the left lateral position, the gastroenterologist performed GIE according to the order of gastroscopy and colonoscopy at the target sedation depth. All involved endoscopists had a mean annual case volume exceeding 1500 procedures.

Follow-up

Follow-up was conducted via blinded telephone interviews by research staff. At 24 hours post-endoscopy, patients were asked to report the occurrence of coughing, fever (axillary temperature ≥ 37.3 °C), and SOB. Those reporting ≥ 2 concomitant respiratory symptoms after 24 hours were classified as having “respiratory complications”[6]. Furthermore, patients with respiratory complications underwent extended follow-up to assess symptom duration, worsening, therapeutic interventions, and hospitalization until symptoms completely resolved. Apart from standard follow-up procedures, patients received no additional planned interventions following discharge. The study protocol did not mandate routine post-discharge testing such as, pulse oximetry, complete blood count, or chest radiography, for any participant. All subsequent healthcare visits and diagnostic evaluations were initiated voluntarily by the patients themselves based on their clinical needs. Additionally, fever was defined as an axillary temperature of ≥ 37.3 °C, and SOB was defined as a score of ≥ 1 on the modified Medical Research Council dyspnea scale[14]. Cough severity was measured by the Numeric Rating Scale (0-10), with a score ≥ 4 indicating a clinically significant cough. Aspiration pneumonia was defined as the presence of gravity-dependent lung segmental infiltration on chest X-ray or computed tomography (CT), along with a characteristic clinical history or risk factors for aspiration or postoperative respiratory complications[15,16].

Outcome

Our primary endpoint was the incidence of GIE-associated respiratory complications under MAC. The secondary endpoints were the duration of respiratory complications, clinical outcomes, and the association between sedation depth (MS vs DS/GA), and the respiratory complications.

Statistical analysis

Histogram distribution and the Kolmogorov-Smirnov test were used to determine normally distributed variables. Normally distributed and skewed continuous variables were described by the mean ± SD, and the median and interquartile range, respectively. Categorical variables are presented as n (%). To compare continuous variables among groups, the independent sample Student’s t test or Mann-Whitney U test was employed based on the normality of the distribution. Categorical data were compared using either the χ² test or Fisher’s exact test.

Multivariate logistic regression was used to investigate the associations between different sedation levels and respiratory complications. To assess confounding, we added covariates to a logistic regression model in the basic model, then individually removed them from the complete model and compared the regression coefficients. Those covariates whose initial regression coefficients were altered by > 10% were included. We constructed three models using multivariable logistic regression models. We constructed three models: Model 1 was adjusted for ASA classification. Model 2 was additionally adjusted for model 1 and heart diseases, asthma, hiatus hernia, gastrointestinal ulcer, and GERD, whereas model 3 was additionally adjusted for model 2 and a history of upper gastrointestinal and oral-pharyngeal surgery, as well as regurgitation and aspiration. Additionally, we performed subgroup analyses by sex (men vs women), patient age (≤ 60 years vs > 60 years), ASA classification (≤ II vs > II), BMI (≤ 24 kg/m² vs > 24 kg/m²), type of procedure (EGD vs colonoscopy vs combined) and procedure time (< 10 minutes vs 10-30 minutes vs > 30 minutes), using multivariable logistic regression models.

In addition, several sensitivity analyses were conducted to evaluate the robustness of our results. First, to address potential bias due to loss to follow-up, we performed sensitivity analyses using multiple imputation based on five replications, thereby establishing the robustness of our findings. Second, matching weighting (MW) was applied to balance baseline confounders between groups. Sedation depth was used as the dependent variable, and all collected confounders were included as independent variables in the logistic regression model to estimate the propensity score. Marginal weights were computed, and extreme weights were trimmed. Covariate balance was assessed using standardized mean differences (SMDs), with an SMD < 0.1 considered acceptable.

All analyses were performed using R4.2.1 (http://www.R-project.org, The R Foundation) and Free Statistics version 2.1 software. Statistical significance was defined as a two-sided P value of < 0.05.

RESULTS

Baseline characteristics

A total of 10313 endoscopy procedures were performed between December 2024 and February 2025, with 8359 (81.1%) conducted under MAC. After excluding 836 patients, we included 7230 participants in the analysis (Figure 1). The general characteristics of participants by sedation depth are shown in Table 1. The two groups differed in terms of age, sex, BMI, ASA classification, smoking history, asthma status, chronic obstructive pulmonary disease status, hypertension status, heart disease status, cerebrovascular disease status, depression status, GERD status, hiatus hernia status, gastrointestinal ulcer status, and history of upper gastrointestinal and oral-pharyngeal surgery (all P values < 0.05). No significant differences were observed in the number of participants who drank alcohol, or who had diabetes mellitus, a history of colorectal or pulmonary surgery, duration of procedure, endoscopic type, or intraprocedural adverse events (all P values > 0.05).

Figure 1 Flow diagram showing patient eligibility. GA: General anesthesia; MS: Moderate sedation; DS: Deep sedation; MAC: Monitored anesthesia care.

Table 1 Baseline characteristics of the study population, mean ± SD/n (%).

Variables	Total (n = 7230)	MAC		P value
		MS (n = 1531)	DS/GA (n = 5699)
Age (years)	52.1 ± 12.7	53.1 ± 13.1	51.9 ± 12.6	< 0.001
Sex				0.002
Male	3579 (49.5)	812 (53.0)	2767 (48.6)
Female	3651 (50.5)	719 (47.0)	2932 (51.4)
BMI (kg/m2)	24.7 ± 3.4	25.4 ± 4.4	24.5 ± 3.1	< 0.001
ASA classification				< 0.001
I	3012 (41.7)	452 (29.5)	2560 (44.9)
II	3605 (49.9)	802 (52.4)	2803 (49.2)
≥ III	613 (8.5)	277 (18.1)	336 (5.9)
Smoker	271 (3.7)	76 (5.0)	195 (3.4)	0.005
Alcohol drinker	311 (4.3)	70 (4.6)	241 (4.2)	0.557
Underlying pulmonary disease
Asthma	82 (1.1)	43 (2.8)	39 (0.7)	< 0.001
COPD	93 (1.3)	33 (2.2)	60 (1.1)	< 0.001
Cardiovascular disease
Hypertension	1604 (22.2)	382 (25.0)	1222 (21.4)	0.003
Heart disease	602 (8.3)	258 (16.9)	344 (6.0)	< 0.001
Neurological diseases
Cerebrovascular	104 (1.4)	60 (3.9)	44 (0.8)	< 0.001
Depression	39 (0.5)	0 (0.0)	39 (0.7)	0.001
Diabetes mellitus	683 (9.4)	164 (10.7)	519 (9.1)	0.057
Gastrointestinal disorders
GERD	541 (7.5)	355 (23.2)	186 (3.3)	< 0.001
Hiatus hernia	84 (1.2)	46 (3.0)	38 (0.7)	< 0.001
Gastrointestinal ulcer	343 (4.7)	107 (7.0)	236 (4.1)	< 0.001
Surgical history
Upper gastrointestinal surgery	176 (2.4)	130 (8.5)	46 (0.8)	< 0.001
Colorectal surgery	324 (4.5)	80 (5.2)	244 (4.3)	0.113
Pulmonary surgery	153 (2.1)	37 (2.4)	116 (2.0)	0.357
Oral-pharyngeal surgery	40 (0.6)	32 (2.1)	8 (0.1)	< 0.001
Intraoperative respiratory adverse events
Severe cough	51 (0.7)	9 (0.6)	42 (0.7)	0.536
Regurgitation and aspiration	12 (0.2)	0 (0.0)	12 (0.2)	0.083
Hypoxemia	215 (3.0)	39 (2.5)	176 (3.1)	0.269
Procedure type				0.063
Combined	5465 (75.6)	1123 (73.4)	4342 (76.2)
EGD	1059 (14.6)	249 (16.3)	810 (14.2)
Colonoscopy	706 (9.8)	159 (10.4)	547 (9.6)
Procedure time				0.102
< 10 minutes	1425 (19.7)	326 (21.3)	1099 (19.3)
10-30 minutes	5159 (71.4)	1059 (69.2)	4100 (71.9)
> 30 minutes	646 (8.9)	146 (9.5)	500 (8.8)
Respiratory complications	77 (1.1)	13 (0.8)	64 (1.1)	0.354

ASA: American Society of Anesthesiologists; BMI: Body mass index; COPD: Chronic obstructive pulmonary disease; EGD: Esophagogastroduodenoscopy; GERD: Gastroesophageal reflux disease; MAC: Monitored anesthesia care; MS: Moderate sedation; DS: Deep sedation; GA: General anesthesia.

Respiratory complications

Notably, among the 7230 patients, 215 reported at least one post-endoscopic respiratory symptom, and 77 met the criteria for respiratory complications by presenting at least two concurrent symptoms (Table 2). The overall incidence of respiratory complications under MAC was 1.1%, with rates of 0.8% in the MS group and 1.1% in the DS/GA group (Table 1). Respiratory complications typically appeared 8.0 (4.0, 12.0) hours post-procedure, with a median duration of 1.5 (1.0, 2.5) days. Further medical intervention was required in 23 cases (30.9%). These patients presented with respiratory complications; detailed examinations and treatments are summarized in Supplementary Table 1. One patient (1/7230) who underwent combined upper and lower endoscopy under DS was diagnosed with aspiration pneumonia and hospitalized for one week. Five hours following the procedure, the patient presented with fever (39.5 °C), cough, productive sputum, and dyspnea, necessitating hospitalization. The chest CT image obtained on admission was consistent with left pneumonia (Figure 2A). Following a one-week course of antibiotic treatment during hospitalization, a subsequent CT scan revealed significant radiographic improvement (Figure 2B), and the patient was subsequently discharged. All affected patients achieved complete clinical recovery (Table 2).

Figure 2 Chest computed tomography findings before and after treatment. A: Chest computed tomography image obtained 5 hours post-operatively, showing left lung infiltration; B: Chest computed tomography image obtained after 7 days of treatment, showing marked resolution of the infiltration.

Table 2 Characteristics of gastrointestinal endoscopy-associated respiratory complications, median (interquartile range)/n (%).

Item	Respiratory complications
	Total (n = 77)	Cough + SOB (n = 55)	Cough + fever (n = 16)	Fever + SOB (n = 3)	Cough + fever + SOB (n = 3)
Post-procedure (hours)	8.0 (4.0, 12.0)	10.0 (4.0, 12.8)	5.8 (4.2, 9.8)	6.8 (5.2, 10.9)	4.5 (4.2, 5.0)
Duration (days)	1.5 (1.0, 2.5)	1.3 (0.8, 2.5)	1.5 (1.0, 2.0)	1.0 (0.9, 1.0)	4.0 (3.0, 6.5)
Aspiration pneumonia	1 (1.2)	0 (0.0)	0 (0.0)	0 (0.0)	1 (16.7)
Further medical treatment1	Yes: 23 (29.9)	Yes: 9 (16.4)	Yes: 10 (62.5)	Yes: 1 (33.3)	Yes: 3 (100)
	No: 54 (70.1)	No: 46 (83.6)	No: 6 (37.5)	No: 2 (66.7)	No: 0 (0.0)

¹Pharmacotherapy, hospitalization, or further diagnosis.

SOB: Shortness of breath.

Association between different sedation levels and respiratory complications during GIE

Table 3 displays the results of the multivariate logistic regression model analyses. We could not find any relationship between sedation depth and respiratory complications in the non-adjusted model or model 1. However, models 2 and 3 demonstrated a negative correlation between the DS/GA group and respiratory complications. After adjusting for ASA classification, heart disease, asthma, hiatus hernia, gastrointestinal ulcer, GERD, history of upper gastrointestinal and oral-pharyngeal surgery, and regurgitation and aspiration, the adjusted odds ratio (OR) for respiratory complications in the DS/GA group was 2.75 (95%CI: 1.32-5.71, P = 0.007) compared with that in the MS group.

Table 3 Associations between different sedation levels and respiratory complications during gastrointestinal endoscopy according to a multiple regression model.

Item	Respiratory complications, number/total number		OR (95%CI)	P value
	MS	DS/GA
Crude model	13/1531	64/5699	1.33 (0.73-2.41)	0.356
Model 11	13/1531	64/5699	1.57 (0.84-2.91)	0.155
Model 22	13/1531	64/5699	2.29 (1.16-4.54)	0.017
Model 33	13/1531	64/5699	2.75 (1.32-5.71)	0.007

¹Adjusted for American Society of Anesthesiologists classification.

²Additionally, adjusted for heart disease, asthma, hiatus hernia, gastrointestinal ulcer, and gastroesophageal reflux disease.

³Further adjusted for a history of upper gastrointestinal and oral-pharyngeal surgery, as well as regurgitation and aspiration.

OR: Odds ratio; GA: General anesthesia; MS: Moderate sedation; DS: Deep sedation.

Stratified analyses

Stratified analysis was performed across subgroups to assess potential differences in the relationship between sedation depth and respiratory complications. No significant interactions were detected in any subgroup, including those stratified by age, sex, ASA classification, BMI, procedure type, or duration (Figure 3).

Figure 3 Relationships between sedation and respiratory complications according to general characteristics. In addition to the stratification component itself, each stratification factor was adjusted for all other variables (American Society of Anesthesiologists classification, heart disease, asthma, hiatus hernia, gastrointestinal ulcer, gastroesophageal reflux disease, history of upper gastrointestinal and oral-pharyngeal surgery, and regurgitation and aspiration). ASA: American Society of Anesthesiologists; EGD: Esophagogastroduodenoscopy; GA: General anesthesia; BMI: Body mass index; MS: Moderate sedation; DS: Deep sedation; OR: Odds ratio.

Sensitivity analyses

The results of the sensitivity analyses are summarized in Table 4. We performed multiple imputations to address missing data from 293 participants who were lost to follow-up. Compared with individuals who received MS, those who received with DS/GA had an adjusted OR of 2.86 (95%CI: 1.37-5.98, P = 0.005) for respiratory complications. After MW was applied, the SMDs for all the covariates were less than 0.1, indicating good balance. The SMDs and baseline characteristics before and after weighting are presented in Supplementary Tables 2 and 3. Following MW, the estimated OR was 3.02 (95%CI: 1.49-6.13; P = 0.002).

Table 4 Sensitivity analysis.

Item	Respiratory complications, number/total number		OR (95%CI)	P value
	MS	DS/GA
Multiple imputation for loss to follow-up data
Crude model	13/1585	67/5938	1.32 (0.72-2.40)	0.367
Model 11	13/1585	67/5938	1.54 (0.83-2.86)	0.168
Model 22	13/1585	67/5938	2.25 (1.14-4.46)	0.020
Model 33	13/1585	67/5938	2.86 (1.37-5.98)	0.005
Sensitivity analysis based on MW
Crude	13/1531	64/5699	1.33 (0.73-2.41)	0.356
After MW	-	-	3.02 (1.49-6.13)	0.002

¹Adjusted for American Society of Anesthesiologists classification.

²Additionally, adjusted for heart disease, asthma, hiatus hernia, gastrointestinal ulcer, and gastroesophageal reflux disease.

³Further adjusted for a history of upper gastrointestinal and oral-pharyngeal surgery, as well as regurgitation and aspiration.

-: It indicates that no weighted patient counts are presented, as they do not represent actual clinical numbers; OR: Odds ratio; GA: General anesthesia; MS: Moderate sedation; DS: Deep sedation; MW: Matching weighting.

DISCUSSION

Although GIE-associated adverse events have been extensively studied[1-4], postprocedural delayed respiratory complications remain ambiguous[7,17]. Many patients develop respiratory symptoms hours after the procedure, which are frequently misattributed to other causes. Consequently, these complications might be underestimated in retrospective studies[8]. Sedation is considered an independent risk factor for aspiration pneumonia but not for bowel perforation or splenic injury[18].

In contrast to MS with endoscopist-delivered benzodiazepine-opioid combinations, anesthesiologist-delivered MAC comprises propofol-based DS or GA. Comparative studies have consistently demonstrated higher complication rates with MAC than with endoscopist-directed sedation[2-4,7,9], suggesting a dose-response relationship between sedation depth and adverse outcomes. A prospective multicenter study[6] reported that 0.83% of participants complained of respiratory complications 24 hours post-endoscopy, whereas Samer Ammar et al[19] followed up 393 children who underwent outpatient upper GIE under GA and reported that the incidence of respiratory symptoms was 7.4% after the examinations.

The diagnosis of aspiration pneumonia following GIE presents significant challenges and requires both radiographic evidence and inflammatory biomarkers for confirmation[15,16,20]. However, most outpatients are discharged before the onset of symptoms; respiratory symptoms such as cough, fever, or SOB typically develop after they leave the endoscopy unit. Our definition of respiratory complications (≥ 2 symptoms at > 24 hours) replicated Friedrich’s criteria[6]. Despite nearly 80% of procedures using DS, the incidence of respiratory complications (1.1%) aligned with Friedrich et al’s MS data (0.8%)[6] but was lower than that of Samer Ammar et al[19]. During extended follow-up of patients with respiratory complications, we observed that respiratory complications typically appeared 8.0 (4.0, 12.0) hours after the procedure, with a median duration of 1.5 (1.0, 2.5) days; however, 56 (69.1%) patients recovered spontaneously without any serious sequelae. These favorable outcomes might be attributed to the implementation of a risk-adapted sedation protocol, in which MS was preferentially administered to high-aspiration-risk patients (e.g., those with GERD or hiatal hernia). Although Friedrich et al[6] reported that intraprocedural coughing or vomiting was a strong predictor of subsequent complications, with a 100% association; this correlation was absent in our cohort. Thus, we hypothesize that DS under MAC suppresses protective cough and vomiting reflexes.

Aspiration pneumonia is the most severe respiratory complication following GIE and is associated with increased in-hospital mortality (9% vs 0.8%; P < 0.001) and longer hospitalization (10.54 days vs 4.85 days; P < 0.001)[8]. In our cohort, only one case (0.02%) was radiologically confirmed by CT of the chest, which was lower than that in previous research. These findings are similar to the results of Bielawska et al[18], who retrospectively analyzed data from 3059045 outpatient colonoscopies and reported that 186 (0.06‰) patients with aspiration pneumonia needed hospitalization. Conversely, Yoo et al[21] reported a 30-day risk of aspiration pneumonia of 1.1% following EGD in a large cohort of approximately 5 million outpatients. A nationwide retrospective analysis[8] of hospitalized United States patients revealed a 1.9% incidence of EGD-associated aspiration pneumonia during 2016-2020, with a statistically significant annual increase from 1.6% to 2.1%. Park et al[22] reported a clinically significant 1.5% incidence of aspiration pneumonia following ESD performed under anesthesiologist-administered sedation. Although advanced age and upper endoscopy have been identified as independent risk factors for aspiration pneumonia[8,22], this association was not observed in our cohort. We attribute this discrepancy to enhanced anesthetic protocols following a critical incident analysis of severe aspiration cases at our institution.

Although the unadjusted incidence rates of respiratory complications were comparable between the two groups, multivariable adjusted models revealed a significant association between DS/GA and increased respiratory complications (adjusted OR = 2.75, 95%CI: 1.32-5.71; P = 0.007). This association remained robust in the subgroup and sensitivity analyses, consistent with prior reports[9,18,23]. It is presumed that deeper sedation leads to diminished airway protective reflexes and an increased risk of aspiration pneumonia[3]. Compared with cardiopulmonary complications during examination, postoperative respiratory complications are difficult to overcome. Therefore, we suggest that MS should be preferred in patients with a high risk of reflux aspiration, such as those with obesity, who are treated with GRED, and who are treated with post-gastrointestinal surgery. Like the critical significance of BMI, obstructive sleep apnea, and the Mallampati classification in predicting procedure-related respiratory complications, these reflux aspiration-associated risk factors should likewise be prioritized during pre-sedation evaluation and sedation strategy formulation. This essential consideration has not been explicitly addressed in previous sedation guidelines[24-27].

However, it is important to acknowledge our study’s limitations. First, while rigorous eligibility criteria and multivariable adjustment were applied to address confounders, the observational design inherently limits causal inference due to potential selection bias and residual confounding. This limitation is compounded by diagnostic challenges in attributing outcomes specifically to DS/GA, rather than pre-existing conditions (e.g., cardiopulmonary disease) or intercurrent infections, such as upper respiratory infections. Second, the absence of routine pulmonary imaging or laboratory tests may fail to detect asymptomatic cases, thereby potentially underestimating the risk of aspiration pneumonia. Third, we did not evaluate the effects of specific sedative agents, nor did we account for variables such as the endoscopist’s annual procedure volume or technical proficiency, which should be focused on in subsequent randomized controlled trials. The strength, consistency, and dose-response gradient of the observed associations, coupled with a biologically plausible mechanism, remain persuasive despite the noted limitations, positioning deeper sedation as a key risk marker for respiratory complications. As such, our findings strengthen the rationale for tiered sedation management and serve as a critical reference for clinical practice in high-risk groups. To advance this evidence, further research employing randomized designs with protocolized monitoring is needed to substantiate the causal relationship.

CONCLUSION

Approximately 1% of patients experience significant respiratory symptoms during the post-procedure period and should be advised to seek medical care if these symptoms develop. Although DS/GA is associated with an increased risk of respiratory complications following GIE, MAC with sedation strategies tailored to patients’ risk factors may not significantly increase the incidence of such complications. Hence, we recommend MS as the preferred strategy for high-risk patients undergoing GIE to significantly reduce the incidence of respiratory complications.