A Proof of Concept Study: Esophagogastroduodenoscopy Is an Aerosol-Generating Procedure and Continuous Oral Suction During the Procedure Reduces the Amount of Aerosol Generated.

Aerosol-generating procedures pose a potential threat to health care workers, especially during this COVID-19 (coronavirus disease 2019) pandemic. Esophagogastroduodenoscopy (EGD) was assumed to be an aerosol-generating procedure and recommendations, therefore, reflect evidence generated from nongastrointestinal aerosol-generating procedures, such as bronchoscopy. However, there is no scientific evidence to support this claim. This study aims to provide scientific evidence on whether EGD is an aerosol-generating procedure and to examine ways of decreasing the amount of aerosol generated.

Methods

This study was a prospective observational trial to examine aerosol generation during EGD by applying a quantitative approach (see Supplementary Material for details).

All patients undergoing EGD at the endoscopy center of the Prince of Wales Hospital from May 7, 2020 to June 1, 2020 were included. Procedures were performed with the patient in the left lateral position with a mouthguard, using a 9.9-mm flexible video gastrointestinal scope (GIF-H290; Olympus Hong Kong and China Limited, Kowloon, Hong Kong SAR). Measurements were taken using the portable GT-526S Handheld Particle Counter (Met One Instruments, Inc, Grants Pass, OR). The 6-channel particle sizes were programmed at 0.3 μm, 0.5 μm, 0.7 μm, 1 μm, 5 μm, and 10 μm. The particle counter was placed within 10 cm of the mouth of the patient once the patient entered the room and measured for at least 1 minute before the start of the procedure. The measurement was continued during the procedure until after the patient left the endoscopy suite.

Statistical Analysis

Multilevel modeling was used to test all of the hypotheses in the study. The particle counts per cubic feet of each particle size are presented in dCF. We tested the associations among the use of sedation, dental sucker, and Log(dCF). The interaction terms between sedation and procedure and between dental sucker and procedure were added to the multilevel models. Age, sex, endoscopist seniority, procedure length, diagnostic or therapeutic procedure, and use of biopsy were included as confounders. A random effect of repeated measures was accounted for in the models to capture the intra-subject variability for the changes in dCF over time.

Results

From May 7, 2020 to June 1, 2020, a total of 93 patients were recruited into the study. There were 59 unsedated patients (63.4%) and 34 sedated patients (36.6%). A dental sucker was used for 30 patients (32.3%). Most of the procedures were diagnostic procedures (n = 85 [91.4%]) with a biopsy taken (n = 68 [73.1%]). A multivariate analysis was also performed and the results are summarized in Table 1 .

Table 1

The Association Between Sedation and Log(dCF_0.3, dCF_0.5, dCF_0.7, dCF_1, dCF_5, dCF_10)

Variable	dCF_0.3		dCF_0.5		dCF_0.7		dCF_1		dCF_5		dCF_10
	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI	Estimate (SE)	95% CI
Procedurea	0.18 (0.03)b	0.12 to 0.24	0.30 (0.05)b	0.20 to 0.40	0.28 (0.07)b	0.14 to 0.42	0.22 (0.09)c	0.04 to 0.40	2.49 (0.29)b	1.92 to 3.06	1.53 (0.25)ba	1.04 to 2.02
Sedationd	–0.08 (0.17)	–0.39 to 0.23	–0.10 (0.17)	–0.43 to 0.23	–0.18 (0.12)	–0.42 to 0.06	–0.16 (0.12)	–0.40 to 0.08	–0.02 (0.30)	–0.61 to 0.57	–0.05 (0.26)	–.56 to .46
Dental sucker	0.64 (0.15)b	0.35 to 0.93	0.59 (0.14)b	0.32 to 0.86	0.19 (0.10)	–0.01 to 0.39	–0.07 (0.10)	–0.27 to 0.13	0.91 (0.29)e	0.34 to 1.48	0.71 (0.25)c	.21 to 1.21
Procedure × sedationf	–0.03 (0.04)	–0.11 to 0.04	–0.06 (0.07)	–0.20 to 0.08	–0.05 (0.10)	–0.25 to 0.15	–0.12 (0.12)	–0.26 to 0.12	–0.73 (0.39)	–1.49 to 0.03	–0.42 (0.34)	–1.09 to 0.25
Procedure × dental suckerg	–0.24 (0.04)b	–0.32 to –0.16	–0.36 (0.07)b	–0.50 to –0.22	–0.36 (0.10)b	–0.56 to –0.16	–0.32 (0.13)c	–0.57 to –0.07	–1.14 (0.40)e	–1.92 to –0.36	–0.71 (0.35)c	–1.40 to –0.02

NOTE. Age, sex, endoscopist seniority, procedure length, diagnostic or therapeutic, and whether take biopsy are potential confounders that controlled in the mode.

CI, confidence interval; dCF, differential count.

Coefficient of procedure indicates the expected change in ln(dCF) during procedure.

P < .001.

P < .05.

Coefficient of sedation indicates the expected change in ln(dCF) for sedation without adjustment of procedural time.

P < .01.

Coefficient of procedure × sedation indicate the expected change in ln(dCF) in the sedated group during procedure.

Coefficient of procedure × dental sucker indicate the expected change in ln(dCF) in the dental sucker group during the procedure.

During the EGD procedure, the level of dCF of all sizes was significantly higher than during the baseline period (P < .001 to .02). Use of the dental sucker significantly reduced the number of particles sized 0.3 μm, 0.5 μm, 0.7 μm, 1 μm, 5 μm, and 10 μm expelled during the procedure compared with baseline (P < .001, P < .001, P < .001, P = .02, P < .01, and P = 0.046, respectively). Simple slope tests revealed that when compared with baseline, the number of dCF_0.3, dCF_0.5, dCF _0.7, and dCF_1 during EGD was significantly increased among procedures performed without a dental sucker (P < .01, P < .01, P < .01, and P = .02 respectively), and decreased nonsignificantly among participants when the dental sucker was used. For larger particle sizes of 5 μm and 10 μm, use of the dental sucker was associated with a significantly smaller magnitude of increase in number of particles during the procedure (P < .01 and P < .01, respectively) (Supplementary Figure 1). There were no significant differences in the level of all dCF outcomes between the individuals with or without sedation (P = .13 to .96). Sedation did not affect the association between procedure time and all dCF outcomes. These results are summarized in Table 1.

Supplementary Figure 1

(A) Line graphs of particle counts sizes 0.3, 0.5, 0.7, 1, 5, and 10 μm of an unsedated diagnostic procedure without the use of dental sucker. (B) Line graphs of particle counts sizes 0.3, 0.5, 0.7, 1, 5, and 10 μm of an unsedated diagnostic procedure with the use of dental sucker. (C) The change in LN_dCF during the procedure compared with baseline with or without sucker. The use of dental sucker reduced the association between timing of procedure (before vs during) and dCF_0.3 (estimate = −0.24 [SE = 0.04]; 95% confidence interval [CI], −0.32 to −0.16; P < .001), dCF_0.5 (estimate = −0.36 [SE = 0.07]; 95% CI, −.50 to −.22; P < .001), dCF_0.7 (estimate = −.36 [SE = .10]; 95% CI, −0.56 to −0.16; P < .001), dCF_1 (estimate = −0.32 [SE = 0.13]; 95% CI, −0.57 to −0.07; P = .02), dCF_5 (estimate = −1.14 [SE = .40]; 95% CI, −1.92 to −0.36; P < .01), and dCF_10 (estimate = −0.71 [SE = 0.35]; 95% CI, −1.40 to −.02; P = .046). In other words, the use of dental sucker significantly reduced the amount of particles of all sizes expelled during the procedure compared with baseline. Simple slope tests revealed that when compared with baseline, particles of all sizes during EGD were significantly increased in procedures performed without dental sucker (dCF_0.3 estimate = 0.18, t = 6.30; P < .01; dCF_0.5 estimate = .30, t = 5.54; P < .01; dCF_0.7 estimate = .28, t = 3.89; P < .01; dCF_1 estimate = .22, t = 2.43; P = .02; dCF_5 estimate = 2.49, t = 8.54; P < .01; dCF_10 estimate = 1.53, t = 6.03; P < .01). The number of dCF_0.3, dCF_0.5, dCF_0.7, and dCF_1 during EGD were nonsignificantly decreased among participants when dental sucker was used (dCF_0.3 estimate = −0.06, t = −1.22; P = .23; dCF_0.5 estimate = −0.07, t = −2.35; P = .02; dCF_0.7 estimate = −0.08, t = −1.06; P = .29; dCF_1 estimate = −0.09, t = −0.84; P =.40). For particles dCF_5 and dCF_10, the magnitude of the increase was significantly smaller with the use of dental sucker (dCF_5 estimate = 1.35, t = 3.69; P < .01; dCF_10 estimate = .82, t = 2.78; P < .01 ). Confounders were controlled at the multilevel modeling.

One per-oral endoscopic myotomy was recorded. The procedure was performed under general anesthesia. The data showed that there was a surge in all particle sizes during the initial endoscope intubation and diagnostic EGD (Supplementary Figure 2).

Supplementary Figure 2

The line graph showing the change in dCF during a per-oral endoscopic myotomy procedure performed under general anesthesia with CO2 insufflation. There was a surge seen in all particle sizes during the initial intubation of the endoscope and diagnostic EGD. There was also a significant increase in all particles generated once the energy cutting devices were used. These findings suggest that general anesthesia might not have a protective effect on the amount of aerosol generated and the use of energy cutting devices generates a significant amount of aerosols.

Discussion

Currently, the term droplet is often used to refer to droplets >5 μm in diameter and the term aerosol refers to particles ≤5 μm in diameter that can remain suspended in air for a significant time, allowing them to be transmitted over longer distances of >1 m. The increase of particle size at 0.3 μm, 0.5 μm, 0.7 μm, and 1 μm supports EGD as an aerosol-generating procedure.

Another important finding of this study was that the use of dental sucker in the oral cavity for continuous suction during the procedure significantly decreased the particle counts of all sizes detected during EGD, suggesting the use of the dental sucker be recommended during EGD. The usual practice in our center was to provide intermittent oral suction via suction catheter. The postulation was continuous oral cavity suction reduced pooling of saliva and can therefore reduce aspiration and patients’ coughing. Another reason might be that the dental sucker also partially suctioned out the particles that were generated during the procedure. The smaller particles might be subject to the negative pressure effect generated by the continuous suction of the dental sucker, causing a paradoxical drop in the particle counts during the procedure.

In this study, we found that conscious sedation was not able to reduce the amount of aerosol and droplets generated during the procedure. This may be due to small sample size and the presence of confounding factors. We also found that there was still a surge of all particle sizes during general anesthesia, which suggests that general anesthesia might not eliminate the aerosol generated.

However, the increase in aerosol does not equate to an increase in the infectivity of respiratory viruses during EGD. Despite numerous studies on the transmission routes of respiratory viruses, the results remained inconclusive. The SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) RNA, however, has been found in fecal samples of infected patients for a prolonged period of time. The role of the gastrointestinal tract in the transmission of SARS-CoV-2 remains uncertain.

EGD is an aerosol-generating procedure and the use of a dental sucker can decrease the amount of aerosol generated, which decreases the risk to health care workers. These findings can inform current guidelines in infection control and the most appropriate personal protective equipment during digestive endoscopy.