Abstract
Esophagogastroduodenoscopy (EGD) serves as a standard screening modality for upper gastrointestinal (GI) tumors, though standardized quality metrics remain undefined. Crucially, there was little direct evidence previously demonstrating the impact of the number of images taken during EGD on the detection rates of early upper gastrointestinal cancer (UGIC). This multicenter retrospective study aimed to assess the influence of the number of photographic images on early UGIC detection rates and identify associated risk factors. We retrospectively analyzed the data of 245,161 subjects who underwent EGD either as part of a health screening program or as symptomatic patients at Zhejiang Provincial Hospital of Traditional Chinese Medicine and Zhejiang Provincial People’s Hospital between May 2018 and October 2024. Cases were stratified by image count: < 20 (low), 20–40 (medium), and > 40 (high). Early UGIC was histologically confirmed. Primary outcomes were the comparison of detection rates across groups; secondary outcomes involved identifying detection-associated factors via logistic regression. Early UGIC detection rates were significantly higher in the high-image group versus the low and medium groups (p < 0.05). Subanalysis revealed optimal detection with > 80 images. Endoscopist-specific detection rates correlated strongly with image count (p < 0.001). Multivariate analysis identified increased image acquisition (OR 1.32, 95% CI 1.18–1.47), advanced age (OR 2.14, 95% CI 1.89–2.42), and male sex (OR 1.48, 95% CI 1.32–1.65) as independent predictors of early UGIC detection. The number of photographic images taken during EGD was significantly correlated with the detection of early UGIC. This parameter can be considered as an effective indicator for assessing the quality of EGD.
Subject terms: Cancer, Diseases, Gastroenterology, Oncology
Introduction
Upper gastrointestinal cancers (UGIC), including gastric cancer (GC) and esophageal cancer (EC), are prevalent malignancies worldwide, causing more than 1.3 million deaths annually1. Despite advancements in the treatment of UGIC in recent years, the five-year survival rate for patients with local or distant metastases remains unfavorably low2. The disease burden of UGIC in China is also significantly high. Gastric cancer ranks as the second most commonly diagnosed cancer and the second leading cause of cancer-related deaths in China3. The mortality and five-year prevalence rates of GC and EC in China are higher than those in most developed countries, with a particularly alarming characteristic being the continued rise in incidence among younger individuals3.
The 5-year survival rate for patients with early-stage UGIC exceeds 85%4, as the majority of early lesions can be radically treated through endoscopic mucosal resection (EMR) or endoscopic submucosal dissection (ESD)5,6. In the majority of countries , esophagogastroduodenoscopy (EGD) stands as the preeminent diagnostic tool for inspecting the upper gastrointestinal tract7, playing a pivotal role in the prevention of upper gastrointestinal cancers (UGIC). High-quality EGD ensures precise examination of the UGIC mucosa. Prior research has revealed that conventional EGD misses 9% to 20% of UGIC cases8–10. Consequently, enhancing the quality of EGD is crucial for reducing the incidence of UGIC. There are numerous factors that can influence the quality of EGD examinations, including the use of sedative medications, preparation prior to endoscopy, duration of the examination, endoscopic biopsy, computer-aided diagnosis, and artificial intelligence systems11–13.
However, despite the existence of a standard procedure for capturing images during EGD, there is a lack of a unified standard globally regarding the number of images taken during these procedures. In Japan, during a standard upper gastrointestinal endoscopy, it is customary to take between 20 and 40 endoscopic images. Yao14, proposed a minimum requirement standard known as the “Systematic Screening Protocol for the Stomach (SSS)”, recommending that at least 22 images should be taken to complete the examination of the stomach. Furthermore, some Asian research centers7 suggest capturing at least 40 digital images to enhance the completeness of the examination.
Nevertheless, currently in the majority of research centers worldwide, only a relatively small number of endoscopic images are taken and stored per examination. Previously, the European Society of Gastrointestinal Endoscopy (ESGE) guidelines15 recommended taking 8 images during EGD examinations, including 4 images of the stomach. It is evident that taking just 4 images is insufficient to obtain detailed information about the entire stomach. Until 2016, the ESGE guidelines16 advocated for a minimum inclusion of 10 anatomical sites in image taking, thereby elevating the recommended number of images to be taken from 8 to 10. In the United States, where the incidence of GC is relatively low, the guidelines do not stipulate a minimum number of images to be taken during an EGD examination17.
Previously, there had been a lack of large-scale clinical studies demonstrating a definitive correlation between the number of images taken during EGD and the detection rate of early UGIC. Therefore, the purpose of this study is to assess whether the number of images taken influences the detection rate of early UGIC in diagnostic EGDs, offering data support for the ongoing improvement of EGD examination standards.
Methods
Study design
This retrospective observational study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committees of all participating institutions, including Zhejiang Provincial People’s Hospital (No. QT2025072) and Zhejiang Provincial Hospital of Traditional Chinese Medicine (No. 2025-KLS-043-01). As the study utilized only de-identified electronic medical records without patient contact, written informed consent was waived. All data were anonymized and handled under strict confidentiality protocols.
Study population
Data were collected from the electronic medical record systems of Zhejiang Provincial Hospital of Chinese Medicine and Zhejiang Provincial People’s Hospital between May 2018 and October 2024. From these databases, we extracted comprehensive demographic information, endoscopic imaging data, and histopathological biopsy results of the study population.
The inclusion criteria were: (i) age older than 18 years; (ii) complete data; and (iii) undergoing pathological biopsy during EGD. Subjects were excluded based on the following criteria: (i) history of upper gastrointestinal surgery or digestive tract cancer; (ii) undergoing therapeutic EGD rather than diagnostic EGD. For patients who underwent multiple EGDs, only the first data set was extracted.
Data collection
The number of images taken was defined as the total count of images taken, starting from the first image taken at the pharynx up to the images of the descending portion of the duodenum. All endoscopic examination images included in this study strictly cover all anatomical regions of the upper gastrointestinal tract. After undergoing inspection and verification, these images were imported into the endoscopic system and automatically entered into the database. In addition to the aforementioned data, the following information was also collected: patient demographics (gender, age, height, weight, BMI, past medical history, etc.), endoscopy-related details (anesthesia status, start time of the examination, biopsy pathology results, etc.), and endoscopist-related information (gender, age, professional title, educational background, years of experience, etc.). Notably, all endoscopists involved in this study received specialized training on the SSS protocol and strictly followed the protocol requirements for image taken during procedures.
Early EC was defined as cancer confined to the mucosal layer, regardless of the presence of lymph node or distant organ metastasis18. For early gastric and duodenal cancers, they were defined as cancers located within the mucosa or submucosa19. All pathological results stemmed from resected specimens or biopsy samples, and were analyzed and diagnosed by two experienced pathologists.
Outcome and measurement
The EGD examinations in the database were categorized into three groups based on the number of images taken: the low-image group (≤ 20 images), the medium-image group (20–40 images), and the high-image group (≥ 40 images). The primary outcome of the study was to compare the detection rates of early UGIC among these three groups.
The secondary outcomes involved comparing the detection rate of early UGIC between the high-image group and the low-image group after propensity score matching (PSM), as well as identifying potential factors that may have influenced the detection rate of early UGIC through logistic regression analysis.
Statistical methods
Statistical analyses were performed using SPSS 27.0, with categorical variables expressed as frequencies (percentages) and continuous variables expressed as medians (interquartile ranges). Nonparametric comparisons utilized the Wilcoxon-Mann–Whitney U test for two-group analyses and the Kruskal–Wallis test with SNK post hoc testing for multiple groups, while categorical data were assessed using chi-square tests. Normality assumptions were verified for the application of parametric ANOVA. Correlation analyses employed Spearman’s rank correlation coefficient. To address potential confounding, propensity score matching was implemented using nearest-neighbor matching with a caliper width of 0.00002 logit units at a 1:1 ratio. Statistical significance was established at p < 0.05 for all analyses.
Results
Baseline characteristics
Between May 2018 and October 2024, a total of 429,973 EGDs were performed by 33 endoscopists. After applying the inclusion and exclusion criteria, 184,812 cases were excluded, leaving a final cohort of 245,161 cases for inclusion, and the research process was illustrated in Fig. 1. Among them, there were 37,669 cases (15.31%) in the low-image group, 92,413 cases (37.56%) in the medium-image group, and 115,079 cases (47.13%) in the high-image group. The median age was 53 years (range: 41–63 years), and 122,917 patients (49.9%) were female. A total of 186,955 patients (76.3%) underwent EGD under sedation. The characteristics of subjects in each group are shown in Table 1.
Fig. 1.
Flowchart for enrollment of patients in this study. 1 Zhejiang Provincial People’s Hospital; 2 The First Affiliated Hospital of Zhejiang Chinese Medical University.
Table 1.
Basic characteristics of each group.
| Total (n = 245,161) | Low-image group (n = 37,669) | Medium-image group (n = 92,413) | High-image group (n = 115,079) | P value | |
|---|---|---|---|---|---|
| Sex | |||||
| Male | 123,144 (50.1%) | 18,070 (48.0%) | 44,032 (47.6%) | 61,042 (52.6%) | < 0.001 |
| Female | 122,917 (49.9%) | 19,599 (52.0%) | 48,381 (52.4%) | 54,937 (47.4%) | |
| Height, median (25–75%) | 163 (158–170) | 164 (158–170) | 163 (157–169) | 164 (158–170) | 0.139 |
| Weight, median (25–75%) | 63.7(55.0–72.5) | 64.0 (56.0–72.5) | 63.8 (56.0–72.0) | 64.0 (56.0–72.0) | 0.2 |
| Age, median (25–75%) | 53 (41–63) | 53 (40–63) | 53 (39–62) | 54 (42–64) | < 0.001 |
| Anaesthesia | |||||
| Consciousness | 58,206 (23.7%) | 12,355 (32.8%) | 25,598 (27.7%) | 20,253 (17.6%) | < 0.001 |
| Sedation | 186,955 (76.3%) | 25,314 (67.2%) | 66,815 (72.3%) | 94,826 (82.4%) | |
| Endoscopist | |||||
| Physician | 20,088 (8.2%) | 678 (1.8%) | 8132 (8.8%) | 11,278 (9.8%) | < 0.001 |
| Attending physician | 48,461 (19.7%) | 10,246 (27.2%) | 20,608 (22.3%) | 17,607 (15.3%) | |
| Associate chief physicia | 71,461 (29.3%) | 2975 (7.9%) | 28,093 (30.4%) | 40,393 (35.1%) | |
| Chief physician | 105,151 (42.8%) | 23,770 (63.1%) | 35,580 (38.5%) | 45,801 (39.6%) | |
| Lesion area | |||||
| Esophagus | 228 (13.5%) | 24 (28.6%) | 84 (21.9%) | 120 (9.8%) | < 0.001 |
| Gastrointestinal | 1380 (81.6%) | 48 (57.1%) | 276 (71.9%) | 1056 (86.3%) | |
| Duodenum | 84 (4.9%) | 12 (14.3%) | 24 (6.2%) | 48 (3.9%) | |
| Detection rate | 1692 (0.69%) | 84 (0.22%) | 384 (0.42%) | 1224 (1.06%) | < 0.001 |
Univariate and multivariate logistic regression analyses
A logistic regression analysis was conducted to identify factors that may affect the detection of early UGIC (Table 2). In univariate analysis, patient age (95% CI 1.067–1.082; p < 0.001), male gender (95% CI 0.328–0.474; p < 0.001), number of images taken (95% CI 1.020–1.024; p < 0.001), medium and high images taken groups (95% CI 0.349–0.530; p < 0.001), endoscopists with the title of associate senior physicians (95% CI 0.593–0.965; p < 0.001), sedation status (95% CI 1.000–1.498; p = 0.005), and the start time of the examinations (95% CI 0.556–0.834; p < 0.001) were significantly associated with the detection of early UGIC.
Table 2.
Univariable and multivariate logistic regression analysis.
| Variables | Positive group (n = 1692) | Negative group (n = 243,469) | Univariate analysis OR (95% CI, P value) | Multivariable analysis OR (95% CI, P value) | |
|---|---|---|---|---|---|
| Patient Gender | Female | 1211 (71.6%) | 121,932 (49.9%) | 0.394 (0.328–0.474, P < 0.001) | 0.492 (0.407–0.594, P < 0.001) |
| Male | 481 (28.4%) | 121,537 (50.1%) | |||
| Patient age | 65(59–73) | 53(41–63) | 1.074 (1.067–1.082, P < 0.001) | 1.071 (1.063–1.079, P < 0.001) | |
| Height | 163(158–170) | 164(158–170) | 0.999 (0.995–1.002, P = 0.440) | ||
| Weight | 64.8(56–72.9) | 64(56–72) | 0.401 (0.998–1.003, P = 0.401) | ||
| Number of pictures taken | 55(37–77) | 39(25–55) | 1.022 (1.020–1.024, P < 0.001) | 1.016 (1.013–1.019, P < 0.001) | |
| Pictures taken group | Low-image group | 84 (5.0%) | 37,494 (15.4%) | ||
| Medium-image group | 385 (22.7%) | 87,645 (35.9%) | 0.218 (0.148–0.319, P < 0.001) | 0.574 (0.374–0.882, P = 0.011) | |
| More pictures group | 1223 (72.3%) | 118,330 (48.7%) | 0.426 (0.349–0.520, P < 0.001) | 0.861 (0.677–1.095, P = 0.223) | |
| Endoscopist title | Physician | 108 (6.4%) | 20,208 (8.3%) | ||
| Attending physician | 264 (15.6%) | 48,216 (19.8%) | 0.742 (0.523–1.054, P = 0.095) | 1.199 (0.835–1.723, P = 0.326) | |
| Associate chief physicia | 564 (33.3%) | 71,448 (29.3%) | 0.756 (0.593–0.965, P = 0.025) | 1.222 (0.948–1.575,P = 0.122) | |
| Chief physician | 756 (44.7%) | 103,597 (42.6%) | 1.091 (0.902–1.318, P = 0.370) | 1.194 (0.979–1.457, P = 0.079) | |
| Anaesthesia | Sedation | 1350 (79.8%) | 186,306 (76.5%) | ||
| Consciousness | 342 (20.2%) | 57,163 (23.5%) | 1.219 (1.000–1.498, P = 0.050) | 1.127 (0.904–1.405, P = 0.288) | |
| EGD start time | 8:00–11:00 | 816 (48.2%) | 138,456 (56.9%) | ||
| 11:00–14:00 | 444 (26.2%) | 56,021 (23.0%) | 0.681 (0.556–0.834, P < 0.001) | 0.749 (0.606–0.926, P = 0.007) | |
| After 14:00 | 432 (25.5%) | 48,992 (20.1%) | 0.915 (0.727–1.152, P = 0.915) | 1.001 (0.791–1.265, P = 0.996) | |
After conducting a further multivariate analysis, we found that a higher number of images taken (95% CI 1.275–2.520; p = 0.001), an older patient age (95% CI 1.013–1.019; p < 0.001), and male gender (95% CI 0.407–0.594; p < 0.001) were significant factors influencing the detection rate of early UGIC.
Detection rate of early UGIC among different groups
Among the 245,161 EGDs conducted, we identified a total of 1692 patients with early UGIC, accounting for 0.69% of the total. Specifically, there were 228 cases of early esophageal cancer (0.09%), 1380 cases of early gastric cancer (0.56%), and 84 cases of early duodenal cancer (0.03%). Among the patients with early gastric cancer, 601 cases had lesions in the gastric antrum (45.15%), 317 in the gastric body (23.87%), 222 in the gastric angle (16.67%), 60 in the gastric fundus (4.50%), and 126 in the cardia (9.50%);8 patients had lesions affecting both the gastric antrum and gastric body, 5 patients had lesions in both the gastric antrum and gastric angle, 5 patients had lesions in both the esophagus and gastric fundus, and 1 patient had lesions involving the esophagus, cardia, and gastric fundus.
As shown in Table 1, 84 early-stage UGIC were detected in the low-image group, with a detection rate of 0.22%; 384 early-stage UGICs were detected in the medium-image group, with a detection rate of 0.42%; and 1224 early-stage UGICs were detected in the high-image group, with a detection rate of 1.06%. Statistically significant differences were observed among the detection rates of the three groups (p < 0.05).
Anatomical site of early UGIC
Stratified analysis by anatomical site revealed distinct patterns in early UGIC detection rates (Table 3): while esophageal (0.06% vs. 0.09% vs. 0.10%, p > 0.05) and duodenal (0.03% vs. 0.03% vs 0.04%, p > 0.05) cancer detection showed no significant differences across the low-, medium-, and high-image groups respectively, GC detection demonstrated a marked stepwise increase (0.13% vs. 0.30% vs. 0.92%, p < 0.05), highlighting the particular value of comprehensive image documentation for early GC identification.
Table 3.
Compares the occurrence of early UGIC among anatomical sites across various groups.
| Lesion area | Low-image group vs. medium-image group | Low-image group vs. high-image group | Medium-image group vs. high-image group | Three groups | ||||
|---|---|---|---|---|---|---|---|---|
| χ2value | P value | χ2value | P value | χ2value | P value | χ2value | P value | |
| Esophagus | 0.795 | 0.373 | 1.652 | 0.199 | 0.311 | 0.577 | 1.699 | 0.428 |
| Gastrointestinal | 10.535 | 0.001 | 82.353 | < 0.001 | 102.618 | < 0.001 | 167.228 | < 0.001 |
| Duodenum | 0.112 | 0.738 | 0.234 | 0.629 | 1.22 | 0.269 | 1.261 | 0.532 |
The detection of early UGIC in the high-image group
In the high-image group, Fig. 2 presents the trend of early UGIC detection rates as the number of images varies. Specifically, when the number of images was between 45 and 49, the detection rate of early UGIC was the lowest, at 0.39%. In contrast, when the number of images exceeded 80, the detection rate of early-stage UGIC rose to 1.85%, a difference that was statistically significant compared to the other groups (p < 0.05). However, when the number of images ranged from 40 to 80, no statistically significant differences were observed in the detection rates of early UGIC among these groups (p > 0.05).
Fig. 2.
Line chart between the number of image taken and the detection rate of early UGIC in the high-image group.
Individual profiles of endoscopic physicians
Table 4 summarizes the demographic characteristics and early UGIC detection rates of 33 endoscopists. The cohort comprised 15 female physicians (45.45%), with 15 holding MD degrees (45.45%) and 11 possessing MS degrees (33.33%). Eleven physicians (33.33%) were full professors, while 20 (60.61%) held associate professor or higher positions. The mean professional experience was 18.06 ± 12.67 years.
Table 4.
Individual profiles of endoscopic physicians.
| Endoscopist | Sex | Experience(y) | Academic qualifications | Title | Number of EGDs | Number of images | Total UGIC | Detection rate (%) |
|---|---|---|---|---|---|---|---|---|
| 1 | Male | 31 | BD | Chief physician | 13,439 | 60.07 ± 23.17 | 131 | 0.95 |
| 2 | Male | 16 | MD | Attending physician | 14,433 | 21.60 ± 8.82 | 60 | 0.41 |
| 3 | Female | 9 | MD | Attending physician | 6972 | 29.82 ± 12.56 | 18 | 0.26 |
| 4 | Female | 35 | MD | Physician | 2064 | 37.24 ± 14.17 | 18 | 0.87 |
| 5 | Male | 15 | MD | Physician | 7164 | 43.25 ± 12.51 | 41 | 0.58 |
| 6 | Male | 7 | BD | Chief physician | 8427 | 30.58 ± 17.12 | 6 | 0.07 |
| 7 | Female | 3 | MD | Attending physician | 4404 | 49.26 ± 15.23 | 24 | 0.55 |
| 8 | Female | 2 | MD | Physician | 4392 | 49.40 ± 15.69 | 30 | 0.68 |
| 9 | Female | 5 | DD | Attending physician | 7668 | 43.41 ± 11.46 | 55 | 0.71 |
| 10 | Male | 8 | DD | Chief physician | 8496 | 10.72 ± 6.43 | 18 | 0.21 |
| 11 | Male | 28 | DD | Chief physician | 10,703 | 49.33 ± 14.81 | 125 | 1.16 |
| 12 | Male | 20 | BD | Associate chief physician | 6660 | 38.14 ± 12.33 | 18 | 0.27 |
| 13 | Male | 14 | DD | Associate chief physician | 10,725 | 30.26 ± 8.41 | 54 | 0.5 |
| 14 | Male | 5 | DD | Associate chief physician | 2784 | 58.03 ± 18.28 | 25 | 0.89 |
| 15 | Male | 20 | DD | Attending physician | 3024 | 43.17 ± 11.49 | 30 | 0.99 |
| 16 | Male | 12 | MD | Chief physician | 13,421 | 23.69 ± 8.70 | 42 | 0.31 |
| 17 | Male | 2 | DD | Attending physician | 2928 | 43.67 ± 11.07 | 7 | 0.24 |
| 18 | Male | 38 | DD | Chief physician | 18,275 | 28.17 ± 12.29 | 150 | 0.82 |
| 19 | Female | 7 | DD | Attending physician | 4488 | 72.11 ± 15.72 | 47 | 1.05 |
| 20 | Female | 42 | BD | Chief physician | 7327 | 42.85 ± 8.34 | 31 | 0.42 |
| 21 | Female | 38 | BD | Associate chief physicia | 8160 | 63.77 ± 20.23 | 96 | 1.18 |
| 22 | Female | 17 | MD | Chief physician | 6276 | 24.15 ± 14.27 | 24 | 0.38 |
| 23 | Female | 8 | MD | Physician | 6574 | 41.41 ± 12.64 | 30 | 0.45 |
| 24 | Male | 25 | MD | Associate chief physician | 11,703 | 51.74 ± 17.51 | 162 | 1.38 |
| 25 | Male | 29 | BD | Associate chief physician | 1320 | 32.36 ± 18.64 | 6 | 0.45 |
| 26 | Female | 29 | BD | Chief physician | 9625 | 85.34 ± 39.72 | 133 | 1.38 |
| 27 | Male | 6 | DD | Chief physician | 2820 | 53.46 ± 10.04 | 18 | 0.64 |
| 28 | Male | 2 | DD | Attending physician | 1031 | 46.21 ± 9.23 | 7 | 0.68 |
| 29 | Male | 42 | DD | Chief physician | 5868 | 70.69 ± 23.18 | 72 | 1.23 |
| 30 | Female | 8 | MD | Associate chief physician | 11,173 | 35.60 ± 13.21 | 30 | 0.26 |
| 31 | Female | 18 | DD | Associate chief physician | 13,537 | 60.93 ± 19.32 | 125 | 0.92 |
| 32 | Female | 4 | DD | Associate chief physician | 5724 | 42.85 ± 9.88 | 47 | 0.82 |
| 33 | Female | 7 | DD | Attending physician | 3551 | 28.56 ± 10.35 | 12 | 0.34 |
| Spearman’s rank correlation coefficient (P value)4 | 0.067(0.710) | 0.367(0.036) | 0.156(0.387) | 0.045(0.804) | 0.049(0.788) | 0.747(< 0.01) |
Analysis demonstrated significant positive correlations between early UGIC detection rates and both the mean number of images taken (R2 = 0.57, p < 0.01) and professional experience (R2 = 0.58, p = 0.036) (Figs. 3 and 4). No significant associations were observed with gender, academic rank, or educational background.
Fig. 3.
Correlation between the number of images taken by different endoscopic physicians and the detection rate of early UGIC.
Fig. 4.
Correlation between the work experience of different endoscopic physicians and the detection rate of early UGIC.
Endoscopists were stratified by the mean number of images taken into low-image (< 40 images, n = 13) and high-image (≥ 40 images, n = 20) groups (Table 5). Early UGIC detection rates were significantly higher in the high-image group (χ2 = 79.681, p < 0.001). Although the proportion of physicians with doctoral degrees was higher in the high-image group (p = 0.157) and the proportion of chief physicians was higher in the low-image group (p = 0.446), these differences were not statistically significant (Figs. 2 and 3).
Table 5.
Basic characteristics of physicians in two groups.
| Total | Low-image group (n > 40) | High-image group (n > 40) | P value | |
|---|---|---|---|---|
| Doctors’s number | 2.3.4.6.10.12.13,16.18.22.25.30.33 (Thirteen people) | 1.5.7.8.9.11.14.15.17.19.20.21.23.24.26.27.28.29.31.32. (Twenty people) | ||
| Patients’gender,female, n(%) | 15(45.5%) | 5(38.5%) | 10(50.00%) | 0.722 |
| Average experience(y) | 16.73 ± 12.79 | 16.92 ± 10.72 | 16.60 ± 14.24 | 0.449 |
| Academic qualifications | 0.348 | |||
| Bachelor of medicine | 7(21.2%) | 3(23.1%) | 4(20.0%) | 0.581 |
| Master of medicine | 11(33.3%) | 6(46.3%) | 5(25.0%) | 0.189 |
| Doctor of medicine | 15(45.5%) | 4(30.8%) | 11(55.0%) | 0.157 |
| Title | 0.861 | |||
| Physician | 4(12.1%) | 1(7.7%) | 3(15.0%) | 0.481 |
| Attending physician | 9(27.3%) | 3(23.1%) | 6(30.0%) | 0.491 |
| Associate chief physician | 9(27.3%) | 4(30.8%) | 5(25.0%) | 0.509 |
| Chief physician | 11(33.3%) | 5(38.5%) | 6(30.0%) | 0.446 |
| UGIC detection rate, n (%) | 564(0.69%) | 152(0.41%) | 412(0.93%) | < 0.001 |
PSM
To minimize selection bias and control for potential confounding factors, we employed PSM to process the study data. Table 6 presents the baseline characteristics of patients before and after PSM. Prior to PSM, significant intergroup differences were observed in age, gender, sedation status, examination start time, endoscopists’ experience, and the number of EGD diagnoses between the two groups (p < 0.001).
Table 6.
Baseline characteristics of patients before and after propensity score matching.
| Variables | All patients (n = 245,161) | Patients before PSM (n = 245,161) | Patients after PSM (n = 189,096) | ||||
|---|---|---|---|---|---|---|---|
| Low-image group (n = 151,509) | High-image group (n = 94,552) | p | Low-image group (n = 94,548) | High-image group (n = 94,548) | p | ||
| Age (years) | 51.89 ± 14.93 | 51.07 ± 15.21 | 53.22 ± 14.38 | < 0.001 | 52.87 ± 14.91 | 53.22 ± 14.38 | 0.064 |
| BMI | 23.89 ± 3.58 | 23.88 ± 3.56 | 23.93 ± 3.62 | < 0.001 | 23.91 ± 3.56 | 23.93 ± 3.62 | 0.346 |
| Anaesthesia | |||||||
| Sedation | 187,644 (76.3%) | 109,137 (72.0%) | 78,507 (83.0%) | < 0.001 | 77,435 (81.9%) | 78,507 (83.0%) | 0.077 |
| Consciousness | 58,392 (23.7%) | 42,372 (28.0%) | 16,044 (16.9%) | 17,113 (19.1%) | 16,041 (16.9%) | ||
| Endoscopist experience (years) | 19.04 ± 12.13 | 17.43 ± 11.64 | 21.63 ± 12.47 | < 0.001 | 21.25 ± 11.34 | 21.63 ± 12.39 | 0.472 |
| The number of EGD diagnoses | 1.25 ± 0.53 | 1.20 ± 0.49 | 1.36 ± 0.61 | < 0.001 | 1.34 ± 0.49 | 1.36 ± 0.57 | 0.603 |
| Star time | 0.393 | ||||||
| a.m. | 168,744 (68.6%) | 104,508 (69.0%) | 64,236 (67.9%) | < 0.001 | 63,610 (67.3%) | 64,236 (67.9%) | |
| p.m. | 77,316 (31.4%) | 47,001 (31.0%) | 30,315 (32.1%) | 29,767 (32.7%) | 30,312 (32.1%) | ||
| Detection rate (%) | 1692 (0.7%) | 609 (0.4%) | 1083 (1.1%) | < 0.001 | 498 (0.5%) | 1083 (1.1%) | < 0.001 |
After applying PSM with a caliper value of 0.00002, a total of 94,548 pairs of patients undergoing EGD were successfully matched. Following the matching process, we reassessed the baseline characteristics of the two patient groups. The results indicated no statistically significant differences between the groups (p > 0.05), demonstrating a good balance in the baseline characteristics of the patients in both groups post-matching. Further analysis revealed that the detection rate of early UGIC was significantly higher in the high-image group compared to the low-image group (p < 0.001).
Other validations
The five most experienced endoscopists (1,2,16,18,31) were designated as the high-proficiency group. Their early UGIC detection rates significantly increased when they obtained > 80 images (Table 7), demonstrating that even skilled operators benefit from comprehensive image documentation.
Table 7.
Early UGIC detection rates of high-proficiency endoscopists in different groups.
| Endoscopist | Total | Pictures ≤ 80 | Pictures > 80 | p | |||
|---|---|---|---|---|---|---|---|
| UGIC | Detection rate, n (%) | UGIC | Detection rate, n (%) | UGIC | Detection rate, n (%) | ||
| 1 | 132 | 0.95 | 66 | 0.59 | 66 | 3.03 | < 0.001 |
| 2 | 60 | 0.41 | 54 | 0.37 | 6 | 100.00 | < 0.001 |
| 16 | 162 | 1.38 | 138 | 1.24 | 24 | 4.54 | < 0.001 |
| 18 | 156 | 0.82 | 138 | 0.76 | 18 | 20.00 | < 0.001 |
| 31 | 125 | 0.92 | 72 | 0.62 | 54 | 2.99 | |
Five endoscopists (11,19,21,26,29) with detection rates > 1% comprised the high-detection group. Their mean image counts exceeded both the high-image threshold (> 40) and the overall average of image counts (Table 8). Comparative analysis revealed no significant detection rate differences between high- and low-image subgroups for most high-detection endoscopists, except for one endoscopist who detected no cases in the low-image group. Notably, Endoscopist 21 showed higher detection rates with fewer images.
Table 8.
Early UGIC detection rates of high-detection-rate endoscopists in different groups.
| Endoscopist | No. of EGDs Pictures | Total | Pictures ≤ 45 | Pictures > 45 | p | |||
|---|---|---|---|---|---|---|---|---|
| UGIC | Detection rate, n (%) | UGIC | Detection rate, n (%) | UGIC | Detection rate, n (%) | |||
| 11 | 49.33 ± 14.81 | 126 | 1.16 | 48 | 1.09 | 78 | 1.22 | 0.583 |
| 19 | 72.11 ± 15.72 | 42 | 1.05 | 0 | 0.00 | 42 | 0.99 | 0.031 |
| 21 | 63.77 ± 20.23 | 96 | 1.18 | 30 | 2.07 | 66 | 1.00 | < 0.001 |
| 26 | 85.34 ± 39.72 | 132 | 1.38 | 12 | 0.93 | 120 | 1.44 | 0.142 |
| 29 | 70.69 ± 23.18 | 72 | 1.23 | 12 | 1.58 | 60 | 1.20 | 0.365 |
These findings suggest that high-detection endoscopists maintain superior examination practices—thorough documentation and meticulous observation—rather than simply relying on lesion prevalence. The number of images reflects their systematic approach to mucosal inspection rather than case complexity.
Discussion
Upper gastrointestinal cancers (UGIC), including gastric cancer (GC) and esophageal cancer (EC), continue to pose significant challenges in the field of oncology. The establishment of endoscopic quality assessment indicators is conducive to setting a unified standard for endoscopists, thereby reducing variations in clinical practice. Despite the publication of multiple guidelines and position statements regarding the quality of EGD (esophagogastroduodenoscopy) examinations16,20, no clear quality assessment criteria have been established, aside from accurate photographic documentation15 and a minimum seven-minute procedure duration for initial diagnostic EGD.
In this retrospective study, which analyzed 245,161 EGD examinations, we discovered that a greater number of images taken during the procedure led to a higher detection rate of early-stage UGIC. Specifically, compared to the medium and low image taken groups, the high-image group (≥ 40 images) exhibited a higher detection rate for lesions. Furthermore, within the high-image group, when the number of images exceeded 80, the detection rate of early-stage UGIC was notably higher than that of the other subgroups within the high-image group. For individual endoscopists, there was a significant, non-linear correlation between the number of images taken and the detection rate of early-stage UGIC. As the number of images increased, the detection rate of early-stage UGIC exhibited an upward trend.
It is important to acknowledge that this study is subject to various confounding factors, such as patient age, gender, BMI, use of sedatives, number of lesions, endoscopist experience, and examination start time, all of which may influence the number of images taken during EGD. To mitigate the impact of these confounders, we performed PSM, which effectively reduced intergroup differences in baseline characteristics, thereby enhancing the reliability of our findings. After controlling for the influence of relevant factors, the early UGIC detection rate in the high-image group remained significantly higher than that in the low-image group, aligning with our previous conclusions.
While limited evidence exists regarding the relationship between the number of EGD images taken and UGIC detection, comprehensive photographic documentation enhances examination quality by providing detailed records and identifying potential blind spots. Previous studies demonstrate that image count independently predicts the detection of esophageal precancerous lesions and early squamous cell carcinomas21. A Japanese observational study22 further established a positive correlation between UGIC detection and either examination duration (> 5 min) or image count (> 50), which is consistent with our findings. These results support incorporating image quantity as an objective EGD quality metric.
The individual habits of endoscopists undoubtedly influence the number of images taken during EGD examinations. Our analysis of 33 endoscopists revealed that both the average number of images taken and the endoscopists’ professional experience were positively correlated with the detection rate of early-stage UGIC. This supports the notion that endoscopists who take more images tend to examine more meticulously, thereby improving the likelihood of detecting occult lesions. Similarly, experienced endoscopists, with their deeper understanding of mucosal abnormalities, demonstrated higher early-stage cancer detection rates. In contrast, factors such as the endoscopists’ gender, age, professional title, and educational background showed no significant correlation with early UGIC detection.
Previous studies have consistently highlighted the importance of endoscopists’ skills and experience in EGD quality. For instance, a multicenter randomized controlled trial23 demonstrated that endoscopists trained through systematic e-learning were more proficient in detecting early gastric cancer. Other studies24–27 have shown that standardized training programs enhance gastrointestinal tumor detection rates. Additionally, ISHIHARA et al.28, reported that experienced endoscopists achieved 100% sensitivity in detecting high-grade dysplasia using NBI screening, outperforming their less experienced counterparts. These findings align with our results, further validating that endoscopists with greater experience and more years of professional practice exhibit higher early-stage UGIC detection rates.
Stratified analysis in this study revealed that the number of images taken significantly influenced the detection rate of early GC but was not significantly correlated with the detection of early EC or duodenal lesions. This discrepancy may be attributed to anatomical and procedural differences among these regions: gastric examinations often require repeated observation using both retroflexion and forward views, coupled with rapid gastric peristalsis, necessitating a higher number of images. In contrast, esophageal lesions are primarily assessed using narrow-band imaging (NBI) and are associated with minimal peristalsis, reducing the need for additional images. Furthermore, the significantly lower incidence of early duodenal cancer compared to EC and GC further diminishes the impact of increased image take on detection rates. Based on these findings, we recommend that, to enhance the detection rate of early UGIC during EGD examinations, the number of images taken for the stomach should be increased, while for the esophagus and duodenum, adherence to standard imaging protocols is sufficient.
Before undergoing EGD, all subjects in this study were administered streptococcal protease to achieve optimal observation conditions. In addition to the number of images taken, the quality of the images also affects the quality of EGD examinations. Several RCTs29 and guidelines4 recommend using mucolytic agents and antifoaming agents during EGD to improve visual clarity. Photographic documentation of all anatomical landmarks is considered credible evidence of a complete examination. Previous research has found a significant correlation between the documentation rate of duodenal papilla photographs and the detection rate of UGIC in asymptomatic patients30. The ESGE recommends capturing images of the duodenal papilla during EGD for all patients16. Unfortunately, the database did not include statistics on papilla photographs, so the present study could not perform a analysis. Among all participants in this study, over 70% chose to undergo EGD under sedation. Prior research21,31 has yielded conflicting results regarding whether sedation affects the detection rate of upper gastrointestinal mucosal lesions. Several randomized controlled trials (RCTs)32–34 demonstrate that sedated EGD significantly improves patient satisfaction, compliance, and repeat-examination acceptance without increasing postprocedural cardiovascular/cerebrovascular risks. While direct evidence linking sedation to enhanced early UGIC detection remains lacking, the Asian Consensus4 recommends routine sedation for EGD in eligible patients without absolute contraindications. Furthermore, the start time of the EGD examination may have an impact on lesion detection. Previous studies35 have shown that as the day progresses, the lesion detection rate and endoscopic biopsy rate tend to decline, potentially due to endoscopist fatigue.
Numerous studies36,37 have established that EGD examination duration is an independent risk factor influencing upper gastrointestinal lesion detection, with longer procedures associated with higher detection rates38. While ESGE recommends documenting examination time in all EGD reports, regional guidelines vary: the Asian Consensus suggests a minimum of 8 minutes4, whereas the British Society of Gastroenterology (BSG) recommends at least 7 minutes20. A multicenter, multi-phase study39 further demonstrated that setting a minimum examination time of 6 min significantly improved lesion detection and overall EGD quality. Although this study did not collect examination times due to unavoidable constraints, a correlation likely exists between the number of images taken and examination duration, as endoscopists who take more images tend to observe more meticulously and spend additional time on the procedure. Previous research22 has even used image count as a proxy for examination time. Clinically, examination time is defined as the interval from capturing the first image at the pharynx or upper esophagus to pressing the exit button post-examination, though manual control introduces potential human error. Standardized protocols (ESGE/BSG) require imaging 10 anatomical landmarks: (1) proximal esophagus, (2) distal esophagus, (3) Z-line, (4) cardia/fundus, (5) anterior gastric body, (6) posterior gastric body, (7) gastric angle, (8) antrum, (9) duodenal bulb, and (10) D2. However, regional variations in gastrointestinal disease burdens lead to differing EGD photography standards. Given these factors, we propose that the number of images taken serves as a reliable and practical indicator for early UGIC detection in research settings.
The primary strength of this study lies in its pioneering multicenter design, which systematically investigates the impact of the number of images taken during endoscopy on the detection rate of early UGIC. The study features a complete cohort and high-quality data, ensuring robust and reliable results. Furthermore, with over 240,000 participants included, this study represents the largest sample size to date in research exploring factors influencing early UGIC detection, significantly enhancing the statistical power and generalizability of the findings. This large-scale sample size provides a solid statistical foundation for in-depth analysis of the relationship between the number of images taken and early UGIC detection rates.
Of course, there are some limitations to this study. As a multicenter retrospective study, there were intergroup differences in some baseline characteristics, and some potential influencing factors may have been difficult to collect. Additionally, due to the lack of random assignment, it was challenging to fully match the number of EGD photographs in the database with the current experimental results. To overcome this limitation, further research, including prospective randomized controlled trials, is needed. Unfortunately, for our study, we were unable to obtain relevant data on the duration of EGD examinations. However, there may be a correlation between examination time and the number of photographs taken, both of which can represent the endoscopist’s focus and the completeness of the EGD examination.
This study found that the number of images taken during EGD is significantly positively correlated with the detection rate of early UGIC, serving as an important factor influencing early cancer detection. Subgroup analysis revealed that endoscopists who took more images had a significantly higher detection rate of early UGIC (p < 0.05), a finding that remained consistent after PSM to eliminate confounding factors. Based on the research evidence, it is recommended to routinely capture at least 40 high-quality images during EGD to ensure the completeness and accuracy of the examination. Furthermore, by incorporating the number of endoscopic image acquisitions into the quality assessment system for endoscopic procedures, this study provides a scientific basis for optimizing endoscopic operation guidelines. Based on the research findings, we recommend the formulation and promotion of standardized endoscopic operation procedures with image quantity as an important indicator, as well as systematic training to enhance the operational standardization and diagnostic capabilities of clinicians. This measure is expected to significantly increase the detection rate of early UGIC, thereby improving patient prognosis, extending survival, and enhancing quality of life.
Author contributions
Shi: conceptualization; Data curation; Formal analysis; Writing-original draft. Zhou: Formal analysis; Visualization. Chen: Methodology; Visualization. Shen: Methodology; Validation. ZhuoMa: Validation; Supervision. Feng:Formal analysis. Sun:Project administration. Zhang: Conceptualization; Resources; Writing—original draft; Writing—review, and Editing.
Funding
Protective effect of isolicorice on ulcerative colitis in mice through NF-κB signaling pathway and its mechanism, 2023ZL266.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Sung, H. et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin.71, 209–249 (2021). [DOI] [PubMed] [Google Scholar]
- 2.Asplund, J., Kauppila, J. H., Mattsson, F. & Lagergren, J. Survival trends in gastric adenocarcinoma: A population-based study in Sweden. Ann. Surg. Oncol.25, 2693–2702 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.B, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin.74, 229 (2024). [DOI] [PubMed] [Google Scholar]
- 4.Chiu, P. W. Y. et al. An Asian consensus on standards of diagnostic upper endoscopy for neoplasia. Gut68, 186–197 (2019). [DOI] [PubMed] [Google Scholar]
- 5.Tao, M., Zhou, X., Hu, M. & Pan, J. Endoscopic submucosal dissection versus endoscopic mucosal resection for patients with early gastric cancer: A meta-analysis. BMJ Open9, e025803 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Liu, Q., Ding, L., Qiu, X. & Meng, F. Updated evaluation of endoscopic submucosal dissection versus surgery for early gastric cancer: A systematic review and meta-analysis. Int. J. Surg.73, 28–41 (2020). [DOI] [PubMed] [Google Scholar]
- 7.Uedo, N. et al. Differences in routine esophagogastroduodenoscopy between Japanese and international facilities: A questionnaire survey. Dig. Endosc.28(Suppl 1), 16–24 (2016). [DOI] [PubMed] [Google Scholar]
- 8.Kamran, U. et al. A root cause analysis system to establish the most plausible explanation for post-endoscopy upper gastrointestinal cancer. Endoscopy55, 109–118 (2023). [DOI] [PubMed] [Google Scholar]
- 9.Alexandre, L., Tsilegeridis-Legeris, T. & Lam, S. Clinical and endoscopic characteristics associated with post-endoscopy upper gastrointestinal cancers: A systematic review and meta-analysis. Gastroenterology162, 1123–1135 (2022). [DOI] [PubMed] [Google Scholar]
- 10.Pimenta-Melo, A. R., Monteiro-Soares, M., Libânio, D. & Dinis-Ribeiro, M. Missing rate for gastric cancer during upper gastrointestinal endoscopy: A systematic review and meta-analysis. Eur. J. Gastroenterol. Hepatol.28, 1041–1049 (2016). [DOI] [PubMed] [Google Scholar]
- 11.Januszewicz, W. & Kaminski, M. F. Quality indicators in diagnostic upper gastrointestinal endoscopy. Therap. Adv. Gastroenterol.13, 1756284820916693 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Li, B. et al. Comparative study on artificial intelligence systems for detecting early esophageal squamous cell carcinoma between narrow-band and white-light imaging. WJG27, 281–293 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Januszewicz, W. et al. Endoscopist biopsy rate as a quality indicator for outpatient gastroscopy: A multicenter cohort study with validation. Gastrointest. Endosc.89, 1141–1149 (2019). [DOI] [PubMed] [Google Scholar]
- 14.Yao, K. The endoscopic diagnosis of early gastric cancer. Ann. Gastroenterol.26, 11–22 (2013). [PMC free article] [PubMed] [Google Scholar]
- 15.Rey, J. F., Lambert, R. & ESGE Quality Assurance Committee. ESGE recommendations for quality control in gastrointestinal endoscopy: guidelines for image documentation in upper and lower GI endoscopy. Endoscopy33, 901–903 (2001). [DOI] [PubMed]
- 16.Bisschops, R. et al. Performance measures for upper gastrointestinal endoscopy: A european society of gastrointestinal endoscopy (ESGE) quality improvement initiative. Endoscopy48, 843–864 (2016). [DOI] [PubMed] [Google Scholar]
- 17.Cohen, J. et al. Quality indicators for esophagogastroduodenoscopy. Gastrointest Endosc63, S10-15 (2006). [DOI] [PubMed] [Google Scholar]
- 18.Mine, S. et al. Japanese classification of esophageal cancer, 12th edition: Part I. Esophagus21, 179–215 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Shitara, K. et al. Pan-Asian adapted ESMO clinical practice guidelines for the diagnosis, treatment and follow-up of patients with gastric cancer. ESMO Open9, 102226 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.B, S. et al. Quality standards in upper gastrointestinal endoscopy: A position statement of the British society of gastroenterology (BSG) and association of upper gastrointestinal surgeons of Great Britain and Ireland (AUGIS). Gut66, 1886 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Liang, M., Xu, C., Zhang, X., Zhang, Z. & Cao, J. Effect of anesthesia assistance on the detection rate of precancerous lesions and early esophageal squamous cell cancer in esophagogastroduodenoscopy screening: A retrospective study based on propensity score matching. Front. Med.10, 1039979 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ksahiwagi, K. et al. Optimal number of images and 2-year interval affect cancer detection in screening esophagogastroduodenoscopy: An observational study. Medicine103, e38774 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yao, K. et al. Development of an e-learning system for the endoscopic diagnosis of early gastric cancer: An international multicenter randomized controlled trial. EBioMedicine9, 140–147 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Unno, S. et al. Assigning a different endoscopist for each annual follow-up may contribute to improved gastric cancer detection rates. Endosc. Int. Open10, E1333–E1342 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Siau, K. et al. Impact of the joint advisory group on gastrointestinal endoscopy (JAG) on endoscopy services in the UK and beyond. Frontline Gastroenterol.10, 93–106 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Zhang, Q. et al. Training in early gastric cancer diagnosis improves the detection rate of early gastric cancer: An observational study in China. Medicine94, e384 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Yamazato, T. et al. Two years’ intensive training in endoscopic diagnosis facilitates detection of early gastric cancer. Intern. Med.51, 1461–1465 (2012). [DOI] [PubMed] [Google Scholar]
- 28.Ishihara, R. et al. Original article: Prospective evaluation of narrow-band imaging endoscopy for screening of esophageal squamous mucosal high-grade neoplasia in experienced and less experienced endoscopists: NBI for esophageal neoplasia. Dis. Esophagus23, 480–486 (2010). [DOI] [PubMed] [Google Scholar]
- 29.Liu, X. et al. Effect of premedication on lesion detection rate and visualization of the mucosa during upper gastrointestinal endoscopy: A multicenter large sample randomized controlled double-blind study. Surg. Endosc.32, 3548–3556 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Park, J. et al. The effect of photo-documentation of the ampulla on neoplasm detection rate during esophagogastroduodenoscopy. Endoscopy51, 115–124 (2019). [DOI] [PubMed] [Google Scholar]
- 31.Liu, Y. & Gu, K. Association between anesthesia assistance and precancerous lesions and early cancer detection during diagnostic esophagogastroduodenoscopy: A propensity score-matched retrospective study. Front. Med.11, 1389809 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Correia, C. et al. Quality standards in upper gastrointestinal endoscopy: Can deep sedation influence it?. GE Port. J. Gastroenterol.31, 101–109 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.McQuaid, K. R. & Laine, L. A systematic review and meta-analysis of randomized, controlled trials of moderate sedation for routine endoscopic procedures. Gastrointest. Endosc.67, 910–923 (2008). [DOI] [PubMed] [Google Scholar]
- 34.Kim, S. Y. et al. The impact of sedation on cardio-cerebrovascular adverse events after surveillance esophagogastroduodenoscopy in patients with gastric cancer: A nationwide population-based cohort study. Gut Liver18, 245–256 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Zhang, Z. et al. Esophagogastroduodenoscopy outcomes variated by the time of the day: A single-center experience. J. Clin. Med.12, 863 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Kawamura, T. et al. Examination time as a quality indicator of screening upper gastrointestinal endoscopy for asymptomatic examinees. Dig. Endosc.29, 569–575 (2017). [DOI] [PubMed] [Google Scholar]
- 37.Dong, L. et al. Relationship between observation time and detection rate of focal lesions in esophagogastroduodenoscopy: A single-center, retrospective study. BMC Gastroenterol.24, 75 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Teh, J. L. et al. Longer examination time improves detection of gastric cancer during diagnostic upper gastrointestinal endoscopy. Clin. Gastroenterol. Hepatol.13, 480-487.e2 (2015). [DOI] [PubMed] [Google Scholar]
- 39.Gao, Y. et al. Setting 6-minute minimal examination time improves the detection of focal upper gastrointestinal tract lesions during endoscopy: A multicenter prospective study. Clin. Transl. Gastroenterol.14, e00612 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.