Original article
The effect of photo-documentation of the ampulla on neoplasm
detection rate during esophagogastroduodenoscopy
Authors
Jae Myung Park1, 2, Chul-Hyun Lim1, 2, Yu Kyung Cho 1, 2, Bo-In Lee 1, 2, Young-Seok Cho1, 2, Ho Jin Song3, Myung-Gyu
Choi1, 2
Institutions
1 Division of Gastroenterology and Hepatology,
Department of Internal Medicine, Seoul St. Mary’s
Hospital, The Catholic University of Korea, Seoul, Korea
2 Catholic Photomedicine Research Institute, Seoul, Korea
3 Department of Health Promotion Medicine, Seoul St.
Mary’s Hospital, The Catholic University of Korea, Seoul,
Korea
submitted 27.2.2018
accepted after revision 12.7.2018
Bibliography
DOI https://doi.org/10.1055/a-0662-5523
Published online: 2018 | Endoscopy
© Georg Thieme Verlag KG Stuttgart · New York
ISSN 0013-726X
Corresponding author
Jae Myung Park, MD, Division of Gastroenterology and
Hepatology, Department of Internal Medicine, Seoul St.
Mary’s Hospital, College of Medicine, The Catholic
University of Korea, 222 Banpo-daero, Seocho-gu, Seoul
137-701, Korea
Fax: +82-2-22582089
parkjerry@catholic.ac.kr
Supplementary Table e1, Table e4, Table e6, Table e7
Online content viewable at:
https://doi.org/10.1055/a-0662-5523
Introduction
Esophagogastroduodenoscopy (EGD) is a common diagnostic
procedure performed to identify upper gastrointestinal (GI)
malignancy. EGD allows the diagnosis of gastric cancer at an
earlier stage, resulting in better survival [1]. However, several
studies have reported on missed or interval upper GI cancers
Park Jae Myung et al. The effect of … Endoscopy
ABSTR AC T
Background A few studies have investigated quality indi-
cators of esophagogastroduodenoscopy (EGD) for identify-
ing upper gastrointestinal (GI) malignancy. The current
study aimed to evaluate whether the rate of ampulla pho-
to-documentation could be associated with the detection
of upper GI neoplasms.
Methods We used data from 111 962 asymptomatic pa-
Downloaded by: University of Michigan. Copyrighted material.
tients who underwent EGD performed by 14 endoscopists
at a health promotion center. The rate of ampulla photo-
documentation was calculated by reviewing EGD photos ar-
chived during each endoscopist’s first year of working at
the center. The detection of neoplasms during a 7-year
period was investigated. We examined the association be-
tween the rate of ampulla photo-documentation and the
rate of neoplasm detection.
Results The mean rate of ampulla photo-documentation
was 49.0 % (range 13.7 % – 78.1 %) during endoscopists’ first
year of working at the center. Endoscopists’ rates of ampul-
la photo-documentation significantly correlated with the
detection of total neoplasms (R2 = 0.57, P = 0.03) and small
neoplasms (R 2 = 0.58, P = 0.03). There was a significant dif-
ference in the detection rates of upper GI neoplasms be-
tween high (n = 7) and low (n = 7) ampulla observers (odds
ratio [OR] 1.31, 95 % confidence interval [CI] 1.03 – 1.68;
P = 0.03). The ampulla photo-documentation rate of each
endoscopist significantly correlated with the examination
time for a normal EGD (R 2 = 0.55; P = 0.04). In multivariate a-
nalysis, high ampulla photo-documentation rate was a pre-
dictor of neoplasm detection (OR 1.33, 95 %CI 1.03 – 1.70).
Conclusions The ampulla photo-documentation rate was
significantly associated with the detection rate for both to-
tal and small upper GI neoplasms. Ampulla photo-docu-
mentation should be considered as a quality indicator of
EGD.
[2 – 5], implying the importance of quality control in EGD exam-
inations.
For a high quality examination, EGD should be evaluated by
appropriate quality indicators, which are divided into three
components: structure, procedure, and outcome [6]. Among
them, the procedure component assesses performance during
endoscopic examination and is related to clinical outcomes
 Original article
such as cancer-related death or neoplasm detection. Guidelines
for colonoscopy have been implemented to comply with several
procedural quality indicators including cecal intubation rate,
observation time, and adenoma detection rate [7]. Compared
with colonoscopy, procedural quality indicators of EGD have
not been well evaluated and validated, even though the guide-
lines recommended several conditions relating to a high stand-
ard procedure [6, 8 – 9]. Recently, a few studies have reported
the importance of observation time, implying that a longer ex-
amination time increases the proportion of neoplasms detect-
ed during EGD [10, 11]. However, more quality indicators are
needed for a complete EGD evaluation.
The ampulla is formed by the union of the pancreatic duct
and the common bile duct. The ampulla is an important land-
mark halfway along the second part of the duodenum that
marks the anatomical transition from foregut to midgut. The
ampulla is located horizontally to the scope and is usually the
most distal portion examined during routine EGD evaluation.
Therefore, this anatomical landmark may be regarded in a sim-
ilar way to the cecum in colonoscopy. Because the ampulla is
not always observed in the front view of EGD [12, 13], it is not
easy to identify this structure when the endoscopist does not
intend to locate it. The reasons include the tangential position
of the ampulla, duodenal folds including the hooding fold, and
the presence of periampullary diverticulum [13]. In capsule
endoscopy, ampulla detection was reported in only half of the
cases [14, 15]. Unlike capsule endoscopy, endoscopists can
control the EGD scope with to-and-fro movement to visually
confirm the ampulla, which may be dependent on the perform-
er’s proficiency or attention during the examination.
In the present study, we investigated whether the rate of
ampulla photo-documentation was related to the rate of upper
GI neoplasm detection.
Methods
Study design and participants
We used data from the comprehensive health screening pro-
gram of the Health Promotion Center of Seoul St. Mary’s Hospi-
tal, which was used in a previous study [10]. These data includ-
ed patient symptoms and physical examinations, blood test re-
sults, low dose chest computed tomography, abdominal ultra-
sonography, and EGD.
In the previous study, we investigated the characteristics of
endoscopists (experience, working days at the center, mean ex-
amination time of EGDs) and health-screening individuals (age,
sex, body mass index, smoking, alcohol, family history of upper
GI cancer, and Helicobacter pylori infection). The endoscopists in
the current study were 14 board-certified gastroenterologists.
They were trained at the same institution and during their fel-
lowship period were taught the importance of observation and
photo-documentation of the ampulla. All endoscopists used
the same endoscope unit (GIF-H260; Olympus, Tokyo, Japan).
The study was conducted in accordance with the Declaration
of Helsinki and the Institutional Review Board of the Seoul St.
Mary’s Hospital, the Catholic University of Korea (IRB No.
KC16RISI0404).
Ampulla photo-documentation rate
For the current study, we looked at the duodenal images stored
in the picture archiving and communication system of EGD ex-
aminations. We defined photo-documentation of the ampulla
as the visualization of the whole area of the ampulla, including
the orifice (▶ Fig. 1a), or partial visualization, when only part of
the area of the ampulla was visualized ( ▶ Fig. 1b). If no part of
the ampulla was found, we did not count it as ampulla photo-
documentation (▶ Fig. 1c). The rate of ampulla photo-docu-
mentation was calculated for each endoscopist during their
first year of working at the health promotion center. The endos-
copists were grouped into quartiles according to ampulla pho-
to-documentation rate in a univariate analysis. In a multivariate
analysis, we divided the endoscopists into two groups by medi-
an photo-documentation rate: high rate of ampulla photo-doc-
umentation (high ampulla observer), and low rate of photo-
documentation (low ampulla observer).
EGD examination time
Downloaded by: University of Michigan. Copyrighted material.
Examination time was also assessed using the picture archiving
and communication system, which shows the time at which the
image was taken. The examination time was defined as the time
the endoscope reached the duodenum to the time it was with-
drawn, as described previously [10]. The endoscopists were
divided into fast and slow examiners according to the mean
duration of all EGD examinations of the 14 endoscopists for a
normal examination without biopsy during their first year of
working at the health promotion center.
Endoscopic findings and histologic reports
Endoscopic findings were extracted from the endoscopy re-
port, as described previously [10]. We reviewed histologic re-
sults from the electronic medical records of biopsy results. We
defined positive histopathologic findings as the detection of
neoplasms on biopsy. The histologic criteria of gastric neoplasia
were defined as per the Vienna classification [16]. Small neo-
plasms were defined as neoplasms of less than 1 cm in diameter
(▶ Fig. 2). We defined early neoplasms as tumors limited to the
mucosal or submucosal layer. If the lesion invaded deeper than
the submucosal layer, we classified it as advanced cancer.
Endoscopist experience was defined as the time from board
certification until participation in the present study.
Outcome and measurement
The primary outcome was the neoplasm detection rate for each
endoscopist, calculated as the total number of neoplastic le-
sions detected divided by the number of individuals screened
and as a proportion of individuals with at least one neoplastic
lesion. The secondary outcome was the rate of small neoplasms
detection for each endoscopist.
We measured the relationship between the rate of ampulla
photo-documentation during EGD examinations and lesion de-
tection rates of the 14 endoscopists. Analysis was per patient,
whereby one positive outcome was registered regardless of
the number of neoplasms diagnosed during the examination.
Park Jae Myung et al. The effect of … Endoscopy
 ▶ Fig. 1 Examples of photo-documentation of the ampulla: a complete; b partial; c nonvisualization.

Downloaded by: University of Michigan. Copyrighted material.

▶ Fig. 2 Representative pictures of small neoplasms. a An erythematous and slightly depressed lesion (arrow) was found in the posterior wall
of the antrum, and was revealed as a tubular adenocarcinoma. This lesion was resected with endoscopic submucosal dissection. The resected
specimen showed a carcinoma, 0.6 cm in size and invading the lamina propria of the mucosal layer. b A flat lesion was found in the cardia
(arrow), and was revealed as low grade dysplasia. This lesion was treated with endoscopic mucosal resection, and the final pathology revealed
a type IIb lesion, 0.8 cm in size.

Statistical methods
We used the Student’s t test to analyze the differences in the
lesion detection rates between high and low ampulla obser-
vers. A comparison of categorical variables was performed
using the chi-squared test or Fisher’s exact test. Data are re-
ported as means ± SD unless otherwise stated. The relationship
between variables was assessed using Pearson’s correlation
coefficient after normality test using the Shapiro-Wilk test
(▶ Supplementary Table e1, available online). P values of less
than 0.05 were considered statistically significant. Univariate
and multivariable logistic regression analyses were used to
identify independent predictors of neoplasm detection. A mul-
tiple logistic regression model was constructed using stepwise
selection with entry criteria of P < 0.1 and stay criteria of P <
0.05. Logistic regression analysis was used to identify predic-
tors of neoplasm detection. To decide the best format of the
factors, we performed Akaike Information Criterion analysis.
All analyses were performed using SAS version 9.4 (SAS Insti-
tute Inc., Cary, North Carolina, USA).
Results
Participants
A total of 111 962 asymptomatic individuals who underwent
EGDs for screening purposes were analyzed, as described pre-
viously [10]. The mean age of the participants was 49.6 ± 10.5
years (range 31 – 98 years), and 64 593 (57.7 %) were male.
Smoking and alcohol history were observed in 17 706 of

Park Jae Myung et al. The effect of … Endoscopy

 Original article
56 696 (31.2 %) and 50 056 of 76 918 (65.1 %) participants,
respectively.
As investigated in the previous study [10], neoplastic lesions
were detected in 262 participants (0.23 %), of whom 101
(0.09 % of total, 38.6 % of neoplasms) had gastric cancers, 139
had gastric dysplasia, 20 had duodenal neoplasms, and 2 had
esophageal cancers during the total study period. There were
differences among endoscopists in their neoplasm detection
rates (range 0.14 % – 0.32 %) ( ▶ Table 2).
Ampulla photo-documentation and neoplasm
detection
The mean rate of ampulla photo-documentation for all EGD ex-
aminations performed by the 14 endoscopists was 49.0 %
(16 022/32 768). There were considerable differences among
endoscopists in the mean rate of ampulla photo-documenta-
tion during their first year of working at the health promotion
center (range 13.7 % – 78.1 %) ( ▶ Table 2). The mean rate of
ampulla photo-documentation significantly and positively
correlated with the detection rate of total upper GI neoplasms
(R 2 = 0.57, P = 0.03) ( ▶ Fig. 3a) and small upper GI neoplasms
(R 2 = 0.58, P = 0.03) ( ▶ Fig. 3b).
When the endoscopists were divided into quartiles accord-
ing to mean rate of ampulla photo-documentation, the neo-
plasm detection rate of endoscopists in the second (n = 4), third
(n = 3), and fourth (n = 4) quartile was higher than in the first
quartile (n = 4), with odds ratios (OR) and 95 % confidence inter-
vals (CIs) of 1.16 (95 %CI 0.80 – 1.69), 1.36 (95 %CI 0.93 – 1.99),
and 1.53 (95 %CI 1.08 – 2.19), respectively. We further analyzed
the data to investigate the possible association between the
ampulla photo-documentation rate and the lesion location. As
shown in ▶ Table 2, the detection rates of the duodenal and an-
tral lesions were positively associated with the ampulla photo-
documentation rate.
Neoplasm detection rate of high vs. low ampulla
observers
Using the median ampulla photo-documentation rate (52 %) as
the cutoff, we grouped seven endoscopists into the high am-
pulla observers group (62.3 % photo-documentation rate) and
seven into the low ampulla observers group (36.0 % photo-doc-
umentation rate). The baseline characteristics of the partici-
pants were not significantly different between the high and
low ampulla observers (▶ Table 3). High ampulla observers had
a higher detection rate for upper GI neoplasms compared with
low ampulla observers (0.26 % vs. 0.20 %; P = 0.03) ( ▶ Table 3).
Furthermore, high ampulla observers found significantly more
dysplasias (0.17 % vs. 0.11 %; P = 0.004) and small neoplasms
(0.14 % vs. 0.09 %; P = 0.01) than low ampulla observers ( ▶ Ta-
ble 3). However, there were no differences in the detection
rates of advanced and early gastric cancers between the two
groups.
Ampulla photo-documentation and examination
time
The mean examination time of all EGDs performed by the 14
endoscopists for an examination without biopsy was 2:58 ±
0:31 minutes:seconds (range 0:47 – 20:27 minutes:seconds)
during their first year of working at the health promotion cen-
ter. There were considerable differences among endoscopists
in their mean examination time (range 1:53 – 3:47 minutes:sec-
onds) ( ▶ Table 2). The photo-documentation rate of each
endoscopist significantly correlated with the examination time
for a normal EGD without biopsy during their first year, as
shown in ▶ Fig. 4 (R 2 = 0.55; P = 0.04).
Ampulla photo-documentation and upper GI
neoplasm detection in fast and slow endoscopists
We further analyzed the effect of ampulla photo-documenta-
tion rate on neoplasm detection rate in the fast (n = 8) and
slow (n = 6) endoscopist groups, respectively, using 3 minutes
Downloaded by: University of Michigan. Copyrighted material.
as a cutoff. We divided the endoscopists in each group by me-
dian rate of ampulla photo-documentation, which was 44.64 %
in the fast endoscopists group and 56.83 % in the slow endos-
copists group. As shown in ▶ Supplementary Table e4 (avail-
able online), there was no difference between the detection
rate of upper GI neoplasms and ampulla photo-documentation
rate in either fast or slow endoscopists.
Ampulla photo-documentation and endoscopist
experience
Endoscopist experience after certification of GI specialty was
1 – 14 years. The ampulla photo-documentation rate did not
correlate with endoscopist experience (R 2 – 0.34, P = 0.23).
Factors associated with detection of upper GI
neoplasms
In univariate analysis, older age, male sex, smoking, high biopsy
rate of the endoscopist, endoscopist with a long mean exami-
nation time, and ampulla photo-documentation rate were
associated with the detection of upper GI neoplasms ( ▶ Ta-
ble 5). In multivariate analysis, ampulla photo-documentation
was independently associated with the diagnosis of upper GI
neoplasms even after adjusting for age, sex, smoking, and
biopsy rate. The examination time could not be included in
this model because of its high multi-collinearity with the am-
pulla photo-documentation rate. An endoscopist with a high
ampulla photo-documentation rate was more likely to detect
upper GI neoplasms (OR 1.33, 95 %CI 1.03 – 1.70) ( ▶ Table 5).
Discussion
In this study, we evaluated whether photo-documentation of
the ampulla could be a potential quality indicator for EGD. We
investigated whether the rate of ampulla photo-documenta-
tion was associated with the rate of neoplasm detection during
EGD. We found positive correlations between ampulla photo-
documentation and the detection of small as well as total upper
GI neoplasms between the high and low ampulla observers.
Park Jae Myung et al. The effect of … Endoscopy
 ▶ Table 2 Ampulla photo-documentation and detection rates for total neoplasms and small lesions for individual endoscopists (N = 111 962).

Endoscopist Ampulla Detection rate, n/N Mean Detection rate according to location, n/N (%)
photo- (%) exami-
docu- nation
Total Small Esopha- Cardia, Mid and Angle Antrum Duode-
menta- time
neo- neo- gus fundus low (n = 49) (n = 104) num
tion (min:
plasms plasms 3 (n = 2) and body (n = 20)
n/N (%) 1 sec) 2
(n = 262) (n = 129) high (n = 52)
body
(n = 35)

A 1151/ 7/3747 4/3747 02:54 0/3747 1/3747 2/3747 0/3747 2/3747 2/3747
2112 (0.19) (0.11) (0.00) (0.03) (0.05) (0.00) (0.05) (0.05)
(54.50)

B 807/ 29/ 18/ 01:53 0/14979 6/14979 4/14979 4/14979 14/ 1/14979
3273 14979 14979 (0.00) (0.04) (0.03) (0.03) 14979 (0.01)
(24.66) (0.19) (0.12) (0.09)

C 384/ 4/1491 0/1491 02:50 0/1491 0/1491 3/1491 0/1491 1/1491 0/1491
1491 (0.27) (0.00) (0.00) (0.00) (0.20) (0.00) (0.07) (0.00)
(25.75)

Downloaded by: University of Michigan. Copyrighted material.

D 292/ 6/4163 0/4163 03:00 0/4163 2/4163 1/4163 3/4163 0/4163 0/4163
2131 (0.14) (0.00) (0.00) (0.05) (0.02) (0.07) (0.00) (0.00)
(13.70)

E 1271/ 6/2390 1/2390 03:39 0/2390 0/2390 1/2390 1/2390 2/2390 2/2390
1754 (0.25) (0.04) (0.00) (0.00) (0.04) (0.04) (0.08) (0.08)
(72.46)

F 2239/ 42/ 19/ 02:33 1/16260 3/16260 8/16260 9/16260 16/ 5/16260
3239 16260 16260 (0.006) (0.02) (0.05) (0.05) 16260 (0.03)
(69.13) (0.26) (0.12) (0.10)

G 907/ 22/ 6/13846 02:36 0/13846 2/13846 6/13846 5/13846 7/13846 2/13846
2032 13846 (0.04) (0.00) (0.01) (0.04) (0.04) (0.05) (0.01)
(44.64) (0.16)

H 1217/ 29/ 17/ 03:06 1/10156 6/10156 5/10156 4/10156 10/ 3/10156
2459 10156 10156 (0.010) (0.06) (0.05) (0.04) 10156 (0.03)
(49.49) (0.29) (0.17) (0.10)

I 1299/ 5/2757 4/2757 02:29 0/2757 2/2757 2/2757 1/2757 0/2757 0/2757
2499 (0.18) (0.15) (0.00) (0.07) (0.07) (0.04) (0.00) (0.00)
(51.98)

J 1236/ 30/ 16/ 03:12 0/10727 5/10727 6/10727 3/10727 14/ 2/10727
2344 10727 10727 (0.00) (0.05) (0.06) (0.03) 10727 (0.02)
(52.73) (0.28) (0.15) (0.13)

K 1508/ 40/ 26/ 03:47 0/12527 7/12527 9/12527 3/12527 19/ 2/12527
1931 12527 12527 (0.00) (0.06) (0.07) (0.02) 12527 (0.02)
(78.09) (0.32) (0.21) (0.15)

L 1206/ 8/4877 2/4877 02:50 0/4877 0/4877 0/4877 4/4877 4/4877 0/4877
2849 (0.16) (0.04) (0.00) (0.00) (0.00) (0.08) (0.08) (0.00)
(42.33)

M 1248/ 23/9498 11/9498 03:40 0/9498 1/9498 2/9498 8/9498 11/9498 1/9498
2196 (0.24) (0.12) (0.00) (0.01) (0.02) (0.08) (0.12) (0.01)
(56.83)

N 1257/ 11/4544 5/4544 03:04 0/4544 0/4544 3/4544 4/4544 4/4544 0/4544
2458 (0.24) (0.11) (0.00) (0.00) (0.07) (0.09) (0.09) (0.00)
(51.14)

Pearson’s cor- 0.57 0.58 0.55 – 0.02 – 0.12 0.00 0.53 0.56
relation coeffi- (0.03) (0.03) (0.04) (0.95) (0.68) (0.99) (0.049) (0.04)
cient (P value) 4

Park Jae Myung et al. The effect of … Endoscopy

 Original article

▶ Table 2 (Continuation)

Endoscopist Ampulla Detection rate, n/N Mean Detection rate according to location, n/N (%)
photo- (%) exami-
docu- nation
Total Small Esopha- Cardia, Mid and Angle Antrum Duode-
menta- time
neo- neo- gus fundus low (n = 49) (n = 104) num
tion (min:
1
plasms plasms 3 2
(n = 2) and body (n = 20)
n/N (%) sec)
(n = 262) (n = 129) high (n = 52)
body
(n = 35)

Spearman’s 0.13
rank correla- (0.66)
tion coefficient
(P value)
1
Ampulla photo-documentation rate during the endoscopists’ first year of work.
2
Examination time without biopsy for 1 year (n = 27 258).
3
Defined as a lesion with a diameter ≤ 1.0 cm.
4
Correlation with the ampulla photo-documentation rate.

Downloaded by: University of Michigan. Copyrighted material.

0.35 Pearson‘s correlation coefficient (P value): 0.57 (0.0338) 0.25 Pearson‘s correlation coefficient (P value): 0.58 (0.0312)

0.30
0.20
0.25
Small neoplams (%)
Total neoplams (%)

0.15
0.20

0.15
0.10

0.10
0.05
0.05

0.00 0.00
0 50 100 0 20 40 60 80 100
a Photo-documentation rate of papilla (%) b Photo-documentation rate of papilla (%)

▶ Fig. 3 Mean rate of neoplasm detection according to photo-documentation of the ampulla during esophagogastroduodenoscopy by 14
endoscopists. a Total neoplasm detection rate. b Small neoplasm detection rate. The correlation between detection rates and ampulla photo-
documentation was calculated using the Pearson’s correlation coefficient

Furthermore, the ampulla photo-documentation rate was sig-
nificantly associated with the neoplasm detection rate even
after adjusting for age, sex, smoking, and biopsy rate. Interest-
ingly, the ampulla photo-documentation rate was positively
associated with the EGD examination time without biopsy.
Endoscopy quality indicators are used to reduce the varia-
tion in procedure performance in clinical practice and to set a
standard among endoscopists. These indicators can be used to
diagnose early neoplasms, which are associated with a good
prognosis. The quality indicators should be measurable param-
eters, so that practice can be evaluated objectively. The best
quality indicator for EGD would be to measure the incidence of
cancer or death after endoscopic examination. However, con-
sidering the relative rarity of upper GI neoplasms and the diffi-
culty in tracking long-term outcomes, surrogate short-term
yield indicators are necessary. In this study, we used the detec-
tion rate of upper GI neoplasms as a surrogate marker.
Photo-documentation of all anatomical landmarks is regard-
ed as a proof of a complete procedure. The European Society of
Gastrointestinal Endoscopy (ESGE) and the Korean Society of
Gastrointestinal Endoscopy (KSGE) recommend taking images
of a minimum of eight anatomical landmarks: two from the
esophagus, four from the stomach, and two from the duode-
num [17 – 18]. Among them, the photo-documentation of the
second part of the duodenum means that a complete examina-
tion has been performed, with the end of the endoscope posi-
tioned near the ampullary area [17]. Textbooks and articles
have described that the complete examination of the ampulla
should form part of a routine examination for early detection
of ampullary and periampullary tumors during screening and

Park Jae Myung et al. The effect of … Endoscopy

 ▶ Table 3 Characteristics of participants and esophagogastroduodenal lesions detected on upper endoscopy stratified by low and high ampulla
observer rates.

Total endoscopists Low ampulla observers High ampulla observers P value1

(N = 14) (n = 7) (n = 7)

Participants 111 962 54 056 57 906

Age, mean ± SD, years 49.7 ± 10.3 49.4 ± 10.2 49.9 ± 10.4 < 0.001

Male, n/N (%) 64 593/111 962 (57.7) 31 156/54 056 (57.6) 33 437/57 906 (57.7) 0.72

BMI, mean ± SD, kg/m 2 23.6 ± 3.1 23.6 ± 3.1 23.7 ± 3.1 0.14

Smoking, n/N (%) 17 706/56 696 (31.2) 8011/25 652 (31.2) 9695/31 044 (31.2) 0.99

Alcohol, n/N (%) 50 056/76 918 (65.1) 24 199/37 109 (65.2) 25 857/39 809 (65.0) 0.45

First-degree family history of upper GI cancer, 5411/56 696 (9.5) 2426/25 652 (9.5) 2985/31 044 (9.6) 0.80
n/N (%)

Helicobacter pylori infection, n/N (%) 5564/56 695 (9.8) 2548/25 651 (9.9) 3016/31 044 (9.7) 0.39

No. of EGDs performed per day, mean ± SD 9.4 ± 1.1 9.5 ± 0.6 9.3 ± 1.5 0.77

Lesions

Downloaded by: University of Michigan. Copyrighted material.

Total neoplasm, n/N (%) 262/111 962 (0.23) 109/54 056 (0.20) 153/57 906 (0.26) 0.03

Dysplasia, n/N (%) 157/111 962 (0.14) 58/54 056 (0.11) 99/57 906 (0.17) 0.004

Early gastric cancer, n/N (%) 79/111 962 (0.07) 38/54 056 (0.07) 41/57 906 (0.07) 0.98

Advanced gastric cancer, n/N (%) 15/111 962 (0.01) 9/54 056 (0.02) 6/57 906 (0.01) 0.36

Esophageal cancer, n/N (%) 2/111 962 (< 0.01) 1/54 056 (< 0.01) 1/57 906 (< 0.01) > 0.99

Lymphoma, n/N (%) 6/111 962 (0.01) 2/54 056 (< 0.01) 4/57 906 (0.01) 0.69
2
Others, n/N (%) 6/111 962 (0.01) 2/54 056 (< 0.01) 4/57 906 (0.01) 0.69
3
Small lesions, n/N (%) 129/111 962 (0.12) 48/54 056 (0.09) 81/57 906 (0.14) 0.01

Gastric adenoma/carcinoma, n/N (%) 233/111 962 (0.21) 99/54 056 (0.18) 134/57 906 (0.23) 0.08

BMI, body mass index; EGD, esophagogastroduodenoscopy; UGI, upper gastrointestinal.

1
P values were calculated using chi-squared test, Fisher’s exact test or independent t test.
2
Neuroendocrine tumor (n = 3) and Barrett’s dysplasia (n = 3).
3
≤ 1 cm in diameter.

surveillance of high risk patients, as well as for diagnosis of
other lesions in the ampullary region [12 – 13, 19 – 21]. Al-
though photo-documentation of the ampulla is not an obliga-
tion in the minimal requirement of the Korean guideline, Kor-
ean endoscopists have been trained during their fellowship
period at their respective Korean training centers to photo-
document the ampulla if possible. This has also been consolida-
ted in regular educational conferences organised by the KSGE
[22 – 23].
Documentation of other upper GI images, including those
deemed part of a fully documented examination, might also
have helped with correlating performance with lesion detec-
tion. The British Society of Gastroenterology recommends that
the presence of an inlet patch should be photo-documented
[8]. Among the recommendations, detection of an inlet patch
can be used as a surrogate maker of a thorough examination
of the esophagus, because an inlet patch can be easily over-
looked when rapidly withdrawing the endoscope. We also ob-
served a similar finding that photo-documentation of the am-
pulla was closely related to the EGD observation time.
Our study showed that ampulla photo-documentation was
significantly associated with the neoplasm detection rate for
each endoscopist. In further analysis, the high ampulla obser-
vers showed a significantly higher detection rate of small neo-
plasms. This could imply that the high ampulla observers exam-
ined the upper GI tract more carefully than the low ampulla ob-
servers. However, there were not enough neoplasm events to
confirm the correlation between ampulla detection rate and lo-
cation, except the antrum and the duodenum. Nevertheless,
the neoplasm detection rate tended to be higher in the high
compared with the low ampulla observers. Considering the
low incidence of neoplasm in the upper GI tract, the total neo-
plasm detection rate would be more appropriate as an outcome
parameter in EGD quality indicator study. Further prospective
studies should be carried out to investigate the causality of am-
pulla observation to upper GI neoplasm detection.
We found one clue to explain why the high ampulla obser-
vers showed higher detection of upper GI neoplasms. As shown
in ▶ Fig. 4, the high ampulla observers performed EGD with a
longer examination time. In our previous study, we found that

Park Jae Myung et al. The effect of … Endoscopy

Original article

the EGD examination time is an independent factor for the

04:00 Pearson‘s correlation coefficient (P value): 0.55 (0.0413) higher quality of EGD [10]. This also suggests that the endos-
copists with longer examination times paid the greater atten-
tion that is needed to detect neoplasms during EGD. The am-
03:30
pulla photo-documentation rate might be associated with the
Examination time (min)

carefulness or concentration of the endoscopist during EGD,

03:00 because the examination time was significantly associated
with this parameter.
We sought to investigate whether the higher ampulla photo-
02:30
documentation rate could be related to endoscopy proficiency.
There have been no indicators to measure this. Therefore, we
02:00 evaluated the experience of the endoscopists after board certi-
fication of GI specialty. However, we found no association be-
tween the experience of the endoscopist and the ampulla pho-
01:30
to-documentation rate. This might be related to the extensive
0 20 40 60 80 100
Photo-documentation rate of papilla (%)
experience gained during the GI fellowship, because the study
endoscopists had to perform more than 1000 EGDs to become
certified during each training period [24]. Therefore, we need
▶ Fig. 4 Mean esophagogastroduodenoscopy (EGD) examination to address whether ampulla photo-documentation could be
time according to photo-documentation of ampulla for 14 endos-

Downloaded by: University of Michigan. Copyrighted material.

copists. a Total neoplasm detection rate. b Small neoplasm detec-
tion rate. The EGD examination time was assessed during endos-
copists’ first year of working at the health promotion center for
procedures in which no biopsies were taken. The correlation be-
tween rates of EGD examination time and ampulla photo-docu-
mentation rate was calculated using the Pearson’s correlation
coefficient.
related to how well endoscopists are trained in EGD during the
fellowship program.
The upper GI working group of the ESGE proposed research
priorities to assess whether reaching any specific anatomical
landmark yielded a better rate of diagnosis [9]. This group
speculated on whether ampulla photo-documentation could
serve as an auditable performance measure for the complete-
ness of the procedure in a patient referred for an EGD. The am-

▶ Table 5 Estimating factors associated with detection of upper gastrointestinal neoplasms.

Total, n (%) Upper GI neoplasms, n (%) OR (95 %CI) P Adjusted OR P

N = 111 962 (95 %CI)*
Absence Presence

Age

▪ ≥ 31 – < 40 years 19 925 19 914 11 (0.06) Reference Reference

(17.80) (99.94)

▪ ≥ 40 – < 60 years 73 299 73 172 127 (0.17) 3.14 (1.70 – 5.82) < 0.001 3.04 (1.64 – 5.64) < 0.001
(65.47) (99.83)

▪ ≥ 60 – < 80 years 18 261 18 150 111 (0.61) 11.07 (5.96 – < 0.001 11.21 (6.01 – 20.89) < 0.001
(16.31) (99.39) 20.58)

▪ ≥ 80 years 477 (0.43) 464 (97.27) 13 (2.73) 50.72 (22.61 – < 0.001 50.54 (22.36 – < 0.001
113.81) 114.22)

Sex

▪ Male 64 593 64 395 198 (0.31) 2.27 (1.71 – 3.01) < 0.001 1.95 (1.45 – 2.63) < 0.001
(57.69) (99.69)

▪ Female 47 369 47 305 64 (0.14) Reference Reference

(42.31) (99.86)

Smoking

▪ No 38 990 38 909 81 (0.21) Reference Reference

(34.82) (99.79)

▪ Yes 17 706 17 639 67 (0.38) 1.83 (1.32 – 2.52) < 0.001 1.60 (1.13 – 2.26) 0.01
(15.81) (99.62)

▪ Unknown 55 266 55 152 114 (0.21) 0.99 (0.75 – 1.32) 0.96 1.24 (0.92 – 1.66) 0.16
(49.36) (99.79)

Park Jae Myung et al. The effect of … Endoscopy

 ▶ Table 5 (Continuation)

Total, n (%) Upper GI neoplasms, n (%) OR (95 %CI) P Adjusted OR P

N = 111 962 (95 %CI)*
Absence Presence

Alcohol

▪ No 26 862 26 798 64 (0.24) Reference

(23.99) (99.76)

▪ Yes 50 056 49 931 125 (0.25) 1.05 (0.78 – 1.42) 0.76

(44.71) (99.75)

▪ Unknown 35 044 34 971 73 (0.21) 0.87 (0.63 – 1.22) 0.43

(31.30) (99.79)

Family history

▪ No 48 958 48 832 126 (0.26) Reference

(43.73) (99.74)

▪ First-degree relative 5411 (4.83) 5394 (99.69) 17 (0.31) 1.22 (0.74 – 2.03) 0.44

▪ Other relatives 2327 (2.08) 2322 (99.79) 5 (0.21) 0.84 (0.34 – 2.04) 0.69

Downloaded by: University of Michigan. Copyrighted material.

▪ Unknown 55 266 55 152 114 (0.21) 0.80 (0.62 – 1.03) 0.09
(49.36) (99.79)

Endoscopist

▪ Fast (≤ 3 minutes) 62 120 61 997 123 (0.20) Reference

(55.48) (99.80)

▪ Slow (> 3 minutes) 49 842 49 703 139 (0.28) 1.41 (1.11 – 1.80) 0.01
(44.52) (99.72)

Biopsy rate

▪ Low (< median, < 17.9) 55 148 55 035 113 (0.20) Reference Reference
(49.26) (99.80)

▪ High (≥ median, ≥ 17.9) 56 814 56 665 149 (0.26) 1.28 (1.00 – 1.64) 0.048 1.39 (1.09 – 1.79) 0.01
(50.74) (99.74)

Experience

▪ ≤ 1 year 40 276 40 178 98 (0.24) Reference

(35.97) (99.76)

▪ > 1 year 71 686 71 522 164 (0.23) 0.94 (0.73 – 1.21) 0.63
(64.03) (99.77)

Ampulla photo-documentation

▪ Low (< 51.98 %) 54 056 53 947 109 (0.20) Reference Reference

(48.28) (99.80)

▪ High (≥ 51.98 %) 57 906 57 753 153 (0.26) 1.31 (1.03 – 1.68) 0.03 1.33 (1.03 – 1.70) 0.03
(51.72) (99.74)

CI, confidence interval; GI, gastrointestinal, OR, odds ratio

* Multiple logistic regression analysis: variables adjusted for age, sex, smoking, biopsy rate, experience of endoscopists.

pulla is located in the most distal portion of the reach of a rou-
tine EGD, analogous to the cecum in colonoscopy. In colono-
scopic examinations, the intubation rate of the cecum, the
most distal site from the insertion inlet, has been used as a
quality indicator [25]. Cecal intubation cannot be achieved in
all colonoscopies, and therefore this measure constitutes an
important quality indicator [26]. However, reaching the second
portion of the duodenum is much easier than intubating the ce-
cum, and this fact might have attributed to the lack of associa-
tion between endoscopist experience and ampulla photo-docu-
mentation in the current study. This means that visualization of
the ampulla is not associated with the examiner’s proficiency.
Based on the current results, we calculated the optimal cut-
off value using the Youden index of receiver operating charac-
teristic curve analysis. The result was then analyzed with uni-
variate and multivariate logistic regression ( ▶ Supplementary

Park Jae Myung et al. The effect of … Endoscopy

 Original article
Table e6 and ▶ Table e7, available online). Therefore, we would
like to recommend a minimal threshold of 50 % for the ampulla
photo-documentation rate when auditing EGD performance
measure.
This study has several limitations, such as its retrospective
design and therefore no allocation for randomization. There-
fore, it is difficult to confer the present result to the standard
rate of ampulla photo-documentation for complete EGD eva-
luations. To overcome this limitation, we need further studies
including prospective clinical trials. In addition, the absence of
ampulla photo-documentation may not indicate that the
endoscopist did not observe the ampulla at all. Finally, the
number of participating endoscopists was small.
In conclusion, our data showed significant correlation be-
tween neoplasm detection and ampulla photo-documentation.
Guidelines should consider recommending this to ensure high
standard procedures. Further larger prospective studies are
warranted to validate our results.
Acknowledgments
The statistical consultation was supported by a grant from the
Korea Health Technology R&D Project through the Korea Health
Industry Development Institute (KHIDI), funded by the Ministry
of Health & Welfare, Republic of Korea (grant number:
HI14C1062).
Competing interests
None
References
[1] Jun JK, Choi KS, Lee HY et al. Effectiveness of the Korean National
Cancer Screening Program in reducing gastric cancer mortality. Gas-
troenterology 2017; 152: 1319 – 1328.e7
[2] Chadwick G, Groene O, Riley S et al. Gastric cancers missed during
endoscopy in England. Clin Gastroenterol Hepatol 2015; 13: 1264 –
1270.e1
[3] Yalamarthi S, Witherspoon P, McCole D et al. Missed diagnoses in pa-
tients with upper gastrointestinal cancers. Endoscopy 2004; 36:
874 – 879
[4] Menon S, Trudgill N. How commonly is upper gastrointestinal cancer
missed at endoscopy? A meta-analysis Endosc Int Open 2014; 2:
E46 – 50
[5] Chadwick G, Groene O, Hoare J et al. A population-based, retrospec-
tive, cohort study of esophageal cancer missed at endoscopy. Endos-
copy 2014; 46: 553 – 560
[6] Park WG, Shaheen NJ, Cohen J et al. Quality indicators for EGD. Am J
Gastroenterol 2015; 110: 60 – 71
[7] Rex DK, Schoenfeld PS, Cohen J et al. Quality indicators for colonos-
copy. Am J Gastroenterol 2015; 110: 72 – 90
[8] Beg S, Ragunath K, Wyman A et al. Quality standards in upper gas-
trointestinal endoscopy: a position statement of the British Society of
Gastroenterology (BSG) and Association of Upper Gastrointestinal
Surgeons of Great Britain and Ireland (AUGIS). Gut 2017; 66: 1886 –
1899
[9] Bisschops R, Areia M, Coron E et al. Performance measures for upper
gastrointestinal endoscopy: a European Society of Gastrointestinal
Endoscopy (ESGE) Quality Improvement Initiative. Endoscopy 2016;
48: 843 – 864
[10] Park JM, Huo SM, Lee HH et al. Longer observation time increases
proportion of neoplasms detected by esophagogastroduodenoscopy.
Gastroenterology 2017; 153: 460 – 469.e1
[11] Teh JL, Tan JR, Lau LJ et al. Longer examination time improves detec-
tion of gastric cancer during diagnostic upper gastrointestinal
endoscopy. Clin Gastroenterol Hepatol 2015; 13: 480 – 487.e2
[12] Hew WY, Joo KR, Cha JM et al. Feasibility of forward-viewing upper
endoscopy for detection of the major duodenal papilla. Dig Dis Sci
2011; 56: 2895 – 2899
[13] Choi YR, Han JH, Cho YS et al. Efficacy of cap-assisted endoscopy for
routine examining the ampulla of Vater. World J Gastroenterol 2013;
19: 2037 – 2043
[14] Clarke JO, Giday SA, Magno P et al. How good is capsule endoscopy for
Downloaded by: University of Michigan. Copyrighted material.
detection of periampullary lesions? Results of a tertiary-referral cen-
ter Gastrointest Endosc 2008; 68: 267 – 272
[15] Kong H, Kim YS, Hyun JJ et al. Limited ability of capsule endoscopy to
detect normally positioned duodenal papilla. Gastrointest Endosc
2006; 64: 538 – 541
[16] Schlemper RJ, Riddell RH, Kato Y et al. The Vienna classification of
gastrointestinal epithelial neoplasia. Gut 2000; 47: 251 – 255
[17] Rey JF, Lambert R. ESGE recommendations for quality control in gas-
trointestinal endoscopy: guidelines for image documentation in up-
per and lower GI endoscopy. Endoscopy 2001; 33: 901 – 903
[18] Cha JM. [Quality improvement of gastrointestinal endoscopy in Korea:
past, present, and future]. The Korean Journal of Gastroenterology =
Taehan Sohwagi Hakhoe chi 2014; 64: 320 – 332
[19] Kim MH, Lee SK, Seo DW et al. Tumors of the major duodenal papilla.
Gastrointest Endosc 2001; 54: 609 – 620
[20] Block B, Schachschal G, Schmidt H. Endoscopy of the upper GI tract: a
training manual. New York: Thieme; 2004
[21] Canard JM, Letard JC, Palazzo L et al. Gastrointestinal endoscopy in
practice. London: Churchill Livingstone; 2011
[22] Kim Y, Kim Y, Chun H. [Standard procedure of upper endoscopy: in-
sertion and observation]. In: Spring Seminars of Korean Society of
Gastrointestinal Endoscopy. Seoul, Korea: The Korean Society of Gas-
trointestinal Endoscopy; 2001: 39 – 44
[23] Chang YW. [Photodocumentation and Data Storage]. In: Spring
Seminars of Korean Society of Gastrointestinal Endoscopy. Seoul,
Korea: The Korean Society of Gastrointestinal Endoscopy; 2003: 76 –
78
[24] Moon HS, Choi EK, Seo JH et al. Education and Training Guidelines for
the Board of the Korean Society of Gastrointestinal Endoscopy. Clin
Endosc 2017; 50: 345 – 356
[25] Rex DK, Petrini JL, Baron TH et al. Quality indicators for colonoscopy.
Gastrointest Endosc 2006; 63: S16 – 28
[26] Baxter NN, Sutradhar R, Forbes SS et al. Analysis of administrative
data finds endoscopist quality measures associated with postcolono-
scopy colorectal cancer. Gastroenterology 2011; 140: 65 – 72
Park Jae Myung et al. The effect of … Endoscopy