Introduction
Colorectal cancer (CRC) represents 13.1% of all malignant tumors and is the second
leading cause of cancer mortality worldwide [1]. The implementation of opportunistic colorectal cancer screening programs in Austria
has led to decreases in incidence and mortality of colorectal cancer over the last
decades [2]. In most European countries, including Austria, screening colonoscopy (SC) is the
gold standard to prevent colorectal cancer by the detection and removal of precursor
lesions [3]. Recent studies showed that colonoscopies are more effective in the distal than
in the proximal colon [4]
[5]. Therefore, the adenoma detection rate (ADR) is the most important quality indicator
in SC. Studies from Kaminski et al. [6] and Corley et al. [7] have shown that a higher ADR is associated with a lower risk of interval cancer.
Because of these studies, guidelines recommend an overall ADR ≥25%, a minimum of 30%
for men and 20% for women [8]
[9].
In the last 20 years, serrated polyps, which are characterized by a sawtooth-like
appearance in the epithelium, have been increasingly recognized. According to the
World Health Organization (WHO) [10] serrated polyps are subclassified into hyperplastic polyps (HPs), sessile serrated
lesions (SSLs) and traditional serrated adenomas (TSAs). In contrast to conventional
adenomas, serrated polyps develop via an alternative pathway, called the serrated
pathway. Furthermore, serrated polyps account for approximately 20% to 35% of colorectal
cancers and may be responsible for a high number of interval cancers [11]
[12]
[13]
[14]. The serrated pathway is characterized by a CpG island methylator phenotype (CIMP),
BRAF mutation, and microsatellite instability (MSI) or microsatellite stable (MSS) [11]
[15]. A study by Arain et al. [13] about interval cancer has shown that serrated polyps are associated with a significantly
higher incidence of CIMP (57% vs. 33%) and MSI (29% vs.11%) and preferentially occur
in the proximal colon (63% vs 39%) in contrast to non-interval cancers. For these
reasons serrated polyps are getting more and more attention during colonoscopies and
a benchmark for serrated polyps should be recommended.
Based on their flat morphology, serrated polyps and especially SSLs are difficult
to detect during colonoscopy. Further, SSLs are typically larger than 5 mm, often
covered by a yellow mucous cap, and located proximal [16]
[17]. In contrast, HPs are the most common, smaller than SSLs, distally located, and
have a low malignant potential. TSAs are rare, mostly distally located, and pedunculated
or sessile [16]
[17].
Because there is currently no established benchmark for identifying serrated polyps,
we have created three separate detection rates to address this issue. The first rate,
known as the serrated polyp detection rate (SDR), encompasses all serrated polyps
(including HPs, SSLs, and TSAs) regardless of their size or location. The second rate,
the serrated polyp and advanced detection rate (SPADR), includes all serrated polyps
with malignant potential (SSLs and TSAs). Finally, the comprehensive serrated polyp
detection rate (CRSDR) identifies all SSLs and TSAs, as well as HPs larger than 10
mm in any area of the colon or more than 5 mm proximal to the sigmoid [18]. The aim of the study was to investigate whether the rates are appropriate as additional
quality indicators for screening and surveillance colonoscopies.
Patient and methods
This retrospective analysis of prospective collected data included 212,668 colonoscopies
between 2012 and September 2018 performed by 290 endoscopists in the Austrian quality
assurance program.
Our database of bowel preparation, based on the Aronchick scale (excellent, good,
fair, poor, poor only in the right colon, not sufficient) was implemented in 2012/2013
and until 2014, we used the term “serrated” for both sessile and traditional serrated
polyps.
In brief, in 2007, the OEGGH (Austrian Society of Gastroenterology and Hepatology)
in cooperation with the HBV (Association of Austrian Social Security Institutions)
and ÖKH (Austrian Cancer Aid) launched the project “Qualitätszertifikat Darmkrebsvorsorge”
(Austrian Certificate of Quality Colonoscopy Screening). Forty-three percent of all
Austrian endoscopists, consisting of specialists in internal medicine as well as surgeons
participate in the project. Endoscopists can participate in the project if they fulfill
the conditions and quality standards of the OEGGH, including information and consulting,
offer premedication and sedation, complete video colonoscopy (cecal intubation), postoperative
care, electronic documentation, and review of the results. Approximately 48% of participating
endoscopists use high-definition endoscopes. In addition, a minimum of 200 complete
colonoscopies (including cecal intubation) and 50 polypectomies under supervision
as well as a minimum of 100 complete colonoscopies and 10 polypectomies per year are
required. As part of the project, an annual hygiene control for endoscopes of participants
is required. Further information about this project has been published in prior studies
[19]
[20]
[21].
All men and women aged 50 to 100 years undergoing screening or surveillance colonoscopy
between 2012 and September 2018 were included in our study. Furthermore, endoscopists
had to perform more than 30 colonoscopies during the study period. Also, data from
patients older than age 30 years who obtained colonoscopies were included. The study
was approved by the Ethics Committee of the Medical University of Vienna as voting
number 1510/2017.
Definitions of quality indicators
The endoscopists' ADR was defined as examinations during which at least one conventional
adenoma (tubular, tubulovillous, villous) was found divided by the total number of
endoscopies.
We calculated the SDR as the number of examinations in which at least one serrated
polyp (HP, SSL, TSA) was found divided by the total number of colonoscopies performed.
The SPADR was defined as the number of examinations in which at least one SSL or TSA
was found divided by the total number of performed colonoscopies.
The CRSDR was defined as the number of examinations in which at least one SSL, TSA,
or HP >10 mm anywhere in the colon or HP > 5 mm proximal to the sigmoid was found
divided by the number of performed colonoscopies.
The sessile serrated lesion detection rate was defined as the number of examinations
in which at least one SSL was found divided by the total number of colonoscopies.
Statistical analysis
Categorical variables are described with absolute and relative frequencies. For continuous
variables, the arithmetic mean and standard deviation (SD) was used. Detection rates
were measured, and Spearman’s rank-order correlation was used to assess whether there
was a correlation between ADR and SDR, SPADR, SSLDR or CRSDR. Furthermore, endoscopists
were grouped into ADR <25% and ≥25% and a Mann-Whitney U-test was used to evaluate
differences in the SDR, SPADR, SSLDR, and CRSDR between the groups with low and high
ADRs. Statistical significance was defined as P ≤ 0.05. Statistical analysis was performed using IBM SPSS Statistics version 25.0,
Microsoft Excel.
Results
A total of 212,668 colonoscopies (2012: n=25.459; 2013: n=29.247; 2014: n=33.969;
2015: n=32.990; 2016: n=32.002; 2017: n=35.800; 2018: n=23.201;) performed by 290
physicians between 2012 and September 2018 were included. Of the patients, 107,725
(50.7%) were female and the mean age was 63.91 years (SD 9.40 years) ([Table 1]).
Table 1 Patient characteristics in all colonoscopies and for ADRs <25% and ≥25%.
|
Patient characteristic
|
All colonoscopies
n=212.668
|
ADR <25%
n=145,406
|
ADR ≥25%
n=67,262
|
|
ADR, adenoma detection rate; SD, standard deviation.
|
|
Female – no (%)
|
107.725 (50.7)
|
74.228 (51.0)
|
33.497 (49.8)
|
|
Male – no (%)
|
104.943 (49.3)
|
71.178 (49.0)
|
33.765 (50.2)
|
|
Mean age (SD)
|
63.91 (9.40)
|
64.05 (9.51)
|
63.75 (9.15)
|
|
Cecal intubation (%)
|
206.303 (97.0)
|
140.315 (96.5)
|
65.988 (98.1)
|
|
Sedation (%)
|
190.163 (89.4)
|
129.236 (88.9)
|
60.924 (90.6)
|
The endoscopists (n=290) were separated into office-based internists (44.5%; n=129),
clinic-based internists (19.6%; n=57), clinic-based surgeons (4.5%; n=13), office-based
surgeons (28.6%; n=83), and interdisciplinary endoscopists (2.8%; n=8). The clinic-based
surgeons performed a total of 4,008 colonoscopies (1.9%) and the office-based surgeons
performed 70,530 (33.2%). Conversely, the clinical-based internists performed 41,566
colonoscopies (19.5%), the office-based internists performed 92,589 (43.5%), and the
interdisciplinary endoscopists performed 3,975 (1.9%).
The cecum was reached in 97.0% of all cases (n=206,303) and in 89.4% (n=190,163),
sedation was used ([Table 1]). In 35.2% (n=67,315), 48.8% (n=93,274), and 11.9% (n=22,739) of all colonoscopies
the bowel preparation was reported to be excellent, good, and fair, respectively.
In contrast, 4.1% of all bowel preparation (n=7,775) was described as poor, poor only
in the right colon, or not sufficient ([Table 1]). Polyps and adenomas were found in 40.9% (n=86,970) and 23.7% (n=50,450) of all
colonoscopies, respectively. Polypectomy was performed during 83,979 screening colonoscopies
(39.5%). They were performed in 66.34% of patients (n=55,591) with forceps, in 20.25%
(n=17,079) with snare, and in 13.41% (n=11.309) with both forceps and snare.
Quality parameters
ADR
At least one conventional adenoma was found in 21.5% of screening colonoscopies
(n=45,784) and the endoscopists’ overall mean ADR was 21.78% (SD 9.27) ([Table 2]). The prevalence of conventional adenomas was 26.9% (n=28,256) for men and 16.7%
(n=17,528) for women. Patients with conventional adenomas had a mean age of 65.98
years
(SD 9.26 years). At 26.92%, (SD 11.25) the mean male ADR was significantly higher
than the
mean female ADR at 16.59 (SD 7.90) (P <0.001) ([Table 3]). One hundred endoscopists had an ADR ≥ 25% with a mean ADR of 31.96% (SD 4.88).
In contrast, the mean ADR for the 190 endoscopists with an ADR < 25% was 16.43% (SD
5.93) ([Table 4]). Regarding to the time trend for the overall ADR, we observed no difference
between 2012 and 2017 (22.03% vs 22.59%, P=0.534) ([Fig. 1]). We found the same results for men 27.16% in 2012 vs. 28.06% in 2017; P=0.693; [Fig. 2] and women (16.09% 2012 vs. 17.16% in 2017; P=0,686)
([Table 5] and [Fig. 3]). The highest overall ADR was found in the group aged >90 years and for men and
women in the group aged 80 to 89 years.
Table 2 Endoscopist characteristics by age group.
|
Age
|
|
|
|
<50
|
50–59
|
60–69
|
70–79
|
80–89
|
>90
|
P value
|
|
ADR, adenoma detection rate; SDR, serrated detection rate; SPADR, serrated adenoma
detection rate; CRSDR, clinically serrated polyp detection rate; SD, standard deviation.
|
|
Overall
|
|
No. SC
|
3.513
|
65.658
|
74.718
|
50.891
|
16.753
|
1.135
|
|
|
ADR % (SD)
|
12.06 (20.00)
|
16.89 (8.30)
|
21.71 (10.14)
|
27.15 (12.43)
|
29.89 (17.51)
|
30.36 (32.74)
|
<0.001
|
|
SDR % (SD)
|
19.52 (24.78)
|
20.86 (12.09)
|
22.65 (12.57)
|
21.01 (12.20)
|
15.73 (12.75)
|
14.03 (23.91)
|
<0.001
|
|
SPADR % (SD)
|
5.46 (18.07)
|
2.64 (3.38)
|
2.15 (2.70)
|
1.88 (2.46)
|
1.46 (6.83)
|
0.70 (3.98)
|
<0.001
|
|
CRSDR % (SD)
|
7.26 (18.85)
|
4.39 (4.63)
|
3.89 (3.97)
|
3.42 (3.39)
|
2.16 (7.00)
|
1.74 (7.57)
|
<0.001
|
|
Male
|
|
No. SC
|
1.661
|
32.894
|
36.456
|
24.869
|
8.487
|
576
|
|
|
ADR % (SD)
|
15.48 (23.86)
|
20.85 (10.54)
|
27.18 (13.22)
|
33.42 (15.08)
|
36.75 (23.40)
|
36.28 (38.17)
|
<0.001
|
|
SDR % (SD)
|
20.27 (26.58)
|
22.94 (13.19)
|
24.10 (13.74)
|
22.21 (13.22)
|
17.09 (16.57)
|
16.34 (29.71)
|
<0.001
|
|
SPADR % (SD)
|
4.53 (15.61)
|
2.44 (3.33)
|
2.25 (4.28)
|
1.82 (2.93)
|
1.16 (3.88)
|
0.80 (5.46)
|
<0.001
|
|
CRSDR % (SD)
|
6.80 (18.60)
|
4.34 (4.58)
|
4.08 (5.40)
|
3.57 (4.28)
|
2.02 (4.55)
|
1.98 (9.18)
|
<0.001
|
|
Female
|
|
No. SC
|
1.852
|
32.764
|
38.262
|
26.022
|
8.266
|
559
|
< 0.001
|
|
ADR % (SD)
|
2.30 (8.00)
|
12.52 (7.68)
|
16.74 (9.48)
|
21.00 (12.70)
|
23.40 (18.66)
|
19.80 (28.52)
|
< 0.001
|
|
SDR % (SD)
|
3.45 (10.07)
|
18.51 (12.46)
|
21.11 (12.96)
|
19.98 (13.79)
|
14.16 (14.34)
|
12.58 (25.40)
|
< 0.001
|
|
SPADR % (SD)
|
0.42 (1.89)
|
2.84 (4.49)
|
2.16 (3.26)
|
2.02 (3.52)
|
0.24 (2.14)
|
0.63 (4.47)
|
< 0.001
|
|
CRSDR % (SD)
|
1.42 (7.96)
|
4.33 (5.85)
|
3.77 (4.75)
|
3.32 (4.34)
|
1.80 (6.75)
|
1.68 (9.69)
|
< 0.001
|
Table 3 Differences between men and women according to age and detection rates.
|
|
All
|
Men
|
Women
|
P value
|
|
SC, screening colonoscopy; TSA, traditional serrated adenoma; SSA/P, sessile serrated,
adenoma/polyp.
|
|
SC
|
212,668
|
104,943 (49.3%)
|
107,725 (50.7%)
|
|
|
Age
|
63.91 (9.40)
|
63.91 (9.43)
|
665.68 (9.37)
|
=0.920
|
|
Conventional adenomas
|
45.784
|
28.256 (61.72%)
|
17.528 (38.28%)
|
<0.001
|
|
Age (yr)
|
65.98 (9.26)
|
65.91 (9.25)
|
66.08 (9.26)
|
=0.500
|
|
Detection rate
|
21.78% (9.27)
|
26.92 (11.25)
|
16.59 (7.90)
|
<0.001
|
|
Serrated adenoma (TSA or SSA/P)
|
4.666
|
2.293
|
2.373
|
=0.543
|
|
Age (yr)
|
63.61 (8.97)
|
63.63 (9.00)
|
63.58 (8.94)
|
=0.823
|
|
Detection rate
|
2.19 (2.49)
|
2.10 (2.38)
|
2.30 (3.09)
|
=0.815
|
|
Serrated polyps
|
43.475
|
23.052
|
20.423
|
<0.001
|
|
Age (yr)
|
63.81 (8.78)
|
63.64 (8.81)
|
64.01 (8.74)
|
<0.001
|
|
Detection rate
|
21.08% (11.44)
|
22.72 (11.90)
|
19.43 (11.69)
|
<0.001
|
|
Clinically significant serrated polyp
|
8.124
|
4.197
|
3.927
|
<0.001
|
|
Age (yr)
|
64.03 (9.04)
|
64.13 (9.01)
|
63.93 (9.07)
|
=0.313
|
|
Detection rate
|
3.81% (3.40)
|
3.89 (3.53)
|
3.73 (3.96)
|
=0.062
|
Table 4 Endoscopist characteristics for all colonoscopies and with ADR <25% and ≥25%.
|
Endoscopists characteristic
|
All colonoscopies
n=212,668
|
ADR <25%
|
ADR ≥25%
|
OR
|
P value
|
|
ADR, adenoma detection rate; M, male; F, female; SDR, serrated detection rate; SPADR,
serrated adenoma, detection rate; CRSDR, clinically serrated polyp detection rate;
SD, standard deviation.
|
|
Number of endoscopists
|
290
|
190
|
100
|
|
|
|
Mean ADR % (SD)
|
21.78% (9.27)
|
16.43 (5.93)
|
31.96 (4.88)
|
2.20 (2.06–2.35)
|
<0.001
|
|
Mean M-ADR % (SD)
|
26.92 (11.25)
|
20.72 (7.90)
|
38.86 (5.71)
|
2.63 (2.46–2.81)
|
<0.001
|
|
Mean F-ADR % (SD)
|
16.59 (7.90)
|
12.20 (4.90)
|
24.94 (5.37)
|
2.54 (2.37–2.73)
|
<0.001
|
|
Mean SDR % (SD)
|
21.08% (11.44)
|
16.15 (8.15)
|
30.43 (10.95)
|
2.25 (2.10–2.42)
|
<0.001
|
|
Mean M-SDR % (SD)
|
22.72 (11.90)
|
17.87 (8.80)
|
31.93 (11.65)
|
2.12 (1.98–2.27)
|
<0.001
|
|
Mean F-SDR % (SD)
|
19.43 (11.69)
|
14.45 (8.30)
|
28.89 (11.38)
|
2.44 (2.29–2.61)
|
<0.001
|
|
Mean SPADR % (SD)
|
2.19 (2.49)
|
1.49 (1.87)
|
3.52 (2.96)
|
5.42 (4.27–6.88)
|
<0.001
|
|
Mean M-SPADR % (SD)
|
2.10 (2.38)
|
1.56 (1.87)
|
3.14 (2.86)
|
2.41 (1.92–3.03)
|
<0.001
|
|
Mean F-SPADR % (SD)
|
2.30 (3.09)
|
1.43 (2.21)
|
3.95 (3.78)
|
2.97 (2.43–3.63)
|
<0.001
|
|
Mean CRSDR % (SD)
|
3.81% (3.40)
|
2.68 (2.47)
|
5.96 (3.87)
|
4.06 (3.37–4.91)
|
<0.001
|
|
Mean M-CRSDR % (SD)
|
3.89 (3.53)
|
2.92 (2.62)
|
5.71 (4.26)
|
2.71 (2.25–3.27)
|
<0.001
|
|
Mean F-CRSDR % (SD)
|
3.73 (3.96)
|
2.43 (2.82)
|
6.20 (4.60)
|
4.48 (3.72–5.40)
|
<0.001
|
Fig. 1 Detection rate per year.
Fig. 2 Detection rate per year in male patients.
Table 5 Trends in detection rates per year.
|
|
2012
|
2013
|
2014
|
2015
|
2016
|
2017
|
2018
|
P value (2012–2017)
|
|
F, female; M, male; ADR, adenoma detection rate; SDR, serrated detection rate; SPADR,
serrated polyp detection rate; CRSDR, clinically serrated polyp detection rate; SD,
standard deviation.
|
|
No. SC
|
25,459
|
29,247
|
33,969
|
32,990
|
32,002
|
35,800
|
23,201
|
|
|
ADR % (SD)
|
22.03 (13.29)
|
20.75 (10.48)
|
21.99 (12.39)
|
22.29 (12.48)
|
22.93 (12.20)
|
22.59 (10.01)
|
22.16 (13.22)
|
0.534
|
|
F-ADR % (SD)
|
16.09 (12.52)
|
15.21 (10.15)
|
17.20 (13.28)
|
15.04 (10.15)
|
17.09 (11.50)
|
17.16 (9.94)
|
16.56 (12.85)
|
0.686
|
|
M-ADR % (SD)
|
27.16 (15.77)
|
26.04 (13.93)
|
25.15 (14.71)
|
27.18 (15.76)
|
27.85 (14.56)
|
28.06 (12.95)
|
27.27 (17.02)
|
0.693
|
|
SDR % (SD)
|
19.20 (12.57)
|
19.43 (12.73)
|
20.10 (12.71)
|
21.35 (13.18)
|
22.48 (13.94)
|
21.90 (13.18)
|
21.28 (14.85)
|
0.213
|
|
F-SDR % (SD)
|
17.32 (13.93)
|
17.30 (13.36)
|
17.69 (13.04)
|
19.74 (13.71)
|
20.48 (14.61)
|
20.51 (14.09)
|
19.34 (16.08)
|
0.246
|
|
M-SDR % (SD)
|
20.95 (13.86)
|
21.31 (14.44)
|
22.39 (15.28)
|
23.08 (14.72)
|
24.06 (15.29)
|
23.47 (14.78)
|
22.59 (17.08)
|
0.605
|
|
SPADR % (SD)
|
0.83 (1.58)
|
1.62 (2.68)
|
1.78 (2.47)
|
2.07 (2.93)
|
2.49 (4.45)
|
3.01 (3.81)
|
2.68 (3.62)
|
<0.01
|
|
F-SPADR % (SD)
|
0.82 (2.13)
|
1.52 (3.10)
|
1.69 (2.70)
|
2.03 (3.36)
|
2.46 (4.67)
|
3.13 (5.10)
|
2.68 (4.26)
|
<0.01
|
|
M-SPADR % (SD)
|
0.83 (1.80)
|
1.75 (2.97)
|
1.85 (2.90)
|
2.08 (3.52)
|
2.38 (4.68)
|
2.86 (3.76)
|
2.62 (4.15)
|
<0.01
|
|
CRSDR % (SD)
|
2.59 (3.67)
|
3.26 (4.13)
|
3.35 (3.66)
|
3.49 (3.74)
|
4.25 (5.52)
|
4.48 (4.61)
|
4.19 (4.75)
|
<0.01
|
|
F-CRSDR % (SD)
|
2.14 (3.71)
|
2.97 (4.72)
|
2.97 (3.93)
|
3.09 (3.96)
|
3.96 (5.50)
|
4.39 (6.03)
|
4.08 (5.53)
|
<0.01
|
|
M-CRSDR % (SD)
|
2.87 (5.01)
|
3.58 (4.75)
|
3.69 (4.30)
|
3.78 (5.00)
|
4.36 (6.26)
|
4.53 (4.63)
|
4.19 (5.72)
|
<0.01
|
Fig. 3 Detection rate per year in female patients.
SDR
In 20.44% of all colonoscopies (n=43.475) a HP, SSL, or TSA was found. The prevalence
of serrated polyps in men of 21.97% (n=23,052) was higher than in women at 21.74%
(n=20,423). The overall mean male SDR was higher than the female SDR (22.72% vs. 19.43%;
P < 0.001) ([Table 3]). Endoscopists with ADR ≥ 25% had a mean SDR of 30.43% (10.95), which was reached
by 23.29% of them (n=68). In contrast, the mean SDR for endoscopists with an ADR <
25% was 16.15% (SD 8.15%) ([Table 4]). No differences were found between 2012 and 2017 for overall, male, or female SDR
([Table 3]). [Table 2] shows the SDR for different age groups.
SPADR and SSLDR
At least one SSL or TSA was found in 2.19% of colonoscopies (n=4,666). The overall
mean SPADR was 2.19% (SD 2.49). Only SSLs were found in 1.78% of all colonoscopies
(n=3,784). The mean SSLDR was 1.75 (SD 2.12). The prevalence of serrated lesions was
2.2%
(2,293) for men and 2.2% (n=2,373) for women. There was no difference between the
mean
male and female SPADRs (2.10% vs. 2.30%; P=0.815) ([Table 3]). Endoscopists with ADRs ≥ 25% and < 25% had mean SPADRs of 3.52% (SD 2.96) and
1.49% (SD 1.87), respectively. Of the endoscopists, 19.17% (n=56) had reached a SPADR
higher than 3.61%. The overall SPADR in 2012 was 0.83% (SD 1.58) and increased to
3.01%
(SD 3.82) in 2017 ([Fig. 1]). Furthermore, we observed an increase by 2.03% from 0.83% (SD 1.80) in 2012 to
2.86% (SD 3.76) in 2017 for men ([Fig. 2]), and for women, from 0.82% (SD 2.13) in 2012 to 3.13% (SD 5.10) in 2017 (all
P < 0.001) ([Table 5] and [Fig. 3]). [Table 2] shows the SPADR in different age groups.
CRSDR
Clinically relevant serrated polyps were found in 3.82% of all examinations (n=8,124).
The overall mean CRSDR was 3.81% (SD 3.40%). For men, the prevalence of clinically
relevant serrated polyps was 4.00% (n=4,197), and for women, it was 2.65% (n=3,927).
Men had a mean CRSDR of 3.89% (SD 3.93) and women a mean CRSDR of 3.73% (SD 3.96)
([Table 3]). The mean CRSDR for endoscopists with an ADR ≥ 25% was 5.96% (SD 3.87). This detection
rate was reached by 16.44% of endoscopists (n=48). In contrast, the mean CRSDR for
endoscopists with an ADR < 25% was 2.68% (SD 2.47). The overall CRSDR increased by
1.92% from 2.59% (SD 3.67) in 2012 to 4.48% (SD 4.61) in 2017 (P < 0.001). For men, the CRSDR increased from 2.87% (SD 5.01) in 2012 to 4.53% (SD
4.63) in 2017 (P < 0.001) ([Fig. 2]). For women the CRSDR increased from 2.14% (SD 3.71) in 2012 to 4.39% (SD 6.03)
in 2017 (P < 0.001) ([Fig. 3]). [Table 2] shows the CRSDR in different age groups.
Comparison of the ADR and serrated polyp detection rates
Spearman rank order showed significant correlations between the ADR and SDR, SPADR,
SSLDR, and CRSDR (rho=0.730 vs. rho=0.508 vs. rho=0.508 vs. rho=0.630; all P <0.01) ([Fig. 4], [Fig. 5], [Fig. 6]) ([Table 6]). Furthermore, the ADR is significantly correlated with the SDR, SPADR, SSLDR, and
CRSDR in the ADR < 25% group (rho=0.580 vs. rho=0.522 vs. rho=0.417 rho=0.512; all
P <0.01) ([Fig. 7], [Fig. 8], [Fig. 9]). In contrast, the ADR was significantly correlated only with the SDR (rho=0.355;
P <0.01) but not with the SPADR, SSLDR, or CRSDR in the ADR ≥25% group (rho=-0.085;
P=0.400 vs. rho=-0.037; P=0.677; rho=0.064; P=0.529) ([Fig. 10], [Fig. 11], [Fig. 12]).
Fig. 4 Correlation of ADR and SPADR.
Fig. 5 Correlation of ADR and SDR.
Fig. 6 Correlation of ADR and CRSDR.
Table 6 Correlation between ADR and SDR and SPADR and CRSDR.
|
|
ADR
|
ADR <25%
|
ADR ≥25%
|
|
ADR, adenoma detection rate; SDR, serrated detection rate; SPADR, serrated polyp,
detection rate; CRSDR, clinically serrated polyp detection rate; SSLDR, sessile serrated,
lesion detection rate.
|
|
|
Spearman coefficient
|
P value
|
Spearman coefficient
|
P value
|
Spearman coefficient
|
P value
|
|
SDR
|
0.730
|
< 0.01
|
0.580
|
< 0.01
|
0.355
|
< 0.01
|
|
SPADR
|
0.508
|
< 0.01
|
0.552
|
< 0.01
|
-0.085
|
=0.400
|
|
CRSDR
|
0.630
|
< 0.01
|
0.512
|
< 0.01
|
0.064
|
=0.529
|
|
SSLDR
|
0.508
|
< 0.01
|
0.417
|
< 0.01
|
-0.037
|
=0.677
|
Fig. 7 Correlation of ADR and SDR in endoscopists with an ADR <25%.
Fig. 8 Correlation of ADR and SPADR in endoscopists with an ADR <25%.
Fig. 9 Correlation of ADR and CRSDR in endoscopists with an ADR <25%.
Fig. 10 Correlation of ADR and SDR in endoscopists with an ADR >25%.
Fig. 11 Correlation of ADR and SPADR in endoscopists with an ADR >25%.
Fig. 12 Correlation of ADR and CRSDR in endoscopists with an ADR >25%.
According to the cut-off of ADR ≥25%, we found that SDR, SPADR, and SSLDR and CRSDR
were significantly higher overall for female and male endoscopists with a mean ADR
≥25% than among those with an ADR <25% (all P <0.01, Mann-Whitney-U-Test).
Discussion
In our study, we included 212,668 screening colonoscopies performed by 290 endoscopists.
We evaluated different serrated polyp detection rates and found strong variability
between endoscopists. The mean SDR, SPADR, and CRSDR were 21.07%, 2.19%, and 3.81%,
respectively, and we documented strong significant correlations between the detection
rates. Furthermore, practitioners in the higher ADR group (ADR ≥ 25%) had significantly
higher serrated detection rates in comparison with the lower ADR group (ADR <25%).
A study by Anderson et al. [22] reported the results of 45,996 screening and surveillance colonoscopies. They investigated
potential benchmarks for SDR based on the 25% and 35% ADR cut-offs. For endoscopists
with ADRs > 25% and > 35%, the corresponding median SDR was 6.8% (interquartile range
[IQR] 4.3%-8.6%) and 10.0% (IQR 8.5%-13.1%), respectively. CRSDR and ADR showed a
significant correlation with a Spearman coefficient of 0.690 (P < 0.01). Two recent studies measured serrated polyp detection rates during surveillance
and/or screening colonoscopies. First, Schramm et al. [23] analyzed 4,161 screening colonoscopies. They found at least one clinically relevant
serrated polyp in 4.7% (95% CI 2.3%-7.2%) of all cases and a SDR of 19.4% (95% CI
13.5%-25.4%). A practitioner’s ADR correlated significantly with the CRSDR (rho=0.54;
P < 0.05). Second, a multicenter study with 104,618 colonoscopies found an overall
mean SPADR of 5.1% (SD 3.8) with a greater than 18-fold difference between the highest
and lowest endoscopist (range 0%-18.8%). A significant correlation was found between
ADR and SPADR (rho=0.540; P < 0.01) [24]. Finally, a cohort study with two primary colonoscopy and three fecal occult blood
tests (FOBT) screening cohorts showed detection rates for serrated polyps between
15.1% and 27.2% (median 29,5%) and for clinically relevant serrated polyps between
2.1% to 7.8% (median 4.6%). [18]
Today, ADR is an accepted indicator for screening colonoscopies and a minimum of 25%
is recommended (men: 30%, women: 20%). Based on the 25% ADR cut-off from the American
College of Gastroenterology and the American Society of Gastrointestinal Endoscopy,
we measured a mean ADR of 21.79% (SD 9.25), a SDR of 21.08% (SD 11.42), a SPADR of
2.19% (SD 2.49), and a CRSDR of 3.81% (SD 3.40). Similar to other studies, we demonstrated
significant correlations between the ADR and the SDRs (SDR: rho=0.730 vs. SPADR: rho=0.508
vs. CRSDR: rho=0.630; all P <0.01). However, a retrospective study by Liang et al. [25] showed no significant correlation between ADR and SDR (rho=0.571; P=0.237). In comparison with other studies, our detection rates were lower than all
of them except the SDR in Schramm et al. [23]. Our study showed a significant difference between the SDR, SPADR, and CRSDR of
endoscopists with ADRs <25% and ≥25% (all P <0.01).
[Table 4] shows that endoscopists with an ADR ≥ 25% detect more conventional adenomas and
serrated lesions. Endoscopists with an ADR ≥ 25% had a 2-fold increase in finding
conventional adenomas, a 5-fold increase in finding serrated lesions, and a 4-fold
increase in finding clinically relevant serrated polyps compared with the endoscopists
in the lower ADR group. These findings were similar for men and women. These results
could be explained by the fact that endoscopists with a higher ADR have increased
skills, experience, and eventually more knowledge about serrated polyps. Increased
detection of serrated lesions also may explain why a higher ADR is associated with
a lower risk for interval cancer.
In the current study, we also observed strong variability in detection rates, which
may have been caused by different levels of knowledge. On the one hand, some endoscopists
disregard serrated polyps and do not biopsy or remove them because of their lack of
knowledge. On the other hand, some physicians diagnose serrated polyps, especially
diminutive HPs, as clinical irrelevant and correctly leave them behind.
We found that detection rates for serrated adenomas and clinically relevant serrated
polyps increased significantly between 2012 and 2017. The same results were found
for both genders (all P <0.01). The overall SPADR increased by 1.85% from 0.83% (SD 1.58) to 2.68% (SD 3.81)
and the overall CRSDR by 1.89% from 2.59% (SD 3.67) to 4.48% (SD 4.61). These positive
trends are the result of many factors, including increasing knowledge about the importance
of serrated adenomas in particular and differences in how endoscopists evaluate HPs
and determine their clinical relevance.
In Austria, SC was recommended at age 50 years for both sexes at the time of this
study. With regard to the group aged 50 to 59 years, we documented an ADR of 16.89%
(SD 8.30%), an SDR of 20.86% (SD 12.57%), an SPADR of 2.64% (SD 3.38%), and an CRSDR
of 4.39% (SD 4.63%). Interestingly, we found the highest overall detections rates
for SPADR and CRSDR in the group aged younger than 50 years.
Regarding gender, we observed significant differences between men and women for conventional
adenomas (27.16% (SD 15.77%) vs. 16.0% (SD 12.52%); P < 0.001) and serrated polyps, including HPs, SSLs, and TSAs (20.95% (SD 13.86%) vs.
17.32% (SD 13.93%); P < 0.001).
The strengths of the study were the large number of colonoscopies and endoscopists,
as well as the fact that all colonoscopies were performed as part of the quality assurance
screening program of OEGGH. Our study has some limitations. First, it was a retrospective
analysis of data that were prospectively collected but not using the design of this
study. The second limitation is that until 2013, SSLs and TSAs were classified as
“serrated adenomas.” However, all detection rates were defined in such a way that
differentiating between SSAs and TSAs was unnecessary.
Conclusions
In summary, our study suggests that ADR correlates significantly with different SDRs
and endoscopists with higher ADRs have significantly higher rates of serrated polyp
detection. Studies evaluating the impact of artificial intelligence on the detection
of serrated lesions by endoscopists with an ADR < 25% would be of great interest for
the future. A study by Zessner-Spitzenberg et al. showed that proximal SDR is associated
with a reduction in post-colonoscopy CRC similar to endoscopists’ ADR [26]. This suggests that quality improvement in endoscopist ADR is necessary to avoid
interval cancer. Also, improving detection of serrated lesions in the right colon
should be emphasized. Further studies are necessary to assess whether the serrated
polyp detection rate should be implemented as an independent quality parameter for
screening colonoscopy.
