- Email: [email protected]
Adenoma detection rate varies greatly during colonoscopy training
ORIGINAL ARTICLE: Clinical Endoscopy
Adenoma detection rate varies greatly during colonoscopy training Sascha C. van Doorn, MD, Robert B. Klanderman, MD, Yark Hazewinkel, MD, Paul Fockens, MD, PhD, Evelien Dekker, MD, PhD Amsterdam, the Netherlands
Background: The adenoma detection rate (ADR) is considered the most important quality indicator for colonoscopy and varies widely among colonoscopists. It is unknown whether the ADR of gastroenterology consultants can already be predicted during their colonoscopy training. Objective: To evaluate the ADR of fellows in gastroenterology and evaluate whether this predicts their ADR as gastroenterology consultants. Design: Retrospective observational study. Setting: Academic and regional centers. Patients: Symptomatic patients undergoing colonoscopy. Main Outcome Measurements: The variance in ADR among 7 gastroenterology fellows during their training (between May 2004 and March 2012) and of the same fellows after they registered as consultants (between October 2011 and April 2014) was evaluated. Multivariate logistic regression was performed to compare the highest detector (endoscopist with highest ADR) with the individual fellows and to evaluate whether an ADR of 20% or higher during the training was predictive of a high ADR as a consultant. Results: During training, ADRs ranged from 14% to 36% (P! .001). Compared with the highest detector, the OR for detecting an adenoma ranged from 0.64 (95% CI, 0.40-1.03) to 0.29 (95% CI, 0.17-0.48). After registration, ADR ranged from 19.8% to 40.2% (P Z .066). Compared with the highest detector during consultancy, the OR ranged from 0.64 (95% CI, 0.34-1.21) to 0.26 (95% CI, 0.13-0.52). Only 2 fellows significantly improved their ADR after completing their training. An ADR lower than 20% during training was associated with a lower ADR as a consultant (OR 0.51; 95% CI, 0.30-0.87). Limitations: Retrospective study. Conclusions: Variance in ADR is already present during the endoscopy training of gastroenterology fellows. Most fellows do not improve their ADR after completing their training. These findings suggest that the ADR can be predicted during colonoscopy training, and we suggest that feedback and benchmarking should be implemented early during training of fellows in an effort to improve ADR in future daily practice as a consultant. (Gastrointest Endosc 2015;82:122-9.)
Colorectal cancer (CRC) arises from precursor colorectal polyps. Detecting and resecting these lesions during colonoscopy decrease the CRC incidence.1 At the same time, it is well known that polyps are missed during
colonoscopy, and this can result in interval cancers in the years after colonoscopy.2-7 Although it is the ultimate quality indicator, the rate of interval cancers is a parameter that can only be used on a long-term basis and is therefore
Abbreviations: AMC, Academic Medical Center; ADR, adenoma detection rate; CI, confidence interval; CIR, cecal intubation rate; CRC, colorectal cancer; OR, odds ratio.
Received September 29, 2014. Accepted December 7, 2014.
DISCLOSURE: Dr. Fockens disclosed that he has received a research grant from Olympus. All other authors disclosed no financial relationships relevant to this article. Copyright ª 2015 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2014.12.038
122 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
Current affiliations: Department of Gastroenterology and Hepatology, Academic Medical Centre, University of Amsterdam, Amsterdam, the Netherlands. Reprint requests: Evelien Dekker, MD PhD, Department of Gastroenterology and Hepatology, Academic Medical Centre, Meibergdreef 9 1105 AZ, Amsterdam, the Netherlands.
www.giejournal.org
van Doorn et al
not practical for monitoring quality in daily colonoscopy practice. Kaminski et al8 showed that the adenoma detection rate (ADR) of individual colonoscopists, ie, the proportion of colonoscopies in which at least 1 histologically confirmed adenoma is detected, is associated with the risk of future interval carcinomas. Recently, Corley et al9 confirmed the association reported by Kaminski et al and demonstrated that the ADR was inversely associated with death caused by interval colorectal cancer (CRC). Therefore, the ADR is considered an important quality indicator of colonoscopy, and current guidelines demand ADRs of at least 20% on screening colonoscopies.10-14 In the current literature, a wide variation in ADRs among colonoscopists is reported.12,15-19 Measurement and benchmarking of this quality indicator will perhaps improve awareness and training in an effort to continuously increase quality in colonoscopy. During endoscopy training of gastroenterology fellows, the most important competency for colonoscopy is safe cecal intubation.20-22 Only a few studies evaluated the effect of fellow participation in colonoscopy screening programs on ADR,23-28 but the literature totally lacks information on the ADR of individual gastroenterology fellows during the years of their colonoscopy training. It is unknown whether a variation in ADR is already present during training and whether measurement and benchmarking of ADR would improve ADR during training and would give fellows a better start for daily practice. Possibly the ADR of a gastroenterologist consultant can already be predicted during training. If consultants have a high ADR during training and are already aware of the importance of adenoma detection, the colonoscopy training could form the basis for an adequate ADR in daily practice. The aim of our study was to evaluate the variance in ADRs among gastroenterology fellows during the course of their colonoscopy training and to evaluate whether individual ADRs during training predict ADRs in later practice as a gastroenterology consultant.
Adenoma detection rate varies during colonoscopy training
who performed their entire endoscopy training in the AMC from May 2004 to March 2012 were included. Seven fellows fulfilled these criteria. For all 7 fellows, data on every colonoscopy that they performed were obtained from the colonoscopy database (version 11.0 of ENDOBASE, Olympus, Winter & Ibe GmbH) and reviewed. For all 7 fellows, reports of every colonoscopy that they performed during their training in the study period (May 2004-March 2012) were retrieved and reviewed from our colonoscopy database (ENDOBASE). Accompanying histopathology reports and electronic patient charts were also searched. All reports of colonoscopies performed during the study period 2004 to 2012 by 1 of the 7 fellows were reviewed. Subsequently, data on consecutive colonoscopies that the fellows performed after they were registered as gastroenterology consultants and practicing were collected. These data were collected from October 2011 to April 2014.
Colonoscopy procedure Colonoscopies performed by fellows during training at the AMC were either high-definition or standarddefinition white-light colonoscopies. Patients prepared by taking 4 L of a polyethylene glycol solution (Kleanprep; Norgine bv, Amsterdam, the Netherlands) or 2 L of a polyethylene glycol solution containing ascorbic acid (Moviprep; Norgine bv) with an additional 2 L of liquids. Procedures were performed with patients under conscious sedation with midazolam and/or fentanyl or under deeper sedation (propofol) when indicated. An Olympus 180 series colonoscope (Olympus Optical Co, Ltd, Tokyo, Japan) was used for the majority of procedures. For the colonoscopies performed as gastroenterology consultants, the colonoscope type, bowel preparation, and sedation use were unknown.
Colonoscopy data
We conducted a retrospective study in gastroenterology fellows trained at the Academic Medical Center (AMC) in Amsterdam, the Netherlands. Permission of the research protocol by an institutional review board was not required per the agreement regarding medical research involving human subjects. The endoscopy department of the AMC is a tertiary care facility and teaching hospital. In the Netherlands, gastroenterology fellows often perform part of their training at a teaching hospital in addition to a few years in one of the academic hospitals. At the AMC, gastroenterology fellows are randomly assigned to training at either the AMC or a teaching hospital. All gastroenterology fellows
For each colonoscopy, demographic information on the patient was collected from the electronic patient chart: age, sex, previous CRC and/or colonic resection, and medical history regarding inflammatory bowel disease, polyposis, or Lynch syndrome. Data collected on each colonoscopy included indication for colonoscopy (screening, surveillance, or diagnostic), bowel preparation (if available with Boston Bowel Preparation score, otherwise scored as good, suboptimal, poor, or unknown), sedatives used, cecal intubation rate (CIR), and deepest point of colonoscope insertion and reason for incomplete colonoscopy if the cecum was not reached. The number of detected and removed polyps was recorded, as were the location in the colon, polypectomy method, and histopathology result (adenomas: tubular, tubulovillous, or villous and grade of dysplasia: low or high; serrated polyps subdivided in hyperplastic polyps, sessile serrated polyps, and traditional serrated polyps; adenocarcinomas), if applicable.
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 123
METHODS Study design and data collection
Adenoma detection rate varies during colonoscopy training
van Doorn et al
Inclusion and exclusion criteria for colonoscopies included in the analysis
and adjusted for poor bowel preparation or a stricture) were calculated for all colonoscopies in an intact colon, including excluded colonoscopies for the detection analyses. P ! .05 was considered statistically significant. SPSS software (version 20.0; IBM Corporation, Somers, NY) was used for analysis.
In the Dutch colonoscopy training system, the first 100 procedures are completely supervised by a consultant present in the endoscopy suite, followed by a period of “on-demand” supervision. On-demand supervision means that a consultant is available whenever the fellow asks for supervision, eg, when the fellow has difficulty reaching the cecum or when a therapeutic procedure like polypectomy is performed. The colonoscopies during training were divided in 3 distinct periods: first 100 supervised colonoscopies, first 100 autonomous colonoscopies, and the remaining colonoscopies during the training. The first 100 colonoscopies, which were probably closely supervised, were excluded from analyses. To ensure a homogeneous cohort of patients per fellow or consultant, exclusion criteria were applied after stratification into 1 of the 3 training periods or the consultancy period. We excluded colonoscopies performed in patients younger than 18 years of age, emergency procedures as well as colonoscopies performed in patients with conditions that predispose to either a very high or very low probability of detecting adenomas: inflammatory bowel disease, polyposis syndromes, Lynch syndrome, referral for adenoma removal, referral after a positive screening test result, eg, fecal occult blood test or CT colonography. In addition, patients with an incomplete colonoscopy were excluded from the detection analyses. To correct for differences in colonoscopy indications, procedure indications were divided into 2 groups: symptomatic, including patients with anemia, melena, hematochezia, changed bowel habits, diarrhea, and abdominal pain and surveillance/screening, including average-risk screening, familial predisposition for CRC, and adenoma or cancer surveillance.
RESULTS Patient characteristics
ADR was defined as the proportion of colonoscopies in which at least 1 histologically confirmed adenoma was detected. c2 analyses were performed to evaluate the variance in ADR among the 7 fellows during their training and during their consultancy and the difference in ADR during training and consultancy per individual fellow. Univariate logistic regression modeling was performed with the following variables: patient age, patient sex, and grouped colonoscopy indication, ie, symptoms or screening and surveillance. The difference in ADR between fellows and consultants was evaluated with multivariate logistic regression to compare the highest detector (the endoscopist with the highest ADR) with the other individual endoscopists, adjusted for significant variables in the univariate regression model. Logistic regression was performed to evaluate whether an ADR of 20% or higher during training was predictive of a high ADR during the consultancy period. Individual fellow CIRs (unadjusted
During the study period, 7 fellows performed a total of 3252 colonoscopies (Table 1). After exclusion of the first 100 procedures per fellow and exclusion of colonoscopies that met the exclusion criteria, 1472 colonoscopies (45% of all colonoscopies) were available for analysis. The mean age of the patients who underwent colonoscopy during the training period was 56 years, and 49% of the patients were male. After exclusion, the mean age was 58 years and 48% were male. Most of the colonoscopy reports did not report on presence of a supervisor, and we assumed a supervisor was only present briefly during procedures if a polypectomy was performed or when called by the fellow for another reason and did not contribute to the ADR. In some reports, it was stated that the supervisor took over the colonoscopy, for example, if the fellow could not intubate the cecum or if a large polyp was removed. In cases in which no supervisor was present during the whole procedure, this was also stated in the report. Data on consecutive colonoscopies that the fellows performed after they were registered as gastroenterology consultants and practicing were collected. Two consultants practiced at an academic center and 5 at a regional hospital. Five of 7 gastroenterology consultants collected data on at least 100 nonselected consecutive colonoscopies that they submitted for accreditation for the Dutch FIT-based CRC screening program, and these data were used for this study. To prevent selection bias, the data that the consultants had submitted for accreditation for the screening program, we verified whether the procedures were indeed consecutive. We visited their practices to review the patient charts. For the 2 other consultants, we collected data on consecutive colonoscopies from the first years of consultancy from colonoscopy databases. After exclusion based on age, indication, and incomplete colonoscopies, 692 colonoscopies (75% of all colonoscopies) were available for analysis (Table 1). During the consultancy period, the mean patient age was 59 years, and 44% were male. After exclusion, the mean patient age remained 59 years and 43% of the patients were male. The mean patient age and percentage of male patients did not differ between fellows and consultants. During training, the unadjusted CIR ranged from 86.4% to 96.0% and the adjusted CIR (excluding procedures with poor bowel preparation and obstruction) from 94.0% to 99.1%. The unadjusted CIRs during the consultancy period
124 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
Study outcomes and statistical analyses
van Doorn et al
Adenoma detection rate varies during colonoscopy training
TABLE 1. Number of procedures during training and consultancy
Fellow 1
Training period
Training period after exclusion* (% included from total number of procedures)
Consultancy period
Consultancy period after exclusion* (% included from total number of procedures)
433
144 (33%)
136
109 (80%)
166
106
81
68
Surveillance
52
38
41
41
Other
215
Symptoms
Fellow 2
14
389
147 (38%)
171
111 (65%)
Symptoms
175
102
39
34
Surveillance
59
45
78
77
479
207 (43%)
100
97 (97%)
Symptoms
177
126
84
84
Surveillance
109
81
14
13
Other
193
Other Fellow 3
Fellow 4
155
54
2
612
333 (54%)
107
91 (85%)
312
245
70
64
Surveillance
99
88
27
27
Other
201
Symptoms
Fellow 5
10
415
194 (47%)
107
89 (83%)
Symptoms
202
132
61
57
Surveillance
82
62
33
32
417
224 (54%)
101
94 (93%)
Symptoms
225
167
74
73
Surveillance
89
57
21
21
507
223 (44%)
206
101 (49%)
Symptoms
199
150
81
77
Surveillance
106
73
25
24
Other
202
Other Fellow 6
Other Fellow 7
131
13
103
6
100
Symptoms
1456
1028
478
457
Surveillance
596
444
231
235
Other
1200
1472 (45%)
107
692 (75%)
Total
3252
928
*Excluded were patients younger than 18 years of age, incomplete colonoscopies, other indication (acute colonoscopy, positive fecal occult bleed test result, positive findings on CT colonography, polyposis syndrome, Lynch syndrome, inflammatory bowel disease [IBD]) and excluding first 100 procedures.
ranged from 89.3% to 99.0% and the adjusted CIRs from 94.2% to 100%. No significant increase in CIR was detected after the first 100 colonoscopies or compared with their practice when being a consultant.
Colonoscopies during training During the overall training period, the ADR ranged from 13.5% to 36.1% and differed significantly among fellows (P! .001) (Table 2). Univariate logistic regression analyses showed a significant association between the OR of finding at least 1 adenoma and the patient’s age, sex, and www.giejournal.org
colonoscopy indication (P ! .001 for all 3 associations). After adjusting for these variables, the OR to detect at least 1 adenoma ranged from 0.64 (95% CI, 0.40-1.03) to 0.29 (95% CI, 0.17-0.48) among all fellows compared with the highest detector, although not significantly for fellows 2 and 3 (Table 2).
Colonoscopies during consultancy The ADR among the 7 gastroenterology consultants ranged from 19.8% to 40.2%; the difference was not statistically significant (P Z .066) (Table 3). Again, univariate Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 125
Adenoma detection rate varies during colonoscopy training
van Doorn et al
TABLE 2. Comparison of ADR during training (ranked by OR)* N [ 1472
ADR, %
OR* (95% CI)
P value
Overall
24.6
Fellow 1
36.1
1
Fellow 3
30.4
0.63 (0.39-1.02)
Fellow 2
27.9
0.60 (0.36-1.01)
.056
Fellow 4
25.2
0.59 (0.38-0.92)
.019
Fellow 6
22.3
0.54 (0.33-0.88)
.013
Fellow 5
21.6
0.44 (0.27-0.74)
.002
Fellow 7
13.5
0.22 (0.13-0.38)
<.001
.060
ADR, Adenoma detection rate; OR, odds ratio; CI, confidence interval. Statistically significant P values are in bold. *Adjusted for patient age, sex, and indication, after exclusion.
fellows with an ADR higher than 20% during training. We also evaluated whether an ADR lower than 20% during the first 100 colonoscopies autonomously performed was associated with an increased ADR during the remainder of the training. Fellows 5, 6, and 7 had ADRs lower than 20% during the first 100 procedures (18.8%, 18.5%, and 8.6%, respectively) (Table 4), and this was associated with a significantly lower ADR during the remainder of the training (OR 0.69; 95% CI, 0.51-0.93), but not with a lower ADR during consultancy (P Z .68).
DISCUSSION
Only for gastroenterology consultants 5 and 6, the ADR was significantly higher during consultancy than it was during the years of their training (Fig. 1). We evaluated whether an ADR lower than 20% during the overall training period was associated with a lower ADR as a gastroenterology consultant, adjusted for patient age, sex, and colonoscopy indication. During the overall training period, only fellow 7 had an ADR lower than 20% (13.5%), and this was associated with a significantly lower ADR during consultancy (OR 0.51; 95% CI, 0.30-0.87) compared with
The ADR is considered the most important surrogate quality indicator for colonoscopy and varies widely among colonoscopists.12,15-19 In this study, we evaluated whether a variation in ADR is already present during colonoscopy training and whether individual ADRs during training predict ADR in later practice. After correction for age, sex, and colonoscopy indication, we detected a significant variance in ADR among 7 fellows during their training (range 13.5%-36.1%, P ! .001). Although the difference in ADR among these fellows when they were gastroenterology consultants did not reach statistical significance, after adjusting for age, sex, and colonoscopy indication, when comparing consultants with the highest detector, this was significant for 5 of the 6 gastroenterology consultants. Only 2 of the fellows improved their ADR after their training (from 21.6% to 38.2%, P Z .004 and from 22.3% to 36.2%, P Z .011). Furthermore, our data suggest that an ADR lower than 20% during training is associated with a lower ADR when a gastroenterology consultant. This study has several limitations. There were only 7 gastroenterology fellows included in this study, as we chose to include only fellows who performed their entire endoscopy training at the AMC to ensure that all fellows received the same training, level of supervision, and followed the same protocols. The fact that only fellows performing their training at the AMC were included may limit the generalizability of the results. The small number of participating fellows could have led to a bias because we could only compare 1 fellow with an ADR lower than 20% with the other fellows in our regression model. A future study should include more fellows, preferably from different teaching hospitals. Because of the retrospective design of this study, not all quality indicators were reported during colonoscopies. We could not correct the ADRs for poor bowel preparation, a potential confounder, as this information was not always included in the colonoscopy reports because the importance of reporting on quality indicators of colonoscopies was not as much acknowledged at that time. The retrospective design, however, also has an advantage as it eliminates possible observer bias (Hawthorn effect; colonoscopists might be more aggressive in polyp detection than usual
126 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
TABLE 3. Comparison of ADR during consultancy (ranked by OR ratio)* N [ 672 Overall
ADR, %
OR* (95% CI)
P value
33.3
Gastroenterology consultant 3y
40.2
1
Gastroenterology consultant 5
38.2
0.64 (0.34-1.21)
Gastroenterology consultant 6
36.2
0.51 (0.27-0.98)
.044
Gastroenterology consultant 4
33.0
0.45 (0.23-0.86)
.015
Gastroenterology consultant 2
32.1
0.39 (0.21-0.74)
.004
Gastroenterology consultant 1
32.4
0.32 (0.17-0.62)
.001
Gastroenterology consultant 7
19.8
0.26 (0.13-0.52)
<.001
.172
ADR, Adenoma detection rate; OR, odds ratio; CI, confidence interval. Statistically significant P values are in bold. *Adjusted for patients age, sex, and indication, after exclusion yGastroenterology consultants are numbered in the same order as during the training.
logistic analyses were performed to calculate the OR of adenoma detection corrected for age, sex, and colonoscopy indication, which were all statistically significant related to ADR (P ! .001). Although the difference in ADR did not reach statistical significance, after adjusting for age, sex, and colonoscopy indication, comparison with the highest detector was significant for 5 of the 6 gastroenterology consultants (OR ranged from 0.64 [95% CI, 0.341.21] to 0.26 [95% CI, 0.13-0.52]) (Table 3).
Association between ADR during training and consultancy
van Doorn et al
Adenoma detection rate varies during colonoscopy training
p< .05
p< .01
Fellow 4
Fellow 5
Fellow 6
Fellow 7
Adenoma detection rate
40%
30%
20%
10%
0% Fellow 1
Fellow 2
Fellow 3
Training
36.1%
27.9%
30.4%
25.2%
21.6%
22.3%
13.5%
Consultancy
32.1%
32.4%
40.2%
33.0%
38.2%
36.2%
19.8%
Figure 1. For each fellow, the ADR during training and during consultancy are shown. Only for fellow 5 and 6 the ADR was significantly higher during consultancy than it was during the years of training.
when they are aware that their detection rates are being recorded29). Another limitation is the heterogeneity of colonoscopy indications. Unfortunately, this is was inevitable because we aimed to include all colonoscopies the fellows performed during the training and because of the retrospective design of the study. The heterogeneity of colonoscopy indications, however, seemed equal among the different fellows. Only 45% of all colonoscopies performed during training were included (1472 of 3235 colonoscopies) in the analyses compared with 75% (692 of 928 colonoscopies) during consultancy. These rates were based on exclusion of the first 100 supervised colonoscopies, incomplete colonoscopies, emergency procedures,
positive results on CT colonography, and procedures in patients with polyposis syndrome, which were more frequent during training compared with consultancy. For consultants 2 and 7, relatively more patients were excluded than for the others as they performed, respectively, many colonoscopies in patients with polyposis or Lynch syndrome or in patients with inflammatory bowel disease. The number of colonoscopies performed by the consultants was relatively small compared with the number that they performed when they were fellows. It was difficult to collect data from the consultancy period, as they all worked at different hospitals with different reporting systems and electronic patient files. We chose to include a random selection of consecutive colonoscopies that they performed in their practice to obtain a reliable impression of the detection rates of all individual consultants. Finally, in the analyses, the limit of a low detector was set at lower than 20% because this was found to be associated with an increased risk of interval carcinomas.8 However, the original study by Kaminski et al8 was performed in a colonoscopy screening population. In this study, only diagnostic and surveillance colonoscopies were included, for which a minimal ADR of 20% might be too low. In the scarce literature, a recent study reported that the ADR stratified by colonoscopy indication was highest for surveillance colonoscopies, followed by screening, and then diagnostic,30 and ADRs in symptomatic and surveillance populations vary from 17% to 49%.31,32 We decided to use 20% as a lower threshold as the fellows performed the colonoscopies between 2004 and 2012 when colonoscopists were less aware of the
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 127
TABLE 4. ADR during first 100 autonomous colonoscopies No. of procedures*
No. of procedures with R1 adenoma
ADR, %
Fellow 1
34
11
32.4
Fellow 2
55
23
41.8
Fellow 3
52
20
38.5
Fellow 4
68
16
23.5
Fellow 5
64
12
18.8
Fellow 6
65
12
18.5
Fellow 7
58
5
8.6
Total
396
99
25.0
*After exclusion of patients younger than 18 years of age, incomplete colonoscopies, and other indication (acute colonoscopy, positive fecal occult bleed test result, positive findings on CT colonography, polyposis syndrome, Lynch syndrome, inflammatory bowel disease).
Adenoma detection rate varies during colonoscopy training
van Doorn et al
importance of the ADR. Future studies are needed to establish the threshold ADR for other populations than screening. The ADR is considered the most important surrogate quality indicator of colonoscopy because it is inversely associated with the risk of interval carcinomas.8,9 To our knowledge, our study is the first to evaluate individual ADRs of gastroenterology fellows during the years of their colonoscopy training. Most previous studies evaluating the ADR of fellows focused on comparison of the ADR of gastroenterology consultants alone with the ADR in colonoscopies in which a fellow was involved.23,25-27 Two studies reported a significantly higher ADR in colonoscopies in which a fellow was involved and observed an increase with each year of training.23,26 Conversely, a meta-analysis from 2013 did not observe an effect on ADR by fellow involvement, and only 1 study in this meta-analysis observed a decrease in the ADR with fellow involvement.24 Only 2 studies evaluated ADR during the course of training and observed no increase in ADR over time; individual fellow ADRs were not reported.20,22 To be able to report on actual independent fellow adenoma detection, in our study we only included colonoscopies for which no supervisor was present during withdrawal. We observed a large variance in ADR already present during training (13.5%-36%). In a small study of only 218 patients, Munroe et al33 observed a decrease in adenoma miss rates with increasing experience, in a tandem colonoscopy study in which withdrawal was first performed by the trainee and then by an attending physician. No individual ADRs of fellows were reported.33 In contrast with these results, we observed that having an ADR lower than 20% during the first 100 autonomously performed colonoscopies in training was associated with a significantly lower ADR during the remainder of the training. We excluded the first 100 colonoscopies because these procedures were completely supervised, and the fellows did not perform individual withdrawal. We do not have data regarding the supervisors at the time; however, the group of staff members in our department that provide supervision of fellows has not changed much during the study period. Whether the emphasis of the supervising mentor is on polyp detection might affect a fellow’s ability to achieve a high ADR in later practice. Prospective studies should investigate the role of supervisors in the ADR of fellows in training. A second observation was that the single fellow who had an ADR lower than 20% during the complete training period also had a low ADR during consultancy. Only 2 fellows improved their ADR significantly after their training. The number of fellows that we could include is too small to draw strong conclusions; however, our data suggest that a high ADR of gastroenterology consultants may already be predicted during their training. Larger studies evaluating colonoscopy quality of more fellows throughout their entire colonoscopy training are needed to confirm our findings.
Previous studies evaluating competence in colonoscopy focused on cecal intubation.20-22,34 The average number of colonoscopies that fellows should perform before they reach an adequate CIR of at least 90% varied greatly in these studies, ranging from 200 to 500 colonoscopies.20-22 Although we were unable to correct for the help of a supervisor, all fellows attained a CIR of 85%, and an adjusted CIR of 90% or higher during their training. As the CIR is an important quality indicator of colonoscopy, this should be monitored throughout training in colonoscopies in which the cecum was independently reached without help from a supervisor. A recent British study used trainee e-portfolios from the national Joint Advisory Group on GI Endoscopy (JETS) database to determine the number of colonoscopies needed to be performed to attain competency and discovered that this number was different per trainee.35 In response to this study, Bretthauer and Brandrud36 therefore question the number of colonoscopies performed as the most important measure of competency and emphasize that quality indicators such as the CIR and ADR should be measured during training instead. Although not statistically significant, the ADR varied widely among the gastroenterology consultants (range 19.8%-40.2%, P Z .066). This finding is in line with previous studies describing ranges in ADR of 7% to 44% and 22% to 44.7% in screening colonoscopies16,19 and 15.5% to 41.1% in diagnostic colonoscopies.17 Previous studies showed that simple monitoring of quality parameters, together with feedback and benchmarking, can improve colonoscopy quality and increase the ADR.29,37-39 During the training of the fellows in our study (2004-2012), no feedback was provided, and at that time, there was not as much awareness of the importance of quality indicators of colonoscopy. Because this study suggests that the ADR may already be predicted during colonoscopy training, feedback and benchmarking of all quality parameters including ADR should already be implemented early during training from the moment the fellow performs colonoscopies independently. This would result in better performing colonoscopists with higher ADRs in daily practice as a consultant. This allows comparison of quality of colonoscopy during training and specific actions for improvement of underperformers. Feedback can only be provided if colonoscopy reports include the key quality indicators. The first step would be to implement mandatory registration of quality indicators for all colonoscopies performed during training, preferably achieved by a structured colonoscopy report.36,40 In conclusion, this study demonstrated that variance in ADR is already present during the colonoscopy training of gastroenterology fellows. The majority of the fellows did not improve their ADR after completing their training. However, our data suggest that a low ADR during training is predictive of also detecting fewer adenomas as a consultant. These findings indicate that the ADR may already be predicted during colonoscopy training. We suggest
128 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
van Doorn et al
Adenoma detection rate varies during colonoscopy training
1. Zauber AG, Winawer SJ, O’Brien MJ, et al. Colonoscopic polypectomy and long-term prevention of colorectal-cancer deaths. N Engl J Med 2012;366:687-96. 2. Van Rijn JC, Reitsma JB, Stoker J, et al. Polyp miss rate determined by tandem colonoscopy: a systematic review. Am J Gastroenterol 2006;101:343-50. 3. Bressler B, Paszat LF, Chen Z, et al. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132:96-102. 4. Singh H, Turner D, Xue L, et al. Risk of developing colorectal cancer following a negative colonoscopy examination: evidence for a 10year interval between colonoscopies. JAMA 2006;295:2366-73. 5. Baxter NN, Goldwasser MA, Paszat LF, et al. Association of colonoscopy and death from colorectal cancer. Ann Intern Med 2009;150:1-8. 6. le Clercq CMC, Bouwens MWE, Rondagh EJ, et al. Postcolonoscopy colorectal cancers are preventable: a population-based study. Gut 2014;63:957-63. 7. Singh H, Nugent Z, Demers AA, et al. Rate and predictors of early/missed colorectal cancers after colonoscopy in Manitoba: a population-based study. Am J Gastroenterol 2010;105:2588-96. 8. Kaminski MF, Regula J, Kraszewska E, et al. Quality indicators for colonoscopy and the risk of interval cancer. N Engl J Med 2010;362: 1795-803. 9. Corley DA, Jensen CD, Marks AR, et al. Adenoma detection rate and risk of colorectal cancer and death. N Engl J Med 2014;370:1298-306. 10. Rex DK. Quality in colonoscopy: cecal intubation first, then what? Am J Gastroenterol 2006;101:732-4. 11. Rex DK, Petrini JL, Baron TH, et al. Quality indicators for colonoscopy. Gastrointest Endosc 2006;63(4 Suppl):S16-28. 12. Lee TJW, Rutter MD, Blanks RG, et al. Colonoscopy quality measures: experience from the NHS Bowel Cancer Screening Programme. Gut 2012;61:1050-7. 13. Chilton A, Rutter M. Quality assurance guidelines for colonoscopy. National Health Service Bowel Cancer Screening Programmes Publ No 6. 2011;(6). 14. National Institute for Public Health and the Environment. Protocol for the authorization and auditing of colonoscopy centres and endoscopists. The Netherlands. 2012;(June). http://www.rivm.nl/en/Documents_and_ publications/Professional_Serviceable/Protocols/Disease_Prevention_and_ Healthcare/Protocol_for_quality_assurance_of_FIT_analyses. 15. De Jonge V, Sint Nicolaas J, Cahen DL, et al. Quality evaluation of colonoscopy reporting and colonoscopy performance in daily clinical practice. Gastrointest Endosc 2012;75:98-106. 16. Greenspan M, Rajan KB, Baig A, et al. Advanced adenoma detection rate is independent of nonadvanced adenoma detection rate. Am J Gastroenterol 2013;108:1286-92. 17. Chen SC, Rex DK. Endoscopist can be more powerful than age and male gender in predicting adenoma detection at colonoscopy. Am J Gastroenterol 2007;17:856-61. 18. Piette C, Manfredi S. Variations between endoscopists in rates of detection of colorectal neoplasia and their impact on a regional screening program based on colonoscopy after fecal occult blood testing. Gastrointest Endosc 2010;71:335-41.
19. Imperiale TF, Glowinski EA, Juliar BE, et al. Variation in polyp detection rates at screening colonoscopy. Gastrointest Endosc 2009;69:1288-95. 20. Lee S-H, Chung I-K, Kim S-J, et al. An adequate level of training for technical competence in screening and diagnostic colonoscopy: a prospective multicenter evaluation of the learning curve. Gastrointest Endosc 2008;67:683-9. 21. Sedlack RE. Training to competency in colonoscopy: assessing and defining competency standards. Gastrointest Endosc 2011;74: 355-66.e1-2. 22. Spier BJ, Benson M, Pfau PR, et al. Colonoscopy training in gastroenterology fellowships: determining competence. Gastrointest Endosc 2010;71:319-24. 23. Peters SL, Hasan AG, Jacobson NB, et al. Level of fellowship training increases adenoma detection rates. Clin Gastroenterol Hepatol 2010;8:439-42. 24. Oh YS, Collins CL, Virani S, et al. Lack of impact on polyp detection by fellow involvement during colonoscopy: a meta-analysis. Dig Dis Sci 2013;58:3413-21. 25. Buchner AM, Shahid MW, Heckman MG, et al. Trainee participation is associated with increased small adenoma detection. Gastrointest Endosc 2011;73:1223-31. 26. Rogart JN, Siddiqui UD, Jamidar PA, et al. Fellow involvement may increase adenoma detection rates during colonoscopy. Am J Gastroenterol 2008;103:2841-6. 27. Eckardt AJ, Swales C, Bhattacharya K, et al. Does trainee participation during colonoscopy affect adenoma detection rates? Dis Colon Rectum 2009;52:1337-44. 28. Chalifoux SL, Rao DS, Wani SB, et al. Trainee participation and adenoma detection rates during screening colonoscopies. J Clin Gastroenterol 2014;48:524-9. 29. Corley DA, Jensen CD, Marks AR. Can we improve adenoma detection rates? A systematic review of intervention studies. Gastrointest Endosc 2011;74:656-65. 30. Kahi CJ, Vemulapalli KC, Johnson CS, et al. Improving measurement of the adenoma detection rate and adenoma per colonoscopy quality metric: the Indiana University experience. Gastrointest Endosc 2014;79:448-54. 31. Kaltenbach T, Friedland S, Soetikno R. A randomised tandem colonoscopy trial of narrow band imaging versus white light examination to compare neoplasia miss rates. Gut 2008;1406-12. 32. Adler A, Pohl H, Papanikolaou IS, et al. A prospective randomised study on narrow-band imaging versus conventional colonoscopy for adenoma detection: does narrow-band imaging induce a learning effect? Gut 2008;57:59-64. 33. Munroe CA, Lee P, Copland A, et al. A tandem colonoscopy study of adenoma miss rates during endoscopic training: a venture into uncharted territory. Gastrointest Endosc 2012;75:561-7. 34. Chung JI, Kim N, Um MS, et al. Learning curves for colonoscopy: a prospective evaluation of gastroenterology fellows at a single center. Gut Liver 2010;4:31-5. 35. Ward ST, Mohammed MA, Walt R, et al. An analysis of the learning curve to achieve competency at colonoscopy using the JETS database. Gut 2014;63:1746-54. 36. Bretthauer M, Brandrud AS. Endoscopy training: time to stop counting procedures? Gut 2014;63:1686-7. 37. Coe SG, Crook JE, Diehl NN, et al. An endoscopic quality improvement program improves detection of colorectal adenomas. Am J Gastroenterol 2013;108:219-26; quiz 227. 38. Lin OS, Kozarek RA, Arai A, et al. The effect of periodic monitoring and feedback on screening colonoscopy withdrawal times, polyp detection rates, and patient satisfaction scores. Gastrointest Endosc 2010;71: 1253-9. 39. Harewood GC, Petersen BT, Ott BJ. Prospective assessment of the impact of feedback on colonoscopy performance. Aliment Pharmacol Ther 2006;24:313-8. 40. Van Doorn SC, van Vliet J-FP, Dekker E. A novel colonoscopy reporting system enabling quality assurance. Endoscopy 2014:181-7.
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 129
that feedback and benchmarking of all quality indicators including the ADR be systematically implemented early during the entire training. We also suggest that each report during training should include the level of supervision during colonoscopy. Future prospective and multicenter studies should investigate the possibilities of quality monitoring of gastroenterology fellows and its influence in improving quality.
REFERENCES
Adenoma detection rate varies greatly during colonoscopy training Sascha C. van Doorn, MD, Robert B. Klanderman, MD, Yark Hazewinkel, MD, Paul Fockens, MD, PhD, Evelien Dekker, MD, PhD Amsterdam, the Netherlands
Background: The adenoma detection rate (ADR) is considered the most important quality indicator for colonoscopy and varies widely among colonoscopists. It is unknown whether the ADR of gastroenterology consultants can already be predicted during their colonoscopy training. Objective: To evaluate the ADR of fellows in gastroenterology and evaluate whether this predicts their ADR as gastroenterology consultants. Design: Retrospective observational study. Setting: Academic and regional centers. Patients: Symptomatic patients undergoing colonoscopy. Main Outcome Measurements: The variance in ADR among 7 gastroenterology fellows during their training (between May 2004 and March 2012) and of the same fellows after they registered as consultants (between October 2011 and April 2014) was evaluated. Multivariate logistic regression was performed to compare the highest detector (endoscopist with highest ADR) with the individual fellows and to evaluate whether an ADR of 20% or higher during the training was predictive of a high ADR as a consultant. Results: During training, ADRs ranged from 14% to 36% (P! .001). Compared with the highest detector, the OR for detecting an adenoma ranged from 0.64 (95% CI, 0.40-1.03) to 0.29 (95% CI, 0.17-0.48). After registration, ADR ranged from 19.8% to 40.2% (P Z .066). Compared with the highest detector during consultancy, the OR ranged from 0.64 (95% CI, 0.34-1.21) to 0.26 (95% CI, 0.13-0.52). Only 2 fellows significantly improved their ADR after completing their training. An ADR lower than 20% during training was associated with a lower ADR as a consultant (OR 0.51; 95% CI, 0.30-0.87). Limitations: Retrospective study. Conclusions: Variance in ADR is already present during the endoscopy training of gastroenterology fellows. Most fellows do not improve their ADR after completing their training. These findings suggest that the ADR can be predicted during colonoscopy training, and we suggest that feedback and benchmarking should be implemented early during training of fellows in an effort to improve ADR in future daily practice as a consultant. (Gastrointest Endosc 2015;82:122-9.)
Colorectal cancer (CRC) arises from precursor colorectal polyps. Detecting and resecting these lesions during colonoscopy decrease the CRC incidence.1 At the same time, it is well known that polyps are missed during
colonoscopy, and this can result in interval cancers in the years after colonoscopy.2-7 Although it is the ultimate quality indicator, the rate of interval cancers is a parameter that can only be used on a long-term basis and is therefore
Abbreviations: AMC, Academic Medical Center; ADR, adenoma detection rate; CI, confidence interval; CIR, cecal intubation rate; CRC, colorectal cancer; OR, odds ratio.
Received September 29, 2014. Accepted December 7, 2014.
DISCLOSURE: Dr. Fockens disclosed that he has received a research grant from Olympus. All other authors disclosed no financial relationships relevant to this article. Copyright ª 2015 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 http://dx.doi.org/10.1016/j.gie.2014.12.038
122 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
Current affiliations: Department of Gastroenterology and Hepatology, Academic Medical Centre, University of Amsterdam, Amsterdam, the Netherlands. Reprint requests: Evelien Dekker, MD PhD, Department of Gastroenterology and Hepatology, Academic Medical Centre, Meibergdreef 9 1105 AZ, Amsterdam, the Netherlands.
www.giejournal.org
van Doorn et al
not practical for monitoring quality in daily colonoscopy practice. Kaminski et al8 showed that the adenoma detection rate (ADR) of individual colonoscopists, ie, the proportion of colonoscopies in which at least 1 histologically confirmed adenoma is detected, is associated with the risk of future interval carcinomas. Recently, Corley et al9 confirmed the association reported by Kaminski et al and demonstrated that the ADR was inversely associated with death caused by interval colorectal cancer (CRC). Therefore, the ADR is considered an important quality indicator of colonoscopy, and current guidelines demand ADRs of at least 20% on screening colonoscopies.10-14 In the current literature, a wide variation in ADRs among colonoscopists is reported.12,15-19 Measurement and benchmarking of this quality indicator will perhaps improve awareness and training in an effort to continuously increase quality in colonoscopy. During endoscopy training of gastroenterology fellows, the most important competency for colonoscopy is safe cecal intubation.20-22 Only a few studies evaluated the effect of fellow participation in colonoscopy screening programs on ADR,23-28 but the literature totally lacks information on the ADR of individual gastroenterology fellows during the years of their colonoscopy training. It is unknown whether a variation in ADR is already present during training and whether measurement and benchmarking of ADR would improve ADR during training and would give fellows a better start for daily practice. Possibly the ADR of a gastroenterologist consultant can already be predicted during training. If consultants have a high ADR during training and are already aware of the importance of adenoma detection, the colonoscopy training could form the basis for an adequate ADR in daily practice. The aim of our study was to evaluate the variance in ADRs among gastroenterology fellows during the course of their colonoscopy training and to evaluate whether individual ADRs during training predict ADRs in later practice as a gastroenterology consultant.
Adenoma detection rate varies during colonoscopy training
who performed their entire endoscopy training in the AMC from May 2004 to March 2012 were included. Seven fellows fulfilled these criteria. For all 7 fellows, data on every colonoscopy that they performed were obtained from the colonoscopy database (version 11.0 of ENDOBASE, Olympus, Winter & Ibe GmbH) and reviewed. For all 7 fellows, reports of every colonoscopy that they performed during their training in the study period (May 2004-March 2012) were retrieved and reviewed from our colonoscopy database (ENDOBASE). Accompanying histopathology reports and electronic patient charts were also searched. All reports of colonoscopies performed during the study period 2004 to 2012 by 1 of the 7 fellows were reviewed. Subsequently, data on consecutive colonoscopies that the fellows performed after they were registered as gastroenterology consultants and practicing were collected. These data were collected from October 2011 to April 2014.
Colonoscopy procedure Colonoscopies performed by fellows during training at the AMC were either high-definition or standarddefinition white-light colonoscopies. Patients prepared by taking 4 L of a polyethylene glycol solution (Kleanprep; Norgine bv, Amsterdam, the Netherlands) or 2 L of a polyethylene glycol solution containing ascorbic acid (Moviprep; Norgine bv) with an additional 2 L of liquids. Procedures were performed with patients under conscious sedation with midazolam and/or fentanyl or under deeper sedation (propofol) when indicated. An Olympus 180 series colonoscope (Olympus Optical Co, Ltd, Tokyo, Japan) was used for the majority of procedures. For the colonoscopies performed as gastroenterology consultants, the colonoscope type, bowel preparation, and sedation use were unknown.
Colonoscopy data
We conducted a retrospective study in gastroenterology fellows trained at the Academic Medical Center (AMC) in Amsterdam, the Netherlands. Permission of the research protocol by an institutional review board was not required per the agreement regarding medical research involving human subjects. The endoscopy department of the AMC is a tertiary care facility and teaching hospital. In the Netherlands, gastroenterology fellows often perform part of their training at a teaching hospital in addition to a few years in one of the academic hospitals. At the AMC, gastroenterology fellows are randomly assigned to training at either the AMC or a teaching hospital. All gastroenterology fellows
For each colonoscopy, demographic information on the patient was collected from the electronic patient chart: age, sex, previous CRC and/or colonic resection, and medical history regarding inflammatory bowel disease, polyposis, or Lynch syndrome. Data collected on each colonoscopy included indication for colonoscopy (screening, surveillance, or diagnostic), bowel preparation (if available with Boston Bowel Preparation score, otherwise scored as good, suboptimal, poor, or unknown), sedatives used, cecal intubation rate (CIR), and deepest point of colonoscope insertion and reason for incomplete colonoscopy if the cecum was not reached. The number of detected and removed polyps was recorded, as were the location in the colon, polypectomy method, and histopathology result (adenomas: tubular, tubulovillous, or villous and grade of dysplasia: low or high; serrated polyps subdivided in hyperplastic polyps, sessile serrated polyps, and traditional serrated polyps; adenocarcinomas), if applicable.
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 123
METHODS Study design and data collection
Adenoma detection rate varies during colonoscopy training
van Doorn et al
Inclusion and exclusion criteria for colonoscopies included in the analysis
and adjusted for poor bowel preparation or a stricture) were calculated for all colonoscopies in an intact colon, including excluded colonoscopies for the detection analyses. P ! .05 was considered statistically significant. SPSS software (version 20.0; IBM Corporation, Somers, NY) was used for analysis.
In the Dutch colonoscopy training system, the first 100 procedures are completely supervised by a consultant present in the endoscopy suite, followed by a period of “on-demand” supervision. On-demand supervision means that a consultant is available whenever the fellow asks for supervision, eg, when the fellow has difficulty reaching the cecum or when a therapeutic procedure like polypectomy is performed. The colonoscopies during training were divided in 3 distinct periods: first 100 supervised colonoscopies, first 100 autonomous colonoscopies, and the remaining colonoscopies during the training. The first 100 colonoscopies, which were probably closely supervised, were excluded from analyses. To ensure a homogeneous cohort of patients per fellow or consultant, exclusion criteria were applied after stratification into 1 of the 3 training periods or the consultancy period. We excluded colonoscopies performed in patients younger than 18 years of age, emergency procedures as well as colonoscopies performed in patients with conditions that predispose to either a very high or very low probability of detecting adenomas: inflammatory bowel disease, polyposis syndromes, Lynch syndrome, referral for adenoma removal, referral after a positive screening test result, eg, fecal occult blood test or CT colonography. In addition, patients with an incomplete colonoscopy were excluded from the detection analyses. To correct for differences in colonoscopy indications, procedure indications were divided into 2 groups: symptomatic, including patients with anemia, melena, hematochezia, changed bowel habits, diarrhea, and abdominal pain and surveillance/screening, including average-risk screening, familial predisposition for CRC, and adenoma or cancer surveillance.
RESULTS Patient characteristics
ADR was defined as the proportion of colonoscopies in which at least 1 histologically confirmed adenoma was detected. c2 analyses were performed to evaluate the variance in ADR among the 7 fellows during their training and during their consultancy and the difference in ADR during training and consultancy per individual fellow. Univariate logistic regression modeling was performed with the following variables: patient age, patient sex, and grouped colonoscopy indication, ie, symptoms or screening and surveillance. The difference in ADR between fellows and consultants was evaluated with multivariate logistic regression to compare the highest detector (the endoscopist with the highest ADR) with the other individual endoscopists, adjusted for significant variables in the univariate regression model. Logistic regression was performed to evaluate whether an ADR of 20% or higher during training was predictive of a high ADR during the consultancy period. Individual fellow CIRs (unadjusted
During the study period, 7 fellows performed a total of 3252 colonoscopies (Table 1). After exclusion of the first 100 procedures per fellow and exclusion of colonoscopies that met the exclusion criteria, 1472 colonoscopies (45% of all colonoscopies) were available for analysis. The mean age of the patients who underwent colonoscopy during the training period was 56 years, and 49% of the patients were male. After exclusion, the mean age was 58 years and 48% were male. Most of the colonoscopy reports did not report on presence of a supervisor, and we assumed a supervisor was only present briefly during procedures if a polypectomy was performed or when called by the fellow for another reason and did not contribute to the ADR. In some reports, it was stated that the supervisor took over the colonoscopy, for example, if the fellow could not intubate the cecum or if a large polyp was removed. In cases in which no supervisor was present during the whole procedure, this was also stated in the report. Data on consecutive colonoscopies that the fellows performed after they were registered as gastroenterology consultants and practicing were collected. Two consultants practiced at an academic center and 5 at a regional hospital. Five of 7 gastroenterology consultants collected data on at least 100 nonselected consecutive colonoscopies that they submitted for accreditation for the Dutch FIT-based CRC screening program, and these data were used for this study. To prevent selection bias, the data that the consultants had submitted for accreditation for the screening program, we verified whether the procedures were indeed consecutive. We visited their practices to review the patient charts. For the 2 other consultants, we collected data on consecutive colonoscopies from the first years of consultancy from colonoscopy databases. After exclusion based on age, indication, and incomplete colonoscopies, 692 colonoscopies (75% of all colonoscopies) were available for analysis (Table 1). During the consultancy period, the mean patient age was 59 years, and 44% were male. After exclusion, the mean patient age remained 59 years and 43% of the patients were male. The mean patient age and percentage of male patients did not differ between fellows and consultants. During training, the unadjusted CIR ranged from 86.4% to 96.0% and the adjusted CIR (excluding procedures with poor bowel preparation and obstruction) from 94.0% to 99.1%. The unadjusted CIRs during the consultancy period
124 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
Study outcomes and statistical analyses
van Doorn et al
Adenoma detection rate varies during colonoscopy training
TABLE 1. Number of procedures during training and consultancy
Fellow 1
Training period
Training period after exclusion* (% included from total number of procedures)
Consultancy period
Consultancy period after exclusion* (% included from total number of procedures)
433
144 (33%)
136
109 (80%)
166
106
81
68
Surveillance
52
38
41
41
Other
215
Symptoms
Fellow 2
14
389
147 (38%)
171
111 (65%)
Symptoms
175
102
39
34
Surveillance
59
45
78
77
479
207 (43%)
100
97 (97%)
Symptoms
177
126
84
84
Surveillance
109
81
14
13
Other
193
Other Fellow 3
Fellow 4
155
54
2
612
333 (54%)
107
91 (85%)
312
245
70
64
Surveillance
99
88
27
27
Other
201
Symptoms
Fellow 5
10
415
194 (47%)
107
89 (83%)
Symptoms
202
132
61
57
Surveillance
82
62
33
32
417
224 (54%)
101
94 (93%)
Symptoms
225
167
74
73
Surveillance
89
57
21
21
507
223 (44%)
206
101 (49%)
Symptoms
199
150
81
77
Surveillance
106
73
25
24
Other
202
Other Fellow 6
Other Fellow 7
131
13
103
6
100
Symptoms
1456
1028
478
457
Surveillance
596
444
231
235
Other
1200
1472 (45%)
107
692 (75%)
Total
3252
928
*Excluded were patients younger than 18 years of age, incomplete colonoscopies, other indication (acute colonoscopy, positive fecal occult bleed test result, positive findings on CT colonography, polyposis syndrome, Lynch syndrome, inflammatory bowel disease [IBD]) and excluding first 100 procedures.
ranged from 89.3% to 99.0% and the adjusted CIRs from 94.2% to 100%. No significant increase in CIR was detected after the first 100 colonoscopies or compared with their practice when being a consultant.
Colonoscopies during training During the overall training period, the ADR ranged from 13.5% to 36.1% and differed significantly among fellows (P! .001) (Table 2). Univariate logistic regression analyses showed a significant association between the OR of finding at least 1 adenoma and the patient’s age, sex, and www.giejournal.org
colonoscopy indication (P ! .001 for all 3 associations). After adjusting for these variables, the OR to detect at least 1 adenoma ranged from 0.64 (95% CI, 0.40-1.03) to 0.29 (95% CI, 0.17-0.48) among all fellows compared with the highest detector, although not significantly for fellows 2 and 3 (Table 2).
Colonoscopies during consultancy The ADR among the 7 gastroenterology consultants ranged from 19.8% to 40.2%; the difference was not statistically significant (P Z .066) (Table 3). Again, univariate Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 125
Adenoma detection rate varies during colonoscopy training
van Doorn et al
TABLE 2. Comparison of ADR during training (ranked by OR)* N [ 1472
ADR, %
OR* (95% CI)
P value
Overall
24.6
Fellow 1
36.1
1
Fellow 3
30.4
0.63 (0.39-1.02)
Fellow 2
27.9
0.60 (0.36-1.01)
.056
Fellow 4
25.2
0.59 (0.38-0.92)
.019
Fellow 6
22.3
0.54 (0.33-0.88)
.013
Fellow 5
21.6
0.44 (0.27-0.74)
.002
Fellow 7
13.5
0.22 (0.13-0.38)
<.001
.060
ADR, Adenoma detection rate; OR, odds ratio; CI, confidence interval. Statistically significant P values are in bold. *Adjusted for patient age, sex, and indication, after exclusion.
fellows with an ADR higher than 20% during training. We also evaluated whether an ADR lower than 20% during the first 100 colonoscopies autonomously performed was associated with an increased ADR during the remainder of the training. Fellows 5, 6, and 7 had ADRs lower than 20% during the first 100 procedures (18.8%, 18.5%, and 8.6%, respectively) (Table 4), and this was associated with a significantly lower ADR during the remainder of the training (OR 0.69; 95% CI, 0.51-0.93), but not with a lower ADR during consultancy (P Z .68).
DISCUSSION
Only for gastroenterology consultants 5 and 6, the ADR was significantly higher during consultancy than it was during the years of their training (Fig. 1). We evaluated whether an ADR lower than 20% during the overall training period was associated with a lower ADR as a gastroenterology consultant, adjusted for patient age, sex, and colonoscopy indication. During the overall training period, only fellow 7 had an ADR lower than 20% (13.5%), and this was associated with a significantly lower ADR during consultancy (OR 0.51; 95% CI, 0.30-0.87) compared with
The ADR is considered the most important surrogate quality indicator for colonoscopy and varies widely among colonoscopists.12,15-19 In this study, we evaluated whether a variation in ADR is already present during colonoscopy training and whether individual ADRs during training predict ADR in later practice. After correction for age, sex, and colonoscopy indication, we detected a significant variance in ADR among 7 fellows during their training (range 13.5%-36.1%, P ! .001). Although the difference in ADR among these fellows when they were gastroenterology consultants did not reach statistical significance, after adjusting for age, sex, and colonoscopy indication, when comparing consultants with the highest detector, this was significant for 5 of the 6 gastroenterology consultants. Only 2 of the fellows improved their ADR after their training (from 21.6% to 38.2%, P Z .004 and from 22.3% to 36.2%, P Z .011). Furthermore, our data suggest that an ADR lower than 20% during training is associated with a lower ADR when a gastroenterology consultant. This study has several limitations. There were only 7 gastroenterology fellows included in this study, as we chose to include only fellows who performed their entire endoscopy training at the AMC to ensure that all fellows received the same training, level of supervision, and followed the same protocols. The fact that only fellows performing their training at the AMC were included may limit the generalizability of the results. The small number of participating fellows could have led to a bias because we could only compare 1 fellow with an ADR lower than 20% with the other fellows in our regression model. A future study should include more fellows, preferably from different teaching hospitals. Because of the retrospective design of this study, not all quality indicators were reported during colonoscopies. We could not correct the ADRs for poor bowel preparation, a potential confounder, as this information was not always included in the colonoscopy reports because the importance of reporting on quality indicators of colonoscopies was not as much acknowledged at that time. The retrospective design, however, also has an advantage as it eliminates possible observer bias (Hawthorn effect; colonoscopists might be more aggressive in polyp detection than usual
126 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
TABLE 3. Comparison of ADR during consultancy (ranked by OR ratio)* N [ 672 Overall
ADR, %
OR* (95% CI)
P value
33.3
Gastroenterology consultant 3y
40.2
1
Gastroenterology consultant 5
38.2
0.64 (0.34-1.21)
Gastroenterology consultant 6
36.2
0.51 (0.27-0.98)
.044
Gastroenterology consultant 4
33.0
0.45 (0.23-0.86)
.015
Gastroenterology consultant 2
32.1
0.39 (0.21-0.74)
.004
Gastroenterology consultant 1
32.4
0.32 (0.17-0.62)
.001
Gastroenterology consultant 7
19.8
0.26 (0.13-0.52)
<.001
.172
ADR, Adenoma detection rate; OR, odds ratio; CI, confidence interval. Statistically significant P values are in bold. *Adjusted for patients age, sex, and indication, after exclusion yGastroenterology consultants are numbered in the same order as during the training.
logistic analyses were performed to calculate the OR of adenoma detection corrected for age, sex, and colonoscopy indication, which were all statistically significant related to ADR (P ! .001). Although the difference in ADR did not reach statistical significance, after adjusting for age, sex, and colonoscopy indication, comparison with the highest detector was significant for 5 of the 6 gastroenterology consultants (OR ranged from 0.64 [95% CI, 0.341.21] to 0.26 [95% CI, 0.13-0.52]) (Table 3).
Association between ADR during training and consultancy
van Doorn et al
Adenoma detection rate varies during colonoscopy training
p< .05
p< .01
Fellow 4
Fellow 5
Fellow 6
Fellow 7
Adenoma detection rate
40%
30%
20%
10%
0% Fellow 1
Fellow 2
Fellow 3
Training
36.1%
27.9%
30.4%
25.2%
21.6%
22.3%
13.5%
Consultancy
32.1%
32.4%
40.2%
33.0%
38.2%
36.2%
19.8%
Figure 1. For each fellow, the ADR during training and during consultancy are shown. Only for fellow 5 and 6 the ADR was significantly higher during consultancy than it was during the years of training.
when they are aware that their detection rates are being recorded29). Another limitation is the heterogeneity of colonoscopy indications. Unfortunately, this is was inevitable because we aimed to include all colonoscopies the fellows performed during the training and because of the retrospective design of the study. The heterogeneity of colonoscopy indications, however, seemed equal among the different fellows. Only 45% of all colonoscopies performed during training were included (1472 of 3235 colonoscopies) in the analyses compared with 75% (692 of 928 colonoscopies) during consultancy. These rates were based on exclusion of the first 100 supervised colonoscopies, incomplete colonoscopies, emergency procedures,
positive results on CT colonography, and procedures in patients with polyposis syndrome, which were more frequent during training compared with consultancy. For consultants 2 and 7, relatively more patients were excluded than for the others as they performed, respectively, many colonoscopies in patients with polyposis or Lynch syndrome or in patients with inflammatory bowel disease. The number of colonoscopies performed by the consultants was relatively small compared with the number that they performed when they were fellows. It was difficult to collect data from the consultancy period, as they all worked at different hospitals with different reporting systems and electronic patient files. We chose to include a random selection of consecutive colonoscopies that they performed in their practice to obtain a reliable impression of the detection rates of all individual consultants. Finally, in the analyses, the limit of a low detector was set at lower than 20% because this was found to be associated with an increased risk of interval carcinomas.8 However, the original study by Kaminski et al8 was performed in a colonoscopy screening population. In this study, only diagnostic and surveillance colonoscopies were included, for which a minimal ADR of 20% might be too low. In the scarce literature, a recent study reported that the ADR stratified by colonoscopy indication was highest for surveillance colonoscopies, followed by screening, and then diagnostic,30 and ADRs in symptomatic and surveillance populations vary from 17% to 49%.31,32 We decided to use 20% as a lower threshold as the fellows performed the colonoscopies between 2004 and 2012 when colonoscopists were less aware of the
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 127
TABLE 4. ADR during first 100 autonomous colonoscopies No. of procedures*
No. of procedures with R1 adenoma
ADR, %
Fellow 1
34
11
32.4
Fellow 2
55
23
41.8
Fellow 3
52
20
38.5
Fellow 4
68
16
23.5
Fellow 5
64
12
18.8
Fellow 6
65
12
18.5
Fellow 7
58
5
8.6
Total
396
99
25.0
*After exclusion of patients younger than 18 years of age, incomplete colonoscopies, and other indication (acute colonoscopy, positive fecal occult bleed test result, positive findings on CT colonography, polyposis syndrome, Lynch syndrome, inflammatory bowel disease).
Adenoma detection rate varies during colonoscopy training
van Doorn et al
importance of the ADR. Future studies are needed to establish the threshold ADR for other populations than screening. The ADR is considered the most important surrogate quality indicator of colonoscopy because it is inversely associated with the risk of interval carcinomas.8,9 To our knowledge, our study is the first to evaluate individual ADRs of gastroenterology fellows during the years of their colonoscopy training. Most previous studies evaluating the ADR of fellows focused on comparison of the ADR of gastroenterology consultants alone with the ADR in colonoscopies in which a fellow was involved.23,25-27 Two studies reported a significantly higher ADR in colonoscopies in which a fellow was involved and observed an increase with each year of training.23,26 Conversely, a meta-analysis from 2013 did not observe an effect on ADR by fellow involvement, and only 1 study in this meta-analysis observed a decrease in the ADR with fellow involvement.24 Only 2 studies evaluated ADR during the course of training and observed no increase in ADR over time; individual fellow ADRs were not reported.20,22 To be able to report on actual independent fellow adenoma detection, in our study we only included colonoscopies for which no supervisor was present during withdrawal. We observed a large variance in ADR already present during training (13.5%-36%). In a small study of only 218 patients, Munroe et al33 observed a decrease in adenoma miss rates with increasing experience, in a tandem colonoscopy study in which withdrawal was first performed by the trainee and then by an attending physician. No individual ADRs of fellows were reported.33 In contrast with these results, we observed that having an ADR lower than 20% during the first 100 autonomously performed colonoscopies in training was associated with a significantly lower ADR during the remainder of the training. We excluded the first 100 colonoscopies because these procedures were completely supervised, and the fellows did not perform individual withdrawal. We do not have data regarding the supervisors at the time; however, the group of staff members in our department that provide supervision of fellows has not changed much during the study period. Whether the emphasis of the supervising mentor is on polyp detection might affect a fellow’s ability to achieve a high ADR in later practice. Prospective studies should investigate the role of supervisors in the ADR of fellows in training. A second observation was that the single fellow who had an ADR lower than 20% during the complete training period also had a low ADR during consultancy. Only 2 fellows improved their ADR significantly after their training. The number of fellows that we could include is too small to draw strong conclusions; however, our data suggest that a high ADR of gastroenterology consultants may already be predicted during their training. Larger studies evaluating colonoscopy quality of more fellows throughout their entire colonoscopy training are needed to confirm our findings.
Previous studies evaluating competence in colonoscopy focused on cecal intubation.20-22,34 The average number of colonoscopies that fellows should perform before they reach an adequate CIR of at least 90% varied greatly in these studies, ranging from 200 to 500 colonoscopies.20-22 Although we were unable to correct for the help of a supervisor, all fellows attained a CIR of 85%, and an adjusted CIR of 90% or higher during their training. As the CIR is an important quality indicator of colonoscopy, this should be monitored throughout training in colonoscopies in which the cecum was independently reached without help from a supervisor. A recent British study used trainee e-portfolios from the national Joint Advisory Group on GI Endoscopy (JETS) database to determine the number of colonoscopies needed to be performed to attain competency and discovered that this number was different per trainee.35 In response to this study, Bretthauer and Brandrud36 therefore question the number of colonoscopies performed as the most important measure of competency and emphasize that quality indicators such as the CIR and ADR should be measured during training instead. Although not statistically significant, the ADR varied widely among the gastroenterology consultants (range 19.8%-40.2%, P Z .066). This finding is in line with previous studies describing ranges in ADR of 7% to 44% and 22% to 44.7% in screening colonoscopies16,19 and 15.5% to 41.1% in diagnostic colonoscopies.17 Previous studies showed that simple monitoring of quality parameters, together with feedback and benchmarking, can improve colonoscopy quality and increase the ADR.29,37-39 During the training of the fellows in our study (2004-2012), no feedback was provided, and at that time, there was not as much awareness of the importance of quality indicators of colonoscopy. Because this study suggests that the ADR may already be predicted during colonoscopy training, feedback and benchmarking of all quality parameters including ADR should already be implemented early during training from the moment the fellow performs colonoscopies independently. This would result in better performing colonoscopists with higher ADRs in daily practice as a consultant. This allows comparison of quality of colonoscopy during training and specific actions for improvement of underperformers. Feedback can only be provided if colonoscopy reports include the key quality indicators. The first step would be to implement mandatory registration of quality indicators for all colonoscopies performed during training, preferably achieved by a structured colonoscopy report.36,40 In conclusion, this study demonstrated that variance in ADR is already present during the colonoscopy training of gastroenterology fellows. The majority of the fellows did not improve their ADR after completing their training. However, our data suggest that a low ADR during training is predictive of also detecting fewer adenomas as a consultant. These findings indicate that the ADR may already be predicted during colonoscopy training. We suggest
128 GASTROINTESTINAL ENDOSCOPY Volume 82, No. 1 : 2015
www.giejournal.org
van Doorn et al
Adenoma detection rate varies during colonoscopy training
1. Zauber AG, Winawer SJ, O’Brien MJ, et al. Colonoscopic polypectomy and long-term prevention of colorectal-cancer deaths. N Engl J Med 2012;366:687-96. 2. Van Rijn JC, Reitsma JB, Stoker J, et al. Polyp miss rate determined by tandem colonoscopy: a systematic review. Am J Gastroenterol 2006;101:343-50. 3. Bressler B, Paszat LF, Chen Z, et al. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132:96-102. 4. Singh H, Turner D, Xue L, et al. Risk of developing colorectal cancer following a negative colonoscopy examination: evidence for a 10year interval between colonoscopies. JAMA 2006;295:2366-73. 5. Baxter NN, Goldwasser MA, Paszat LF, et al. Association of colonoscopy and death from colorectal cancer. Ann Intern Med 2009;150:1-8. 6. le Clercq CMC, Bouwens MWE, Rondagh EJ, et al. Postcolonoscopy colorectal cancers are preventable: a population-based study. Gut 2014;63:957-63. 7. Singh H, Nugent Z, Demers AA, et al. Rate and predictors of early/missed colorectal cancers after colonoscopy in Manitoba: a population-based study. Am J Gastroenterol 2010;105:2588-96. 8. Kaminski MF, Regula J, Kraszewska E, et al. Quality indicators for colonoscopy and the risk of interval cancer. N Engl J Med 2010;362: 1795-803. 9. Corley DA, Jensen CD, Marks AR, et al. Adenoma detection rate and risk of colorectal cancer and death. N Engl J Med 2014;370:1298-306. 10. Rex DK. Quality in colonoscopy: cecal intubation first, then what? Am J Gastroenterol 2006;101:732-4. 11. Rex DK, Petrini JL, Baron TH, et al. Quality indicators for colonoscopy. Gastrointest Endosc 2006;63(4 Suppl):S16-28. 12. Lee TJW, Rutter MD, Blanks RG, et al. Colonoscopy quality measures: experience from the NHS Bowel Cancer Screening Programme. Gut 2012;61:1050-7. 13. Chilton A, Rutter M. Quality assurance guidelines for colonoscopy. National Health Service Bowel Cancer Screening Programmes Publ No 6. 2011;(6). 14. National Institute for Public Health and the Environment. Protocol for the authorization and auditing of colonoscopy centres and endoscopists. The Netherlands. 2012;(June). http://www.rivm.nl/en/Documents_and_ publications/Professional_Serviceable/Protocols/Disease_Prevention_and_ Healthcare/Protocol_for_quality_assurance_of_FIT_analyses. 15. De Jonge V, Sint Nicolaas J, Cahen DL, et al. Quality evaluation of colonoscopy reporting and colonoscopy performance in daily clinical practice. Gastrointest Endosc 2012;75:98-106. 16. Greenspan M, Rajan KB, Baig A, et al. Advanced adenoma detection rate is independent of nonadvanced adenoma detection rate. Am J Gastroenterol 2013;108:1286-92. 17. Chen SC, Rex DK. Endoscopist can be more powerful than age and male gender in predicting adenoma detection at colonoscopy. Am J Gastroenterol 2007;17:856-61. 18. Piette C, Manfredi S. Variations between endoscopists in rates of detection of colorectal neoplasia and their impact on a regional screening program based on colonoscopy after fecal occult blood testing. Gastrointest Endosc 2010;71:335-41.
19. Imperiale TF, Glowinski EA, Juliar BE, et al. Variation in polyp detection rates at screening colonoscopy. Gastrointest Endosc 2009;69:1288-95. 20. Lee S-H, Chung I-K, Kim S-J, et al. An adequate level of training for technical competence in screening and diagnostic colonoscopy: a prospective multicenter evaluation of the learning curve. Gastrointest Endosc 2008;67:683-9. 21. Sedlack RE. Training to competency in colonoscopy: assessing and defining competency standards. Gastrointest Endosc 2011;74: 355-66.e1-2. 22. Spier BJ, Benson M, Pfau PR, et al. Colonoscopy training in gastroenterology fellowships: determining competence. Gastrointest Endosc 2010;71:319-24. 23. Peters SL, Hasan AG, Jacobson NB, et al. Level of fellowship training increases adenoma detection rates. Clin Gastroenterol Hepatol 2010;8:439-42. 24. Oh YS, Collins CL, Virani S, et al. Lack of impact on polyp detection by fellow involvement during colonoscopy: a meta-analysis. Dig Dis Sci 2013;58:3413-21. 25. Buchner AM, Shahid MW, Heckman MG, et al. Trainee participation is associated with increased small adenoma detection. Gastrointest Endosc 2011;73:1223-31. 26. Rogart JN, Siddiqui UD, Jamidar PA, et al. Fellow involvement may increase adenoma detection rates during colonoscopy. Am J Gastroenterol 2008;103:2841-6. 27. Eckardt AJ, Swales C, Bhattacharya K, et al. Does trainee participation during colonoscopy affect adenoma detection rates? Dis Colon Rectum 2009;52:1337-44. 28. Chalifoux SL, Rao DS, Wani SB, et al. Trainee participation and adenoma detection rates during screening colonoscopies. J Clin Gastroenterol 2014;48:524-9. 29. Corley DA, Jensen CD, Marks AR. Can we improve adenoma detection rates? A systematic review of intervention studies. Gastrointest Endosc 2011;74:656-65. 30. Kahi CJ, Vemulapalli KC, Johnson CS, et al. Improving measurement of the adenoma detection rate and adenoma per colonoscopy quality metric: the Indiana University experience. Gastrointest Endosc 2014;79:448-54. 31. Kaltenbach T, Friedland S, Soetikno R. A randomised tandem colonoscopy trial of narrow band imaging versus white light examination to compare neoplasia miss rates. Gut 2008;1406-12. 32. Adler A, Pohl H, Papanikolaou IS, et al. A prospective randomised study on narrow-band imaging versus conventional colonoscopy for adenoma detection: does narrow-band imaging induce a learning effect? Gut 2008;57:59-64. 33. Munroe CA, Lee P, Copland A, et al. A tandem colonoscopy study of adenoma miss rates during endoscopic training: a venture into uncharted territory. Gastrointest Endosc 2012;75:561-7. 34. Chung JI, Kim N, Um MS, et al. Learning curves for colonoscopy: a prospective evaluation of gastroenterology fellows at a single center. Gut Liver 2010;4:31-5. 35. Ward ST, Mohammed MA, Walt R, et al. An analysis of the learning curve to achieve competency at colonoscopy using the JETS database. Gut 2014;63:1746-54. 36. Bretthauer M, Brandrud AS. Endoscopy training: time to stop counting procedures? Gut 2014;63:1686-7. 37. Coe SG, Crook JE, Diehl NN, et al. An endoscopic quality improvement program improves detection of colorectal adenomas. Am J Gastroenterol 2013;108:219-26; quiz 227. 38. Lin OS, Kozarek RA, Arai A, et al. The effect of periodic monitoring and feedback on screening colonoscopy withdrawal times, polyp detection rates, and patient satisfaction scores. Gastrointest Endosc 2010;71: 1253-9. 39. Harewood GC, Petersen BT, Ott BJ. Prospective assessment of the impact of feedback on colonoscopy performance. Aliment Pharmacol Ther 2006;24:313-8. 40. Van Doorn SC, van Vliet J-FP, Dekker E. A novel colonoscopy reporting system enabling quality assurance. Endoscopy 2014:181-7.
www.giejournal.org
Volume 82, No. 1 : 2015 GASTROINTESTINAL ENDOSCOPY 129
that feedback and benchmarking of all quality indicators including the ADR be systematically implemented early during the entire training. We also suggest that each report during training should include the level of supervision during colonoscopy. Future prospective and multicenter studies should investigate the possibilities of quality monitoring of gastroenterology fellows and its influence in improving quality.
REFERENCES