Introduction
Endoscopic retrograde cholangiopancreatography (ERCP) is one of the most advanced
and technically demanding therapeutic procedures in endoscopy, requiring extensive
hands-on training before achieving competency and undertaking independent practice.
In most training centers threshold numbers are still used as a surrogate for competency,
although there is a growing body of evidence showing that mere caseload does not guarantee
competency in ERCP, with many trainees failing to reach predefined thresholds at the
end of their training [1]. Current data support a paradigm shift, from the traditional volume-based approach
to a competency-based approach to training in ERCP [2].
In addition to the change underway in hands-on training, simulator-based training
using mechanical, digital or animal models has been proposed as a solution to shorten
the learning curve in endoscopy [3]
[4]
[5]. While several models for ERCP have been described [6]
[7], the Boškoski-Costamagna mechanical trainer (BCMT) (Cook Medical) is, to our knowledge,
the first mechanical model that has been independently validated and which was deemed
to show good face validity as a training model for novice endoscopists in an evidence-based
fashion [8].
A novel application of this model [9], which was shown to be capable of fulfilling innovative tasks such as writing or
drawing with the aid of a duodenoscope and a modified ERCP catheter, was adopted by
our group as the basis for motion-training. Thinking outside the box, we hypothesized
that trainees unfamiliar with the handling of a side-viewing scope and accessories,
could better acquire basic skills in ERCP (i. e. positioning in front of the papilla,
scope and accessory handling, selective cannulation etc.) by first practicing more
familiar motor tasks such as writing their names or drawing simple shapes using the
BCMT and a modified ERCP catheter.
In our current study, we aimed to assess whether motion training offers an additional
benefit to standard training on the BCMT in achieving the skills required for successful
cannulation.
Methods
We conducted a multicenter, randomized clinical trial of trainee performance on the
mechanical simulator, comparing the cannulation skills of trainees receiving standard
training on the model versus those undergoing motion training prior to attempting
cannulation on the BCMT.
Participant selection and group allocation
Endoscopy trainees without prior experience in ERCP (including simulator training)
but with experience in upper and lower GI endoscopy (> 100 cases each) from seven
centers in Europe (Bucharest, Rome, Zagreb, Bologna, Oxford, Brussels, and Strasbourg)
were invited to participate in the study. Trainees included in the study were randomized
to undergo motion training followed by standard cannulation training (motion training
group) or standard cannulation training alone (standard training group) on the mechanical
simulator; both trainees and trainers were aware of the group allocation of the trainees
during the study ([Fig. 1]). Because this was an investigator-driven initiative that initially included only
two centers, block randomization could not be implemented and sequentially numbered
sealed opaque envelopes prepared at each individual center were used to allocate participants
to their respective study group.
Fig. 1 Flowchart of the E-motion study (according to the CONSORT criteria).
Training protocol
Cannulation training was performed using the Boškoski-Costamagna mechanical endoscopic
simulator (Cook Medical, Limerick, Ireland) consisting of a lightweight metal cage
and plastic components to represent the upper gastrointestinal tract. The simulator
provides trainees with a suitable environment to learn and practice key maneuvers
in ERCP training such as scope handling and correct positioning in front of the papilla,
cannulation of the desired duct, stone extraction or stent placement [9]. Selective cannulation, one of the cornerstones of successful ERCP as well as one
of the most complex parts of the procedure is simulated with the use of a non-biological
replica of the papilla made out of latex with a transparent bilio-pancreatic duct
system attached. There are several types of papillae available (small orifice, large
orifice, common channel or parallel ducts for biliary and pancreatic access) [10] and adjustment of its position is possible for easier or more difficult access,
with the intention of simulating known anatomical variants ([Fig. 2]). The model is ergonomic and quick to assemble and it uses actual duodenoscopes
and accessories, making it easy and convenient to use in variable teaching settings.
Fig. 2 The Boškoski-Costamagna ERCP trainer (Cook Medical, Limerick, Ireland) a Non-biological cannulation models consisting of a papilla and biliopancreatic ducts
in various anatomical designs (for easy, medium and difficult cannulation). b Cannulation attempt of the papilla using a standard duodenoscope and sphincterotome.
c Complete simulator setup consisting of: endoscopy trolley, standard duodenoscope,
the Boškoski-Costamagna ERCP trainer and simulated fluoroscopy unit with the aid of
a HD camera and monitor
Motion training
Trainees randomized to the motion training group initially underwent motion training,
which consisted of completing five tasks, such as writing and drawing, using the mechanical
simulator, duodenoscope, and a specially modified 10Fr catheter with a graphite tip
end before standard training and any attempt at cannulation of the papilla replica.
For this purpose, the papilla on the model was replaced with a piece of white paper
([Fig. 3]) and the trainees were asked to try writing their names and draw simple shapes of
their own choice. For each task, they were allotted 5-minute slots and they did not
receive any verbal or hands-on instruction from their supervisors.
Fig. 3 Motion training. Attempt at writing using a standard duodenoscope with a modified
catheter with a graphite tip on a white piece of paper.
Standardized cannulation training
Both groups received standardized cannulation training using the model. This was performed
at each center by an expert endoscopist (> 2000 procedures), with the aim of explaining
the practical aspects of selective common bile duct cannulation.
Briefly, the expert at each participating center explained the basic theory of scope
handling with an emphasis on using a side-viewing scope, positioning in front of the
papilla, and use of various dedicated accessories for ERCP. This theoretical briefing
was followed by a demonstration of cannulation technique by the expert operator, using
the mechanical simulator and highlighting the essential steps of cannulation: scope
insertion, positioning in front of the papilla, use of the duodenoscope, wheels, shaft,
and elevator, and selective cannulation using a sphincterotome and a 0.035i guidewire
([Video 1]).
Video 1 The Emotion Study: Motion training, theoretical briefing and cannulation attempts
performed on the Boskoski-Costamagna ERCP trainer.
Following this demonstration, trainees proceeded to perform attempt cannulation on
the mechanical model, under the supervision of the expert endoscopist.
Cannulation attempts
Trainees in both groups attempted deep cannulation of the common bile duct using a
standard sphincterotome and a 0.035i guidewire with the local expert handling the
accessories.
The trainees were required to complete a total of 20 cannulation attempts on four
different papilla models, each attempt being timed from the moment the sphincterotome
was out of the duodenoscope channel to the moment when deep cannulation of the common
bile duct was confirmed by direct visualization of the guidewire inside the transparent
duct. If cannulation was not achieved after 5 minutes, the attempt was considered
a failure. Four different difficulty levels of cannulation were simulated by adjusting
the papilla type and angulation ([Table 1] and [Fig. 2a])
Table 1
Papilla types used for cannulation attempts.
|
Attempts 0–5
|
Attempts 6–10
|
Attempts 11–15
|
Attempts 16–20
|
|
Common bilio-pancreatic channel, normal inclination
|
Common bilio-pancreatic channel, difficult inclination
|
Separate bile duct channel, normal inclination
|
Separate bile duct channel, difficult duct anatomy
|
Each cannulation was graded by the supervisor using a 4-point scoring system taken
from The endoscopic ultrasound (EUS) and ERCP Skills Assessment Tool (TEESAT). This
validated scale uses a 4-point scoring system: 1 (superior), achieves task independently;
2 (advanced), achieves task with minimal verbal instruction; 3 (intermediate), achieves
task with multiple verbal instructions; and 4 (novice), unable to complete the procedure,
and was previously validated as an assessment tool for ERCP trainees [11]. For each timed cannulation, trainees were allowed 1 minute of unassisted attempts,
following which, if deep cannulation of the desired duct was not achieved, verbal
instructions were offered to assist with cannulation.
Outcome measures
The main outcome measure was time to successful selective cannulation of the common
bile duct on the mechanical model. Trainee performance while cannulating, as assessed
by the TEESAT score, was the secondary outcome measure of the study.
Sample size estimation
Based on data from the validation study of the Boškoski-Costamagna model [8] we assumed a mean cannulation time for the novice endoscopists of 120 seconds. We
calculated that 170 procedures would be required to detect a 25 % decrease in mean
cannulation times with 90 % power at a 0.05 significance level. However, taking into
account the fact that each trainee would attempt multiple procedures, we adjusted
the sample size to account for the inflation effect [4] and assumed that a minimum of 16 trainees in each group (32 in total) should perform
20 attempts per trainee to achieve adequate statistical power as well as a clinically
relevant study design, in accordance with previous studies of ERCP training [4].
Data management and statistical analysis
All participating centers collected data prospectively using a standard report form
(Appendix 1), completed by the trainer after each attempt at cannulation. Data were collected
in a centralized database and analyzed using SPSS v. 22.
Bivariable analysis between the two study groups was carried out using the Mann Whitney
U and Kruskall-Wallis tests for comparing non-parametric variables and the chi-square
test for dichotomic variables. Multivariate analysis using logistic and linear regression
was carried out to account for the potential confounders arising from differences
in training methods between the various participating centers. P values were considered statistically significant if < 0.05.
Ethical concerns
The study did not include any patients and all cannulation attempts were made using
a mechanical simulator; approval from the local ethical committee was obtained by
each participating center prior to study inception in accordance with the Declaration
of Helsinki. Both trainees and local experts signed an informed consent regarding
the collection and storage of personal data.
Results
Thirty-six trainees without prior ERCP experience from seven ERCP training centers
were included in the study (Appendix 2). Sixteen trainees (44.4 %) were allocated to the motion training group and 20 (55.6 %)
were allocated to the standard training group; they completed a total of 720 supervised
attempts of selective cannulation of the common bile duct on the BCMT.
Successful cannulation of the bile duct was achieved in 698 of 720 attempts (96.9 %),
with no significant difference between the two study groups (96.5 % motion training
group vs. 97.25 % Standard cannulation group, P = 0.66). Multivariable analysis using logistic regression showed that there was a
statistically significant difference between centers in terms of successful cannulation,
independent of study group and papilla type, with trainees from one individual center
performing significantly better in terms in successful cannulation (OR 11.7, 95 %CI
1.2–107.6).
Trainees in the motion training group had significantly lower median cannulation times
compared to trainees in the standard group (36 vs. 48 seconds, minimum cannulation
time of 3 and 4 seconds respectively, maximum cannulation time of 300 seconds for
both groups; P < 0.001). This finding was confirmed in multivariable analysis using linear regression,
where group allocation and type of papilla were identified as independent risk factors
for prolonged cannulation time ([Table 2]).
Table 2
Potential risk factors for delayed cannulation.
|
Factor
|
Standardized Coefficient (95 %CI)
|
P value
|
|
Study group (E-motion/standard)
|
0.077 (0.513–19.515)
|
0.039[1]
|
|
Type of papilla (1–4)
|
0.077 (0.235–8.678)
|
0.039[1]
|
|
Training center
|
0.059 (–0.521–4.917)
|
0.113
|
1 Statistically significant in multivariable analysis (linear regression).
On subgroup analysis ([Table 3]), this difference was only statistically significant for the first papilla configuration
(median times 31 s vs. 42 s, P = 0.002), with further timed cannulation attempts not differing significantly between
the two study groups.
Table 3
Comparison of cannulation times between the two study groups according to the type
of papilla.
|
Papilla type
|
1
|
2
|
3
|
4
|
|
Group
|
MT
|
ST
|
MT
|
ST
|
MT
|
ST
|
MT
|
ST
|
|
Median cannulation time (s)
|
31.5
|
42
|
33.5
|
47
|
38.5
|
51
|
41
|
48
|
|
Minimum(s)
|
3
|
9
|
3
|
6
|
12
|
5
|
5
|
12
|
|
Maximum(s)
|
300
|
300
|
300
|
300
|
300
|
300
|
300
|
300
|
|
P value[1]
|
0.002[2]
|
0.094
|
0.298
|
0.052
|
MT, motion training group; ST, standard training group.
1 Mann-Whitney U test.
2 Statistically significant using the Mann Whitney U test.
Overall, time to cannulation increased as the level of complexity increased, with
median times of 36 seconds, 41.5 seconds, 46.5 seconds and 45 seconds for types 1
to 4, respectively, across the entire study population (P = 0.005).
Trainees in the motion training group had better technical performance than the standard
training group, according to the supervisors’ assessment, achieving TEESAT scores
of 1 or 2 in 255 of 320 attempts (79.7 %) compared to 284 of 400 (71 %) in the standard
training group (P = 0.009 chi-square). Multivariable analysis using logistic regression confirmed this
finding, with standard group allocation being a risk factor for achieving less satisfactory
TEESAT scores (OR 1.86, 95 % CI 1.29–2.72) after accounting for potential confounders
such as papilla type and training center.
Discussion
The main focus of our study was to assess whether an innovative, out-of-the-box but
intuitive method of training on a mechanical simulator can help improve a trainee’s
basic understanding of the procedure and speed up the learning curve for such complex
technical skills as selective cannulation of the bile duct.
Our main finding was that trainees receiving prior motion training performed better
than the standard training group, both in terms of cannulation times and overall performance
as assessed by their supervisors. This effect was, however, restricted to the early
part of the training, suggesting a plateau effect whereby the standard training group
could catch up to the motion training group after the first cannulation attempts.
This plateau effect has also been reported by some other trials focused on simulator
training in endoscopy, particularly in the field of colonoscopy [12]
[13].
Traditional training following the “master-apprentice” method where trainees learn
by trial and error under the supervision of an expert endoscopist is, in most centers,
still the only teaching method available. Despite the trainees having experience in
forward-viewing endoscopy, such skills may prove to be insufficient when handling
the side-viewing duodenoscope and associated accessories, which puts the patient at
risk while resulting in longer operating room time [14] and higher costs.
Although it has gained popularity, simulator training in advanced endoscopy is yet
to be included in training programs, because of the lack of published experience to
indicate its impact on clinical outcome. However, there is evidence that simulator
training in ERCP improves basic skills and translates into higher successful cannulation
rates [3]
[4] and faster cannulation time [4] with overall higher performance scores [3].
Different simulators are available to aid ERCP training, such as mechanical and computer-based
simulators and animal models. Mechanical simulators are relatively cheap and quick
to install, without requiring special preparation or facilities, and display realism
to some extent, despite not offering the same tactile sensation as animal models [15]. However, mechanical simulators provide the opportunity of unlimited repetition
of specific tasks while adjusting difficulty levels to suit more advanced trainees
and they also allow the use of real scopes and accessories. Because of their versatility,
mechanical simulators have outperformed both computer-based and ex-vivo models with
respect to trainee preference and basic skill development across several studies [16]
[17]
[18].
New innovative teaching methods are gradually being introduced to simulation-based
training. In a recent study, Perretta et al. showed the feasibility of modifying a
laparoscopy trainer to allow flexible endoscopy training on this model. Their model
adopted some basic tasks derived from laparoscopy training to simulate endoscopic
procedures such as snaring, injection, clipping and even selective cannulation [19].
To our knowledge, our trial is the first to explore alternative ways of using mechanical
simulators with the aim of improving trainee motor skills in the early phases of training
in ERCP. The concept of using unfamiliar instruments (such as an endoscope and a simulator)
to conduct more familiar tasks is also the focus of research in virtual reality simulators
in endoscopy, but there are currently no long-term data about the impact of such training
in real-life practice.
Our study has several strengths underscoring the validity of the results. First, this
was a large, randomized, multicenter trial that recruited trainees and trainers from
very different practice backgrounds, which supports the generalizability of our findings.
Furthermore, our study used a validated model that was independently shown to have
good face and construct validity [8], as well as a validated tool to assess trainee performance, the TEESAT score. Inasmuch
as ERCP training, and selective cannulation of the bile duct in particular, can be
standardized, our study achieved a controlled environment where trainees could be
objectively assessed.
Several limitations inherent in our study design need to be taken into account when
interpreting the results. First, the supervisors were not blinded to the group allocation
of the trainees, which could potentially induce a bias in evaluating trainee performance.
However, we believe the potential bias of this limitation is mitigated by the fact
that the main outcome measure was represented by the cannulation time, which is not
subject to evaluation bias. Furthermore, the differences found between the motion
training group and the standard group were consistent for both the primary and secondary
outcome measure, which suggests the risk of bias is very low. Second, the study design
meant that trainees in the motion training group were allowed more time on the model
(up to 25 minutes in total) compared to the standard group, which could account for
the early advantage in performance on the model. However, there is no way to mitigate
this bias other than looking at long-term outcomes of the trainees, which was outside
the scope and possibilities of this study. Third, although statistically significant,
it is unclear whether the difference in median cannulation time between the two study
groups could be clinically significant when trainees start their hands-on training.
This aspect is particularly important because the median cannulation times on our
model were quite short in both study groups (below 60 s in both groups), significantly
shorter than those reported by van der Wiel at al. in their trial on the BCMT [8] which was the basis for our study design. While we acknowledge the difference is
very unlikely to have any clinical impact on its own, and while also underscoring
the fact that faster might not actually translate into better cannulation skills in
real-life practice, we believe it to represent a surrogate marker for better skill
acquisition in the early phases of training, warranting further exploration of the
utility of this type of simulator-based training. Another potential limitation stems
from the design of our study, which lacked block randomization and relied on a simple
randomization process at each study site, resulting in an unbalanced distribution
of trainees in the two study groups (16 vs 20). We aimed to address the potential
bias of this aspect by conducting multivariable regression analysis for each of the
outcome variables to minimize the risk of bias and the results of this analysis were
consistent with the findings of our univariate analysis. Consequently, despite the
aforementioned limitations, we believe that the findings of our investigator-driven
study are encouraging and warrant further investigation to assess how this innovative
training method could further be refined and incorporated into everyday practice.
Conclusion
In conclusion, we have shown that motion training on the Boškoski-Costamagna mechanical
trainer leads to swifter cannulation times and better overall performance in first-time
trainees in ERCP techniques. Future studies should assess the long-term impact of
such innovative training methods by looking at trainee performance during the hands-on
training period and beyond, using clearly defined outcomes such as cannulation rates,
technical success rates, and validated assessment scores such as the TEESAT score.
