Introduction
In recent years, quality assurance of endoscopic retrograde cholangiopancreatography
(ERCP) procedures has gained increasing interest from healthcare professionals and
patient organizations. This is not without reason, as ERCP nowadays is primarily a
minimally invasive therapeutic procedure that is associated with potentially severe
complications [1–3]. Significant training and experience are required to maximize
procedural success and minimize the complication risks [4].
In 2018, the European Society of Gastrointestinal Endoscopy (ESGE) and United European
Gastroenterology (UEG) set up an initiative and published a list of key quality performance
measures for ERCP and endoscopic ultrasound (EUS) [5]. These performance measures
are intended to set a minimum standard for quality and the outcome of ERCP and EUS. Both
societies encourage healthcare professionals to implement these performance targets
on a national basis.
Quality assurance in ERCP has also gained significant awareness in the Netherlands.
Since January 2016, endoscopists in the Netherlands have been required to register
several procedural steps and the outcomes of all ERCP procedures in a mandatory nationwide
quality registry, using the Rotterdam Assessment Form for ERCP (RAF-E), a self-assessment
registry tool that provides insight into ERCP performance [6].
One of the reasons for this mandatory registry was the finding of Ekkelenkamp et al.
in 2014 that the overall procedural success rate of ERCP, using the RAF-E, was only
85.8 % [7]. Interestingly, since the implementation of the mandatory quality registry,
it has been noted that fewer endoscopists perform more ERCPs (unpublished finding),
which is likely beneficial with regard to procedural success and quality outcomes.
The current study aimed to evaluate the procedural success of ERCP after the implementation
of the mandatory nationwide registry and to determine whether performance measures,
according to the ESGE standards, were met.
Methods
Database and data collection
Data was retrieved from a prospectively maintained gastrointestinal (GI) endoscopy
database (Trans.IT database; Rotterdam, The Netherlands). All centers participating
in the database use a uniform structured tool to report endoscopy findings and all
endoscopy reports are then uploaded into the database. The database has recently been
described in detail elsewhere [8]. In short, this anonymized multicenter database
was initiated in 2012 and currently collects GI endoscopy data from 19 Dutch hospitals
(three academic and 16 nonacademic hospitals), distributed over 9 of the 12 provinces
of the Netherlands.
The database also automatically creates a RAF-E form based on the findings in the
ERCP report and uploads this to the mandatory national registry. The RAF-E includes
several items, such as the indication for performing the ERCP, a priori ERCP degree
of difficulty (Schutz score), presence of virgin papillary anatomy, outcome of common
bile duct cannulation, and procedural success, amongst others [9]. Some endoscopy
centers decided to use the Trans.IT database as an intermediary to report ERCPs to
the national registry. Other centers report directly to the website of the national
registry, in which case a RAF-E form is not created in the Trans.IT database. In the
current study, 11 of 19 centers use the Trans.IT database to create a RAF-E form.
Therefore, data from 11 centers (two academic, nine nonacademic, distributed over
8 of the 12 provinces in the Netherlands) were available for analysis.
Registration of procedural indication and procedural outcome were used as an indicator
for the completeness of data, with only the years in which at least 90 % of the performed
ERCPs were completely registered considered for inclusion. This was achieved from
2019 onwards. All ERCP reports of procedures in the Trans.IT database performed between
January 2019 to December 2020 from the 11 centers were analyzed. ERCP reports were
included if the indication and procedural outcome of the procedure were in fact registered.
ERCP procedures for patients aged younger than 16 years were excluded.
Outcomes and definitions
The primary outcome was the overall procedural outcome, defined as “success” or “failure”
of the ERCP, which was identical to the primary outcome measure described in the prospective
voluntary evaluation by Ekkelenkamp et al. [7]. The ESGE performance measures that
could be analyzed based on outcome findings in the database were: (a) the percentage
of successful bile duct cannulations in patients with virgin papillary anatomy (and
a biliary indication); (b) the percentage of successful stent placements for a biliary
obstruction located below the liver hilum; and (c) complete removal of bile duct stones
(stone size < 10 mm). The outcomes of academic centers were compared with those of
nonacademic centers.
The ESGE performance measures that could not be analyzed were the percentage of patients
with adequate administration of prophylactic antibiotics before ERCP (when indicated)
and the rate of post-ERCP pancreatitis, because these data are not included in the
nationwide ERCP registry.
Additionally, we evaluated the success of cannulation in patients with a virgin papilla
in relation to the type of sedation used and to the American Society of Anesthesiologists
(ASA) classification. The type of sedation was categorized into either propofol sedation
or general anesthesia in one group, or conscious sedation with midazolam and fentanyl
in a second group. The ASA classifications were grouped into ASA classes 1 and 2,
and ASA classes 3 and 4.
Statistical analysis
The statistical analysis was based on descriptive analyses, using frequencies (%)
for categorical variables, and mean (standard deviation [SD]) for normally distributed
continuous variables or median (interquartile range [IQR]) for non-normally distributed
continuous variables. Categorical variables were analyzed using the chi-squared test
and continuous variables were analyzed using the Mann–Whitney U test. A two-sided P value of < 0.05 was considered statistically significant.
All data were exported in comma-separated value files (CSV) from the Trans.IT database
and imported into SPSS software for statistical analysis (SPSS 25.0; IBM Corp., Armonk,
New York, USA).
Ethical considerations
The collection of patient data in the Trans.IT database has been approved by the privacy
officer of the Erasmus Medical Center in accordance with the Dutch Personal Data Protection
Act. All patient data are anonymously stored in a secure environment and are therefore
exempt from formal ethics approval. All included hospitals provided written consent
for participation.
Results
From January 2019 to December 2020, a total of 5671 ERCP procedures were registered
by 57 endoscopists in 11 centers (2064 in an academic and 3607 in a nonacademic setting).
The median number of ERCPs per endoscopists was 95 in this period (IQR 23–129). Not
all endoscopists performed ERCPs during the full study period either because of retirement
or starting as a newly registered gastroenterologist. The median number of ERCPS per
month per endoscopist, corrected for months of participation, was 4.7 (IQR 3.5–6.3).
Of the 5671 ERCPs, 21 ERCPs (0.4 %) were excluded because patients were under 16 years
of age, 173 (3.1 %) because the indication was not registered, and 182 (3.2 %) because
the procedural outcome was not registered. Therefore, 5295 ERCPs (93.4 %) were available
for analysis.
The overall procedural success rate was 89.1 %. [Table 1 ] shows procedural success according to indication and degree of difficulty as per
the classification on the RAF-E form. The ESGE target standards for performance measures
and the study outcomes overall and per degree of difficulty are shown in [Table 2 ].
Table 1
Procedural success rates for the different indications and degrees of difficulty.
Voluntary registry[* ]
Mandatory registry
n (%)
Procedural success, % (n)
n (%)
Procedural success, % (n)
Indication
4388 (51.2)
85.2 (3740)
2439 (46.1)
89.5 (2183)
1829 (21.3)
86.2 (1576)
1021 (19.3)
90.2 (921)
545 (6.4)
87.3 (476)
669 (12.6)
87.4 (585)
292 (3.4)
87.7 (256)
174 (3.3)
93.7 (163)
186 (2.2)
78.5 (146)
243 (4.6)
83.5 (203)
1335 (15.6)
88.0 (1175)
749 (14.1)
88.8 (665)
8575 (100)
85.8 (7360)
5295 (100)
89.1 (4720)
Degree of difficulty
5676 (66.3)
88.1 (4999)
3162 (60.8)
91.2 (2885)
1989 (23.2)
84.3 (1676)
1444 (27.8)
87.5 (1263)
890 (10.4)
75.2 (669)
595 (11.4)
82.5 (491)
CBD, common bile duct.
* As reported by Ekkelenkamp et al. (2014) [6].
Table 2
Target standards for the ESGE performance measures and outcomes for the different
degrees of difficulty in academic and nonacademic centers.
n
Target standard, %
Success, % (n)
Overall
Schutz 1
Schutz 2
Schutz 3
Successful biliary cannulation in patients with a virgin papilla
2007
90
90.3 (1813)
91.2 (1291)
90.9 (430)
72.1 (62)
473
95
92.4 (437)
92.0 (275)
93.0 (120)
92.7 (38)
Appropriate stent placement biliary obstruction, after successful biliary cannulation
694
95
97.0 (673)
99.1 (453)
93.2 (178)
93.1 (27)
791
95
98.2 (777)
98.4 (421)
97.8 (310)
100 (36)
Bile duct stone extraction, < 10 mm
1313
90
97.9 (1286)
97.9 (1286)
–
–
190
95
97.9 (186)
97.9 (186)
–
–
ESGE, European Society of Gastrointestinal Endoscopy.
Successful biliary cannulation in patients with a virgin papilla was achieved in 90.3 %
of ERCPs in nonacademic centers (ESGE target standard 90 %) and in 92.4 % of ERCPs
in academic centers (ESGE target standard 95 %).
Successful stent placement in patients with a biliary obstruction located below the
liver hilum was achieved in 97.0 % of ERCPs in nonacademic centers (ESGE target standard
95 %) and 98.2 % of ERCPs in academic centers (ESGE target standard 95 %).
Successful extraction of bile duct stones smaller than 10 mm was achieved in 97.9 %
in both nonacademic centers (ESGE target standard 90 %) and in academic centers (ESGE
target standard 95 %).
[Fig. 1 ] shows the individual endoscopist cannulation rates in patients with a virgin papilla
and the number of ERCPs performed in the study period. The rates of successful cannulation
in patients with a virgin papilla were not significantly different between ERCPs performed
with the patient under general anesthesia or propofol sedation and those under conscious
sedation, both for patients with an ASA class of 1 or 2 and those with an ASA class
of 3 or 4 ([Table 3 ]).
Fig. 1 Individual endoscopist cannulation rates for patients with a virgin papilla and number
of endoscopic retrograde cholangiopancreatographies (ERCPs) performed.
Table 3
Cannulation outcomes in patients with a virgin papilla for the different sedation
types and ASA classifications.
n
Common bile duct cannulation success, % (n)
ASA classification 1 or 2
928
92.0 (854)
526
92.6 (487)
ASA classification 3 or 4
667
88.3 (589)
141
89.4 (126)
ASA, American Society of Anesthesiologists.
Discussion
In this study, we evaluated whether the ERCP outcome data of 11 Dutch hospitals collected
within a mandatory ERCP registration database met the quality performance targets,
as defined by the ESGE. We found an overall procedural success rate of 89.1 % and
five of the six ESGE target standards that could be assessed from the database were
met. Successful bile duct cannulation in patients with virgin papillary anatomy was
not met in the academic centers, with a rate of 92.4 % compared with the required
95 %. The target standards were met for successful bile duct cannulation in patients
with virgin papilla in nonacademic centers, successful stent placements for an obstruction
located below the liver hilum in academic and nonacademic centers, and successful
removal of bile duct stones in academic and nonacademic centers. This mandatory registry
permits valuable insights into the performance of Dutch gastroenterologists and provides
source information to help improve the overall quality of ERCPs.
This is the first study that has reported on the procedural outcomes of ERCP in the
Netherlands since the implementation of the nationwide mandatory quality registry
in 2016. A previous study by Ekkelenkamp et al. that was based on voluntary registration,
including approximately 50 % of all ERCP procedures performed in the Netherlands in
2014, reported an overall procedural success rate of 85.8 % [7]. In the current study,
the overall procedural success rate was higher at 89.1 %. Although a direct comparison
is difficult to make, it is at least reassuring that, despite mandatory registration,
results have numerically improved. It is therefore tempting to speculate that as a
consequence of the implementation of the mandatory registry, endoscopists performing
ERCPs are more conscious and critical about their own performance. This may be reflected
by the observation that, compared with the study of Ekkelenkamp, currently fewer endoscopists
perform more ERCPs in the majority of Dutch centers.
We aimed to compare our results with the ESGE quality performance measures, which
were published in order to improve the outcome and quality of endoscopy. The current
study is the first to investigate whether the ESGE quality performance measures for
ERCP procedures are being met in daily clinical practice. The results show that monitoring
of the ESGE quality performance measures is not only feasible but also provides valuable
insight into the performance level of individual endoscopists, centers, and ultimately
a country.
Our results for successful biliary stenting and stone extraction were comparable to
an Austrian nationwide benchmarking program, in which 28 of 140 ERCP sites participated.
Biliary stenting was successful in 97.8 % vs. 97.0 %–98.2 % in our study and stone
extraction in 98.6 % vs. 97.9 % in our study [10]. In the current study, the target
standard of successful biliary cannulation in patients with a virgin papilla in academic
centers was not met, with 92.4 % compared with the 95 % target. In this regard, it
is important to take into consideration that the ESGE has stated that the quality
of the evidence used to develop the target measure for biliary cannulation was graded
as low. Further evaluation is needed as to whether the target standard of 95 % for
expert centers is realistic, taking into consideration that potentially more primarily
failed and difficult ERCPs are referred to academic centers, for which sometimes more
advanced selective cannulation techniques are required.
Compared with our study, the successful cannulation rate in patients with a virgin
papilla was found to be lower in a study from the UK (84 % vs. 90.3 %–92.4 %) [11].
A nationwide study from Sweden [12] and a multicenter study from Norway, including
11 hospitals [13], reported common bile duct cannulation rates of 92 % and 91.1 %,
respectively, but these studies did not report selectively on cannulation rates in
patients with a virgin papilla. The ERCP outcome data for the nine nonacademic centers
in the current study are in line with a study from the USA that reported on ERCP outcomes
from eight community hospitals [2].
A notable finding in our study was the low cannulation rate in nonacademic centers
for cases with a Schutz 3 difficulty score (72.1 %). It should be noted however that
this finding is based on only 62 procedures and that more evidence is needed to establish
whether this observation holds true. However, we believe this is an excellent example
of the strength of a national registry, giving the opportunity to detect these trends
and making it possible to intervene on both a personal and national level. Further
actions, such as additional training of endoscopists in nonacademic centers or maybe
centralizing the more difficult ERCP cases in academic centers, should be explored.
The funnel plot provided in [Fig. 1 ] is another example of the strength of a national registry and shows how the national
registry permits the identification of low performing endoscopists who may benefit
from additional training in ERCP.
An additional finding of our study was that cannulation rates in patients with a virgin
papilla were similar whether the ERCP was performed with the patient under general
anesthesia or propofol sedation, or under conscious sedation with midazolam and fentanyl,
regardless of the ASA classification. A prospective nationwide study from Sweden that
reported on the impact of sedation types on cannulation rates in patients with a virgin
papilla, in a total of 31 001 ERCP procedures, found a statistically significant difference
based on the type of sedation, with a cannulation success rate of 89.0 % for propofol
sedation vs. 86.7 % for midazolam sedation, although this small difference seems to
carry limited clinical relevance [14].
The Trans.IT database was chosen for this project because it currently offers more
transparency than the national database, which is a key strength of this study. The
national registry only registers procedures that are submitted to the registry and
does not record how many ERCPs are actually performed in each center. The Trans.IT
database includes all ERCP procedures performed in a center and an overview of the
quantity of data registered for each procedure. This allowed us to control for the
completeness of data for all ERCP procedures performed within a certain time period,
which is not possible with the national database. We attempted to minimize bias by
analyzing only the time period in which at least 90 % of the performed ERCPs were
completely registered. As such, the outcomes of our study are a reliable representation
of everyday clinical practice. This is however a Dutch study, which makes it potentially
difficult to generalize our results. For example, in the Netherlands, it is common
practice that gastroenterologists perform endoscopic procedures, such as ERCP, in
combination with direct patient care management in both an inpatient and outpatient
setting. The median number of 95 ERCPs per endoscopist during the study period may
suggest that further concentration of ERCP procedures should be considered, as other
studies have shown that endoscopists may benefit from higher yearly volumes of ERCPs
to achieve core skills [15].
The limitations of this study that need to be addressed are firstly the fact that
this is not a strictly nationwide report. Although the 11 centers included in the
current study are distributed over eight of the 12 provinces in the Netherlands and
represent both academic and nonacademic centers, it is not certain that the outcomes
from these 11 centers is representative of the whole of the Netherlands. Nonetheless,
more than 5000 ERCPs were included, which amounts to approximately 12.5 % of the total
number of ERCPs performed in the Netherlands in this period. Second, not all centers
participating in the Trans.IT database could be included, because eight of the 19
participating hospitals registered ERCPs in the Trans.IT database without using the
RAF-E form.
Third, not all years during which the reporting of ERCPs was mandatory were included.
Owing to start-up problems and the time required to train endoscopists to correctly
register ERCPs during the first years of the national registry, a sizeable percentage
of ERCPs were not registered completely (registration rates for procedural indications
and outcomes of 63.1 % and 56.0 %, respectively, in 2017 vs. 95.4 % and 93.7 %, respectively,
in 2019). Registration problems occurred not only in the Trans.IT database but were
seen also on a nationwide level in the national registry. We believe that including
these years with low registration rates could have potentially induced bias and therefore
we decided to exclude these years from the analysis. A correct registration in these
years would have allowed us to perform a time-trend analysis and to assess whether
procedural success increased each year during mandatory registration. Finally, not
all ESGE quality performance measures could be evaluated because the RAF-E form that
is currently used does not include information on antibiotic prophylaxis or post-ERCP
pancreatitis.
In conclusion, the Dutch national mandatory centralized registry of ERCP reporting
offers the opportunity to evaluate and improve the quality of ERCP. Comparison with
the ESGE quality performance measures is feasible and showed that the overall quality
of ERCP in Dutch ERCP centers is high. Five out of six ESGE quality performance measures
were achieved successfully, but the 95 % target for successful biliary cannulation
of a virgin papilla in academic centers was not met.
