Introduction
At the turn of the millennium, the introduction of the disruptive technologies of
small-bowel capsule endoscopy (SBCE) and device-assisted enteroscopy (DAE) led to
a paradigm shift in the diagnosis and management of small-bowel disease. While SBCE
offers a noninvasive means to visualize of the entire length of the small bowel, DAE
allows for detailed, direct endoscopic assessment of suspected lesions, the acquisition
of targeted biopsies, and the delivery of endotherapy [1 ]. These two technologies, along with similar advances in dedicated small-bowel cross-sectional
imaging, are complementary. Their tandem use, with SBCE and/or cross-sectional imaging
often guiding the need for DAE, has revolutionized our minimally invasive endoscopic
approach to the management of small-bowel pathology.
Since first described by Yamamoto et al. over two decades ago, the repertoire of DAE
techniques has expanded to include double-balloon enteroscopy (DBE), single-balloon
enteroscopy (SBE), manual spiral enteroscopy, and latterly motorized-spiral enteroscopy
(MSE) [2 ]
[3 ]
[4 ]. DAE has become well established as the procedure of choice for patients requiring
further endoscopic evaluation, biopsy, or endotherapy of small-bowel pathology. However,
despite its increasing use in clinical practice, large-scale data regarding performance
measures for DAE remain limited.
Key performance indicators (KPIs) have been established to improve and standardize
the quality of care for several endoscopic procedures [5 ]
[6 ]
[7 ]. These KPIs provide endoscopy services and endoscopists with measurable benchmarks
to audit their performance, identify areas for improvement, and ensure the delivery
of consistent, high quality care. In 2019, the European Society of Gastrointestinal
Endoscopy (ESGE) proposed a set of quality performance measures for small-bowel endoscopy
[8 ]. Nonetheless, to date, no large-scale studies have evaluated the utility and impact
of these performance measures in clinical practice.
The aim of the “DEEP-UK” quality improvement project was to evaluate the performance
measures for DAE across the UK against the quality benchmarks proposed by the ESGE.
Methods
Study design and participants
This was a multicenter retrospective quality improvement project conducted at 12 enteroscopy
centers in the UK, including England, Scotland, and Northern Ireland ([Fig. 1 ]). All consecutive adult patients (≥18 years of age) who underwent DAE for diagnostic
and/or therapeutic indications between January 2017 and December 2022 were included
in the analysis. Endoscopy data from the participating centers are prospectively collected
and uploaded to the National Endoscopy Database [9 ]. Demographic, clinical, and endoscopic data were extracted from electronic endoscopy
databases in each center using a standardized Excel spreadsheet (Microsoft Co., Redmond,
Washington, USA). Anonymized data from all centers were collated into a single database
for the final analysis. The audit departments of all of the included sites approved
the study (host site, Sheffield Teaching Hospitals; registration no. CEU 7073).
Fig. 1 A map of the UK showing the 10 cities where the 12 participating sites were located.
Performance measures
We compared the performance measures of DAE against the quality benchmarks proposed
by the ESGE [8 ], and among the different centers to evaluate variations in practice. The performance
measures of DAE in this study included: (i) indication for DAE; (ii) diagnostic and
therapeutic yields; (iii) completeness of procedure; (iv) patient comfort; and (v)
adverse events (AEs).
Patient comfort levels were recorded by the endoscopists immediately after the procedures
and were assessed on a scale from 0 (comfortable) to 4 (severe discomfort). AEs during
or after the procedures were classified as minor and major AEs. Minor AEs included
self-limited or transient symptoms that did not require extended treatment or hospitalization.
Major AEs included severe complications, such as perforation, bleeding, and pancreatitis,
or unplanned hospital admission related to the procedure.
Statistical analysis
We used descriptive statistics to present the patient characteristics and procedural
outcomes. Continuous data were expressed as the median and interquartile range (IQR),
and categorical data were expressed as counts and percentages. Comparisons between
categorical outcomes were performed using the Fisher’s exact test or the chi-squared
test, and between continuous variables using the Mann–Whitney U test. A two-tailed P value of <0.05 was considered significant. The variations in practice between different
centers were assessed using funnel plots. No correction was done for multiple testing
owing to the exploratory nature of the study and the high value of reporting rare
outcomes. Statistical analyses were performed with Stata version 17 (StataCorp., College
Station, Texas, USA).
Results
Patient characteristics
Between January 2017 and December 2022, a total of 1663 patients (53.1% men) underwent
2005 DAE procedures; their median age was 60 years (range 18–93 years), and 22.0%
of patients had an American Society of Anesthesiologists (ASA) score of ≥3. The most
common medical co-morbidities were: hematological disease (44.0%), cardiovascular
disease (27.4%), and respiratory disease (11.5%). In addition, surgically altered
anatomy was present in 235 patients (14.1%), of whom 98 (41.7%) had undergone small-bowel
resections. The characteristics of the patients are listed in [Table 1 ].
Table 1 Characteristics of the 1663 patients who underwent a total of 2005 device-assisted
enteroscopies (DAEs), and the procedure indications.
Patient characteristics, and indications
IQR, interquartile range; ASA, American Society of Anesthesiologists; PEJ, percutaneous
endoscopic jejunostomy; ERCP, endoscopic retrograde cholangiopancreatography.
Sex, n (%)
883 (53.1)
780 (46.9)
60 (44–71)
ASA score, n (%)
595 (35.8)
701 (42.2)
348 (20.9)
19 (1.1)
Co-morbidities, n (%)
732 (44.0)
455 (27.4)
192 (11.5)
48 (2.9)
184 (11.1)
81 (4.8)
93 (5.6)
18 (1.1)
59 (3.5)
235 (14.1)
Medication, n (%)
147 (8.8)
Indication for DAE, n (%)
775 (38.7)
423 (21.1)
289 (14.4)
63 (3.1)
31 (1.5)
89 (4.4)
70 (3.5)
19 (0.9)
1 (0.05)
245 (12.2)
DAE trends and indications
Royal Free Hospital, St. Mark’s Hospital, and Sheffield Teaching Hospitals were the
high volume centers. There was a large reduction in the number of annual procedures
during the COVID-19 pandemic, which subsequently displayed a recovery trend and exceeded
the prepandemic level by 2022 ([Fig. 2 ]). Almost all procedures (98.1%, 95%CI 97.4%–98.6%) were performed for appropriate
indications as published in international guidelines. Antecedent SBCE was performed
in 56.1% of cases, and cross-sectional or magnetic resonance imaging in 51.1% of cases.
The route of insertion was decided on the basis of prior imaging in 84.1% of cases.
The most common indications for DAE were small-bowel bleeding (38.7%), tumors or polyps
(21.1%), and suspected Crohn’s disease (14.4%) ([Table 1 ]).
Fig. 2 Line graph of the total number of device-assisted enteroscopies performed per year.
DAE procedures and technical success rate
DBE was used for most procedures (82.0%), followed by SBE (17.2%) and spiral enteroscopy
(14 manual spiral enteroscopies and one MSE) (0.7%). The antegrade and retrograde
routes
were used in 73.7% and 25.9% of cases, respectively. Only eight procedures (0.4%)
were
performed with laparoscopic assistance ([Table 2 ]). Hyoscine n-butyl bromide (median dose 20 mg) was used as an intravenous
antispasmodic agent in 802 procedures (40%); glucagon (median dose 1 mg) was used
as an
alternative antispasmodic in 65 procedures (3.2%). The overall technical success rate
was
98.0%. Failed procedures were more likely to be retrograde than antegrade (5.7% vs.
1.0%;
P <0.001), owing to poor bowel preparation or excess
looping.
Table 2 Procedure characteristics and performance measures for device-assisted enteroscopy
(DAE).
Procedure characteristics
IQR, interquartile range; PEJ, percutaneous endoscopic jejunostomy.
* Type of sedation not documented (n=31).
‡ Procedures performed under general anesthesia or where sedation type was not
documented were excluded; patient comfort score not documented (n=7).
Type of enteroscopy, n (%)
1645 (82.0)
345 (17.2)
15 (0.7)
Route of insertion, n (%)
1477 (73.7)
520 (25.9)
8 (0.4)
Type of sedation, n (%)*
956 (47.7)
404 (20.1)
614 (30.6)
Patient comfort score, n (%)‡
604 (59.7)
214 (21.1)
137 (13.5)
38 (3.7)
18 (1.7)
Depth of insertion, median (IQR), cm
180 (120–220)
90 (50–150)
Procedure time, median (IQR), minutes
55 (39–76)
50 (34–80)
Diagnoses, n (%)
510 (25.4)
489 (24.4)
338 (16.9)
36 (1.8)
12 (0.6)
606 (30.2)
Therapeutic interventions, n (%)
444 (22.1)
244 (12.2)
184 (9.2)
76 (3.8)
70 (3.5)
49 (2.4)
19 (0.9)
Adverse events, n (%)
32 (1.6%)
13 (0.6%)
Diagnostic and therapeutic yield
The overall diagnostic yield was 70.0% (95%CI 67.8%–72.0%), with evidence of variations
between centers ([Fig. 3 ]). Inflammatory lesions, such as ulcers, erosions, and strictures, were the most
common findings (25.4%), followed by vascular lesions, such as small-bowel angioectasias
and
Dieulafoy lesions (24.4%), and polyps (11.0%) or other mass lesions (5.8%). Anatomical
alterations, such as small-bowel/Meckel’s diverticula, accounted for 1.8% of findings
([Table 2 ]). Biopsies were obtained in 35.8% (95%CI 33.7%–38.0%) of procedures, and detected
lesions were marked with a submucosal tattoo of sterile carbon particles in 78.0%
(95%CI
69.4%–85.0%) of cases. There were no significant differences in the diagnostic yield
between
antegrade and retrograde procedures (70.4% vs. 68.2%; P =0.37),
nor between DBE and SBE procedures (70.0% vs. 69.3%; P =0.83). The
diagnostic yield of procedures performed for ESGE-guided indications for DAE was
significantly higher than those performed for other nonspecific indications (P <0.001) (Table 1s , see online-only
Supplementary material).
Fig. 3 Funnel plot showing the variation in the diagnostic yield between centers, with the
blue line representing the ESGE minimum quality standard and the orange line representing
the overall diagnostic yield (centers with <10 procedures were excluded).
Therapeutic interventions were performed in 855 procedures (42.6%, 95%CI 40.4%–44.8%)
procedures and included argon plasma coagulation, endoscopic clipping, polypectomy,
adrenaline injection, stricture dilation, direct percutaneous endoscopic jejunostomy
insertion, and foreign body retrieval ([Table 2 ]). There were variations in the therapeutic yield between centers ([Fig. 4 ]). The overall therapeutic success rate was 96.6% (95%CI 95.1%–97.7%), based on intention
to treat.
Fig. 4 Funnel plot showing variation in the therapeutic yield between centers, with the orange
line representing the overall therapeutic yield (centers with <10 procedures were
excluded).
DAE procedure time and extent
The total procedure time was documented in 50.2% (95%CI 40%–52.4%) of cases. Where
it
was documented, the median time for procedures was of 54 minutes (IQR 37–78). Procedure
time
was not significantly different between antegrade and retrograde procedures (P =0.30). The extent of the procedure was documented in 73.4% (95%CI
71.4%–75.3%) of cases, with a median depth of insertion of 160 cm (IQR 90–200). The
estimated depth of insertion was 180 cm and 90 cm for the antegrade and retrograde
procedures, respectively (P <0.001). Total enteroscopy was
achieved in 17 procedures (0.8%, 95%CI 0.5%–1.3%), of which six were in patients with
previous small-bowel resections. The maximum point of insertion was marked with a
tattoo in
34.5% (95%CI 32.4%–36.6%) of procedures.
Sedation and patient comfort
Almost half the DAE procedures were performed with the patient under general anesthesia
(47.7%), while deep sedation with propofol and conscious sedation were used in 20.1%
and
30.6% of cases, respectively ([Table 2 ]). The median propofol dose was 902.5 mg (IQR 678.5–1148), with higher doses
administered to younger patients (<70 years of age) compared with older patients (1055
mg
vs. 784 mg; P =0.002). The median midazolam dose was 4 mg (IQR
3–5) and, similarly, younger patients received higher doses than older patients (4
mg vs. 3
mg; P <0.001). Additionally, the median dose of fentanyl was 75
mcg (IQR 50–100), with higher doses used for younger patients compared with older
patients
(P <0.001). Poor patient tolerance limited 6% of the
procedures performed under conscious sedation, despite a median midazolam dose of
4 mg and a
median fentanyl dose of 75 mcg being used. Overall, patient comfort (comfort scores
0–1) was
significantly better with the use of deep sedation than with conscious sedation (99.7%
vs.
68.5%; P <0.001).
Adverse events
Minor AEs occurred in 32 cases (1.6%, 95%CI 1.0%–2.2%) and were mainly self-limited
oxygen desaturation or bradycardia and hypotension secondary to sedation. Major AEs
occurred in 0.6% (95%CI 0.3%–1.1%) of procedures; these included six cases of perforation,
three of pneumonia requiring hospitalization, two of post-polypectomy bleeding, and
one each of severe pancreatitis and unstable cardiac arrhythmia. Only one case of
perforation was related to therapy (post-polypectomy); two occurred during the insertion
or removal of the endoscope at the upper esophagus, and three after biopsy of friable
malignant tissue.
Comparison between high volume and low volume centers
The overall diagnostic and therapeutic yields were comparable between the three high
volume centers (>50 annual procedures) and the low volume centers (<50 annual procedures).
However, at the high volume centers, a greater proportion of cases were found to have
an appropriate indication, and to have had the depth of insertion and detected lesions
marked with tattoos, and the procedure extent documented (le 2s ).
Only 19.4% of procedures at high volume centers were performed with the patient under
conscious sedation compared with 51.4% of procedures at low volume centers. Patient
comfort at high volume centers was better than at low volume centers when conscious
sedation was used (78.0% vs. 62.2%), but not when deep sedation was used (99.7% vs.
100%); however, deep sedation was used in only 3.7% of cases at low volume centers
compared with 29.8% at high volume centers.
Discussion
This is the largest study to report DAE performance and outcomes, and the first multicenter
study to evaluate the performance measures for DAE against the ESGE quality benchmarks.
We included 1663 patients who underwent 2005 DAE procedures for diagnostic and therapeutic
purposes across 12 enteroscopy centers in the UK. Although DAE had high diagnostic
and therapeutic yields, with a low incidence of AEs, there was evidence of variations
in practice and room for improvement in optimizing sedation practices, increasing
standardization of depth of insertion documentation, and adopting marking techniques
to aid in the follow-up of detected or treated lesions.
The overall diagnostic yield for DAE in our study was 70%, and all participating centers
exceeded the ESGE minimum standard of 50%. A meta-analysis of early studies evaluating
the performance of DBE over its first decade of use, reported a similar pooled diagnostic
yield of 68.1% [10 ]. Furthermore, our findings confirm the results of a more recent meta-analysis by
Lipka et al. [11 ], who found a comparable diagnostic yield between DBE and SBE procedures. The high
diagnostic yield in our study can be attributed to two main factors: appropriate patient
selection and the use of DAE as a second-line modality after abnormal SBCE or dedicated
cross-sectional imaging findings. However, we observed variations in the diagnostic
yield for DAE among centers, which ranged from 53.3% to 81.2%. This is consistent
with the diagnostic yields for DAE reported in recent studies, which range from 59%
to 76.5% [12 ]
[13 ]
[14 ]
[15 ]. These variations are likely a reflection of the differences in diagnostic yield
among the various indications for DAE and the different levels of experience between
endoscopists. In the current study, adherence to the list of ESGE-guided indications
for DAE was associated with significantly higher diagnostic yields compared with procedures
performed for other nonspecific indications.
The rate of therapeutic intervention was not included in the ESGE performance measures
owing to a lack of supporting data [8 ]. We found that therapeutic interventions were performed in 42.6% of procedures,
with variations among centers, although only one center had a therapeutic yield of
less than 20%. A recent multicenter study in the USA reported a higher rate of therapeutic
interventions at 49.5% [14 ]. Therefore, the present data suggest that centers might aim for a minimum therapeutic
yield of 20%.
There is a paucity of evidence regarding the number of DAE procedures required to
achieve competence during training and the minimum number of annual procedures required
to maintain competence [8 ]. An interesting finding in the current study is that the diagnostic and therapeutic
yields of DAE were not directly related to the volume of procedures in each center,
as shown in [Fig. 3 ] and [Fig. 4 ]. This suggests that other factors, beyond procedural volume, such as the endoscopists’
experience in advanced upper and lower gastrointestinal endoscopy and appropriate
patient selection may have a major influence on the outcomes of DAE.
The estimated depth of enteroscope insertion was reported in 73.4% of procedures,
compared with the minimum ESGE standard of ≥80%. This benchmark was however based
on very low quality evidence as the estimated depth of insertion is often inaccurate
in clinical practice and varies between endoscopists, despite early promising results
of accurate estimations in porcine models [16 ]
[17 ]. Similarly, submucosal tattooing of the point of maximal insertion in ≥80% of procedures
was proposed as a quality benchmark, but was performed in only 34.5% of procedures
in our study. Moreover, tattooing of lesions that might require surgical intervention
was performed in 78% of cases, compared with the minimum ESGE standard of ≥95% [8 ]. These findings highlight areas for improvement in procedure documentation and marking
techniques, to aid in the follow-up of detected or treated lesions.
The invasive nature of DAE coupled with the relatively long procedure time requires
high doses of sedation to ensure patient comfort. A study of 956 patients undergoing
different endoscopic modalities under conscious sedation showed that the tolerability
of DBE was worse than other endoscopic modalities, including ERCP, despite high doses
of sedation [18 ].
Conscious sedation was regularly used in our study; however, even with relatively
high doses of midazolam and fentanyl, approximately a third of procedures were poorly
tolerated by patients (comfort scores ≥2), and 6% of procedures had to be terminated
early because of poor patient tolerance and withdrawal of consent. Conversely, almost
all procedures performed under deep propofol sedation were well tolerated. In two
previous multicenter studies in Portugal and the Netherlands, most DAE procedures
were performed under anesthesiologist-administered propofol sedation [13 ]
[19 ]. This highlights the varying practice approach in the UK with respect to using moderate
sedation compared with other countries where monitored anesthesia care (MAC) is the
preferred approach. While performing all DAE procedures under general anesthesia or
propofol sedation poses a challenge in the UK because of the logistical complexities
associated with the provision of MAC for advanced endoscopy, our results suggest that
DAE should not be routinely performed with patients under conscious sedation.
We observed a low rate of major AEs (0.6%), which is well below the 5% benchmark proposed
by the ESGE. This rate is comparable with pooled AE rates from meta-analyses and more
recent multicenter studies from Portugal and the USA [10 ]
[11 ]
[13 ]
[14 ]. Such findings reinforce the high safety profile and low complication rate of DBE
and SBE procedures, and support adopting a lower major AE benchmark in future guidelines.
There was only one MSE procedure in our study, and this resulted in an upper esophageal
perforation on withdrawal of the device. MSE has not been used in the UK following
this unfortunate incident. A case-matched study comparing MSE and DBE found that MSE
did not offer diagnostic or therapeutic advantages over DBE, and was associated with
more frequent AEs [20 ]. More recently, serious safety concerns, including fatalities, with regard to the
MSE device have led to its withdrawal and recall from clinical practice and the market
worldwide [21 ].
There are several strengths to our study. First, we provide a comprehensive overview
of small-bowel endoscopy practices in the UK, which captures the impact of the COVID-19
pandemic on enteroscopy services and the road to recovery over the last 3 years. Second,
this is the first study to evaluate performance measures for DAE against the ESGE
quality benchmarks with >2000 procedures performed across 12 different sites. Third,
we showed that variations in practice exist between different centers. Fourth, we
provided evidence to support the use of general anesthesia or deep sedation over conscious
sedation for DAE. Importantly, the results of this study set an evidence-based framework
for the development of future performance measures and challenge some of the existing
quality benchmarks for DAE, akin to the precedent set with SBCE [22 ].
The limitations of this study include the retrospective nature of the analysis, which
has inherent limitations, such as the presence of confounding factors and selection
bias. The lack of adequate sedation in a proportion of procedures in the current study
may have influenced the length of small-bowel explored and, consequently, the diagnostic
and therapeutic outcomes of these procedures. Furthermore, our study focused on assessing
patient comfort levels during DAE procedures as recorded by the endoscopists, which
may have introduced further bias. Future prospective studies should use validated
questionnaires, such as the Global Rating Scale (GRS), to gain a more comprehensive
understanding of overall patient experience [23 ].
Another important limitation that should be considered when interpreting the results
of this study is that some patients underwent repeated DAE procedures, which means
that not all observations were statistically independent. However, each endoscopic
procedure is inherently unique and all established performance measures in endoscopy
account for the total number of procedures, including repeated procedures, as certain
findings, therapies, or complications may have occurred in only one of several endoscopies
a patient has had over an extended period of time. Therefore, correcting for multiple
testing might have provided misleading results for rare outcomes such as DAE complications.
Finally, we were unable to assess the rates of accurate photodocumentation and the
quality of bowel preparation, both of which are minor performance indicators for DAE
[8 ].
In conclusion, DEEP-UK is the first study to evaluate the performance measures for
DAE against the ESGE quality benchmarks. DAE performance measures in the UK meet the
ESGE standards with high diagnostic and therapeutic yields, and a low complication
rate; however, variations in practice exist between different centers, highlighting
potential areas for quality improvement.