Keywords capsule endoscopy - small bowel transit time - anemia - obscure gastrointestinal bleed
Introduction
Gastrointestinal bleed (GIB) is a common GI presentation.[1 ] Small bowel bleed specifically constitutes 5 to 10% of all GIB and cannot be detected
by upper and lower endoscopy.[2 ] Capsule endoscopy (CE) is an advanced methodology of detecting small bowel overt
and obscure GI bleeding (OGIB). It is a relatively noninvasive test that visualizes
the small bowel and is considered the test of choice after adequate upper endoscopy
and colonoscopy.[3 ]
[4 ] Small bowel capsule endoscopy (SBCE) is superior to push enteroscopy and computed
tomography (CT) angiography (CE detected inflammatory lesions by 9% more than push
enteroscopy in one study and it identified a bleeding source in 50 vs. 24% of patients
in another study).[3 ]
[4 ] The technique can be improved in the future if a breakthrough is achieved with tissue
sampling or therapeutic intervention. SBCE sensitivity can fluctuate between 40 and
93%.[3 ]
[4 ] Its sensitivity depends on several factors including bowel preparation, type of
capsule used, bleeding amount, and transit time/bowel motility in addition to time
from bleeding.[3 ]
[4 ] Furthermore, some small bowel mucosa might not be visualized because of the turbulent
pathway of the capsule.
Despite its clinical significance, CE is costly. Its cost differs greatly between
countries and the type of capsule used.[4 ]
[5 ] Its financial burden hinders the ability of patients to repeat the test once negative
results are reported, even if no bleeding source was detected. This can lead to complications
if the patient is discharged home without proper identification and control of the
bleeding source. A possible cheaper alternative is rereading SBCE.
Reading CE is time-consuming, leading to a faster reading of the recording and possibly
skipping some lesions.[6 ]
[7 ] Previous study protocols used the two-reader approach in their methodologies.[8 ]
[9 ] We study the sensitivities of SBCE when read by two different physicians in comparison
with the studies repeating the test. We hypothesize that rereading SBCE improves its
sensitivity, making it, at least, comparable to repeating the test, hence cost-effective.
In addition, we studied the association between small bowel transit time (SBTT) and
diagnostic yield. Our second hypothesis is that the longer the SBTT, the better the
diagnostic yield.
Materials and Methods
This is a retrospective cohort study on capsule endoscopies conducted between 2018
and 2019 in a tertiary care center in Lebanon. A total of 42 patients with anemia
or OGIB were included for SBCE after a negative evaluation with upper and lower GI
endoscopy. The results were read by two physicians at different times. The second
reader was blinded from the first reader's results but had access to the chief complaint,
patient characteristics, and the capsule videos (single blinded study). We compared
the sensitivity of the two readings and calculated the p -value to check for a statistical significance.
Inclusion criteria: Patients with anemia and/or OGIB were included. Documentation of capsule reaching
the cecum was required.
Exclusion criteria: Patients with Crohn's disease, small bowel obstruction, pacemakers, and implantable
defibrillators were excluded.
The participating outpatients were cleared by their cardiologists or primary physicians
to hold their anticoagulation/antiplatelet medications if any. However, due to the
subjective nature of this study, there was no direct control of their medications.
All the patients fasted for 8 hours prior to capsule ingestion and were prepared with
a total of two sodium picosulfate sachets followed by 4 L of water given in split
doses (1 sachet followed by 2 L the evening before and 1 sachet followed by 2 L the
morning of the procedure). Alverine citrate/simethicone was also given the night before
capsule ingestion and the morning of the procedure. Patients were instructed to resume
clear liquids 2 hours after capsule ingestion.
The type of the capsule used was a CapsoCam Plus, which provides a 360-degree panoramic
view of the small bowel mucosa using four cameras at a sequential of 90-second intervals
in its midsection, allowing a better visualization of the small bowel mucosa, thus
improving the diagnostic yield.[9 ] It is able to capture images at variable rates between three and five frames per
second per camera with an operating time exceeding 15 hours depending on its transit
speed.[10 ] The capsule video was read at a maximum speed of 10 frames per second, per single
view mode, as per the European Society of Gastrointestinal Endoscopy (ESGE) quality
recommendation.[11 ] Both endoscopists have more than 10 years of experience in SBCE. Informed consent
was obtained from all participants.
Patients were compared for positive findings, SBTT, bowel cleanliness, age, and sex.
The findings of the SBCE were classified according to the small bowel results. Findings
were considered positive when ulcer, erosion, arteriovenous malformation, diverticula,
bleed, prominent vessel, or small bowel lesions were detected. Findings were labeled
negative when normal results were found. To note, only positive findings in the small
bowel contributed to positive results and positive diagnostic yield.
Results
In this study, out of a total of 42 patients, the first reading detected 18 positive
tests compared with 31 in the second reading. Comparing the sensitivity of the two
readings, the first reader had a 43% diagnostic yield, while the second had a 74%
diagnostic yield ([Table 1 ]). Using the Mann–Whitney U test, p -value was statistically significant at 0.0043 ([Table 1 ]). For the first reader, out of the 18 positive tests, 17 (94.44%) were detected
again by the second physician. In addition, the second reader reported 14 (33.33%)
new tests to be positive ([Table 2 ]).
Table 1
Sensitivity of the two readings
Positive results
Negative results
p -value
First reading
43%
57%
0.0043
Second reading
74%
26%
Table 2
New findings detected in the second reading
Case number
Indication
First reader
Second reader
1
Anemia
Normal
AVM + ulcer
2
Anemia
Normal
Large AVM + active bleed
3
Anemia
Normal
Ulcer
4
Anemia
Normal
Ulcer
5
Anemia
Normal + poor preparation
Ulcer
6
Anemia + melena
Normal
Ulcer
7
Anemia
Normal
Hemangioma
8
Anemia
Normal
Erosions + hemangioma
9
Anemia
Normal
AVM
10
Anemia
Normal
Ulcer + AVM + active bleed
11
Anemia
Normal
Erosions + AVM/hemangioma
12
Anemia
Normal
Ulcer + AVM
13
Anemia
Normal
Ulcer
14
Anemia
Normal
Ulcer
Abbreviation: AVM, arteriovenous malformation.
SBTT is defined as the time from first duodenal image to first cecal image.[12 ] In our study, we excluded nine patients who did not have a documented SBTT. Therefore,
33 of 42 patients were included and stratified into five groups (0–2, 2–4, 4–6, 6–8,
and >8 hours). The results were as follows. The majority (11 patients) had an SBTT
between 4 and 6 hours ([Table 3 ]). The average SBTT was 5.48 hours. We studied the association between SBTT and the
detection of positive findings reported by either reader, as shown in [Table 4 ]. Odds ratio (OR) and the corresponding p -values were calculated to see if a correlation exists between the two entities. The
0- to 2-hour group was our control. We concluded that all groups had a positive OR
(OR > 1) suggesting a positive correlation between a longer SBTT and a higher diagnostic
yield, but no clinical significance was demonstrated (p > 0.05).
Table 3
Number of patients according to their small bowel transit time (SBTT)
SBTT (h)
Number (%)
0–2
2 (6%)
2–4
7 (21.2%)
4–6
11 (33.3%)
6–8
8 (24.2%)
>8
5 (15.2%)
Table 4
Association between small bowel transit time and reader detection of a positive finding
Small bowel transit time (h)
Odds ratio (OR)
95% confidence interval
p -value
4–6
2.66
0.12–57.62
0.26
6–8
3.00
0.12–73.64
0.25
>8
4.00
0.11–136.95
0.22
The results in [Table 4 ] clearly show the increment of OR in association with the longer SBTT, being the
highest in the time group of greater than 8 hours.
In [Table 5 ], we studied the association between SBTT, age, and sex. Among the 33 patients who
had a documented SBTT, the average transit time was the same (∼5 hours) for both males
and females and across all age groups except for the 40- to 60-year age group who
had a small sample size and a longer SBTT with a statistical significance (p < 0.05).
Table 5
Association between SBTT, age, and sex
Age (y)/sex
Number
Average SBTT (h)
p -value
Age (y)
<40
19
5:12:28
0.67
40–60
2
10:10:21
0.016
>60
12
5:44:37
0.74
Sex
Females
13
5:23:45
0.49
Males
20
5:54:12
0.89
Abbreviation: SBTT, small bowel transit time.
Discussion
In a previous retrospective report, Svarta et al[13 ] studied 676 patients, of which 82 patients (12%) had repeat CE with a diagnostic
yield of 55%. Comparing the data, we had a 74% diagnostic yield for rereading CE and
one-third (33.3%) of the previously negative tests were found positive in the second
reading. Rereading CE was statistically significant (p = 0.0043). These results were comparable, if not higher than, to the sensitivity
of repeating CE in the study conducted by Svarta et al.[13 ]
In contrast, Blanco-Velasco et al[14 ] applied a two-reader approach on 100 SBCEs of various indications (48 tests for
small bowel bleeding and 52 for other indications); a nonsignificant increase of 6.3%
in the diagnostic yield was obtained after a second reading of the small bowel bleed
subset. Nonetheless, our findings suggest that having two readers interpret the CE
provides a significant alternative to repeating this costly test. However, it is evident
that rereading CE might not be useful in cases of bad preparation, capsule retention,
and if some significant mucosal visualization was missed because of the turbulent
capsule pathway. In our study, poor bowel preparation or active bleed might explain
the difference in the positive findings between the two endoscopists, hence the nonconcordance.
Therefore, emphasis on bowel cleanliness is advised. However, recent systematic reviews
and meta-analyses evaluating the effect of bowel preparation on SBCE results showed
no clear advantage.[15 ]
[16 ]
[17 ] Moreover, intestinal bleeding is usually intermittent, and in certain cases, physicians
need to repeat the test to better localize the bleeding focus and decide on therapeutic
interventions (in case of rebleed, overt bleeding, or if blood transfusion is needed).
Furthermore, the time from bleeding can greatly affect the sensitivity of CE. The
closer the study is to the bleeding episode, the higher the sensitivity, dropping
within 1 week to 66.6% and to less than 10% in 2 weeks.[18 ] In patients with occult bleeding who had CE on the same day or the day after fecal
immunochemical test (FIT) turned positive, the prevalence of small bowel disease was
significantly higher.[18 ]
SBTT is defined as the time from first duodenal image to first cecal image.[12 ] Evidence is growing regarding the positive correlation between a slower SBTT and
a higher diagnostic yield. However, there is still some controversy regarding the
specific SBTT that is defined appropriate to have the best diagnostic yield.
In a retrospective study of 212 patients with anemia or OGIB, Buscaglia et al[19 ] showed a twofold increase in the diagnostic yield in patients with an SBTT longer
than 6 hours.
In a prospective study including 1,433 patients with OGIB, Girelli et al[20 ] also concluded that a longer SBTT was associated with a higher detection rate of
significant lesions. This is probably related to better visualization with slower
transit time. However, they did not stratify the SBTT into groups to check for the
specific time above which there was statistical significance for a higher diagnostic
yield. But the mean transit time for the group who had significant findings was 4.7 ± 1.75 hours
compared with 4.48 ± 1.63 hours for the group who had normal or negligible findings,
and this was statistically significant (p < 0.05). This was comparable to the average transit time in our study (∼5 hours;
[Table 5 ]). We found a positive OR in all the following SBTT stratification groups: 4 to 6,
6 to 8, and >8 hours ([Table 4 ]). We noticed that OR increased in association with a longer SBTT (>8 hours) although
no statistical significance was obtained ([Fig. 1 ]). Due to the small sample size, the range of the 95% confidence interval (CI) was
wide ([Table 4 ]). Further studies and larger samples are needed. There is also no difference in
the SBTT between both sexes and between different age groups.
Fig. 1 Graph of odds ratio forest plot of small bowel transit time (SBTT) and sensitivity.
This study has its limitations. First, a retrospective study limits the availability
of additional patient information or test data (e.g. documentation of SBTT). The small
sample size (42 patients) might not increase the statistical power. The preparation
for CE in an outpatient setting is sometimes suboptimal, decreasing the diagnostic
yield. All patients with Crohn's disease were excluded. Also, we were unable to further
stratify the OGIB as overt or occult. In addition, there was a variable time span
between the two reads, and at times it was up to 18 months, explaining why there is
no effect of the findings of the second read on the clinical management, and making
the second physician more prone to detect or overread the findings. Another prospective
study where two independent readers translate the data to the treating physician will
be ideal, making the results and their effects more applicable to clinical situations.
Conclusion
For patients with an initial negative SBCE, a second evaluation of the test is recommended
to increase its diagnostic power and provide a cheaper alternative to repeating the
test. Our rereading-based study results were comparable/superior to both the repeating-based
study conducted by Svarta et al.[13 ] and the two-reader approach adopted by Blanco-Velasco et al.[14 ]
A longer SBTT is possibly correlated with a higher diagnostic yield. However, more
data with higher recruitment numbers and further group stratification according to
transit time should be implemented in future studies.
We also suggest adding “time from bleeding” in case of overt bleed and studying the
usefulness of the FIT test in case of occult bleeding.
