Does Capsule Endoscopy Improve Outcomes in Obscure Gastrointestinal Bleeding? Randomized Trial Versus Dedicated Small Bowel Radiography
Article Outline
- Abstract
- Materials and Methods
- Results
- Discussion
- Acknowledgments
- Supplementary material
- References
- Copyright
Background & Aims
Capsule endoscopy improves the diagnostic yield in patients with obscure gastrointestinal (GI) bleeding, but whether it improves outcomes is uncertain.
Methods
Patients with obscure GI bleeding and negative upper endoscopy, colonoscopy, and push enteroscopy were randomly assigned to capsule endoscopy or dedicated small bowel contrast radiography. Patients returned at 1, 2, 3, 6, 9, and 12 months for follow-up visits and to check hemoglobin level. The primary endpoint was further bleeding.
Results
The predefined sample size of 136 patients (54 overt bleeding, 82 occult bleeding) was enrolled. Diagnostic yield was 20 (30%) with capsule vs 5 (7%) with radiography (difference = 23%; 95% CI: 11%–36%). Further bleeding with capsule versus radiography occurred in 20 (30%) versus 17 (24%) (difference, 6%; 95% confidence interval [CI], −9% to 21%), subsequent diagnostic or therapeutic interventions for bleeding were performed in 17 (26%) versus 15 (21%) (difference, 4%; 95% CI, −10% to 19%), subsequent hospitalizations for bleeding were required in 8 (12%) versus 4 (6%) (difference, 6%; 95% CI, −3% to 16%), and subsequent blood transfusions were given in 5 (8%) versus 4 (6%) (difference, 2%; 95% CI, −7% to 10%). Further bleeding was more common in patients presenting with overt bleeding than in those with occult bleeding (21/54 [39%] vs 16/82 [20%]; difference, 19%; 95% CI, 4% to 35%).
Conclusions
The significant improvement in diagnostic yield with capsule endoscopy may not translate into improved outcomes in a population with obscure GI bleeding. Most patients do well whether or not abnormalities are identified, and additional diagnostic or therapeutic interventions may be required whether or not capsule endoscopy identifies a source of bleeding.
Keywords: Gastrointestinal Hemorrhage, Capsule Endoscopy, Small Bowel Radiography
Abbreviations used in this paper: CI, confidence interval, GI, gastrointestinal
See editorial on page 1655.
Obscure gastrointestinal (GI) bleeding, defined as persistent or recurrent GI bleeding without a source identified by standard evaluation of the upper and lower GI tract, is an uncommon presentation of GI bleeding that can be difficult to diagnose and frustrating to manage for patients and gastroenterologists.1, 2 The most common site identified in patients with obscure GI bleeding is the small intestine.2 The difficulty in examining the entire small intestine has made assessment for small intestinal sources of bleeding problematic. However, the recent advent of video capsule endoscopy has provided the opportunity to obtain images from the entire length of the small intestine in most patients. Capsule endoscopy approximately doubles the diagnostic yield of push enteroscopy in patients with obscure GI bleeding, and its yield is also far superior to that of small bowel radiography.3
The capsule improves visualization and reported diagnostic yield, but it has no means for external control, preventing the ability to visualize all areas of the small intestine and not allowing prolonged examination or reexamination of questionable or poorly seen areas. Furthermore, the video capsule has no means to acquire specimens for diagnostic testing or to apply therapy. In the early days of upper endoscopy for upper GI bleeding, the marked improvement in diagnostic yield with upper endoscopy did not translate into a benefit in clinical outcomes when randomized controlled trials were conducted.4, 5, 6, 7 Thus, it is important to determine whether the use of capsule endoscopy, with improved visualization but without the ability to take diagnostic specimens or to apply therapy, will translate into improvements in management and outcomes in patients.
We therefore performed a randomized controlled trial to assess whether capsule endoscopy influences the management and improves the outcomes of patients presenting with obscure GI bleeding. Our hypothesis was that the improved diagnostic yield of video capsule endoscopy would lead to improved outcomes in patients with obscure GI bleeding as measured by a primary endpoint of the proportion of patients with further bleeding.
Materials and Methods
Adult patients seen at the Los Angeles County + University of Southern California Medical Center for initial evaluation of obscure GI bleeding were eligible for this study. Obscure GI bleeding was defined as either overt (persistent or recurrent melena or hematochezia plus anemia) or occult (iron-deficiency anemia in men or postmenopausal women; iron-deficiency anemia plus positive fecal occult blood test in premenopausal women) and required a nondiagnostic upper endoscopy, colonoscopy, and push enteroscopy. Exclusion criteria included known or suspected GI tract obstruction, severe motility disorders (eg, achalasia, gastroparesis, pseudoobstruction), pregnancy (because of x-ray exposure in patients assigned to small bowel radiography), or cardiac pacemaker or implanted electromedical device.
Subjects were enrolled and randomly assigned by study investigators to capsule endoscopy (PillCam SB1 with RAPID software versions 2.0 to 5.0; Given Imaging Ltd, Yoqneam, Israel) or dedicated small bowel contrast radiography. The dedicated small bowel examination was chosen because it is more accurate than the standard small bowel follow-through and has been used at our institution in place of enteroclysis because of reports that it has sensitivity and specificity comparable to enteroclysis with greater ease, less time required, fewer side effects, and less radiation exposure.8 The randomization schedule was computer generated with concealed allocation and was prepared by someone uninvolved in any aspect of the study. In addition, randomization was stratified based on overt versus occult bleeding.
Patients having capsule endoscopy had preparation with 2 L of polyethylene glycol–based lavage solution orally the night before the capsule endoscope; clear liquids were permitted beginning 2 hours after capsule ingestion. Prokinetic medications were not used, and no capsules were placed into the small intestine endoscopically. Patients having small bowel radiography took nothing by mouth after midnight before their examination. Patients undergoing dedicated small bowel contrast radiography rapidly ingested barium, and the radiologist followed the head of the column of barium through the length of the small bowel to the ileocecal valve with fluoroscopy. Serial radiographic images, including areas of suspected pathology, were obtained during the procedure.
The capsule endoscopies were first reviewed by a senior GI fellow at a maximum speed of 10 frames/s for single-image reading or 15 frames/s in dual-image reading of the small intestine. The study was subsequently read by a faculty member (L.L.), focusing on the abnormalities noted by the first reader, but also reviewing the entire small intestinal examination at speeds ≤25 frames/s in dual-image mode. Dedicated small bowel radiography was read by a radiology resident and faculty attending physician.
Physicians not involved in the study managed the patients and made all subsequent decisions about diagnostic testing and therapeutic interventions. The results of the patients' capsule endoscopy or dedicated small bowel radiography study were provided to the physicians managing the patient. These measures were used to simulate standard clinical practice and to avoid bias that might be introduced if study investigators made clinical management decisions about study subjects. The study follow-up was predefined as 1 year after random assignment. Follow-up included monthly visits with complete blood count for the first 3 months and then every 3 months thereafter for a total of 1 year. Patients were queried at follow-up visits about symptoms of bleeding and other health care interactions since the last visit. If related care occurred at an outside institution, medical records were obtained. A full-time bilingual study coordinator was available for patients to contact at any time during regular hours, and patients also were given contact information for an on-call, noninvestigator gastroenterologist after hours.
The primary endpoint for the study was further bleeding. Further bleeding was defined as overt or occult bleeding requiring subsequent hospitalization or transfusion, or associated with a drop in hemoglobin level after stabilization of >2 g/dL (with hemoglobin in the anemic range). We believe documenting an improvement in diagnostic yield without a clinical benefit would not adequately justify the use of capsule endoscopy. Rather, documentation that capsule endoscopy reduces further bleeding would provide the most important and clinically relevant evidence of the utility of capsule endoscopy. Secondary endpoints included diagnostic yield (proportion of patients in whom a potential source of bleeding is identified by the study diagnostic test; visualization of blood without any lesion was not considered a positive diagnosis), additional interventions (to diagnose and/or treat the cause of bleeding), blood transfusions after randomization, and subsequent hospitalization for bleeding.
Sample size was chosen with the baseline assumptions that 70% of patients have further bleeding after small bowel radiography and this can be decreased by 25% with the improved diagnostic yield of video capsule endoscopy. With these assumptions, 136 patients were required to show a significant difference with a power of 80% and an α of 0.05. Comparison of proportional data between the study groups was performed with Fisher's exact test, with a 2-sided P value < .05 considered significant.
The study was approved by the University of Southern California Health Sciences Institutional Review Board, and all patients provided informed consent, signing study consent forms. This trial was an investigator-initiated study with partial support provided by an American Society for Gastrointestinal Endoscopy Wireless Video Capsule Endoscopy Clinical Research Award. The study had no other sources of support, and none of the authors have a financial or other relationship with the manufacturers of any capsule endoscopy systems.
Results
The progress of patients through the study is shown in Supplementary Figure 1. The 136 patients were enrolled between 2003 and 2008 and included 54 with overt bleeding and 82 with occult bleeding. Baseline characteristics are shown in Table 1. Forty-nine (91%) of 54 patients with overt bleeding were hospitalized before random assignment, whereas 42 (51%) of 82 with occult bleeding were hospitalized before enrollment. The initial capsule was not performed in 6 patients (1 died before study without recurrent bleeding, 1 withdrew consent, 4 did not return for capsule 3 of these 4 returned for follow-up visits for 1, 6, and 12 months). The capsule reached the cecum while still transmitting images in 48 (80%) of 60 patients. One patient, whose capsule recording ended after <2 hours because of technical malfunction, had an immediate repeat capsule study. No retention of capsules within the GI tract occurred. The initial small bowel radiography study was not performed in 4 patients, who refused the study but returned for follow-up visits for 2, 12, 12, and 12 months. No adverse events associated with capsule endoscopy or small bowel radiography were noted.
Table 1. Baseline Characteristics of the Study Groups
| Capsule (n = 66) | Radiography (n = 70) | |
|---|---|---|
| Mean age, y | 56 | 53 |
| Female, n (%) | 44 (67) | 45(64) |
| Latino, n (%) | 53(80) | 59(84) |
| Asian, n (%) | 7(11) | 4(6) |
| African American, n (%) | 3(5) | 4(6) |
| White, n (%) | 3(5) | 3(4) |
| Overt bleeding, n (%) | 26(39) | 28(40) |
| Occult bleeding, n (%) | 40(61) | 42(60) |
| Nonsteroidal anti-inflammatory drug use, n (%) | 28(42) | 22(31) |
| Hospitalized before enrollment, n (%) | 42(64) | 49(70) |
| Received blood transfusion before enrollment, n (%) | 38(58) | 42(60) |
| Mean hemoglobin (g/dL) | 8.1 | 7.9 |
Sixty-one of the 66 patients in the capsule group were followed through the 1-month visit (overt, 23/26; occult, 38/40), 59 were followed through the 2-month visit (overt, 22/26; occult, 37/40), 57 were followed through 3 months (overt, 21/26; occult, 36/40), 54 through 6 months (overt, 21/26; occult, 33/40), and 47 through 12 months (overt, 18/26; occult, 29/40). Sixty-seven of the 70 patients in the small bowel radiography arm were followed through the 1-month visit and the 2-month visit (overt, 28/28; occult, 39/42), 66 were followed through 3 months (overt, 28/28; occult, 38/42), 64 through 6 months (overt, 27/28; occult, 37/42), and 54 through 12 months (overt, 21/28; occult, 33/42).
Results of predefined endpoints after randomization are shown in Table 2. The diagnostic yield for capsule was significantly better than the yield with small bowel radiography (20/66 [30%] vs 5/70 [7%]); P = .0007). Two additional patients in the capsule arm had blood identified on examination without specific lesions visualized. The increased diagnostic yield with capsule versus radiography was consistent for the subgroup of patients with overt bleeding before random assignment and the subgroup with occult bleeding at enrollment (Table 2). Diagnoses with capsule endoscopy included erosions/ulcers and vascular ectasia, whereas diagnoses with small bowel radiography were small intestinal diverticula, focal ileal thickening, and jejunal nodules (Table 2). The median time from enrollment to capsule endoscopy was 1.5 days in the overt group and 6.5 days in the occult group. The diagnostic yield in the overt group was 2 (18%) of 11 patients if the capsule was performed within 1 day after enrollment and 5 (45%) of 11 if done beyond 1 day. The yield in the occult group was 6 (32%) of 19 if the capsule endoscopy was performed within 6 days of enrollment and 7 (37%) of 19 if done beyond 6 days.
Table 2. Predefined Endpoints for all Patients and for the Subgroups With Overt Bleeding and Occult Bleeding at Enrollment
| Capsule, n/n (%) | Radiography, n/n (%) | Difference, % (95% CI) | |
|---|---|---|---|
| Further bleeding | |||
| All patients | 20/66(30) | 17/70(24) | 6(−9to21) |
| Overt bleeding | 13/26(50) | 8/28(29) | 21(4–47) |
| Occult bleeding | 7/40(18) | 9/42(21) | −4(−21to13) |
| Diagnostic yield | |||
| All patients | 20/66(30) | 5/70(7) | 23(11–36) |
| Overt bleeding | 7/26(27) | 1/28(4) | 23(5–42) |
| Erosions/ulcers | 6/26(23) | 0 | |
| Vascular ectasias | 1/26(4) | 0 | |
| Jejunal nodules | 0 | 1/28(4) | |
| Occult bleeding | 13/40(33) | 4/42(10) | 23(6–40) |
| Erosions/ulcers | 9/40(23) | 0 | |
| Vascular ectasias | 4/40(10) | 0 | |
| Small bowel diverticula | 0 | 3/42(7) | |
| Focal ileal thickening | 0 | 1/42(2) | |
| Subsequent interventions for diagnosis or treatment of bleeding | |||
| All patients | 17/66(26) | 15/70(21) | 4(−10to19) |
| Overt bleeding | 10/26(38) | 8/28(29) | 10(−15to35) |
| Occult bleeding | 7/40(18) | 7/42(17) | 1(−15to17) |
| Subsequent hospitalization for bleeding | |||
| All patients | 8/66(12) | 4/70(6) | 6(−3to16) |
| Overt bleeding | 8/26(31) | 2/28(7) | 24(3–44) |
| Occult bleeding | 0/40 | 2/42(5) | −5(−16to5) |
| Subsequent blood transfusion | |||
| All patients | 5/66(8) | 4/70(6) | 2(−7to10) |
| Overt bleeding | 5/26(19) | 2/28(7) | 12(−6to30) |
| Occult bleeding | 0/40 | 2/42(5) | −5(−16to5) |
Further bleeding, the primary endpoint, occurred in 20 (30%) patients in the capsule arm and 17 (24%) in the small bowel radiography arm (95% confidence interval [CI] of difference, −9% to 21%) (Table 2). Seventeen patients (10 capsule and 7 radiography) had recurrent overt bleeding, and 20 (10 capsule and 10 radiography) had recurrent occult bleeding. Further bleeding tended to be higher in the capsule group than in the radiography group among patients with overt bleeding before random assignment (13/26 [50%] vs 8/28 [29%]) (Table 2), but not among patients with occult bleeding before randomization (7/40 [18%] vs 9/42 [21%]) (Table 2). Further bleeding occurred at approximately double the rate in patients presenting with overt bleeding compared with occult bleeding in the overall population (21/54 [39%]) vs 16/82 [20%]; difference, 19%; 94% CI, 4%–35%).
Among the 123 patients followed for ≥3 months all but 14 (11%) (6 capsule [overt, 1; occult, 5], 8 radiography [overt, 3; occult, 5]) had an increase of hemoglobin ≥ 2 g/dL (or into normal range). An additional 14 patients (capsule, 6; radiography, 8) had this increase but then had a subsequent decrease back to <2 g/dL increase (or below normal range). Differences in need for transfusion, subsequent hospitalization, and additional interventions for diagnosis or treatment of bleeding between the study arms were not significant (Table 2).
Deaths occurred in 5 (8%) patients in the capsule arm (1 overt; 4 occult), but none were related to bleeding (3 cardiac, 1 pancreatic cancer, 1 unknown cause). No deaths occurred during the study period in the radiography arm, although one patient was reported by family members to have died at 2 years after enrollment at an outside facility. The difference in 1-year mortality was 8% (95% CI, 1%–14%).
Results of the predefined endpoints based on positive versus negative findings on initial capsule endoscopy or small bowel dedicated radiography are shown in Table 3. The number of patients having positive tests was small, making comparisons of uncertain utility, but no significant differences were seen for positive versus negative initial testing. Further interventions tended to be more common in patients with a positive initial study than a negative initial study in the overall population (10/25 [40%] vs 22/111 [20%]; difference = 20%; 95% CI, 0%–41%). Listings of additional diagnostic or therapeutic interventions for bleeding and diagnoses made of obscure bleeding sources subsequent to initial study diagnostic testing are shown in Table 4 in relation to the initial diagnostic yield (positive vs negative capsule or small bowel radiography) and bleeding presentation (overt vs occult).
Table 3. Number of Patients With Predefined Endpoints Based on Positive Versus Negative Findings on Original Capsule Endoscopy or Small Bowel Radiography
| Capsule endoscopy | Small bowel radiography | |||
|---|---|---|---|---|
| Positive (n = 20) | Negative (n = 46) | Positive (n = 5) | Negative (n = 65) | |
| Further bleeding | 5(25) | 15(33) | 2(40) | 15(23) |
| Subsequent interventions for diagnosis or treatment of bleeding | 8(40) | 9(20) | 2(40) | 13(20) |
| Subsequent hospitalization for bleeding | 2(10) | 6(13) | 1(20) | 3(5) |
| Subsequent blood transfusion | 2(10) | 3(7) | 1(20) | 3(5) |
Table 4. Number of Patients With Additional Diagnostic or Therapeutic Interventions for Bleeding and Subsequent Diagnoses Made of Bleeding Sources, Related to Positive vs Negative Initial Testing and Overt vs Occult Bleeding
| Additional interventions | Subsequent diagnoses | |
|---|---|---|
| Positive capsule, | ||
| Overt bleeding | 3/7 | 0/7 |
| Upper endoscopy = 1 | ||
| Colonoscopy = 1 | ||
| Double-balloon enteroscopy = 2 | ||
| Occult bleeding | 5/13 | 1/13 |
| Upper endoscopy = 1 | Vascular ectasia | |
| Push enteroscopy = 2a | ||
| Colonoscopy = 1 | ||
| Double-balloon enteroscopy = 2 | ||
| Negative capsule | ||
| Overt bleeding | 7/19 | 4/19 |
| Small bowel radiography = 1 | GI stromal tumor = 2 | |
| Computed tomography = 2a | Diverticula = 1 | |
| Bleeding scintigraphy = 1 | Ulcer = 1 | |
| Angiography = 2a | ||
| Push enteroscopy = 2a | ||
| Colonoscopy = 2 | ||
| Double-balloon enteroscopy = 2 | ||
| Transjugular intrahepatic portosystemic = 1 | ||
| Shunt = 1 | ||
| Surgery = 2a | ||
| Occult bleeding | 2/27 | 0/27 |
| Colonoscopy = 1 | ||
| Double-balloon enteroscopy = 1 | ||
| Positive radiography | ||
| Overt bleeding | 0/1 | 0/1 |
| Occult bleeding | 2/4 | 0/4 |
| Upper endoscopy = 1 | ||
| Colonoscopy = 2 | ||
| Capsule endoscopy = 1 | ||
| Double-balloon enteroscopy = 1 | ||
| Negative radiography | ||
| Overt bleeding | 8/27 | 4/27 |
| Computed tomography = 1a | GI stromal tumor = 1 | |
| Bleeding scintigraphy = 1 | Vascular ectasia = 1 | |
| Angiography = 1a | Inflammatory bowel disease (1 y after initial negative colonoscopy) = 1 | |
| Upper endoscopy = 2a | Gastric ulcer (7 mo after initial testing negative) = 1 | |
| Push enteroscopy = 1 | ||
| Flexible sigmoidoscopy = 1 | ||
| Colonoscopy = 3a | ||
| Capsule endoscopy = 2 | ||
| Double-balloon enteroscopy = 2 | ||
| Surgery = 1a | ||
| Occult bleeding | 5/38 | 2/38 |
| Upper endoscopy = 2 | Erosions = 2 | |
| Push enteroscopy = 1 | ||
| Colonoscopy = 1 | ||
| Capsule endoscopy = 4a | ||
aDiagnostic tests that subsequently identified source of bleeding. All 3 GI stromal tumors were visualized on computed tomography and were confirmed at surgery. |
Discussion
Capsule endoscopy for evaluation of obscure GI bleeding has been evaluated in a number of case series and cohort studies. A meta-analysis showed that capsule endoscopy had a significantly greater diagnostic yield than push enteroscopy with pooled yields from 14 studies of 63% versus 28%.3 Capsule endoscopy was also superior to small bowel radiography: pooled yields from 3 studies were 67% versus 8%.3 A number of these case series also suggest that the findings of the capsule endoscopy led to meaningful changes in clinical management in many patients, and some also suggest an improvement in outcomes such as bleeding activity.9, 10, 11, 12 However, the true result of any change in management related to a diagnosis made with capsule endoscopy cannot be assessed appropriately in a case series or single cohort. A randomized trial is required to have a suitable control group for comparison.
A MEDLINE search in May 2009 of the term “capsule endoscopy” restricted to randomized controlled trials and to clinical trials shows only one prior randomized controlled trial of capsule endoscopy versus another diagnostic test.13 This trial compared capsule endoscopy with push enteroscopy in 89 patients presenting with obscure GI bleeding. Diagnostic yield in the all-patients randomly assigned analysis was higher with capsule endoscopy (43% vs 21%; P = .05). However, because all patients with a negative result on their first test (capsule or push enteroscopy) were crossed over to receive the other diagnostic test, the clinical outcomes related to capsule endoscopy compared with a control cannot be determined. No significant differences were seen in diagnostic yield, therapeutic effect, additional explorations, or clinical remission at 1 year between the capsule-first and push enteroscopy–first strategies.
We confirmed that capsule endoscopy significantly improves diagnostic yield. At first glance, our 30% diagnostic yield with capsule endoscopy may seem low compared with prior studies. However, our study's inclusion criteria required a negative push enteroscopy, thereby removing many patients who may have had a positive capsule. As mentioned, the pooled yield for push enteroscopy versus capsule is 28% versus 63%. This suggests that, after removing the 28% of patients who would have a diagnosis with push enteroscopy, one would expect the yield of capsule to be 35%, very close to our 30% yield. In addition, in contrast to some other studies, we did not include the finding of blood without a lesion as a positive finding.
Despite the better diagnostic yield of capsule compared with small bowel radiography, we found no significant difference between capsule and radiography in the number of patients with further bleeding or in the number of subsequent interventions for bleeding, subsequent hospitalizations for bleeding, or transfusions. We speculate that the improved diagnostic yield of capsule endoscopy did not translate into an improvement in clinical outcomes because (1) most patients with obscure bleeding do well, whether or not abnormalities are identified, and (2) subsequent interventions may be required even when the initial test is positive because the current capsule endoscope cannot obtain specimens or apply therapy. In addition, merely visualizing a lesion on capsule (or radiography) does not document that the lesion is the cause of bleeding unless active bleeding or stigmata of recent hemorrhage are also identified. For example, Haghighi et al14 reported that vascular ectasias, ulcers, and erosions were at least as common on capsule endoscopy in healthy subjects as in patients with obscure bleeding.
We did not perform an economic analysis as part of our study. Repeat hospitalizations and subsequent diagnostic and therapeutic interventions are the primary drivers of cost. Because subsequent hospitalizations and interventions occurred at low and similar rates in the 2 study groups, our results suggest capsule endoscopy was not likely to have provided a cost advantage in our population.
The rate of further bleeding in our population was low, as were the rates of subsequent interventions, hospitalizations, and transfusions. The low rate of further bleeding was noted regardless of whether a patient had a capsule or small bowel radiography and regardless of whether their initial diagnostic test was positive or negative. Low rates of further bleeding also have been reported previously in nonrandomized trials, with some series noting a higher rate of further bleeding in those with positive capsule studies compared with negative studies.11, 15, 16 Some series have also indicated that diagnostic yield may be higher in patients presenting with overt obscure bleeding compared with occult obscure bleeding,12 whereas others did not.15 We found no suggestion of a difference in the diagnostic yield between patients with overt and occult bleeding. However, patients with overt bleeding in our study were twice as likely to have further bleeding compared with patients with occult bleeding.
Current recommendations from an American Gastroenterological Association Medical Position Statement suggest that patients with obscure GI bleeding undergo a capsule endoscopy after a negative upper endoscopy and colonoscopy, that patients with a positive capsule have subsequent interventions directed by the findings, and that those without a bleeding source identified can either be observed or, if their clinical course mandates, undergo further diagnostic testing.2 Our study did not directly assess this algorithm because we evaluated capsule endoscopy in patients after a negative upper endoscopy, colonoscopy, and push enteroscopy. We decided to include push enteroscopy before capsule because it can obtain diagnostic specimens and provide therapy and also should identify nearly half the abnormalities seen by capsule.3 Future randomized trials will need to assess whether push enteroscopy or capsule should be the first test after negative upper endoscopy and colonoscopy and whether capsule endoscopy would improve outcomes if performed before push enteroscopy.
Although our results suggest that capsule endoscopy may not improve outcomes in an overall population of patients with obscure bleeding after a negative push enteroscopy, individual patients may benefit from capsule endoscopy. Future studies should attempt to identify clinical characteristics that aid in stratifying the use of capsule endoscopy and other interventions. For example, our results suggest that patients with occult obscure bleeding may have a more benign course than patients with overt obscure bleeding.
The rate of further bleeding in this study was much lower than the baseline assumption of 70% used in designing the study. This can decrease the chance of showing an observed difference to be statistically significant. Although further bleeding was slightly more common with capsule endoscopy than radiography, the 95% CIs around the difference in observed rates of further bleeding suggest that capsule endoscopy could be associated at most with a 9% lower rate of further bleeding than radiography in our population. It is uncertain whether clinicians would consider this 9% difference clinically important, but sample size calculations indicate that to document this difference as statistically significant would require a study of 650 patients.
Examination of the population studied is important in assessing the generalizability of study results. For example, factors such as age and concurrent medications may affect the causes of bleeding and rates of further bleeding. Younger patients reportedly are more likely to have small intestinal tumors and inflammatory bowel disease, whereas older patients are more likely to have vascular lesions and nonsteroidal anti-inflammatory drug-induced erosions or ulcers.2 In addition, the greater use of antithrombotic therapy among older patients may increase the risk of obscure bleeding from preexisting, previously silent lesions. The mean age in our study was approximately 55 years (within the range of mean ages from the late 40s to late 60s in other studies17, 18), and nearly 40% of our patients took nonsteroidal anti-inflammatory drugs (including aspirin). The main difference in our study population compared with other studies is the high proportion of Latino subjects. However, to our knowledge, racial or ethnic background is not documented to be an independent factor influencing the source or outcome of obscure bleeding.
In conclusion, the significant improvement in diagnostic yield with capsule endoscopy may not translate into improved outcomes in a population with obscure GI bleeding. Most patients do well whether or not abnormalities are identified, and additional diagnostic or therapeutic interventions may be required whether or not the capsule identifies a source of bleeding. The ability to control the capsule and to perform diagnostic and therapeutic interventions may be necessary to significantly improve clinical outcomes with capsule endoscopy in a population with obscure GI bleeding.
Acknowledgments
This trial was registered at www.clinicaltrials.gov as NCT01006824.
Supplementary material
Supplementary Figure 1.
Flow diagram showing progress of patients through the study.
References
- AGA technical review on the evaluation and management of occult and obscure gastrointestinal bleeding. Gastroenterology. 2000;118:201–221
- American Gastroenterological Association (AGA) technical review on obscure gastrointestinal bleeding. Gastroenterology. 2007;133:1697–1717
- A meta-analysis of the yield of capsule endoscopy compared to other diagnostic modalities in patients with obscure gastrointestinal bleeding. Am J Gastroenterol. 2005;100:2407–2418
- Prospective, randomized study of diagnosis and outcome in acute upper-gastrointestinal bleeding: endoscopy versus conventional radiography. Am J Dig Dis. 1975;20:1103–1109
- Outcome of endoscopy and barium radiography for acute upper gastrointestinal bleeding: controlled trial in 1037 patients. Br Med J. 1982;284:545–548
- . Limited value of early endoscopy in the management of acute upper gastrointestinal bleeding (Prospective controlled trial). Am J Surg. 1980;140:284–290
- Routine early endoscopy in upper-gastrointestinal-tract bleeding: a randomized, controlled trial. N Engl J Med. 1981;304:925–929
- Detailed per-oral small bowel examination vs. enteroclysis (Part II: radiographic accuracy). Radiology. 1985;155:31–34
- Impact of capsule endoscopy on outcome in mid-intestinal bleeding: a multicentre cohort study in 285 patients. Eur J Gastroenterol Hepatol. 2008;20:971–977
- Diagnostic yield and clinical outcomes after capsule endoscopy in 100 consecutive patients with obscure gastrointestinal bleeding. Eur J Gastroenterol Hepatol. 2006;18:881–888
- Capsule endoscopy versus standard tests in influencing management of obscure digestive bleeding: results from a German multicenter trial. Am J Gastroenterol. 2005;100:1736–1742
- Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology. 2004;126:643–653
- Capsule endoscopy or push enteroscopy for first-line exploration of obscure gastrointestinal bleeding?. Gastroenterology. 2007;132:855–862
- Comparison of capsule endoscopy (CE) findings of healthy subjects (HS) to an obscure gastrointestinal bleeding (OGIB) patient population [abstract]. Gastrointest Endosc. 2005;61:AB104
- Long-term follow-up of patients with obscure gastrointestinal bleeding after negative capsule endoscopy. Am J Gastroenterol. 2006;101:1224–1228
- . Negative capsule endoscopy in patients with obscure GI bleeding predicts low rebleeding rates. Gastrointes Endosc. 2008;68:1122–1127
- Diagnostic yield and therapeutic impact of double-balloon enteroscopy in a large cohort of patients with obscure gastrointestinal bleeding. Am J Gastroenterol. 2006;101:2011–2015
- A single-center experience of 260 consecutive patients undergoing capsule endoscopy for obscure gastrointestinal bleeding. Am J Gastroenterol. 2007;102:89–95
This article has an accompanying continuing medical education activity on page e11. Learning Objective: Upon completion of reading this article, successful learners will be able to discuss the diagnostic evaluation of obscure gastrointestinal bleeding.
Conflicts of interest The authors disclose no conflicts.
Funding This work was supported by a grant from the American Society for Gastrointestinal Endoscopy Wireless Video Capsule Endoscopy Clinical Research Award (Project no. 16; 2004).
PII: S0016-5085(10)00156-3
doi:10.1053/j.gastro.2010.01.047
© 2010 AGA Institute. Published by Elsevier Inc. All rights reserved.
Refers to article:
- Continuing Medical Education Exam 1, May 2010 , 22 March 2010
