Cost-effectiveness analysis of management strategies for obscure GI bleeding

ORIGINAL ARTICLE: Clinical Endoscopy

Cost-effectiveness analysis of management strategies for obscure GI bleeding Lauren Gerson, MD, MSc, Ahmad Kamal, MD, MSc Stanford, California, USA

Background and Aims: Of patients who are seen with GI hemorrhage, approximately 5% will have a smallbowel source. Management of these patients entails considerable expense. We performed a decision analysis to explore the optimal management strategy for obscure GI hemorrhage. Methods: We used a cost-effectiveness analysis to compare no therapy (reference arm) to 5 competing modalities for a 50-year-old patient with obscure overt bleeding: (1) push enteroscopy, (2) intraoperative enteroscopy, (3) angiography, (4) initial anterograde double-balloon enteroscopy (DBE) followed by retrograde DBE if the patient had ongoing bleeding, and (5) small-bowel capsule endoscopy (CE) followed by DBE guided by the CE findings. The model included prevalence rates for small-bowel lesions, sensitivity for each intervention, and the probability of spontaneous bleeding cessation. We examined total costs and quality-adjusted life years (QALY) over a 1-year time period. Results: An initial DBE was the most cost-effective approach. The no-therapy arm cost $532 and was associated with 0.870 QALYs compared with $2407 and 0.956 QALYs for the DBE approach, which resulted in an incremental cost-effectiveness ratio of $20,833 per QALY gained. Compared to the DBE approach, an initial CE was more costly and less effective. The initial DBE arm resulted in an 86% bleeding cessation rate compared to 76% for the CE arm and 59% for the no-therapy arm. The model results were robust to a wide range of sensitivity analyses. Limitations: The short time horizon of the model, because of the lack of long-term data about the natural history of rebleeding from small-intestinal lesions. Conclusions: An initial DBE is a cost-effective approach for patients with obscure bleeding. However, capsuledirected DBE may be associated with better long-term outcomes because of the potential for fewer complications and decreased utilization of endoscopic resources. (Gastrointest Endosc 2008;68:920-36.)

Copyright ª 2008 by the American Society for Gastrointestinal Endoscopy 0016-5107/$34.00 doi:10.1016/j.gie.2008.01.035

vascular lesions, including Dieulafoy’s lesions; or hemorrhage associated with portal hypertension. The natural history of bleeding from AVMs remains poorly understood. There appears to be a spontaneous cessation rate from AVMs of approximately 40% per year.6,7 Current options for the diagnosis and management of small-bowel lesions include push enteroscopy, capsule endoscopy (CE), intraoperative enteroscopy, and doubleballoon enteroscopy (DBE). Sonde enteroscopy, a technique that involves the peroral placement of a long (3 m) enteroscope, with a distal balloon that is propelled by peristaltic activity into the distal small bowel has been largely abandoned because of the long duration of the examination (mean 7 hours), patient discomfort, the need for additional endoscopy staff, and the inability to perform therapeutic procedures.8 The diagnostic yield of push enteroscopy in patients with obscure overt bleeding ranges from 3% to 70%, with AVMs as the most common lesions identified in 7% to 60% of examinations.2,9,10 Although

920 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Approximately 5% of patients who are seen with GI hemorrhage have no source found by an upper endoscopy and a colonoscopy.1 In approximately 75% of these patients, responsible lesions can be detected in the small bowel.2-4 Small-bowel arteriovenous malformations (AVMs) are the most common finding in 20% to 40% of patients who are seen with obscure overt bleeding (defined as the presence of recurrent melena or hematochezia with normal evaluation by upper endoscopy and colonoscopy).5 Other potential causes of bleeding, in order of frequency, include ulcerations; primary or metastatic tumors and Abbreviations: AVM, arteriovenous malformation; CE, capsule endoscopy; CPT, current procedural terminology code; DBE, double-balloon enteroscopy; ICER, incremental cost-effectiveness ratio; QALY, qualityadjusted life year; QOL, quality of life; RBC, red blood cell.

Gerson & Kamal

a CE does not offer therapeutic potential for small-bowel lesions, multiple studies to date demonstrated a 25% to 55% higher diagnostic yield of CE compared with push enteroscopy in patients who were seen with obscure GI hemorrhage.11,12 Intraoperative enteroscopy, which involved the insertion of an endoscope through an incision in the mid small intestine (enterotomy), is associated with a diagnostic yield between 58% and 84%.13,14 Although the success rate of reaching the ileum approaches 90%, the use of intraoperative enteroscopy does not appear to be associated with lower rates of recurrent GI bleeding and carries a significant risk for postoperative complications, including mortality in up to 2% to 4% of cases. DBE, first described by Yamamoto et al15 in 2001, allows complete visualization of the small intestine by using a 200-cm enteroscope (Fujinon Inc, Saitama, Japan) equipped with a 140-cm overtube. In the major studies published on DBE to date, obscure bleeding was the indication for the DBE in approximately 36% to 100% of examinations, and the overall diagnostic yield from DBE ranged from 43% to 80%.16-18 Diagnostic or therapeutic success was reached in 55% to 75% of examinations, comparable with other diagnostic modalities for the small bowel. Recently, the International Conference on Capsule Endoscopy produced a consensus statement on obscure GI bleeding and concluded that ‘‘capsule endoscopy was the preferred test for mucosal imaging of the entire small intestine and should be part of the initial evaluation in patients with obscure bleeding.’’19 The proposed algorithm from this consensus conference proposed an initial CE followed by other potential diagnostic tests, including a DBE, angiography, or intraoperative enteroscopy, if there were positive findings on a CE. Although a CE allows for visualization of the entire small bowel in a noninvasive fashion, disadvantages associated with the use of a CE include the potential for missed lesions20 and an inability to perform therapeutic intervention. We, therefore, used decision analysis to estimate the potential cost-effectiveness of an initial CE compared with other diagnostic modalities, including DBE, push enteroscopy, angiography, or intraoperative enteroscopy, for patients with bleeding from small-bowel lesions.

Management strategies for obscure GI bleeding

Capsule Summary What is already known on this topic d

Double balloon enteroscopy (DBE) allows for diagnostic imaging and endoscopic therapy in the small bowel but is time consuming and expensive.

What this study adds to our knowledge d

In a decision analysis that compared the costeffectiveness of an initial DBE to other imaging modalities for obscure bleeding, the DBE approach was a cost-effective approach, but an initial capsule endoscopy reduced the number of DBE procedures and had fewer associated complications.

CE followed by a DBE guided by the CE findings. Patients in the CE arm would only proceed to a DBE if they had persistent obscure bleeding after the CE examination. Patients with normal CE examinations and ongoing GI hemorrhage would undergo an initial anterograde DBE, with the assumption that a lesion might have been missed on the prior CE examination. We performed our analysis from a third-party–payer perspective over a 1-year time horizon by using 1-month cycles for probabilities of hemorrhage, bleeding cessation, and alterations in healthrelated quality of life (QOL).

Base-case patient The base-case patient was a 50-year-old man with a 6-month history of recurrent melena and associated iron-deficiency anemia. The patient had a prior normal upper endoscopic examination, including retroflexed views of the gastric cardia and the fundus, a colonoscopy, and a small-bowel series. He required approximately 2 units of packed red blood cells (RBC) per month for ongoing GI bleeding. He denied the use of aspirin, nonsteroidal anti-inflammatory agents, or warfarin. There was no personal history of malignancy or prior surgeries, except for a routine appendectomy. The patient was considered to be a candidate for a CE, as well as endoscopic or surgical therapy, for presumed small-bowel sources of the bleeding.

PATIENTS AND METHODS

Competing strategies

Cost-effectiveness analysis is a quantitative method used to evaluate the outcomes and costs of interventions designed to improve health.21 We used decision analysis software (TreeAge Pro 2005 Suite; TreeAge Software, Boston, Mass) to create a decision tree to compare no therapy (reference arm) to 5 competing modalities for a 50-yearold patient with obscure overt bleeding: (1) push enteroscopy, (2) intraoperative enteroscopy, (3) angiography, (4) initial anterograde DBE, followed by a retrograde DBE if the patient had ongoing bleeding, and (5) small-bowel

A truncated version of the decision tree is depicted in Figure 1. A patient who is seen with obscure GI bleeding would initially undergo a diagnostic upper endoscopy, a colonoscopy, and a small-bowel series that would be nondiagnostic for a potential bleeding source. Obscure bleeding would continue, and the patient would then enter the decision tree, at which time the following therapeutic options outlined above could be undertaken. We did not model the possibility of obtaining additional diagnostic or therapeutic interventions if the patient continued to bleed after an initial procedure. The exception

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 921

Management strategies for obscure GI bleeding

Gerson & Kamal

Figure 1. The model schema. Patients with small-intestinal bleeding from AVMs may enter the no-therapy arm (supportive blood transfusions, reference arm) or 1 of 5 treatment arms, including push enteroscopy, intraoperative enteroscopy, angiography, initial oral DBE, or CE followed by DBE if patients continued to bleed after CE. Patients with normal capsule examinations and who were ongoing bleeding underwent subsequent anterograde DBE. For each arm, the probability of spontaneous cessation, cessation after endoscopic therapy, and procedural complications were modeled.

to this was in the CE arm, in which patients who had positive findings on a CE and ongoing bleeding would proceed to DBE examination, with the route determined by the location of the lesion during the CE. A DBE would subsequently be performed by the opposite approach only if the patient had persistent bleeding after the initial DBE. Our model assumed that there could be single or multiple small-bowel lesions and that the diagnostic test would locate and treat multiple lesions if present in the area of small bowel examined. For example, we modeled the probability that patients could continue to bleed from distal AVMs after an anterograde DBE located and treated multiple proximal small-bowel lesions. After each intervention, we modeled the probability of cessation or continuation of bleeding after endoscopic therapy as shown by a circular ‘‘chance node.’’ Patients who were undergoing diagnostic and potentially therapeutic procedures were also modeled to sustain a chance of complication related to the intervention, including hemorrhage, perforation, and/or death from the complication. Patients with lesions missed by the examination were modeled to have either spontaneous cessation of bleeding from the lesion or ongoing GI hemorrhage that resulted in ongoing transfusions and a possibility of death from ongoing hemorrhage (Fig. 2). We did not model repeated or alternative procedures if the patients continued to experience GI bleeding.

We modeled that patients in the no-therapy arm would not undergo any endoscopic interventions. AVMs are ectatic blood vessels that consist of thin walls with or without endothelial lining. The factors that trigger bleeding from AVMs and the natural history of vascular lesions in the small bowel have not been well characterized. Estimates about the spontaneous cessation rate were derived primarily from clinical trials that compared

922 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Potential lesions in the small bowel that accounted for bleeding included AVMs (prevalence of 40%), ulcerations (20%), neoplasms (5%), and other vascular lesions (20%), including Dieulafoy’s lesions of the small bowel. We also assumed that 15% of the patients would have missed lesions in the stomach or the colon, including Cameron’s lesions, gastric antral vascular ectasia, and AVMs of the colon and/or malignancies. For each of these lesions, we modeled the prevalence of each lesion and the sensitivity of each intervention for the diagnosis of the lesion based on a comprehensive literature review. Given the limitations of the published literature, we assumed 100% specificity for each test. Limited long-term data exist about the recurrence of AVMs after therapeutic intervention and rates of spontaneous cessation for these lesions over time. We, therefore, limited our analysis to a 1-year time horizon based on the current evidence from the available medical literature.

No-therapy arm

Gerson & Kamal

Management strategies for obscure GI bleeding

Patients Without Bleeding

Patients With Obscure Bleeding

Intervention

Complication

Death Surgery

Figure 2. A Markov state diagram for patients with obscure GI bleeding. Patients enter the model with obscure bleeding. During each 3-month cycle, patients can either remain in a state of bleeding (recursive arrow) or progress to a new health state, as demonstrated by a straight arrow. Patients may stop bleeding spontaneously or via an endoscopic or surgical intervention. Transition rates between health states were derived from the literature (see Table 1).

hormonal therapy with a placebo for the treatment of AVMs in the GI tract. The spontaneous cessation rate in these patients was reported to range from 40% to 50% per year (Table 1).6,7 For example, in a 1992 study by Lewis et al,7 44% of 34 patients with small bowel AVMs who were untreated did not receive any transfusions over a mean follow-up period of 13.4 months. We, therefore, assumed that the rate of spontaneous cessation of bleeding from AVMs would be 45% per year. We modeled that patients enrolled in the no-therapy arm would receive approximately 2 units of packed RBCs per month. This estimate was based on data from a 1996 retrospective study,22 in which 83 patients who had GI bleeding from small-bowel AVMs demonstrated on both push enteroscopy and Sonde enteroscopy were observed for outcomes after their examinations. Of these patients, 55 (66%) underwent electrocautery of their AVMs, whereas 28 (34%) received only medical therapy. Follow-up assessments of these patients revealed that, over a mean (SD) of 26 months, the noncauterized group continued to bleed, which required 2.2  3.9 units of packed RBC per month, which did not change significantly after diagnosis, whereas the cauterized group significantly improved from 2.4  3 units per month pretreatment to 0.32  0.91 units per month after cauterization (P ! .001). There has been a paucity

of published literature about the natural history of bleeding from small-bowel AVMs in patients who do not receive endoscopic or medical therapy. We did not model the administration of medical therapy for the treatment of AVMs, because studies to date have not demonstrated benefit of hormonal therapy compared with a placebo.23 Although there is likely to be formation of additional AVMs over time, we did not model this occurrence because of the short time horizon of 1 year and the fact that no data are available about the rate of this phenomenon. For patients with small-bowel ulcerations, we assumed that these patients would receive 8 weeks of proton pump inhibitor therapy and that most of the lesions would be caused by aspirin or other anti-inflammatory agents.24 We did not model the possibility or treatment of infectious etiologies (such as tuberculosis) as an etiology for smallbowel ulcerations. We also did not model the probability or management of ulcerations that were caused by inflammatory bowel disease. Based on a prospective study by Lai et al,25 in 2006, in which 49 patients were observed for 19 months after a CE was performed for obscure bleeding, we modeled that the rebleeding rate would approximate 25% for small-bowel ulcerations and 50% for patients with active bleeding seen without an identifiable cause (most likely vascular

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 923

Management strategies for obscure GI bleeding

Gerson & Kamal

TABLE 1. Baseline probabilities for the model* Probability

Base-case estimate Range in sensitivity analysis Sensitivity

Reference

Findings in patients with obscure GI bleeding* AVM Overall prevalence

40%

20%-60%

Push enteroscopy

18%

4%-49%

45%

2-4,9-12,28-42

Intraoperative enteroscopy

36%

19%-48%

90%

13,14,30,53-57,84

Angiography

15%

10%-52%

38%

59,60,85

DBE

33%

5%-56%

83%

24,66,67

CE

35%

16%-64%

87%

56,66,67

80%

50%-100%

Overall prevalence

5%

1%-10%

Push enteroscopy

2%

0%-10%

40%

Angiography

4%

4%-10%

45%, 78%

Intraoperative enteroscopy

5%

7%-17%

100%

CE

4%

0%-13%

80%

11,12,35,36,38,41-43,56,61

DBE

4%

4%-20%

80%

24,67,81,88-90

Overall prevalence

20%

10%-40%

Push enteroscopy

4%

0%-15%

20%

2-4,9-12,28-42

Angiography

4%

0%-6%

Intraoperative enteroscopy

18%

0%-27%

90%

13,14,30,53-57

CE

11%

5%-26%

55%

11,12,35,36,38,41-43,56,61

DBE

13%

0%-56%

65%

24,67,81,88-90

20%

0%-20%

Push enteroscopy

3%

0%-49%

15%

2-4,9-12,28-42

Angiography

6%

0%-43%

30%

61,86,87

Intraoperative enteroscopy

10%

3%-13%

50%

13,14,30,53-57

CE

10%

0%-40%

50%

11,12,35,36,38,41-43,56,61

DBE

10%

0%-19%

50%

13,14,30,53-57

15%

5%-20%

Detection rates

Probability that AVM will be within reach of DBE anterograde approach

24

Malignancy

2-4,9-12,28-42 61,86,87 13,14,30,53-57

Ulcerations

20%, 60%

61,86,87

Other vascular lesions Prevalence

Missed lesions on upper or lower endoscopyy

44,91,92

(continued on next page)

924 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Gerson & Kamal

Management strategies for obscure GI bleeding

TABLE 1 (continued ) Probability

Base-case estimate Range in sensitivity analysis Sensitivity

Reference

Rebleeding rates Without endoscopic therapy AVM

0.55

0.22-0.80

25,38,46

Ulceration

0.25

0.10-0.40

25

Tumor

0.90

0.0-0.10

25

Other, including vascular

0.50

0.25-0.80

25,46

Spontaneous cessation rate bleeding AVMs

0.45

0.10-0.80

6,7

AVM

0.20

0.10-0.40

25,46

Ulceration

0.10

0.05-0.25

93-95

Other vascular lesions

0.25

0.0-0.50

96-99

Mortality, GI hemorrhage

0.10

0.04-0.45

Complication after endoscopyz

0.0001

Complication after DBEz

0.02

0.002-0.02

Retention after CE

0.01

0-0.21

Success in capsule retrieval by DBE after retention

0.50

0.25-0.75

Mortality after endoscopic complication

0.00005

Morbidity after intraoperative enteroscopy

0.20

0.01-0.57

13,14,30,53-57

Mortality after intraoperative enteroscopy

0.05

0-0.20

13,14,30,53-57

Complication after angiography

0.05

0.01-0.20

62,63

Mortality after angiography

0.0005

0.0002-0.0010

62,63

Postendoscopic therapy

Complication rates

0.0013-0.000009

0.000021-0.00009

100-102 7,12,28,29 65 42,103-106 71,72 7,12,28,29

QOL estimates No bleeding

1.0

0.9-1.0

Definition

GI hemorrhage

0.80

0.70-1.0

75,76

Hospitalization

0.75

0.40-0.81

77,78

Return to normal QOL after bleeding cessation

1 mo

0-3 mo

Expert opinion

Return to normal QOL after perforation

3 mo

0-6 mo

79

xIncludes reported cases of perforation after DBE; pancreatitis and abdominal pain were not included.*Includes weighted averages from the literature. See Patients and Methods section for more details. yIncludes missed lesions found in the upper-GI tract that would be in reach of a standard endoscope, including AVMs, ulcerations, hiatal hernia with ulcerations (Cameron’s erosions), and other vascular lesions (Dieulafoy’s lesions); missed lesions in the colon, including diverticula, AVMs, and other vascular lesions; and lesions in the small bowel, including Dieulafoy’s lesions, diverticula, radiation enteritis, or vasculitis. We assumed that both a push enteroscopy and a DBE would be 75% effective in identification and treatment of missed lesions. CE and angiography would detect missed lesions 50% of the time, whereas they would not be detected during intraoperative enteroscopy. zComplications after a push enteroscopy or a DBE, including perforation, hemorrhage, cardiopulmonary complications, and adverse effects caused by medication.

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 925

Management strategies for obscure GI bleeding

lesions) and that there would not be any rebleeding for tumors that were identified and resected.

Push enteroscopy Video push enteroscopy, by using adult or pediatric colonoscopes, is commonly used in the diagnosis and management of obscure bleeding. Dedicated push enteroscopes were developed with working lengths of 220 to 250 cm. For patients enrolled in the push enteroscopy arm, we assumed that push enteroscopy, by using a traditional small-bowel video enteroscope, would be performed with the patient under conscious sedation. We assumed that the extent of the examination would be 50 to 150 cm of visualized small bowel distal to the proximal duodenum and that an overtube would be used, based on data that the use of an overtube to prevent looping correlates with significantly greater depth of insertion in 2 casecontrol studies.26,27 Although the overall yield for any pathologic lesions in patients with obscure overt bleeding ranges from 3% to 70% on push enteroscopy, we calculated, by using weighted averages from prior clinical trials, that AVMs would be identified in approximately 18% of push enteroscopies.2-4,9-12,28-43 Once AVMs were identified, endoscopic therapy would be administered. We assumed that the efficacy for endoscopic treatment would be equivalent, regardless of whether ablation occurred when using electrocoagulation, laser therapy, or argon plasma coagulation.9,28,29,31,32,39-42,44,45 We modeled that cessation of bleeding would be associated with improved patient outcomes, including a decreased probability of death from hemorrhage and a decreased cost associated with transfusions. To calculate the probability that a patient would stop bleeding after a push enteroscopy, we used information from clinical trials of endoscopic therapy for smallbowel lesions and cessation rates.9,28,29,31,32,39-42,44,45 Based on the literature, we modeled that there would be a 20% rebleeding rate in a patient who received endoscopic therapy for AVMs, whereas AVMs that were not treated would have a rebleeding rate of 55%.25,46 We also modeled that bleeding AVMs that were not treated would have a spontaneous cessation rate of 45%.6,7 The model included the possibility of a complication associated with push endoscopy of approximately 1 per 10,000 examinations,47 with an associated mortality of 0.005%. We assumed that a perforation that required surgical repair would occur in 50% of the patients with a complication and that hemorrhage that required hospitalization with emergent endoscopic examination would occur in the remaining 50% of patients.48,49 We assumed that all patients with complications would present within 1 to 7 days after the endoscopic procedure.48

Gerson & Kamal

tine (enterotomy), was initially performed in the 1950s with a sterile rigid sigmoidoscope50 and then, in the 1970s, with fiberoptic endoscopes.51 In the 1980s, intraoperative enteroscopy was subsequently performed by passing a fiberoptic colonoscope first orally and then anally while the surgeon would manually telescope the bowel over the tip of the endoscope,30,52 which allowed passage of the enteroscope to the terminal ileum in the majority of cases, while decreasing morbidity associated with an enterotomy. Traditionally, the criterion standard for comparison of the efficacy of an intraoperative enteroscopy was surgical resection, with histopathologic documentation of the pathology. For patients with GI bleeding of obscure origin, the overall diagnostic yield of intraoperative enteroscopy was between 58% and 84%.13,14,30,53-57 The performance characteristics for intraoperative enteroscopy are shown in Table 1. By using weighted probabilities from prior studies, we calculated that the prevalence of AVMs was approximately 36% on an intraoperative enteroscopy, with a sensitivity of 90%. The study by Hartmann et al56 compared a CE with an intraoperative enteroscopy in 47 patients. A CE detected definitive multiple or bleeding angiectasias in 22 of 47 patients (47%). An intraoperative enteroscopy confirmed AVMs in all of these patients (100%). One patient had an AVM found on intraoperative enteroscopy that was not seen on CE. Based on the literature, intraoperative enteroscopy had the lowest miss rate for small-bowel lesions compared with the other diagnostic modalities (Table 1). The calculated postoperative morbidity rate associated with an intraoperative enteroscopy was 20%, with a mortality rate of 5%.13,14,30,53-57

Angiography

Intraoperative enteroscopy, which involves insertion of an endoscope through an incision in the mid small intes-

Angiographic therapy is most successful when patients have acutely bleeding lesions and are bleeding at least 1.0 mL/min. Tagged RBC scans are usually initially attempted, because detection of bleeding sites can occur with bleeding rates of 5 mL/min. Because our base-case patient had obscure bleeding, we did not model initial testing with tagged RBC scans but instead proceeded directly to selective mesenteric angiographic examination, and we assumed that a detected lesion would be treated with selective embolotherapy with either microcoils, gelatin sponge pledgets, or polyvinyl alcohol embolospheres.58 To identify the probability of finding a known AVM when using angiography, we examined prior studies in which subsequent surgical resection was the criterion standard. Because most of these prior studies included small numbers of patients, we used weighted averages to obtain our base-case probability. Based on these studies, the ability of angiography to detect AVMs ranged from 10% to 52%, with a weighted average of 20%.59,60 An angiography had a lower detection rate for AVMs compared with a CE, DBE, and intraoperative enteroscopy, and, similarly, a higher miss rate for other small-bowel lesions that

926 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Intraoperative enteroscopy

Gerson & Kamal

Management strategies for obscure GI bleeding

TABLE 2. Published studies of DBE in patients with obscure GI bleeding Patients with bleeding/DBE examinations

Diagnostic yield (%)

Yamamoto et al, 2004

66/178 (37%)

76

61

86

N/A

Perforation: 1 (0.6%)

May et al,81 2005

90/248 (36%)

80

76

35

N/A

None

17

64/147 (44%)

72

62

16

N/A

None

33/89 (37%)

80

42

44

N/A

None

130/237 (55%)

43

60

0

N/A

Perforation: 1 (0.4%)

60

77

20

20%

None

Study 90

Ell et al,

2005 88

Di Caro et al,

2005

Mehdizadeh et al, 2006

66

Hadithi et al,67 2006

35/35 (100%)

Diagnostic or treatment Total Rebleed success (%) DBE(%)* rate

Complications

24

2006

168/275 (61%)

73

55

42

N/A

Pancreatitis: 3 (1%)

89

2006

32/40 (80%)

48

75

0

N/A

Perforation: 1 (2.5%)

Monkemuller et al, 2006

29/70 (41%)

67

57

30

0%

Polypectomy bleed: 1 (1.4%)

Manabe et al,107 2006

31/31 (100%)

74

74

29

0%

None

Nakamura et al,108 2006

28/28 (100%)

41

43

63

6%

Perforation: 1 (3.6%)

43

43

40

N/A

None

63

60

34

Heine et al,

Kaffes et al,

18

Akahoshi et al,109 2006 Totalsy

20/103 (19%) 726 Patients; 1481 examinations

Perforation: 4 (0.3%); pancreatitis: 3 (0.2%)

N/A, Not available. *Defined as initial DBE in one direction, with a tattoo at the most distal insertion point, followed by a DBE from the opposite direction with the prior tattoo site identified. yCalculated averages for diagnostic yield, treatment success, and total DBE rates.

were not actively bleeding.61 The complication rate from an angiography, including access-site thrombosis or hemorrhage, contrast reactions, and injury to the target vessels, specifically, the superior mesenteric artery, inferior mesenteric artery, and celiac artery, including dissection and distal embolization, approximates 5%, although some centers reported rates as high as 11%.62,63 DBE, first described by Yamamoto et al15 in 2001, allows complete visualization of the small intestine by using a 200-mL enteroscope with an outer diameter of 8.5 mm and equipped with a 140-cm overtube with an outer diameter of 12 mm (Fujinon). Latex balloons at the tip of the enteroscope and the overtube are inflated and deflated with air from a pressure-controlled pump system. By inflating the overtube balloon enough to grip the intestinal wall (which can occur at a balloon pressure of 45 mm Hg), the endoscope can be inserted further without forming redundant loops in the small intestine. The overtube can then be inserted while the endoscope balloon is inflated. This method allows for insertion of the endoscope deep into the small intestine.

Except for rare instances, a total enteroscopy with cecal intubation by the antegrade approach alone is not feasible. In patients without prior imaging modalities to suggest the location of lesions, it is reasonable to investigate the patient who has obscure bleeding initially by using the anterograde approach, because the majority of the lesions would be expected to be located in the proximal to mid small bowel.24 We modeled that an oral DBE would occur if the CE suggested a lesion located within the initial 75% of the small-bowel transit time.64 The majority of patients do not require retrograde DBE after a successful oral procedure.66 A total enteroscopy, defined as an initial DBE procedure, with tattooing of the most distal extent of the examination, followed by a DBE from the opposite approach that identifies the prior tattoo site, can be attempted on different days to minimize the patient discomfort and bowel distension. Published studies that examined outcomes associated with a DBE that were used for the model are shown in Table 2. Summary statistics based on these 726 patients who underwent a total of 1481 DBE examinations demonstrated an average diagnostic yield of 63% and diagnostic or treatment success of 60%. Obscure bleeding was the

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 927

DBE

Management strategies for obscure GI bleeding

Gerson & Kamal

indication for the DBE in approximately 36% to 100% of examinations, and the overall diagnostic yield from DBE ranged from 43% to 80%. Diagnostic or therapeutic success was reached in 55% to 75% of examinations, which is comparable with other diagnostic modalities for the small bowel. Patients with obscure bleeding as an indication for a DBE had higher pathology detection rates. No deaths have been reported to date after a DBE. Intestinal perforation occurred in 4 of 1481 procedures (0.3%). Mild pancreatitis, possibly a sequela of duodenal hypertension from double-balloon inflation, was reported in 3 of the patients (0.2%). A recent multicenter survey study, including 2367 DBE procedures from 10 centers, demonstrated that the perforation rate was 0.3%.65 For patients enrolled in the DBE arm, we modeled that the patients would undergo an empiric anterograde DBE as the initial diagnostic test. We derived the probability of finding lesions based on comparative studies in which patients had undergone prior CE examinations. Three studies to date described results in patients who were undergoing a DBE after a CE.24,66,67 In the study by Heine et al,24 in which 275 patients underwent a DBE, smallbowel AVMs were the indication for DBE in 60 patients (22%) based on a prior CE. The diagnostic yield for a DBE was 48% when an oral DBE was performed alone and 70% when both oral and rectal approaches were performed. In the U.S. multicenter trial,66 CE suggested AVMs in 42 patients (36%) of which 24 (57%) were confirmed by a DBE. In the study by Hadithi et al,67 AVMs were detected during 19 of 35 CE examinations (54%) (primarily located in the jejunum), and DBE diagnosed AVMs in 16 of these 19 patients (84%). By using the weighted probabilities from these studies, we calculated that the probability of finding an AVM on a DBE was 33%. The probability of AVM detection on a DBE was lower than with a CE because a CE allows for complete small-bowel imaging in addition to the possibility of false-positive lesions that may be identified on a CE and classified as nonspecific red spots or other lesions that may not represent true AVMs. For tumors and ulcerations, prior studies showed excellent agreement between CE and DBE that ranged from 96% to 100%.24,66 Although most lesions will be located within the proximal two thirds of the small bowel, it is also possible that patients might be bleeding from isolated or multiple lesions in the distal bowel. To calculate this probability, we used data from the study by Heine et al,24 in which the probability of finding a lesion on rectal DBE if an oral examination was normal was approximately 20%. Information regarding outcomes after endoscopic therapy for AVMs have not been widely published. In a clinical trial by Hadithi et al67 that included 35 patients with obscure GI bleeding, AVMs were present and treated in 16 of the patients (45%). Over a mean follow-up of 5 months (range 2-12 months), 80% of the patients did not report any subsequent GI bleeding. We modeled that the

rebleeding rate after therapy for AVMs would approach 20%. After completion of both anterograde and retrograde DBE, the probability of recurrent bleeding decreased to 5%.67 We modeled the probability of complications and death associated with endoscopic complications for DBE by using a rate of 0.3% for a perforation based on the available medical literature (Table 2) and a large study from Mensink et al.65

928 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Capsule-directed DBE CE, developed by Given Imaging (Given Imaging Ltd, Yoqneam, Israel), involves swallowing a 11  26 mm capsule that subsequently obtains images throughout the GI tract and transmits the data during an 8-hour recording period via radio frequency to a recording device worn about the patient’s waist. Multiple studies demonstrated the superiority of CE compared with push enteroscopy for patients with obscure GI bleeding. The yield of CE ranged from 55% to 75%,11,12,35,36,38,41-43,56,61 with a 25% to 50% higher yield compared with a push enteroscopy. A recent meta-analysis that examined 14 trials comparing CE to push enteroscopy for obscure bleeding demonstrated a 63% yield for CE compared to 28% for push enteroscopy.68 In a recent pooled analysis of raw data from manufacturer-sponsored trials, CE was shown to be superior to push enteroscopy, small-bowel series, and colonoscopy with ileal intubation.69 In this analysis, CE identified pathology in approximately 70% of the 530 pooled examinations. We assumed that patients would undergo a small-bowel CE after a bowel preparation. After the initial CE examination, we modeled a spontaneous cessation rate of hemorrhage for AVMs, ulcerations, tumors, and other vascular lesions according to the published literature.6,7,25,46 If the patients continued to experience bleeding, then a DBE would be performed, with an oral approach initiated if the CE findings occurred within the initial 75% of the examination time, and rectal approach for lesions within the distal 25% of the transit time.64 Patients with a normal CE and ongoing bleeding would undergo empiric DBE via the oral approach. Although up to 1% of patients may not be able to swallow the capsule70 and will require endoscopic placement, we assumed, in the decision analysis, that all patients could ingest the capsule. We did assume that capsule retention would occur in 1% of patients and that a DBE would be successful in retrieval of the capsule in 50% of cases.71,72 Patients who undergo surgical removal of the CE after retention could undergo subsequent DBE if they continued to bleed.

Outcomes The primary focus of our analysis was cost per qualityadjusted life years (QALY). The guidelines on cost-effectiveness analyses suggest that QALYs are the most appropriate health-economic unit for decision analytic

Gerson & Kamal

models.73 Patients with acute GI bleeding were previously shown to have associated diminished health-related QOL when using generalized health surveys, such as the Short Form Medical Outcomes Survey 36.74 Health-state utilities are preference-weighted measures required for cost-effectiveness analyses. Utility measurements are designed to provide a numerical indicator of symptom severity based on a scale of 0 to 1.0, in which 0 represents death or a state equivalent to death, and 1.0 represents ideal health, or a state without a particular disease. A paucity of utility data exists for the state of GI hemorrhage. In a decision analysis that examined anticoagulant therapy to prevent a stroke in patients with atrial fibrillation, the mean utility of GI hemorrhage was 0.84 when derived hypothetically from elderly patients with atrial fibrillation who were asked about a theoretical situation of GI hemorrhage.75 In a 1995 decision analysis that examined the cost-effectiveness of warfarin and aspirin for prophylaxis of stroke in patients with nonvalvular atrial fibrillation,76 74 elderly veterans with atrial fibrillation were interviewed about a hypothetical scenario of GI hemorrhage; the mean time trade-off value was 0.76. Therefore, hypothetical estimates for the utility state associated with the state of acute GI bleeding ranged between 0.76 and 0.84. We assumed that patients without bleeding would have a perfect utility score of 1.0, whereas patients with ongoing bleeding would have a QALY score of 0.80.75,76 We modeled that the QALY associated with hospitalization for perforation would approximate 0.7577,78 and that patients would return to their usual state of health within 1 month after bleeding.79 In patients with obscure bleeding, other important outcomes include costs associated with diagnostic testing and management. Prior studies showed that most patients with obscure GI bleeding undergo a large number of diagnostic tests and hospitalizations before a diagnosis is established and treatment is initiated. In one study, of 39 patients with obscure bleeding, there were a total of 277 diagnostic tests performed before study entry, with an average of 7.3 tests per patient, and 49% of patients continued to have an unknown bleeding source after push enteroscopy.28 In another study, of 14 patients with obscure bleeding who were undergoing intraoperative enteroscopy, there was an average of 5 hospital admissions and 46 units of blood transfused per patient before the examination.80

Data sources

Management strategies for obscure GI bleeding

in clinical trials, we used weighted averages of the available probabilities to estimate the base-case probability for the model. To calculate yearly probabilities, the formula P Z 1  e (rt) was used, where P Z probability, r Z rate, and t Z time.

Estimation of model costs We estimated the costs associated with push enteroscopy, intraoperative enteroscopy, angiography, and CE from the perspective of a third-party payer (Table 3). We only considered direct health care costs and did not account for costs because of non–health care expenditures, such as transportation and/or time lost from work. We used facility and professional fees based on Medicareallowable payments in 2005 to derive costs for endoscopic procedures, GI hemorrhage, and perforation after an endoscopic procedure. We derived an average cost based on Medicare payments for rural and urban areas for 4 major states in the United States (California, Illinois, New York, and Georgia). The costs of inpatient hospital services were obtained by using the 2005 Medicare Prospective Payment System diagnosis related group. We did not include costs of cardiopulmonary complications from conscious sedation, because these complications are rare and are treated by prompt drug reversal. Discounting of costs was not performed because of the short time horizon of the analysis. Because a DBE is a relatively new procedure, having been initiated into the United States in approximately September 2004, current procedural terminology code (CPT) codes are not yet available. The average procedural time per procedure is typically 75 to 80 minutes.81 Currently, most physicians who perform DBE use the CPT code of 44376 for diagnostic anterograde procedures (small-intestinal endoscopy, enteroscopy beyond second portion of duodenum, including ileum; diagnostic, with or without collection of a specimen or specimens by brushing or washing). We estimated that the cost for the DBE would be twice the cost of the 44376 CPT code based on the procedural time. For DBE procedures performed via the rectal approach, most physicians currently use an unlisted CPT code of 44799. However, for the purposes of this decision analysis, we estimated the cost for a rectal DBE examination by combining the cost for a diagnostic colonoscopy (CPT code 44378) with the cost for a small-bowel examination of the ileum and distal jejunum (CPT code 44376). We widely varied the costs for the DBE examinations in the sensitivity analyses.

We performed a thorough search of published articles between 1960 and 2006 from the MEDLINE database to identify relevant English-language publications that pertain to AVMs, push enteroscopy, intraoperative enteroscopy, angiography, CE, and DBE. Studies published only as abstracts were excluded. For assessment of procedural outcomes, case series were not included. Because there was a large variation in the number of patients enrolled

The main outcomes evaluated included overall costs per model arm, QALYs, the number of patients with cessation of bleeding, the frequency of endoscopic complications, and the overall mortality. We estimated the probability of death over the course of the year based on age-specific death rates from 2003 U.S. life tables82 in

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 929

Cost-effectiveness and sensitivity analyses

Management strategies for obscure GI bleeding

Gerson & Kamal

TABLE 3. Costs based on 2007 CPT codes and 2005 Medicare rates* CPT or DRG code

Cost item

Cost, $ ($ range in sensitivity analysis)

91110

CEy

1120 (500-1500)

44360

Push enteroscopy, diagnostic

160 (100-1000)

44366

With hemostasis

282 (100-1200)

DBEz Anterograde approach 44376

Diagnostic

577 (200-2000)

44378

With hemostasis

780 (400-2000)

Rectal approach 45378 þ 44376

Diagnostic

1264 (400-2500)

45378 þ 44378

With hemostasis

1568 (400-3000)

Intraoperative enteroscopy 49000

14,799 (7000-30,000)

Exploratory laparotomy

805

44378

Small intestinal endoscopy with hemostasis

412

00840

Anesthesia

252

174,175

Hospitalizationx

37204

Transcatheter embolization, angiography

36430

Monthly blood transfusionsk Small-bowel resection

13310 1097 (500-3000) 122 (60-500) 14710 (10,000-30,000)

44120

Enterectomy

1148

00840

Anesthesia

252

174,175

Hospitalizationx Endoscopic complication GI hemorrhage

99283

Emergency department visit

13310 27,562 (13,000-52,000) 13,652 70

44378

DBE with hemostasis

174, 175

HospitalizationV

13310

Perforation

41,191

99283

Emergency department visit

44605

Repair of perforation

00840

Anesthesia

148,149

Hospitalizationx Omeprazole OTC 20 mg daily for 8 weeks{

412

70 1212 252 39657 48 (20-200)

*Sources: American Medical Association. Current Procedural Terminology Code Book for CPT codes. 2005. Available at: http://www.ama-assn.org/ama/pub (accessed February 26, 2008). Agency for Healthcare Research and Quality(AHRQ) Web site for 2003 Diagnosis Related Group charges. Available at: http:// hcup.ahrq.gov/HCUPnet (accessed February 26, 2008). The costs shown are the sum of included costs, based on an average of 4 different carrier localities for each CPT code to derive a national average. yCost of the capsule ($450) plus physician interpretation. zCPT code for a DBE is not yet established. For an anterograde DBE, the cost estimated was with the CPT code of 44376, previously used for Sonde enteroscopy including the ileum. The current procedural times average is 75-80 minutes for DBE examinations. The DBE cost was obtained by doubling the reimbursement for the 44376 code. For a retrograde approach, the currently recommended code is the unlisted code of 44799. However, for the model, we estimated costs of diagnostic colonoscopy (CPT code 45378 or $453) plus small-bowel examination (44376 or 44378 if hemostasis was performed). xHospitalization charges were based on an average of charges for hospitalizations with and without complications. The cost for an endoscopic complication was calculated by assuming that 50% would experience hemorrhage and 50% perforation. kThe cost of blood transfusions was $61 per unit. When assuming that the bleeding patient receives 2 units per month, the cost of transfusional therapy was $1464 per year. {Omeprazole over the counter based on retail price of $11 for 14 capsules. Available at: http://www.drugstore.com (accessed February 26, 2008).

930 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Gerson & Kamal

Management strategies for obscure GI bleeding

TABLE 4. Base-case results

Model arm No therapy

Average cost ($)*

Effectivenessy

Incremental cost-effectivenessz

532

0.870

Base case

Push enteroscopy

1025

0.888

-

DBE

2407

0.956

$20,833 per QALY

Angiography

3215

0.899

Capsule-directed DBE

4309

0.942

-

21263

0.939

-

Intraoperative enteroscopy

*Average cost per patient. yQALYs gained per year. zIncremental cost per additional QALY gained compared with no therapy.

addition to the probability of death resulting from endoscopic procedural complications. We calculated the incremental cost-effectiveness ratio (ICER) between the no treatment arm and the most effective strategy. For this model, the ICER represented the additional cost accrued to gain the greatest fraction of QALYs when adopting the more expensive yet more effective strategy. If one of the arms was less costly and more effective than the other arm, then that strategy was said to be dominant. We performed 1-way, 2-way, and multivariable sensitivity analyses to evaluate the effects from varying costs and probabilities over ranges derived from the review of the medical literature.

RESULTS

TABLE 5. Base-case results per 1000 patients

Model arm

No. patients No. patients with bleeding No. still bleeding cessation deaths

No therapy

388

590

22

Push enteroscopy

324

656

19

Angiography

297

684

18

Capsule-directed DBE

235

760

5

Intraoperative enteroscopy

158

821

21

DBE

125

865

10

By using the base-case probabilities shown in Table 1, the no-therapy arm was the least expensive and was associated with the lowest fraction of QALYs. The initial DBE arm was the most effective but more expensive than the no-therapy arm, whereas all of the other strategies (except for a push enteroscopy, which was less costly) were less effective than the DBE arm and more expensive. The notherapy arm cost $532 and was associated with 0.870 QALYs, whereas the DBE arm cost $2407 and was associated with 0.956 QALYs, which resulted in an incremental costeffectiveness ratio of $20,833 per QALY gained (Table 4). Compared with a DBE, an initial CE was more costly and less effective. Based on these results, approximately 86.5% of patients would experience cessation of hemorrhage over the course of a year in the DBE arm compared with 76% in the CE arm and 59% in the no-therapy arm (Table 5). A tornado analysis revealed that the variables with the most impact on the cost-effectiveness ratio (in descending order) were the QOL associated with the well state, the

QOL associated with bleeding, the probability of an AVM, the mortality rate associated with hemorrhage, the probability of missed lesions on upper and lower endoscopies, the DBE complication rate, and the DBE cost. We performed 1-way sensitivity analyses on all parameters of the model. When the QALY of the well state was !0.53, the no-therapy arm was the preferred approach. When this value ranged between 0.6 and 0.8, the initial CE approach was preferred. The effectiveness of the CE arms and the DBE arms were equivalent when (a) the QALY state associated with bleeding O0.875, (b) the miss rate for lesions on upper and/or lower endoscopies exceeded 23%, and (c) when the DBE complication rate was O30%. However, despite equal effectiveness, the DBE arm was less expensive and, therefore, was preferred by extended dominance. When the cost of yearly transfusions exceed $10,000, the DBE arm cost less than the notherapy arm but remained more effective. When the cost of DBE exceeded $5925, the CE arm cost less than the

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 931

Management strategies for obscure GI bleeding

Gerson & Kamal

would require a DBE after the CE examination because of ongoing bleeding, and, subsequently, 47 patients would experience ongoing bleeding and require a DBE via the opposite approach. Therefore, an initial CE was associated with a decreased DBE workload and potential for endoscopic complications, since only patients with ongoing bleeding would undergo subsequent DBE examination.

DISCUSSION

DBE arm but remained less effective. At a probability of AVM !5%, the no-therapy arm was preferred. We performed a Monte Carlo simulation to simultaneously vary all of the key variables identified during the tornado analysis. We assumed a triangular distribution for each parameter entered into the simulation. The mean ICER of these 1000 trials (Fig. 3) was $28,600 per QALY (2.5 and 97.5 percentiles, $27,871 and $41,680, respectively). The percentage of trials that resulted in less than the $50,000 and $100,000 willingness-to-pay thresholds were 86% and 99%. For example, if a third-party payer were willing to pay $50,000 per patient with bleeding cessation for the use of DBE, then 86% of the patients in this simulation would be within the budget. A major limitation of our model was the short time horizon of 1 year. We extended the time frame of our analysis to determine the impact on the current model, with patients remaining in their current states after receiving their initial course of treatment. After 5 years, the results of the analysis did not change. However, after 7 years, the CE arm and DBE arm were equally effective, which resulted in 5.3 QALYs per arm. This exercise was limited, however, by a lack of knowledge about the long-term probabilities of bleeding cessation and the potential recurrence of small-bowel lesions. Because a DBE is a time-consuming and difficult procedure, we also examined the number of DBE procedures required in the initial DBE strategy compared with the initial CE strategy. By assuming a cohort of 1000 patients, all would require an initial oral DBE and then approximately 130 would require subsequent rectal DBE procedures because of ongoing bleeding. In the initial CE arm, all 1000 patients would undergo a CE; approximately 237 patients

Approximately 5% of patients with GI bleeding will have a source localized to the small intestine. Before the advent of CE, the primary tools available for diagnosis and treatment of small-bowel lesions included push enteroscopy, which could primarily detect lesions just distal to the ligament of Treitz, or intraoperative enteroscopy, which had a high success rate for treatment of small-bowel lesions but also carried a high morbidity and mortality rate. With the advent of CE in 2000, the entire small bowel was able to be visualized by using a noninvasive methodology that was associated with a 25% to 50% higher yield compared with small-bowel radiography or push enteroscopy. The major disadvantages associated with CE include a lack of therapeutic options, the potential for missed lesions because of tumbling of the capsule, and a field of view of approximately 140 degrees. The advent of DBE in 2000 and then the introduction of DBE into the United States in 2004 allowed for therapeutic potential in the small bowel without the need for an intraoperative enteroscopy, except in circumstances in which bowel adhesions or other technical factors prevent advancement of the enteroscope. A DBE can be performed in the endoscopy suite with the patient under conscious sedation but generally requires additional staff to help with overtube positioning and, on average, 60 to 80 minutes per procedure. By using a systematic approach to estimate the depth of insertion by advancement cycles, at least 240 cm of small-bowel mucosa can be reached from the oral approach and 140 cm from the ileocecal valve.81 The retrograde approach is more challenging because of the need for adequate colonic preparation, the potential for looping in the colon, and the difficulty with intubation of the ileocecal valve, particularly in a patient with prior adhesions. Even with considerable experience, the failure rates for rectal procedures remain at 20% because of the presence of adhesions from prior pelvic and abdominal surgeries and poor preparation that prevented intubation of the ileocecal valve.83 However, the majority of smallbowel lesions would be expected to be located within the proximal small bowel and within reach of an oral DBE. Although a CE is not associated with any potential for therapeutic maneuvers and can miss lesions, the disadvantages associated with DBE include the long duration of the procedure, potential for complications, the need for additional staffing, and difficulty in missing lesions in the

932 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Figure 3. Probabilistic sensitivity analysis by using 1000 trials. This analysis simultaneously varied all parameters over the full range of plausible values. Each point represents the ICER generated by 1 trial through the simulation. The bold line represents the median ICER of $20,833 per QALY. By definition, 50% of the trials fall on either side of the line. The remaining 2 diagonal lines represent willingness-to-pay thresholds. If a third-party payer was willing to pay $50,000 per patient with cessation of bleeding, then 86% of the patients would fall within the budget.

Gerson & Kamal

Management strategies for obscure GI bleeding

mid small bowel where access is often challenging. Nonetheless, our decision analysis suggests that, in patients with obscure bleeding, DBE would be a cost-effective approach because of the ability to identify and treat lesions during the initial small-bowel examination. Currently, the cost for a DBE is not known. It is anticipated that CPT codes will be established for a DBE within the near future. Many hospital billing departments are recommending that providers use the CPT code 44799 (unlisted procedure small intestine). The long associated procedural time may discourage gastroenterologists from performing this procedure. We found that a DBE remained a cost-effective approach, regardless of the cost for the examination, because patients did not incur the additional costs of the capsule examination, and therapeutic interventions could be performed. However, the limitation of these conclusions is the current lack of availability of a DBE in most centers. An initial DBE would significantly increase the endoscopy workload (including procedural time) for gastroenterologists and would lead to a higher rate of endoscopic complications compared with initial usage of a CE. Benefits of performing an initial CE would include a reduction in the number of subsequent DBE procedures, and localization of small-bowel lesions to the proximal or distal small bowel. Therefore, in hospitals where a DBE is not available, an initial CE would remain a cost-effective approach compared with the other diagnostic modalities included in our model. There are several limitations to our analysis. We derived the probabilities for the model from prior clinical trials, most that included small numbers of patients. Outcomes data about cessation of bleeding after endoscopic therapy, particularly a DBE, are not yet available but are needed to perform an analysis over a longer time period. In addition, the prevalence of small-bowel lesions is likely to be different, depending upon the age of the patient. In patients below the age of 40 years and with obscure GI bleeding, more common etiologies for obscure bleeding would include tumors, Meckel’s diverticulum, Crohn’s disease, or a Dieulafoy’s lesion. In this population, a CE may be more effective in the identification of pathology located in the mid small bowel, although a DBE could offer both diagnostic and therapeutic potential. In patients over the age of 40 years, in whom small-bowel AVMs are more likely to be causal, an initial DBE appears to be cost effective in that diagnosis and management can occur during the same examination. Given the published rates in the literature, we were able to calculate the sensitivity of each test for each type of small-bowel lesion but not the specificity. We, therefore, assumed 100% specificity of each modality. In the preliminary multicenter U.S. DBE study, miss rates for a CE and a DBE for small-bowel lesions were between 20% and 30%.65 It would be expected that higher false-positive rates for lesions on a CE would lead to increased utilization and costs attributable to a DBE and that higher

false-positive rates on any test would lead to higher complication rates. Future studies that derive sensitivity and specificity of each modality for each type of small-bowel lesion are warranted to refine future cost-effectiveness analyses in this area. To derive these probabilities, subjects would be required to undergo comparative diagnostic testing. Additional limitations of our model include the short time horizon of 1 year. As DBE becomes more widely practiced, patients who have undergone the procedure with endoscopic therapy for AVMs could be followed over time to determine the long-term duration of endoscopic therapy. We were able to demonstrate equal effectiveness for the capsule and DBE arms at 7 years. As shown in Table 5, the DBE arm had a 0.5% absolute increase in deaths compared with CE because of procedural complications. If the base-case 50-year-old male patient was expected to live approximately 20 more years, then that difference alone would amount to a 0.1 absolute year advantage over a lifetime for CE, an amount that would exceed the initial 0.014 benefit in QALYs produced by DBE during the 1-year time horizon of the current model. Although we focused our analysis on the impact of bleeding on QOL, a paucity of information exists about the impact of chronic GI bleeding on QOL. To date, health-state utility values exist only for patients with acute GI bleeding, and these values were derived from hypothetical situations.73,74 Because obscure GI bleeding is usually a chronic process, it would be expected that the health-related QOL might be lower in this population, and future studies are warranted in this area. We did not model indirect costs to the patient, such as time lost from work. Although most patients lose a day of work for any endoscopic procedure, a CE does not require sedation and, therefore, could be performed during a working day if needed. This would be unlikely to impact the current decision analysis. However, more information regarding work productivity and chronic bleeding is needed. In summary, we demonstrated that an initial DBE is a costeffective approach for patients with obscure bleeding. However, a capsule-directed DBE may be associated with better long-term outcomes because of the potential for fewer complications and decreased utilization of endoscopic resources. Future research is warranted to understand the natural history of bleeding from these lesions and the impact of obscure bleeding on health-related QOL.

www.giejournal.org

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 933

DISCLOSURE The following author reports no disclosures relevant to this publication: A. Kamal. The following author has disclosed actual or potential conflicts: L. Gerson has received research support, equipment, and honorarium from Fujinon, and has received honorarium from Given Imaging Inc.

Management strategies for obscure GI bleeding

Gerson & Kamal

1. Szold A, Katz LB, Lewis BS. Surgical approach to occult gastrointestinal bleeding. Am J Surg 1992;163:90-2, discussion 92–3. 2. Pennazio M, Arrigoni A, Risio M, et al. Clinical evaluation of push-type enteroscopy. Endoscopy 1995;27:164-70. 3. Chong J, Tagle M, Barkin JS, et al. Small bowel push-type fiberoptic enteroscopy for patients with occult gastrointestinal bleeding or suspected small bowel pathology. Am J Gastroenterol 1994;89:2143-6. 4. Davies GR, Benson MJ, Gertner DJ, et al. Diagnostic and therapeutic push type enteroscopy in clinical use. Gut 1995;37:346-52. 5. Thompson JN, Hemingway AP, McPherson GA, et al. Obscure gastrointestinal haemorrhage of small-bowel origin. Br Med J (Clin Res Ed) 1984;288:1663-5. 6. Junquera F, Feu F, Papo M, et al. A multicenter, randomized, clinical trial of hormonal therapy in the prevention of rebleeding from gastrointestinal angiodysplasia. Gastroenterology 2001;121:1073-9. 7. Lewis BS, Salomon P, Rivera-MacMurray S, et al. Does hormonal therapy have any benefit for bleeding angiodysplasia? J Clin Gastroenterol 1992;15:99-103. 8. Gostout CJ. Sonde enteroscopy. Technique, depth of insertion, and yield of lesions. Gastrointest Endosc Clin N Am 1996;6:777-92. 9. Schmit A, Gay F, Adler M, et al. Diagnostic efficacy of push-enteroscopy and long-term follow-up of patients with small bowel angiodysplasias. Dig Dis Sci 1996;41:2348-52. 10. Descamps C, Schmit A, Van Gossum A. ‘‘Missed’’ upper gastrointestinal tract lesions may explain ‘‘occult’’ bleeding. Endoscopy 1999;31:452-5. 11. Ell C, Remke S, May A, et al. The first prospective controlled trial comparing wireless capsule endoscopy with push enteroscopy in chronic gastrointestinal bleeding. Endoscopy 2002;34:685-9. 12. Mata A, Bordas JM, Feu F, et al. Wireless capsule endoscopy in patients with obscure gastrointestinal bleeding: a comparative study with push enteroscopy. Aliment Pharmacol Ther 2004;20:189-94. 13. Desa LA, Ohri SK, Hutton KA, et al. Role of intraoperative enteroscopy in obscure gastrointestinal bleeding of small bowel origin. Br J Surg 1991;78:192-5. 14. Ress AM, Benacci JC, Sarr MG. Efficacy of intraoperative enteroscopy in diagnosis and prevention of recurrent, occult gastrointestinal bleeding. Am J Surg 1992;163:94-8, discussion 98–9. 15. Yamamoto H, Sekine Y, Sato Y, et al. Total enteroscopy with a nonsurgical steerable double-balloon method. Gastrointest Endosc 2001;53: 216-20. 16. May A, Wardak A, Nachbar L, et al. Influence of patient selection on the outcome of capsule endoscopy in patients with chronic gastrointestinal bleeding. J Clin Gastroenterol 2005;39:684-8. 17. Ell C, May A, Nachbar L, et al. Push-and-pull enteroscopy in the small bowel using the double-balloon technique: results of a prospective European multicenter study. Endoscopy 2005;37:613-6. 18. Monkemuller K, Weigt J, Treiber G, et al. Diagnostic and therapeutic impact of double-balloon enteroscopy. Endoscopy 2006;38:67-72. 19. Pennazio M, Eisen G, Goldfarb N. ICCE consensus for obscure gastrointestinal bleeding. Endoscopy 2005;37:1046-50. 20. Appleyard M, Fireman Z, Glukhovsky A, et al. A randomized trial comparing wireless capsule endoscopy with push enteroscopy for the detection of small-bowel lesions. Gastroenterology 2000;119:1431-8. 21. Russell LB, Gold MR, Siegel JE, et al. The role of cost-effectiveness analysis in health and medicine. Panel on Cost-Effectiveness in Health and Medicine. JAMA 1996;276:1172-7. 22. Askin MP, Lewis BS. Push enteroscopic cauterization: long-term follow-up of 83 patients with bleeding small intestinal angiodysplasia. Gastrointest Endosc 1996;43:580-3. 23. Hodgson H. Hormonal therapy for gastrointestinal angiodysplasia. Lancet 2002;359:1630-1. 24. Heine GD, Hadithi M, Groenen MJ, et al. Double-balloon enteroscopy: indications, diagnostic yield, and complications in a series of 275 patients with suspected small-bowel disease. Endoscopy 2006;38:42-8.

25. Lai LH, Wong GL, Chow DK, et al. Long-term follow-up of patients with obscure gastrointestinal bleeding after negative capsule endoscopy. Am J Gastroenterol 2006;101:1224-8. 26. Benz C, Jakobs R, Riemann JF. Do we need the overtube for push-enteroscopy? Endoscopy 2001;33:658-61. 27. Taylor AC, Chen RY, Desmond PV. Use of an overtube for enteroscopy: does it increase depth of insertion? A prospective study of enteroscopy with and without an overtube. Endoscopy 2001;33:227-30. 28. Foutch PG, Sawyer R, Sanowski RA. Push-enteroscopy for diagnosis of patients with gastrointestinal bleeding of obscure origin. Gastrointest Endosc 1990;36:337-41. 29. Adrain AL, Dabezies MA, Krevsky B. Enteroscopy improves the clinical outcome in patients with obscure gastrointestinal bleeding. J Laparoendosc Adv Surg Tech A 1998;8:279-84. 30. Zaman A, Sheppard B, Katon RM. Total peroral intraoperative enteroscopy for obscure GI bleeding using a dedicated push enteroscope: diagnostic yield and patient outcome. Gastrointest Endosc 1999;50: 506-10. 31. Shackel NA, Bowen DG, Selby WS. Video push enteroscopy in the investigation of small bowel disease: defining clinical indications and outcomes. Aust N Z J Med 1998;28:198-203. 32. Hayat M, Axon AT, O’Mahony S. Diagnostic yield and effect on clinical outcomes of push enteroscopy in suspected small-bowel bleeding. Endoscopy 2000;32:369-72. 33. Sharma BC, Bhasin DK, Makharia G, et al. Diagnostic value of pushtype enteroscopy: a report from India. Am J Gastroenterol 2000;95: 137-40. 34. Landi B, Cellier C, Gaudric M, et al. Long-term outcome of patients with gastrointestinal bleeding of obscure origin explored by push enteroscopy. Endoscopy 2002;34:355-9. 35. Lewis BS, Swain P. Capsule endoscopy in the evaluation of patients with suspected small intestinal bleeding: results of a pilot study. Gastrointest Endosc 2002;56:349-53. 36. Mylonaki M, Fritscher-Ravens A, Swain P. Wireless capsule endoscopy: a comparison with push enteroscopy in patients with gastroscopy and colonoscopy negative gastrointestinal bleeding. Gut 2003;52:1122-6. 37. Saurin JC, Delvaux M, Gaudin JL, et al. Diagnostic value of endoscopic capsule in patients with obscure digestive bleeding: blinded comparison with video push-enteroscopy. Endoscopy 2003;35:576-84. 38. Saurin JC, Delvaux M, Vahedi K, et al. Clinical impact of capsule endoscopy compared to push enteroscopy: 1-year follow-up study. Endoscopy 2005;37:318-23. 39. Keizman D, Brill S, Umansky M, et al. Diagnostic yield of routine push enteroscopy with a graded-stiffness enteroscope without overtube. Gastrointest Endosc 2003;57:877-81. 40. Romelaer C, Le Rhun M, Beaugerie L, et al. Push enteroscopy for gastrointestinal bleeding: diagnostic yield and long-term follow-up. Gastroenterol Clin Biol 2004;28:1061-6. 41. Adler DG, Knipschield M, Gostout C. A prospective comparison of capsule endoscopy and push enteroscopy in patients with GI bleeding of obscure origin. Gastrointest Endosc 2004;59:492-8. 42. Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology 2004;126:643-53. 43. Hartmann D, Schilling D, Bolz G, et al. Capsule endoscopy versus push enteroscopy in patients with occult gastrointestinal bleeding. Z Gastroenterol 2003;41:377-82. 44. Zaman A, Katon RM. Push enteroscopy for obscure gastrointestinal bleeding yields a high incidence of proximal lesions within reach of a standard endoscope. Gastrointest Endosc 1998;47:372-6. 45. Landi B, Tkoub M, Gaudric M, et al. Diagnostic yield of push-type enteroscopy in relation to indication. Gut 1998;42:421-5. 46. Hsu CM, Chiu CT, Su MY, et al. The outcome assessment of doubleballoon enteroscopy for diagnosing and managing patients with obscure gastrointestinal bleeding. Dig Dis Sci 2007;52:162-6.

934 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

REFERENCES

Gerson & Kamal 47. Clarke GA, Jacobson BC, Hammett RJ, et al. The indications, utilization and safety of gastrointestinal endoscopy in an extremely elderly patient cohort. Endoscopy 2001;33:580-4. 48. Sieg A, Hachmoeller-Eisenbach U, Eisenbach T. Prospective evaluation of complications in outpatient GI endoscopy: a survey among German gastroenterologists. Gastrointest Endosc 2001;53:620-7. 49. Schauer PR, Schwesinger WH, Page CP, et al. Complications of surgical endoscopy. A decade of experience from a surgical residency training program. Surg Endosc 1997;11:8-11. 50. Strodel WE, Eckhauser FE, Knol JA, et al. Intraoperative fiberoptic endoscopy. Am Surg 1984;50:340-4. 51. Greenberg GR, Phillips MJ, Tovee EB, et al. Fiberoptic endoscopy during laparotomy in the diagnosis of small intestinal bleeding. Gastroenterology 1976;71:133-5. 52. Bowden TA Jr, Hooks VH 3rd, Teeslink CR, et al. Occult gastrointestinal bleeding: locating the cause. Am Surg 1980;46:80-7. 53. Lopez MJ, Cooley JS, Petros JG, et al. Complete intraoperative smallbowel endoscopy in the evaluation of occult gastrointestinal bleeding using the Sonde enteroscope. Arch Surg 1996;131:272-7. 54. Douard R, Wind P, Panis Y, et al. Intraoperative enteroscopy for diagnosis and management of unexplained gastrointestinal bleeding. Am J Surg 2000;180:181-4. 55. Kendrick ML, Buttar NS, Anderson MA, et al. Contribution of intraoperative enteroscopy in the management of obscure gastrointestinal bleeding. J Gastrointest Surg 2001;5:162-7. 56. Hartmann D, Schmidt H, Bolz G, et al. A prospective two-center study comparing wireless capsule endoscopy with intraoperative enteroscopy in patients with obscure GI bleeding. Gastrointest Endosc 2005;61:826-32. 57. Jakobs R, Hartmann D, Benz C, et al. Diagnosis of obscure gastrointestinal bleeding by intra-operative enteroscopy in 81 consecutive patients. World J Gastroenterol 2006;12:313-6. 58. d’Othee BJ, Surapaneni P, Rabkin D, et al. Microcoil embolization for acute lower gastrointestinal bleeding. Cardiovasc Intervent Radiol 2006;29:49-58. 59. Richardson JD, Max MH, Flint LM Jr, et al. Bleeding vascular malformations of the intestine. Surgery 1978;84:430-6. 60. Rollins ES, Picus D, Hicks ME, et al. Angiography is useful in detecting the source of chronic gastrointestinal bleeding of obscure origin. AJR Am J Roentgenol 1991;156:385-8. 61. Saperas E, Dot J, Videla S, et al. Capsule endoscopy versus computed tomographic or standard angiography for the diagnosis of obscure gastrointestinal bleeding. Am J Gastroenterol 2007;102:731-7. 62. Aina R, Oliva VL, Therasse E, et al. Arterial embolotherapy for upper gastrointestinal hemorrhage: outcome assessment. J Vasc Interv Radiol 2001;12:195-200. 63. Cohn SM, Moller BA, Zieg PM, et al. Angiography for preoperative evaluation in patients with lower gastrointestinal bleeding: are the benefits worth the risks? Arch Surg 1998;133:50-5. 64. Gay G, Delvaux M, Fassler I. Outcome of capsule endoscopy in determining indication and route for push-and-pull enteroscopy. Endoscopy 2006;38:49-58. 65. Mensink PB, Haringsma J, Kucharzik T, et al. Complications of double balloon enteroscopy: a multicenter survey. Endoscopy 2007;39:613-5. 66. Mehdizadeh S, Ross AS, Gerson L, et al. What is the learning curve associated with double-balloon enteroscopy? Technical details and early experience in 6 U.S. tertiary care centers. Gastrointest Endosc 2006;64:740-50. 67. Hadithi M, Heine GD, Jacobs MA, et al. A prospective study comparing video capsule endoscopy with double-balloon enteroscopy in patients with obscure gastrointestinal bleeding. Am J Gastroenterol 2006;101:52-7. 68. Triester SL, Leighton JA, Leontiadis GI, et al. A meta-analysis of the yield of capsule endoscopy compared to other diagnostic modalities in patients with non-stricturing small bowel Crohn’s disease. Am J Gastroenterol 2006;101:954-64.

www.giejournal.org

Management strategies for obscure GI bleeding

69. Lewis BS, Eisen GM, Friedman S. A pooled analysis to evaluate results of capsule endoscopy trials. Endoscopy 2005;37:960-5. 70. Lapalus MG, Dumortier J, Fumex F, et al. Esophageal capsule endoscopy versus esophagogastroduodenoscopy for evaluating portal hypertension: a prospective comparative study of performance and tolerance. Endoscopy 2006;38:36-41. 71. Baichi MM, Arifuddin RM, Mantry PS. What we have learned from 5 cases of permanent capsule retention. Gastrointest Endosc 2006;64: 283-7. 72. Rondonotti E, Herrerias JM, Pennazio M, et al. Complications, limitations, and failures of capsule endoscopy: a review of 733 cases. Gastrointest Endosc 2005;62:712-6, quiz 752, 754. 73. Gold MR, Siegel JE, Russell LB, et al. Cost-effectiveness in health and medicine. Oxford: Oxford University Press; 1996. 74. Fryback DG, Dasbach EJ, Klein R, et al. The Beaver Dam Health Outcomes Study: initial catalog of health-state quality factors. Med Decis Making 1993;13:89-102. 75. Thomson R, Parkin D, Eccles M, et al. Decision analysis and guidelines for anticoagulant therapy to prevent stroke in patients with atrial fibrillation. Lancet 2000;355:956-62. 76. Gage BF, Cardinalli AB, Albers GW, et al. Cost-effectiveness of warfarin and aspirin for prophylaxis of stroke in patients with nonvalvular atrial fibrillation. JAMA 1995;274:1839-45. 77. Tsevat J, Cook EF, Green ML, et al. Health values of the seriously ill. SUPPORT investigators. Ann Intern Med 1995;122:514-20. 78. Tsevat J, Dawson NV, Wu AW, et al. Health values of hospitalized patients 80 years or older. HELP investigators. Hospitalized Elderly Longitudinal Project. JAMA 1998;279:371-5. 79. Joneja JS, Sharma DB, Sharma D, et al. Quality of life after peptic perforation. J Assoc Physicians India 2004;52:207-9. 80. Flickinger EG, Stanforth AC, Sinar DR, et al. Intraoperative video panendoscopy for diagnosing sites of chronic intestinal bleeding. Am J Surg 1989;157:137-44. 81. May A, Nachbar L, Ell C. Double-balloon enteroscopy (push-and-pull enteroscopy) of the small bowel: feasibility and diagnostic and therapeutic yield in patients with suspected small bowel disease. Gastrointest Endosc 2005;62:62-70. 82. Arias E. United States life tables, 2003. Natl Vital Stat Rep 2006;54: 1-40. 83. Mehdizadeh S, Han NJ, Cheng DW, et al. Success rate of retrograde double-balloon enteroscopy. Gastrointest Endosc 2007;65:633-9. 84. Lewis BS, Wenger JS, Waye JD. Small bowel enteroscopy and intraoperative enteroscopy for obscure gastrointestinal bleeding. Am J Gastroenterol 1991;86:171-4. 85. Marx FW Jr, Gray RK, Duncan AM, et al. Angiodysplasia as a source of intestinal bleeding. Am J Surg 1977;134:125-30. 86. Ettorre GC, Francioso G, Garribba AP, et al. Helical CT angiography in gastrointestinal bleeding of obscure origin. AJR Am J Roentgenol 1997;168:727-31. 87. Lau WY, Ngan H, Chu KW, et al. Repeat selective visceral angiography in patients with gastrointestinal bleeding of obscure origin. Br J Surg 1989;76:226-9. 88. Di Caro S, May A, Heine DG, et al. The European experience with double-balloon enteroscopy: indications, methodology, safety, and clinical impact. Gastrointest Endosc 2005;62:545-50. 89. Kaffes AJ, Koo JH, Meredith C. Double-balloon enteroscopy in the diagnosis and the management of small-bowel diseases: an initial experience in 40 patients. Gastrointest Endosc 2006;63:81-6. 90. Yamamoto H, Kita H, Sunada K, et al. Clinical outcomes of doubleballoon endoscopy for the diagnosis and treatment of small-intestinal diseases. Clin Gastroenterol Hepatol 2004;2:1010-6. 91. Chak A, Koehler MK, Sundaram SN, et al. Diagnostic and therapeutic impact of push enteroscopy: analysis of factors associated with positive findings. Gastrointest Endosc 1998;47:18-22. 92. O’Mahony S, Morris AJ, Straiton M, et al. Push enteroscopy in the investigation of small-intestinal disease. QJM 1996;89:685-90.

Volume 68, No. 5 : 2008 GASTROINTESTINAL ENDOSCOPY 935

Management strategies for obscure GI bleeding

Gerson & Kamal

93. Kim JI, Cheung DY, Cho SH, et al. Oral proton pump inhibitors are as effective as endoscopic treatment for bleeding peptic ulcer: a prospective, randomized, controlled trial. Dig Dis Sci 2007;52:3371-6. 94. Lo CC, Hsu PI, Lo GH, et al. Comparison of hemostatic efficacy for epinephrine injection alone and injection combined with hemoclip therapy in treating high-risk bleeding ulcers. Gastrointest Endosc 2006;63:767-73. 95. Chung IK, Ham JS, Kim HS, et al. Comparison of the hemostatic efficacy of the endoscopic hemoclip method with hypertonic saline-epinephrine injection and a combination of the two for the management of bleeding peptic ulcers. Gastrointest Endosc 1999;49:13-8. 96. Sone Y, Kumada T, Toyoda H, et al. Endoscopic management and follow up of Dieulafoy lesion in the upper gastrointestinal tract. Endoscopy 2005;37:449-53. 97. Park CH, Sohn YH, Lee WS, et al. The usefulness of endoscopic hemoclipping for bleeding Dieulafoy lesions. Endoscopy 2003;35: 388-92. 98. Parra-Blanco A, Takahashi H, Mendez Jerez PV, et al. Endoscopic management of Dieulafoy lesions of the stomach: a case study of 26 patients. Endoscopy 1997;29:834-9. 99. Kwan V, Bourke MJ, Williams SJ, et al. C. Argon plasma coagulation in the management of symptomatic gastrointestinal vascular lesions: experience in 100 consecutive patients with long-term follow-up. Am J Gastroenterol 2006;101:58-63. 100. Christensen S, Riis A, Norgaard M, et al. Short-term mortality after perforated or bleeding peptic ulcer among elderly patients: a population-based cohort study. BMC Geriatr 2007;7:8-15. 101. Lecleire S, Di Fiore F, Merle V, et al. Acute upper gastrointestinal bleeding in patients with liver cirrhosis and in noncirrhotic patients: epidemiology and predictive factors of mortality in a prospective multicenter population-based study. J Clin Gastroenterol 2005;39: 321-7. 102. Klebl F, Bregenzer N, Schofer L, et al. Risk factors for mortality in severe upper gastrointestinal bleeding. Int J Colorectal Dis 2005;20: 49-56.

103. Barkin JS, O’Loughlin C. Capsule endoscopy contraindications: complications and how to avoid their occurrence. Gastrointest Endosc Clin N Am 2004;14:61-5. 104. Eliakim R, Fischer D, Suissa A, et al. Wireless capsule video endoscopy is a superior diagnostic tool in comparison to barium follow-through and computerized tomography in patients with suspected Crohn’s disease. Eur J Gastroenterol Hepatol 2003;15:363-7. 105. Fireman Z, Mahajna E, Broide E, et al. Diagnosing small bowel Crohn’s disease with wireless capsule endoscopy. Gut 2003;52: 390-2. 106. Kastin DA, Buchman AL, Barrett T, et al. Strictures from Crohn’s disease diagnosed by video capsule endoscopy. J Clin Gastroenterol 2004;38:346-9. 107. Manabe N, Tanaka S, Fukumoto A, et al. Double-balloon enteroscopy in patients with GI bleeding of obscure origin. Gastrointest Endosc 2006;64:135-40. 108. Nakamura M, Niwa Y, Ohmiya N, et al. Preliminary comparison of capsule endoscopy and double-balloon enteroscopy in patients with suspected small-bowel bleeding. Endoscopy 2006;38:59-66. 109. Akahoshi K, Kubokawa M, Matsumoto M, et al. Double-balloon endoscopy in the diagnosis and management of GI tract diseases: methodology, indications, safety, and clinical impact. World J Gastroenterol 2006;12:7654-9.

936 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 5 : 2008

www.giejournal.org

Received July 18, 2007. Accepted January 17, 2008. Current affiliations: Division of Gastroenterology and Hepatology Stanford University School of Medicine, Stanford, California, USA. Presented at American Society for Gastrointestinal Endoscopy, Digestive Disease Week, May 21-24, 2006, Los Angeles, California (Gastrointest Endosc 2006;63:AB90). Reprint requests: Lauren B. Gerson, MD, Division of Gastroenterology and Hepatology, Stanford University Medical Center, A149, 300 Pasteur Dr, Stanford, CA 94305-5202.