Obscure GI bleeding: is capsule endoscopy sufficient?

For decades, investigation of obscure GI bleeding (OGIB) has often caused frustration in both patients and clinicians. Noninvasive tests, such as small-bowel follow-through, radioisotope-labeled red blood cell scan, and push enteroscopy, have had suboptimal diagnostic yields in the range of 20% to 40%. Invasive methods, such as laparotomy or intraoperative enteroscopy, may improve the yield up to 70%. However, complications such as mucosal laceration, avulsion of mesenteric vessels, prolonged ileus, and wound infection are not uncommon, and death has been reported in the literature. Thanks to the latest advancement in miniaturization and wireless technology, we are now capable of containing a color camera, light-emitting diode, battery, and signal transducer in a miniature plastic capsule. A meta-analysis has shown that capsule endoscopy (CE) is superior to barium radiography and push enteroscopy in OGIB. Furthermore, CE is well known for its safety and acceptance, and the risk of capsule retention is less than 2%. Given its noninvasiveness and satisfactory diagnostic yield, CE is now recommended as the first-line investigation for OGIB. Although most existing trials only focused on the diagnostic value of CE, few studied its clinical impact on the long-term outcome of OGIB. In a 1-year study based on telephone interview, the negative predictive value of CE for clinical rebleeding was 87%. In current issue of Gastrointestinal Endoscopy, Macdonald et al present their long-term follow-up data of patients with OGIB. Altogether, 49 patients underwent CE; 42 (86%) of them had complete capsule examination and at least 1 year of follow-up. The study was regarded as ‘‘positive’’ when at least one highly relevant lesion was found (P2), and ‘‘negative’’ when the abnormality found could not explain for blood loss (P1) or when no abnormality (P0) was seen. The overall diagnostic yield was 57% for P2 lesions, and the most commonly found lesion was angiodysplasia (79%). As expected, there exists a significant difference in the clinical rebleeding rate between the positive and negative groups (42% vs 11%, P ! .01). Of the 18 negative studies, 5 had P1 lesions, and none of them had rebleeding reported. The negative predictive value was 89%. Limited by its inability for biopsy and therapy, the development of CE has been focused on its capacity as a first-line di-