The entire length of the small intestine could not be fully visualized until the introduction of capsule endoscopy (CE) in the year 2000. The invention of double-balloon enteroscopy (DBE) in 2001 revolutionized the management of small intestinal diseases by adding diagnostic as well as therapeutic options [1]. Small bowel bleeding is defined as a bleeding originating from any part of the small intestine. The terminology “mid-gastrointestinal (GI) bleeding” is more specific and refers to bleeding distal to the ampulla of Vater up to the terminal ileum, which can be diagnosed using CE, balloon-assisted enteroscopy (BAE), cross-sectional studies, or angiography [2]. Before the era of CE and BAE, small bowel bleeding was called obscure GI bleeding, as it was considered GI bleeding without an obvious cause, because of not being detected by conventional upper and lower GI endoscopies. However, this terminology is now reserved for cases without an identifiable cause in spite of full visualization of the small bowel by CE, BAE, or other imaging modalities [3]. Generally, small bowel bleeding accounts for approximately 5% of all GI bleeding [4]. It can be classified into overt bleeding which presents as hematemesis, hematochezia or melena; and occult bleeding in which the presentation is persistent anemia and a positive fecal blood test [3].

The causes of small bowel bleeding are generally classified into erosions/ulcers, vascular lesions, and tumors/polyps. Inflammatory lesions such as erosions and ulcers are more common in Eastern countries, while vascular lesions are predominant in Western countries [5].

Small bowel bleeding is more likely due to inflammatory bowel disease, Dieulafoy’s lesion, or Meckel’s diverticulum in young patients, but tends to be due to vascular lesions, malignancies, or ulcers in older patients [6]. Among vascular lesions, angioectasias are dilated blood vessels with or without an endothelial lining. Small arteriovenous communications develop in them owing to incompetent precapillary sphincters. The pathogenesis of these vascular abnormalities is local hypoperfusion, which stimulates sympathetic nerves to initiate smooth muscle relaxation in the vessel wall to increase blood flow and reverse hypoxia, eventually resulting in an arteriovenous communication through the ectasias. Several systemic comorbidities can be associated with local hypoperfusion [4].

The key point while taking a patient’s history is to ask about overt bleeding. If present, the form and color of the bleeding may provide a clue about the location of the bleeding source. Hematemesis usually indicates bleeding above the ligament of Treitz. Tarry stool suggests an upper GI, jejunal, or ileal origin, while bright or dark red stool suggests an ileal or colorectal origin. However, acute massive bleeding in the upper GI tract may cause reddish stools, and scant bleeding from ileal or colonic lesions may cause black stools, especially in constipated patients [4].

Patients with comorbidities are likely to have vascular lesions. Young patients without comorbidities typically have bleeding from Meckel’s diverticulum; Crohn’s disease or, in rare cases, an arteriovenous malformation; or small bowel tumors/ polyps associated with hereditary polyposis syndrome. Older patients are likely to have non-steroidal anti-inflammatory drug (NSAID)-induced enteritis, chronic inflammatory diseases such as intestinal tuberculosis, Behçet’s disease, radiation enteritis, or small bowel tumors such as lymphomas, adenocarcinoma, metastatic tumors, and carcinoid tumors [7]. The presence of chronic epistaxis may point to hereditary hemorrhagic telangiectasia [8].

Ohmiya et al. developed a comorbidity index, referred to as the “Ohmiya index” (Table 1), to predict small bowel bleeding attributed to vascular etiologies [6]. By combining the Ohmiya index and the age of onset of small bowel bleeding, the type of lesion can be predicted. In 53% of young patients aged <50 years with an Ohmiya index score of <2, Meckel’s diverticulum and Crohn’s disease were commonly found. In 72% of elderly patients with an Ohmiya index score of <2, inflammatory diseases, drug-induced injuries, or tumors were found. In 68% of the patients with an Ohmiya index score of ≥2, small bowel vascular lesions were found regardless of age [6].

The result of the initial upper and lower endoscopies may be negative in some patients with upper and lower GI bleeding owing to missed lesions as a result of inappropriate timing of the procedure, difficulty in visualizing the lesion location, massive active bleeding, or poor bowel preparation that results in poor visibility [9]. The diagnostic yield of second-look upper and lower endoscopies were 2%–25% and 6%–23%, respectively [3].

Computed tomographic (CT) enterography/CT enteroclysis, CT angiography, and magnetic resonance enterography are cross-sectional imaging modalities that can be used to evaluate patients with small bowel bleeding. Cross-sectional studies provide a diagnostic yield in patients with small bowel bleeding of 17%–48% [4]. These modalities are beneficial in providing an initial diagnostic/therapeutic guidance before endoscopic intervention, or after the failure to identify the bleeding site by CE. For occult bleeding, CT enterography is more suitable. CT enterography is more sensitive for detecting neoplastic or inflammatory lesions than for detecting vascular lesions or Meckel’s diverticula [10]. CT angiography is suitable for massive overt bleeding at bleeding rates ranging from 0.5 to 1.0 mL/min, which are equivalent to 2–3 units of blood loss in 24 hours. The absence of oral contrast allows tracking of the extravasated blood, which can point to the bleeding site, and angiography enables the process of embolization of the responsible blood vessels to control the bleeding immediately [9]. Obtaining a chest CT scan is also important, as the etiology of small bowel bleeding could be secondary to a lung disease such as intestinal tuberculosis, metastatic lung cancer, or arteriovenous malformations in hereditary hemorrhagic telangiectasia (Rendu-Osler-Weber disease) [8].

CE was the first breakthrough in the diagnosis of small bowel diseases. CE takes two digital images per second (Pill-Cam SB2; Medtronic, Minneapolis, MN, USA, EndoCapsule; Olympus Co., Tokyo, Japan, and OMOM; Jinshan Science & Technology, Chongqing, China), three per second (MiroCam; IntroMedic, Seoul, Korea), or 20 per second (CapsoCam; CapsoVision, Saratoga, CA, USA). PillCam SB3 (Medtronic) has a faster adaptable frame rate system (two or six images per second), depending on the transit speed. PillCam (Medtronic) has new v9.0 software with a “Top 100” feature, which automatically identifies the 100 most clinically relevant images using artificial intelligence algorithms. This facilitates detection of 83.5% of lesions with a high bleeding potential and >95.5% of angioectasias [11]. CE is safe and noninvasive, with a diagnostic yield of 41%–80%. It can also aid in the identification of the nearest route for BAE [4]. The major drawback of CE is retention, with a retention rate of 1.4% when used to identify causes of small bowel bleeding [12]. Retention is defined as failure of capsule excretion after ≥2 weeks. The known risk factors of retention are small intestinal Crohn’s disease, previous abdominal surgery, and abdominal radiotherapy. In patients with these factors, intestinal patency should be assessed by using a dissolving patency capsule from PillCam, which was introduced in 2012, before CE became available. The other disadvantages of CE include the inability to obtain biopsies, inaccurate identification of the lesion site, and false-positive and false-negative results especially in proximal intestinal lesions, diverticula, and submucosal tumors [8].

DBE was the first type of BAE and was introduced in 2001 by Yamamoto et al. [1] BAE allows passage of the endoscope to the distal small bowel using an overtube with a balloon. DBE has an additional endoscope balloon, while single-balloon enteroscopy (SBE) uses only an overtube balloon. In patients with small bowel bleeding, the diagnostic yield of BAE is 55%–78% [8]. BAE has relatively good maneuverability in the distal small bowel, allowing hemostatic procedures to be performed, such as clipping and argon plasma coagulation. Hong et al. analyzed the data of 1,108 BAE procedures and reported a complication rate of 1.1% (12/1,108) [13]. The most common complications were bleeding (n=4), perforation (n=4), pancreatitis (n=2), and post-polypectomy syndrome (n=2) [13]. Aspiration pneumonia with oral BAE has also been mentioned in some reports. A meta-analysis by Wadhwa et al. revealed that the diagnostic yield, therapeutic yield, and incidence of adverse events were comparable between DBE and SBE [14]. However, the rate of complete enteroscopy was significantly higher in DBE than in SBE, although the clinical impact of complete enteroscopy is controversial [14].

A comparison between CT enterography and CE in patients with small bowel bleeding revealed a significantly higher diagnostic yield with CE, mainly due to the better detection of angioectasias by CE, while the sensitivity for tumor detection was comparable, with superiority of CT enterography [15]. In another pooled analysis, CE had a significantly higher diagnostic yield in small bowel bleeding than DBE performed in a single route, but the diagnostic yield of CE was lower than that of complete DBE when both antegrade and retrograde routes were used [16]. To perform BAE, the selection of the route (antegrade or retrograde) is determined on the basis of previous information such as medical history, and CT and CE findings. Prior to BAE, contrast-enhanced CT is recommended to avoid overlooking small bowel tumors such as GI stromal tumors.

The diagnostic yield of CE was significantly higher in the patients with ongoing overt obscure GI bleeding than in those with previous overt and occult obscure GI bleeding [17]. Patients aged ≥60 years have a non-significantly higher diagnosis rate (74.2% vs. 60.6%) than those aged <60 years [18]. A greater number of packed red blood cell (RBC) transfusions was associated with a higher diagnostic yield of CE. One study showed that >50% of patients with a positive CE study had received >3 units of packed RBCs before the procedure, which is more than that given to patients with a negative CE (21.2%) [17]. Other factors associated with increased diagnostic yield include moderate/severe renal disease and treatment with antiplatelet drugs [17]. The low yield associated with late intervention may be due to the healing of small mucosal defects; resolution of bleeding from a Dieulafoy’s lesion, which may then appear as a normal mucosa; and small nonbleeding angioectasias between intestinal folds that are easily missed [4].

We previously reported the importance of the timing and diagnostic yield of DBE in patients with overt obscure GI bleeding [19]. Patients who underwent DBE within 1 month of the most recent episode of overt bleeding had a significantly higher diagnostic yield than other patients (84% vs. 57%, p=0.002). We also reported a significant negative correlation between positive DBE findings and the length of interval since bleeding (p<0.001). Therefore, endoscopic investigations should be performed as soon as possible after overt bleeding.

Otani et al. found that repeat CE (73.2%) had a significantly higher diagnostic yield than DBE (39.6%) in patients with obscure GI bleeding with a previously negative CE finding [20]. Japanese guidelines recommend assessing the need for further workup after a negative CE result is obtained. For patients with a history of frequent hemorrhage/massive bleeding/severe anemia, they recommend BAE. For young patients, they recommend a Meckel’s scan because Meckel’s diverticulum is often missed on CE. For patients with a low likelihood of a false-negative study result, the guidelines recommend follow-up only. If bleeding occurs earlier in the follow-up period, BAE should be performed. If bleeding recurs well into the follow-up period, repeated upper and lower GI endoscopies are advisable. Whenever both imaging studies yield negative results, CE should be repeated [8]. In case of a negative BAE with persistent bleeding, angiography, intraoperative enteroscopy, or a Meckel’s scan should be considered. If the bleeding has spontaneously resolved, follow-up alone is recommended [4].

We previously reported the long-term outcomes of patients with negative DBE findings for overt obscure GI bleeding [21]. Of 42 patients, 16 (38%) developed overt rebleeding during the follow-up period (5.4 years on average) and 14 underwent further investigations. In 10 (71%) of the 14 patients, the bleeding source was identified. Hence, repeated enteroscopy should be performed in patients with overt rebleeding even after a negative enteroscopy finding was obtained.

Angiography can be both diagnostic and therapeutic. It has a variable diagnostic yield depending on the bleeding rate. Yields are high (77%) in overt bleeding and low (<20%) in occult bleeding [9]. Angiography has a significantly lower diagnostic yield than CE in patients with overt small bowel bleeding (20% vs. 53.3%) [22]. Angiography can be used as the first-line diagnostic modality in patients with hemodynamic instability or as the last-line modality in patients with negative BAE despite ongoing bleeding. The complication rate is 10%, with renal impairment, infection, and infarction being the most common. Provocative angiography can be used if the bleeding has stopped and the source could not be identified. This is performed with the use of anticoagulants, thrombolytic agents, and vasodilators. However, it is associated with an increased incidence of bleeding complications [9].

The most frequently used nuclear medicine scan to assess patients with acute obscure GI bleeding is performed with 99m-Tc-labeled autologous RBCs. The minimal bleeding rate needed for positive results is 0.1–0.4 mL/min [23]. The drawbacks are the poor localization of the bleeding source and the lack of immediate availability. The overall diagnostic yield of tagged RBCs is approximately 50% [9]. Meckel’s scan using 99m-Tc-pertechnetate can be used in small bowel bleeding if all previous modalities were non-diagnostic. Positive Meckel’s scans only indicate an ectopic gastric mucosa in the small intestine, but it is a reasonable study if a Meckel’s diverticulum is suspected after failure of other modalities [7]. This is not recommended as a first-choice imaging study.

Intraoperative enteroscopy can be performed by transoral or transrectal insertion, or through an enterotomy. It is performed with upper or lower endoscopy or BAE. The indication for using this approach in patients with obscure GI bleeding is limited to patients with failed BAE and angiography-related therapies, or failed localization of small bowel lesions during surgical exploration. Its diagnostic and therapeutic yields range from 58% to 100%. The incidence rate of adverse events varies with patients’ conditions (1%–50%). The most commonly encountered adverse events include prolonged postoperative ileus, small bowel obstruction, and wound infection [24]. Undoubtedly, it is more invasive than BAE and is not a firstchoice modality.

The American College of Gastroenterology (ACG) guidelines recommend performing CE before BAE in patients with small bowel bleeding [3]. However, the Japanese guidelines recommend performing contrast-enhanced (preferably dynamic) CT as the first step in BAE management, especially in patients with overt bleeding. If the bleeding stopped after a negative CT finding was obtained, CE may be considered [8].

ACG guidelines recommend BAE with a negative CE finding under a high clinical suspicion of a small bowel lesion or active bleeding [3]. Antegrade BAE is recommended as the first approach for overt ongoing bleeding because bloody enteric contents usually progress distally via peristalsis. When bloody enteric content is first observed, a clip should be placed to mark the approximate location of the bleeding lesion in the small intestine. If the bleeding site is suspected to be in the distal ileum on CT, retrograde BAE should be considered first [25].

CE and BAE have revolutionized the diagnosis and treatment of patients with small bowel bleeding. Cross-sectional imaging modalities are beneficial to provide an anatomical road map before enteroscopy. On the basis of the patient’s history and background, the type of lesion can be predicted. The appropriate therapeutic modalities should be used, with understanding of their advantages and disadvantages.