BACKGROUND AND STUDY AIMS
The aim of the study was to assess the feasibility, diagnostic yield, and interobserver agreement of capsule endoscopy in the investigation of patients with obscure or occult gastrointestinal bleeding.
PATIENTS AND METHODS
A total of 64 consecutive patients with occult bleeding (31 %) or overt bleeding (69 %) were assessed using capsule endoscopy after negative upper and lower endoscopy and small-bowel radiology. The quality of visualization of the small-bowel mucosa was scored from 1 (poor) to 4 (excellent). Thirty video capsule recordings with normal or abnormal findings were blindly assessed by four independent endoscopists. Interobserver agreement was evaluated using the kappa index.
RESULTS
The small bowel was completely visualized in 57/64 patients (89 %). Incomplete small-bowel transit was most commonly due to prolonged gastric retention (five patients). The mucosa visualization scores (means) for the proximal, middle, and distal thirds of the small bowel were 3.7, 3.3, and 2.2 respectively. Visualization of the distal ileum was good (> or = 3) in 38 % and a bleeding site was found in 45 % of patients. Push-enteroscopy was also performed in 56 patients. The results of the two techniques were similar in 37 patients, capsule endoscopy was superior in 12 patients, and push-enteroscopy was superior in seven patients. Interobserver agreement was good for bleeding and for angiodysplasia, but poor for ulcers and tumors. Mean interobserver agreement was better among experienced endoscopists than among junior endoscopists.
CONCLUSIONS
Capsule endoscopy allowed the whole small intestine to be explored in 89 % of patients, with good visualization of the mucosa, except distally. Interobserver agreement was better among the experienced endoscopists and was better for red-colored abnormalities (bleeding and angiodysplasia) than for ulcers and tumors.