Introduction Capsule endoscopy might be a suitable non-invasive method to identify patients with suspected upper gastrointestinal (GI) bleeding who need hospital admission and endoscopic therapy.

Method Stable patients with suspected upper GI bleeding were recruited. After swallowing 1L water with simethicone, MACE was performed using a Mirocam Navi (a single camera device controlled using an external handheld magnet (Intromedic Ltd., Korea)) not less than one hour before gastroscopy (OGD), and findings compared. Prior to OGD, a clinical opinion was documented on whether patients could be discharged based on MACE findings.

Results The mean duration of oesophagogastric MACE in 15 patients (78.6% M, mean age 60±14.9) was 29.2±17.4mins: duodenal inspection was usually performed after video download. Overall, mucosal visualisation (worst-best:1–5) was good in the gastric cardia (4.1±1.5), gastric body (4.2±2.6), antrum (4.9±0.2) and D2 (5.0±0); less so in oesophagus (2.9±2.4), fundus (3.1±1.2) and D1 (3.3±1.2). Lesions detected by both modalities accounted for 20.5% of pathologies: one case each of oesophagitis, duodenitis, benign gastric polyps, gastric ulcer, oesophageal and gastric varices and two cases of gastritis. 64.1% were only detected by MACE: oesophagitis (n=3), hiatus hernia (n=3), gastritis (n=5), gastric polyps (n=4), gastric angioectasia (n=1), duodenal erosions (n=3) and D1 ulcers (n=3). 26.9% of lesions were only detected by OGD: Barrett’s oesophagus (n=1), gastric erosion (n=1), early gastric antral vascular ectasia (n=1), duodenitis (n=1) and D1 ulcers (n=2). Both cases of D1 ulcers (neither exhibiting stigmata of recent bleed) missed by MACE were due to rapid transit: the edge of one of the lesions could be seen in retrospect. MACE correctly recommended admission for 3/3 patients: one because of poor views and two due to fresh blood in the stomach (no lesion identified at MACE or OGD) or duodenum (DU missed at MACE). The median hospital stay of those for whom MACE suggested discharge (excluding one with 384 hour admission due to recurrent hypoglycaemia) was 53 hours (range 15–103). MACE correctly suggested safe discharge for 12/12 patients. No complications were seen. Post-procedure pain, discomfort and distress scores were all lower with MACE (p<0.05).

Conclusion MACE was extremely well tolerated and allowed the identification of patients suitable for out-patient management. Both OGD and MACE missed lesions. Sensitivity of the MiroCam Navi could be improved if the mechanism triggering image capture occurred earlier in the oesophagus and a double-ended camera developed to increase imaging of the fundus and D1.