Endoscopy 2013; 45(09): 777
DOI: 10.1055/s-0033-1344149
Letters to the editor
© Georg Thieme Verlag KG Stuttgart · New York

# Capsule endoscopy for acute upper gastrointestinal bleeding: is the cherry ripe yet?

Further Information

### Publication History

Publication Date:
29 August 2013 (online)

We read with interest the prospective cohort study by Gralnek et al. [1], which used the PillCam ESO 2 (Given Imaging Ltd., Yoqneam, Israel) in the emergency department for patients presenting with acute upper gastrointestinal bleeding. This represents part of a limited emerging literature on the subject [2] [3] [4]. Although the feasibility and safety of video capsule endoscopy (VCE) in this clinical setting appear clear, there remain a number of obstacles to be overcome before VCE can become an integral part of the initial emergency department assessment of upper gastrointestinal bleeding.

The first issue is the overall accuracy of VCE as a triage tool to stratify patients with regard to timing of endoscopy, use of intravenous medications (proton pump inhibitors, octreotide, terlipressin), appropriate level of inpatient monitoring and care, or stratifying to outpatient management. The crux of this issue lies in the ability of VCE to detect luminal blood and mucosal lesions and, most importantly, in the sensitivity of VCE for the detection of lesions with high risk stigmata. The finding of 83.3 % sensitivity for luminal blood compared quite favorably with the 33.3 % sensitivity of nasogastric aspirate. Most of this difference can be explained by blood in the duodenal lumen detected by VCE but missed by nasogastric aspirate. However, nasogastric aspiration has been known to have a low sensitivity and specificity for acute upper gastrointestinal hemorrhage [5], so this is a rather low bar to hurdle. Although the authors found no significant difference in the identification of peptic or inflammatory lesions between VCE (67.5 %) and esophagogastroduodenoscopy (87.5 %), the possibility of a type II error is quite plausible given the small numbers of patients completing both studies (n = 41). Most significantly, it will be critical to assess the sensitivity of VCE for the detection of lesions with high risk stigmata, as this would be an independent triage tool to determine the suitability of outpatient management. Failure to detect any high risk lesion could have major clinical consequences with regard to inappropriate outpatient management stratification. We also wonder whether the use of intravenous erythromycin in this setting could have skewed the evidence of gross blood findings affecting the results for nasogastric tube aspiration.

The second issue regards who is most appropriate and available to interpret VCE in the emergency room situation? There are some preliminary data showing that emergency room physicians and gastroenterologists demonstrate excellent agreement and interpretation of VCE in this setting [3]. Alternatively, video images could be interpreted by an off-site gastroenterologist for real-time decision making.

The third issue is cost-effectiveness of VCE in this setting. The mean length of stay for patients discharged from the hospital with a diagnosis of gastrointestinal hemorrhage is 4.5 days and the mean hospital charges are US$26 210 per admission [6]. The physician Medicare national average fee in the USA for PillCam ESO 2 is US$ 755 including the technical and professional fees. VCE must demonstrate considerable diagnostic sensitivity and specificity for triage so that significant inpatient cost savings can be realized or hospitalizations safely avoided, otherwise it will simply add another costly test to the evaluation of upper gastrointestinal hemorrhage. The further question of whether VCE can replace a traditional esophagogastroduodenoscopy or buy time until one can be done will only be realized in further studies.

In conclusion, we commend the investigators for evaluating this interesting and novel strategy of using VCE in the setting of acute upper gastrointestinal bleeding; however, we await more data for its effective implementation in an algorithmic through-the-door approach.