Capsule endoscopy for the diagnosis of oesophageal varices in people with chronic liver disease or portal vein thrombosis

Background Current guidelines recommend performance of oesophago-gastro-duodenoscopy at the time of diagnosis of hepatic cirrhosis to screen for oesophageal varices. These guidelines require people to undergo an unpleasant invasive procedure repeatedly with its attendant risks, despite the fact that half of the people do not have identifiable oesophageal varices 10 years after the initial diagnosis of cirrhosis. Video capsule endoscopy is a non-invasive test proposed as an alternative method for the diagnosis of oesophageal varices. Objectives To determine the diagnostic accuracy of capsule endoscopy for the diagnosis of oesophageal varices in children or adults with chronic liver disease or portal vein thrombosis, irrespective of the aetiology. To investigate the accuracy of capsule endoscopy as triage or replacement of oesophago-gastro-duodenoscopy. Search methods We searched the Cochrane Hepato-Biliary Group Diagnostic Test Accuracy Studies Register (October 2013), MEDLINE (Ovid SP) (1950 to October 2013), EMBASE (Ovid SP) (1980 to October 2013), ACP Journal Club (Ovid SP) (1991 to October 2013), Database of Abstracts of Reviews of Effects (DARE) (Ovid SP) (third quarter), Health Technology Assessment (HTA) (Ovid SP) (third quarter), NHS Economic Evaluation Database (NHSEED) (Ovid SP) (third quarter), and Science Citation Index Expanded (SCIEXPANDED) (ISI Web of Knowledge) (1955 to October 2013). We applied no language or document type restrictions. Selection criteria Studies that evaluated the diagnostic accuracy of capsule endoscopy for the diagnosis of oesophageal varices using oesophago-gastro-duodenoscopy as the reference standard in children or adults of any age, with chronic liver disease or portal vein thrombosis. Data collection and analysis We followed the available guidelines provided in the Cochrane Handbook for Diagnostic Test of Accuracy Reviews. We calculated the pooled estimates of sensitivity and specificity using the bivariate model due to the absence of a negative correlation in the receiver operating characteristic (ROC) space and of a threshold effect. Main results The search identified 16 eligible studies, in which only adults with cirrhosis were included. In one study, people with portal thrombosis were also included. We classifiedmost of the studies at high risk of bias for the 'Participants selection' and the 'Flow and timing' domains. One study assessed the accuracy of capsule endoscopy for the diagnosis of large (high-risk) oesophageal varices. In the remaining 15 studies that assessed the accuracy of capsule endoscopy for the diagnosis of oesophageal varices of any size in people with cirrhosis, 936 participants were included; the pooled estimate of sensitivity was 84.8% (95% confidence interval (CI) 77.3% to 90.2%) and of specificity 84.3% (95% CI 73.1% to 91.4%). Eight of these studies included people with suspected varices or people with already diagnosed or even treated varices, or both, introducing a selection bias. Seven studies including only people with suspected but unknown varices were at low risk of bias; the pooled estimate of sensitivity was 79.7% (95% CI 73.1% to 85.0%) and of specificity 86.1% (95% CI 64.5% to 95.5%). Six studies assessed the diagnostic accuracy of capsule endoscopy for the diagnosis of large oesophageal varices, associated with a higher risk of bleeding; the pooled sensitivity was 73.7%(95% CI 52.4% to 87.7%) and of specificity 90.5%(95% CI 84.1% to 94.4%). Two studies also evaluated the presence of red marks, which are another marker of high risk of bleeding; the estimates of sensitivity and specificity varied widely. Two studies obtained similar results with the use of a modified device as index test (string capsule). Due to the absence of data, we could not performall planned subgroup analyses. Interobserver agreement in the interpretation of capsule endoscopy results and any adverse event attributable to capsule endoscopy were poorly assessed and reported. Only four studies evaluated the interobserver agreement in the interpretation of capsule endoscopy results: the concordance was moderate. The participants' preferences for capsule endoscopy or oesophago-gastro-duodenoscopy were reported differently but seemed in favour of capsule endoscopy in nine of 10 studies. In 10 studies, participants reported some minor discomfort on swallowing the capsule. Only one study identified other significant adverse events, including impaction of the capsule due to previously unidentified oesophageal strictures in two participants. No adverse events were reported as a consequence of the reference standard. Authors' conclusions We cannot support the use of capsule endoscopy as a triage test in adults with cirrhosis, administered before oesophago-gastroduodenoscopy, despite the low incidence of adverse events and participant reports of being better tolerated. Thus, we cannot conclude that oesophago-gastro-duodenoscopy can be replaced by capsule endoscopy for the detection of oesophageal varices in adults with cirrhosis. We found no data assessing capsule endoscopy in children and in people with portal thrombosis.