Randomized sham-controlled trials in endoscopy: a systematic review and meta-analysis of adverse events.

Schulman, Allison R; Popov, Violeta; Thompson, Christopher C

BACKGROUND AND AIMS Sham procedures in endoscopy are used with the intention of controlling for placebo response, potentially allowing more precise evaluation of treatment effect. Nevertheless, this type of study may impose significant risk without potential benefit for those in the sham group. The aim of the current study was to systematically review and analyze the endoscopic literature to assess the safety of sham controls. METHODS MEDLINE and Embase databases were searched for endoscopic sham procedures for all dates to July 2017. Only randomized controlled trials comparing an endoscopic therapy with a sham were included. Primary outcome was adverse events (AEs) categorized as mild, moderate, or severe. Results were combined using a random-effects model. Heterogeneity was assessed with the I2 statistic, and publication bias was assessed with the Egger test and funnel plots. RESULTS Data were extracted from 34 publications (1987-2017; 100% full text), with a total of 2492 procedures (1355 treatment/1137 sham). Sham procedures involved upper endoscopy (31 studies) and ERCP (3 studies). Treatment arms included procedures with the following indications: weight loss (38.2%), GI bleeding (26.5%), GERD (20.6%), sphincter of Oddi dysfunction (8.8%), and dysphagia (6.2%). Overall percentage of severe adverse events (SAEs) in the sham group was 1.7% (19/1137). Of these, the most common SAEs in the sham groups were need for surgery/intensive care unit stay (35.3%), post-ERCP pancreatitis (23.5%), and perforation (11.8%). There was no significant difference in the odds of developing an SAE between the treatment group and the sham group (odds ratio, 1.3; 95% confidence interval [CI], 0.7-2.3). The pooled additional risk incurred from being initially randomized to the sham arm and then receiving a cross-over intervention was significant (RR, 1.33; 95% CI, 1.14-1.56; P < .001), compared with patients initially randomized to the study intervention. CONCLUSION The frequency of AEs in endoscopic sham procedures is substantial, and patients are subjected to considerable morbidity. These results raise a serious ethical dilemma regarding the use of sham-controlled trials.