Risk of ambulatory colonoscopy in patients with cirrhosis: a propensity-score matched cohort study.

Patients with cirrhosis demonstrate alterations in physiology, hemodynamics, and immunity which may increase procedural risk. There exist sparse data regarding the safety of performing ambulatory colonoscopy in patients with cirrhosis. From a population-based sample of three North American states (California, Florida, and New York), we collected data on 3,590 patients with cirrhosis who underwent ambulatory colonoscopy from 2009 to 2014. We created a control cohort propensity score-matched for cirrhotic severity who did not undergo colonoscopy (N = 3,590) in order to calculate the attributable risk for adverse events. The primary endpoint was the rate of unplanned hospital encounters (UHEs) within 14 days of colonoscopy (or from a synthetic index date for the control cohort). Predictors for UHE were assessed in multivariable regression. The attributable risk for any UHE following colonoscopy was 3.1 % (confidence interval [CI] 2.1-4.1 %,  < 0.001). There was increased risk for infection (0.9 %, CI 0.7-1.1 %), spontaneous bacterial peritonitis (0.1 %, CI 0.0-0.3 %), decompensation of ascites (0.3 %, CI 0.2-0.4 %), and cardiovascular event (0.4 %, CI 0.3-0.5 %). There was no increased attributable risk for gastrointestinal bleeding, perforation, or development of the hepatorenal syndrome. The presence of ascites at time of procedure was the only predictor for UHE in the fully-adjusted model (OR 2.6, CI 1.9-3.5,  < 0.001). There is a moderate though detectable increase in risk for adverse event following ambulatory colonoscopy in patients with cirrhosis. The presence of ascites in particular portends higher risk. These data may guide clinicians when counseling patients with cirrhosis on the choice of colorectal cancer screening modality.