Background: For patients with cirrhosis, non-selective beta blockers (NSBB; carvedilol, nadolol and propranolol) reduce portal pressure and have been shown in placebo-controlled trials to prevent liver decompensation. While some data suggest carvedilol may be more effective than nadolol or propranolol, no prior study has directly compared NSBBs in relation to major clinical outcomes.
Objectives: To define the comparative effectiveness of carvedilol, nadolol and propranolol, in cirrhosis.
Methods: Using two US administrative claims datasets (IBM® MarketScan® and Optum® Clinformatics®), we identified three 1:1 propensity score (PS)-matched cohorts with cirrhosis initiating carvedilol, nadolol or propranolol, between 2005-2022 (cohort entry date [CED]=first dispensing date). All patients were required to have cirrhosis diagnosed in the 183 days before CED (covariate assessment period) and no prior NSBB use in that 183-day period. Using database-specific, 1:1 PS-matching, we balanced 106 covariates among initiators in each pairwise cohort. Patients were followed in an as-treated approach for decompensation (i.e., esophageal variceal hemorrhage, ascites or spontaneous bacterial peritonitis, or hepatic encephalopathy) or death. We estimated database-specific, PS-matched rate differences (RD) using generalized linear models, assuming Poisson distributions, and hazard ratios (HR) using Cox proportional hazard regression models, with corresponding 95% CIs.
Results: PS-matched rates of any decompensation event or death were lower among carvedilol initiators, versus nadolol (MarketScan RD=-134.4 [95%CI=-168.9, -99.9], HR=0.71 [95%CI=0.65-0.78], n=4,609 pairs; Optum RD= -54.4 [95%CI=-96.8, -12.1], HR=0.92 [95%CI=0.85-0.99]; n=4,555 pairs), and versus propranolol (MarketScan RD=-214.7 [95%CI=-248.4, -180.9], HR=0.67 [95%CI=0.61-0.72], n=5,923 PS-matched pairs; Optum RD= -184.4 [95%CI=-224.5, -144.3], HR=0.79 [95%CI=0.74-0.85]; n=5,870 pairs). PS-matched rates of individual decompensation events were also significantly lower with carvedilol initiation, as compared to nadolol or propranolol. Findings persisted in subgroups with compensated and decompensated cirrhosis, and when applying a 365-day intention-to-treat (ITT) approach.
Conclusions: In patients with cirrhosis, rates of major liver decompensation events and death were significantly lower with initiation of carvedilol, as compared to nadolol or propranolol. This finding was robust in both compensated and decompensated cirrhosis, suggesting broad potential benefits of carvedilol across the spectrum of cirrhosis.
