Chronic liver diseases are prevalent, particularly among patients with type 2 diabetes, who have a higher incidence of metabolic dysfunction-associated steatotic liver disease (MASLD) and are subject to other liver diseases.1 With no pharmacotherapy for these liver conditions, recent attention was given to glucagon-like peptide-1 receptor agonists (GLP1a) for their effectiveness in treating type 2 diabetes and obesity and in reducing liver enzyme levels and liver fat levels.2 A 72-week, phase II, placebo-controlled randomised trial in patients with non-alcoholic steatohepatitis (NASH) and liver fibrosis found that the GLP1a semaglutide improves the resolution of NASH but does not affect fibrosis regression.2 By its limited size, however, this trial could not provide effectiveness data on major liver outcomes (MALO), such as decompensated cirrhosis, hepatocellular carcinoma, liver transplantation or liver-related death, despite its long-term follow-up. Non-randomised observational studies can be useful to address such questions.

In Gut, Wester et al 3 present an observational study that evaluates the effectiveness of GLP1a in the long-term incidence of MALO in patients with chronic liver disease and type 2 diabetes. They use data from Swedish healthcare registers between 2010 and 2020 to identify all Swedish adult residents with diagnoses of chronic liver disease and type 2 diabetes and who filled at least one prescription of metformin, an indication for second-line treatment with a GLP1a (liraglutide, semaglutide, dulaglutide, exenatide and lixisenatide).

Within this base cohort, defined as of January 2010, initiation and continuation of a GLP1a were compared with non-initiation. The outcome was the first MALO event to occur during follow-up, namely decompensated cirrhosis (variceal bleeding, ascites, portal hypertension or hepatorenal syndrome), hepatocellular carcinoma, need for liver transplantation or MALO-related death. The authors emulated a hypothetical pragmatic randomised controlled trial (RCT), a so-called ‘target trial’, by generating a series of 131 separate ‘trials’, …