Table 1

Protocol of a target trial and an emulated trial using observational data

Protocol component	Target trial	Emulated trial using observational data
Eligibility criteria	Inclusion Swedish residents ≥18 years of age between January 2010 and November 2020 with any chronic liver disease and type 2 diabetes previously diagnosed at any point, who are currently using at least 1 g of metformin per day and have used metformin for at least 6 months. Exclusion Any previously filled prescription of GLP1 agonists. History of any contraindications to GLP1 agonists (pancreatitis, inflammatory bowel disease or severe chronic kidney disease). History of major adverse liver outcomes.	Same as the target trial. Patients are required to have filled at least one prescription of metformin the last year, corresponding to a daily dose of 1 g.
Treatment strategies	Initiation of a GLP1 agonist at baseline and continued treatment during the follow-up, unless contraindicated. No initiation of a GLP1 agonist at baseline and continuation of no treatment during the follow-up, unless indicated.	Same as the target trial.
Treatment assignment	Random unblinded assignment at baseline.	Patients are classified as initiators or non-initiators according to what their data at baseline are compatible with. Randomisation is emulated by adjusting for baseline confounders.
Outcome	Major adverse liver outcomes, which will be a composite outcome including decompensated cirrhosis (variceal bleeding, ascites, portal hypertension or hepatorenal syndrome), hepatocellular carcinoma, need for liver transplantation or major adverse liver outcome-related death.	Same as the target trial.
Follow-up	Start (baseline) Any calendar month in which all eligibility criteria are met. Patients can be enrolled in several different target trials. End The calendar month of first outcome, emigration from Sweden, 10 years of follow-up or December 2020, whichever occurs first.	Same as the target trial.
Causal contrasts	Intention-to-treat and per-protocol effects.	Observational analogues of the intention-to-treat and per-protocol effects.
Statistical analyses	Intention-to-treat analysis Estimate the risk curves in each group defined by assigned treatment strategy via a parametric pooled logistic model with an indicator for treatment group, a flexible time-varying intercept, product terms between treatment group and time, and a trial indicator. Per-protocol analysis Same as above, but individuals will be censored when they deviate from their assigned strategy and inverse-probability weights will be applied to adjust for baseline and time-varying covariates associated with adherence. The effect of initiating GLP1 agonists will be studied in prespecified subgroups according to liver disease aetiology (MASLD, other than MASLD) and liver disease severity at baseline (compensated cirrhosis, no cirrhosis).	Same as the target trial with sequential emulation (starting in each calendar month of the study period) and additional adjustments for baseline covariates associated with treatment initiation in both the intention-to-treat and per-protocol analyses.

.GLP1, glucagon-like peptide-1 receptor; MASLD, metabolic dysfunction-associated steatotic liver disease.