*p < 0.0001 for change vs baseline. †Data from the EAS. ‡Data from the FAS. Primary analyses of change in HbA1c and primary and sensitivity analyses of change in body weight analysed using an adjusted ANCOVA model with baseline HbA1c (A only), body weight (B only), T2D duration, age, BMI, pre-initiation use of GLP-1RA (yes/no), pre-initiation use of DPP-4i (yes/no), pre-initiation use of insulin (yes/no), number of OADs use pre-initiation (0–1/2+) and sex as covariates. Sensitivity analyses of change in HbA1c used a mixed model for repeated measurements with baseline HbA1c, T2D duration, age, BMI, time and time-squared as covariates and pre-initiation use of GLP-1RA (yes/no), pre-initiation use of DPP-4i (yes/no), pre-initiation use of insulin (yes/no), number of OADs use pre-initiation (0–1/2+) and sex as fixed factors with random intercept and random coefficient for time (slope). On-treatment period represents the time period in which patients were considered treated with semaglutide. In-study period represents the time period during which patients were considered to be participating in the study, regardless of semaglutide treatment status. Mean (±SD) baseline values for HbA1c and body weight were 7.9 ± 1.3% (62.9 ± 14.2 mmol/mol) and 102.0 ± 21.2 kg, respectively, in the EAS and 7.9 ± 1.4% (62.9 ± 14.9 mmol/mol) and 101.5 ± 21.1 kg, respectively, in the FAS. Error bars represent upper and lower 95% CIs. n-Values indicate patients contributing to the analysis. ANCOVA, analysis of covariance; BMI, body mass index; CI, confidence interval; DPP-4i, dipeptidyl peptidase-4 inhibitor; EAS, effectiveness analysis set; EOS, end of study; FAS, full analysis set; GLP-1RA, glucagon-like peptide-1 receptor agonist; OAD, oral antidiabetic drug; T2D, type 2 diabetes.