AimTo compare the risks of hospitalization for heart failure (HHF) associated with sulfonylurea (SU), dipeptidyl peptidase-4 inhibitor (DPP-4i), and thiazolidinedione (TZD) as add-on medications to metformin (MET) therapy using the data of Korean adults with type-2 diabetes from the Korean National Health Insurance database.MethodsWe identified 98,383 people who received SU (n = 42,683), DPP-4i (n = 50,310), or TZD (n = 5,390) added to initial treatment of MET monotherapy in patients with type-2 diabetes. The main outcome was the hospitalization for HHF. Hazard ratios for HHF by type of second-line glucose-lowering medication were estimated by Cox-proportional hazard models. Sex, age, duration of MET monotherapy, Charlson Comorbidity Index and additional comorbidities, and calendar year were controlled as potential confounders.ResultsThe observed numbers (rate per 100,000 person-years) of HHF events were 1,129 (658) for MET+SU users, 710 (455) for MET+DPP-4i users, and 110 (570) for MET+TZD users. Compared to that for MET+SU users (reference group), the adjusted hazard ratios for HHF events were 0.76 (95% confidence interval 0.69–0.84) for MET+DPP-4i users and 0.96 (95% confidence interval 0.79–1.17) for MET+TZD users.ConclusionDPP-4i as an add-on therapy to MET may lower the risks of HHF compared with SU.