AbstractAimsTo examine whether early treatment intensification using dipeptidyl‐peptidase 4 inhibitors (DPP4i) delays insulin initiation in Chinese patients diagnosed with type 2 diabetes for less than 5 years.Materials and MethodsIn a territory‐wide prospective cohort study, patients with type 2 diabetes initiating DPP4i at diabetes duration <2 years (early intensification) and 3–5 years (late intensification) were matched using 1:1 propensity‐score matching (n = 908 in each arm). We used Cox regression to compare the risk of insulin initiation between the two groups. We explored the interactive and mediation effects of glycated haemoglobin (HbA1c) variability score (HVS), defined as the percentage of HbA1c varying by ≥0.5% compared with preceding values.ResultsOf 1816 patients (60.7% men, mean age 54.4 ± 11.9 years), 92.4% and 71.9% were treated with metformin and sulphonylureas respectively at DPP4i initiation. Early DPP4i intensification [hazard ratio (HR) 0.71, (95% CI 0.58–0.68)] and low HVS (<50%) (HR = 0.40, 0.33–0.50) were associated with delayed insulin initiation during a median 4.08 years of follow‐up. Early intensification with low HVS had the lowest risk versus late intensification with high HVS (HR = 0.30, 0.22–0.40) (p_interaction = 0.013). HVS mediated 19.5% of the total effect of early DPP4i intensification on delaying insulin initiation. The late and early intensification groups had similar HbA1c at month 0 (8.4 ± 1.3% vs. 8.4 ± 1.5%) and month 3 (7.6 ± 1.2% vs. 7.6 ± 1.3%) after DPP4i initiation. By month 12, HbA1c in the late intensification group deteriorated (7.9 ± 1.4%) but remained stable in the early intensification group (7.6 ± 1.4%, p = 0.001) with persistent between‐group difference over 72 months (8.2 ± 1.7% vs. 7.7 ± 1.6%, p = 0.001).ConclusionsIn type 2 diabetes, early DPP4i intensification delayed insulin initiation, partially explained by reduced glycaemic variability.