OBJECTIVES: A key objective of treating type 2 diabetes mellitus (TD2M) is maintaining glycaemic control, (glycated haemoglobin-HbA1c) with targets set between 6.5%-7.5%. Doctors frequently fail to intensify treatment in uncontrolled patients and a study was designed to assess the long term clinical effects of this inertia.  The study involved Classification and Regression Tree (CART) analyses to evaluate the timepoint that intensification should occur to gain glycaemic control. METHODS: Incident T2DM patients were identified in the UK CPRD database between 1 January 2000-31 August 2014 and followed for 5 years. Patients initiated on metformin monotherapy and did not achieved HbA1c<7% after 90 days. CART was applied, with Hba1c level as the dependent variable and three explanatory effects: time to intensification class (TTIc), Year and Year by TTIc interaction. For each iteration TTIc was assigned per subject (i.e. those patients with TTI before or after TTI cut-point). The goodness of fit, the Akaike Information Criteria (AIC), was obtained for the model and the optimal cut point was the one resulting in the lowest AIC value.  This created two classes of subjects, rapid and delayed TTI, whose HbA1c values were compared RESULTS: The model identified that the optimal intensification time was ≤244 days post the first HbA1c≥7% (rapid TTI). 50.7% of patients with rapid TTI achieved HbA1c target<7%, compared to only 14.0% in delayed TTI, despite rapid patients starting with higher average HbA1c levels (8.65%) than delayed patients (7.85%). The effects of rapid intensification caused immediate and maintained reduction in glucose levels, not observed in the delayed group. CONCLUSIONS: CART analysis identified the optimal timing for treatment intensification post loss of glycaemic control as ≤244 days. This analytical method could be used in future database studies to aid in group definition by treatment exposure and provide valuable clinical information for physicians.