RESUMO
OBJECTIVE: Therapeutic guidelines recommend the combination of drugs as necessary to control type 2 diabetes (T2D). This research assessed the effectiveness of pioglitazone (Pio), metformin (Met) and sulfonylurea (SU) combinations in the routine clinical practice. RESEARCH DESIGN AND METHODS: A nationwide, 12-month prospective, observational cohort study was performed in 2294 patients with T2D (50.3% females, mean age: 61.1 years, mean body mass index: 30.2 kg/m(2), mean time since diagnosis: 8.5 years) who started, at the discretion of treating physician, oral antihyperglycaemic treatment with either Pio + SU, Pio + Met or SU + Met because of inadequate control with previous therapy. Fasting plasma glucose (FPG), glycohaemoglobin (HbA1c), lipids, blood pressure, and anthropometric parameters were measured, and 10-year cardiovascular risk was estimated. RESULTS: FPG, HbA1c and total cholesterol at baseline had mean values (184.6 mg/dl, 8.5% and 246.0 mg/dl, respectively) associated with an excess of micro- and macrovascular risk. The mean changes from baseline in the Pio + SU, Pio + Met and SU + Met cohorts were, respectively, -37.9, -32.7 and -25.8 mg/dl for FPG; -1.1, -1.0 and -0.7% for HbA1c; -30.7, -38.7 and -17.1 mg/dl for triglycerides; and +2.3, +2.5 and +0.6 mg/dl for HDL cholesterol. In consequence, the estimated 10-year cardiovascular risk decreased more in the Pio cohorts, particularly with Pio + Met (1.7% versus 1.4% Pio + SU and 1.0% SU + Met -Framingham equation- and 0.6% versus 0.4% SU + Met - Systematic Coronary Risk Evaluation model-). Related adverse events were significantly (p = 0.016) more frequent in Pio cohorts (4.7% with Pio + SU, 5.1% with Pio + Met) than in the SU + Met cohort (2.4%). CONCLUSIONS: In patients with T2D failing therapy, mostly SU or Met monotherapy, pioglitazone add-on treatment was associated with a significant improvement of micro- and macrovascular risk estimations. These results from real-life clinical conditions support the findings of prior randomised trials, although they should be interpreted with caution because of the observational, nonrandomised design.