Purpose: To test the hypothesis that cumulative exposure to sulphonylurea (SU) or metformin (MET) have different effects on mortality when taken as a replacement or add-on of one for the other.
Methods: All consecutive diabetes patients aged over 20 years were screened at their first diabetes outpatient visit between 2001 and 2008 (n = 79869). Only patients on MET (n = 11374) or SU (n = 18502) monotherapy were retained. All patients were followed up for death until December 31, 2011, but censored at first exposure to anything else besides MET/SU. Adjusted time-dependent Cox regression and competing risk regression analysis, with daily updates of treatment modalities were performed.
Results: Mean age was 62.1 ± 11.2 years and follow-up was 4.6 ± 3.2 years (138496 person-years). Adjusted all-cause and cardiovascular mortality rates were significantly higher in MET as compared with SU group. All-cause mortality hazard ratios (HR) for cumulative time exposure were as follows: HR 0.956 (95%CI 0.951-0.962, p < 0.001) for SU added to MET, HR 1.092 (95%CI 1.087-1.096, p < 0.001) for SU replacing MET, HR 0.979 (95%CI 0.975-0.983, p < 0.001) for MET added to SU, and HR 1.127 (95%CI 1.118-1.136, p < 0.001) for MET replacing SU.
Conclusion(s): The effect on all-cause mortality was beneficial for MET+SU combined therapy, but deleterious for either SU replacing MET, or MET replacing SU. There were no major outcome differences when analyzing individual SU, or specific mortality.
Keywords: Metformin; mortality; sulphonylurea.