A Claims Database Analysis of Dose-Dependency of Metformin and Incidence of Lactic Acidosis in Japanese Patients with Type 2 Diabetes

Abstract

Introduction: Patients with type 2 diabetes (T2D) in Japan are prescribed a lower dose of metformin that their counterparts in Western countries due to concerns for the risk of lactic acidosis incidence. Here we report our study on the association between high-dose metformin administration and the incidence of lactic acidosis in Japanese patients with T2D. Methods: A Japanese claims database (April 2008-November 2018) was analyzed. Factors associated with the incidence of lactic acidosis were first identified from the database records by conducting a case-control study, and these were then used as confounding factors in subsequent analyses. The association between high-dose metformin administration (C 1000 mg/day) and the incidence of lactic acidosis was compared with that between lowdose metformin (\ 1000 mg/day) or no metformin administration and lactic acidosis incidence by using the following approaches: a logistic regression analysis hypothesizing that metformin-associated lactic acidosis is short term; a time-dependent proportional hazard model hypothesizing that the influence of metformin is cumulative; and a case-control study in which lactic acidosis incidence was the case and metformin administration within 3 months prior to the incidence of lactic acidosis (or corresponding date for the control) was the exposure. Results: Prescriptions for biguanide and vitamin B complex and volume depletion were identified as factors associated with the incidence of lactic acidosis. The incidence rate was higher in patients prescribed metformin than in those not receiving metformin; however, it was not higher in those prescribed high-dose metformin compared to those prescribed low-dose metformin. The estimated regression coefficient for high-dose metformin administration was 0.816 (p \ 0.001); this was not higher than those for low-dose metformin (1.047), vitamin B complex (2.725) and volume depletion (3.301). The time-dependent proportional hazard analysis did not indicate any effect of metformin prescription.