Objective: Metformin has been associated with the serious side-effect lactic acidosis. However, it remains unclear whether the use of metformin was a cause or a coincidence in lactic acidosis.
Design: A literature search of the Index Medicus (1959-66) and of the databases Embase, Medline, Medline Express (1966-99) was performed using the keywords metformin, biguanides and lactic acidosis. All articles of cases with metformin-induced lactic acidosis (MILA) were cross-referenced.
Subjects: Cases were included for analysis if they met the following criteria: serum pH < or =7.35, lactate concentration > or =5 mmol L(-1).
Intervention: A forum of six experts in intensive care medicine independently categorized the cases in MILA unlikely (score 0), possible MILA (score 1) or probable MILA (score 2).
Main outcome measures: Statistical analysis included the paired interobserver agreement (kappa) and multivariate regression analysis.
Results: Of 80 reported cases, 33 were excluded because of insufficient quality. The forum scores of the remaining 47 cases were distributed normally with a mean score of 7 (range 2-10). The kappa-value was 0.041 (SD = 0.24, range -0.514, 0.427). Neither lactate concentration nor mortality correlated with serum metformin concentrations.
Conclusions: Given the low interobserver agreement and the lack of any relationship between metformin levels and outcome parameters, the concept that there is a simple, causal relationship between metformin use and lactic acidosis in diabetic patients has to be reconsidered.