Safety and efficacy of metformin in a restricted formulary

To assess the efficacy and safety of metformin (MET) in the Veterans' Administration Northern California Health Care System during the period from June 1995 through April 1996 when its use required approval by Endocrinology.

A retrospective review of patient charts and computerized pharmacy and laboratory records. Patients served as their own historical controls.

Patients receiving MET (n = 251) were identified from the pharmacy database. On-line laboratory data, including the intermediate outcome variable HbA1c, were retrieved by computer for the interval 4 months prior to the initial prescription to May 1996. Clinical data including weight and blood pressure were obtained from chart review.

Of 228 patients whose charts were available for review, 29 reported side effects, and 12 discontinued use due to these side effects. No patients were identified with lactic acidosis. Both baseline and treatment data on HbA1c were available on 164 patients. Mean HbA1c (%) data (unpaired), expressed as mean +/- SE, were as follows: between 4 months pretreatment and 1 month pretreatment, 9.41 +/- 0.19 (n = 103 tests); between 1 month pretreatment and baseline, 9.41 +/- 0.19 (n = 110 tests); 3 months of treatment, 8.79 +/- 0.16 (n = 157 tests, P < 0.05); 6 months of treatment, 8.30 +/- 0.17 (n = 79 tests, P < 0.0001); 9 months of treatment, 8.72 +/- 0.24 (n = 70 tests, P < 0.05), compared to pretreatment values. Similar analysis of unpaired weight and blood pressure data in 152 patients did not reveal any reduction in these clinical parameters over this treatment period. Serum lipids were unchanged on treatment (by paired analysis), but the number of tests was limited.

In this setting, MET provided sustained beneficial effects on glycemic control and was well tolerated. Any effects on weight, blood pressure, and serum lipids were not demonstrable in this analysis. We conclude that MET can substantially improve outcome of diabetes care.