Glycemic and weight changes after persistent use of incident oral diabetes therapy: a Veterans Administration retrospective cohort study.

PURPOSE: Systematic reviews have reported that sulfonylureas and metformin were as effective in reducing hemoglobin A1c (A1C) as other oral antidiabetic drugs (OADs) in clinical trial populations. Data on comparative effectiveness of OADs in other populations is limited. The objective was to compare the effectiveness of incident OAD regimens in reducing A1C and to compare the effect of OADs on body mass index (BMI).
METHODS: Retrospective cohort study using data from the Veterans Affairs Mid-South network (2001-2007). Of 18 205 veterans who filled 19 511 incident OAD prescriptions, 2096 had complete covariates, persisted on their incident treatment for 12 months, and had baseline and 12-month A1C values. For the BMI analysis, 2484 patients had complete information. Incident OAD regimens included metformin and sulfonylureas. Primary outcomes were 12-month A1C and BMI, which were compared controlling for demographic characteristics, baseline A1C and BMI, psychiatric diagnoses, and healthcare utilization.
RESULTS: Median [interquartile range (IQR)] A1C decreased from 7.1% [6.5, 7.8] at baseline to 6.5% [6.0, 7.0] at 12 months. Twelve month-A1C in sulfonylurea users was similar to metformin users. The median [IQR] BMI decreased from 31.1 [27.8, 34.9] to 30.7 [27.5, 34.5] kg/m(2). Sulfonylureas were associated with a significantly higher 12-month BMI than metformin (12-month adjusted mean difference: 1.05 kg/m(2), 95%CI: 0.90-1.20, p < 0.0001).
CONCLUSIONS: These analyses support the use of metformin as first choice of OAD because of similar glycemic control but improved BMI when compared to sulfonylureas.