The effect of incident antidiabetic regimens on lipid profiles in veterans with type 2 diabetes: a retrospective cohort

Abstract

Objective: Effects of oral antidiabetic drugs (OADs) on lipids may influence cardiovascular outcomes. Our aim was to compare time to initiation of lipid lowering medication (LLM) and 12-month lipid profiles among new OAD users.

Methods: We identified a retrospective cohort of 17,774 veterans who received care at Veterans Administration (VA) Mid-South Network with a first OAD from 1 January 2000 to 31 December 2007. There were 6917 patients (38.9%) not on a LLM at baseline, and 3871 (56%) had complete covariates. Incident users of sulfonylurea and combination metformin + sulfonylurea were compared to metformin users for time to LLM initiation. Incident users of these OADs and thiazolidendiones were included in comparison of 12-month low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides (TGs), and total cholesterol. All analyses adjusted for demographics, lipids, HbA1C, healthcare utilization, and cardiovascular disease at baseline.

Results: The median time to starting LLM was 2.35 years (interquartile range 0.96, 4.6) following metformin initiation and not statistically different for users of sulfonylureas, or combination OADs. Compared to metformin users, 12-month HDL was 1.35 mg/dl (95%CI: -2.01, -0.72) lower and TGs were 5.7% higher (95%CI: 1.5%, 10.0%) for sulfonylurea users; TGs were 24.8% (95%CI: 0.7%, 54.5%) higher for thiazolidinedione users. Statin users had LDL and total cholesterol 16.7 mg/dl (95%CI: -19.9, -13.5) and 18.6 mg/dl (95%CI: -22.1, -15.1) lower than non-statin users, respectively.

Conclusions: Time to LLM initiation was similar between OADs. Metformin use resulted in more favorable lipids at 12 months compared to sulfonylureas or thiazolidinediones.

Figure 1.

Patient flow and eligibility for study cohort

Figure 2.

Cumulative incidence of prescription for a new lipid lowering medication among each OAD exposure group.

Figure 3.

Forest plots of difference in each lipid level compared to the average metformin user for each OAD exposure and with the use of statin and non-statin lipid lowering medication use. *The intercept represents an average white male metformin user without cardiovascular disease, not using any lipid lowering medication, who entered the cohort in 2000 and adjusted for the average baseline HbA1c, total cholesterol, HDL, and triglycerides