Changing Incident Diabetes Regimens
A Veterans Administration cohort study from 2000 to 2005
- Mary Margaret Huizinga, MD12,
- Christianne L. Roumie, MD12,
- Tom A. Elasy, MD123,
- Harvey J. Murff, MD124,
- Robert Greevy, PHD15,
- Xulei Liu, MS15,
- Theodore Speroff, PHD12356 and
- Marie Griffin, MD126
- 1Veterans Administration Tennessee Valley Healthcare System, Nashville, Tennessee
- 2Division of General Internal Medicine and Public Health, Vanderbilt University Medical Center, Nashville, Tennessee
- 3Diabetes Research and Training Center, Vanderbilt University Medical Center, Nashville, Tennessee
- 4Vanderbilt Epidemiology Center, Vanderbilt University Medical Center, Nashville, Tennessee
- 5Biostatistics, Vanderbilt University Medical Center, Nashville, Tennessee
- 6Preventive Medicine, Vanderbilt University Medical Center, Nashville, Tennessee
- Address correspondence to Mary Margaret Huizinga, MD, Vanderbilt University Medical Center, 1215 21st Ave. South, 6100 Medical Center East, North Tower, Nashville, TN 37232. E-mail:
Introduced in the mid-1950s, sulfonylureas have historically been the first line of treatment for type 2 diabetes (1). Metformin was introduced to the U.S. market in 1994 and has since experienced increasing market share (2). Metformin has many advantages over the sulfonylureas, including decreased weight gain and less risk of hypoglycemia (3). Despite some debate, the American Diabetes Association (ADA) recommended in 2006 that metformin be used as the first-line agent unless contraindicated (4). Whereas metformin use as a first-line agent was increasing before the ADA recommendations, no clear estimate of incident glycemic control regimens is available.
A retrospective cohort of patients receiving a glucose-lowering medication from 1 October 1999 to 30 June 2005 was constructed from the administrative and pharmacy databases of the Veterans Administration (VA) Mid-South Network. For veterans who were also eligible for Medicare, these data were obtained through the VA Information Resource Center (VIReC) and merged with the analytical database (5).
Over 65,000 patients receiving glucose-lowering medications were identified. From this initial cohort, incident glucose-lowering medication prescriptions were delineated. Incident prescriptions were defined as the first glucose-lowering medication prescription after at least 365 days of active use of the VA pharmacy services without a glucose-lowering agent. The five most common drug exposure categories identified were metformin, sulfonylurea, thiazolidinedione, insulin, and metformin plus sulfonylureas. The number of incident prescriptions of each drug exposure category during calendar years 2000 to 2005 was determined.
A total of 13,333 incident glucose-lowering prescriptions in the five exposure categories were identified between 2000 and 2005. Of these, 45% were metformin (n = 6,057), 41% were a sulfonylurea (n = 5,405), 1.4% were a thiazolidinedione (n = 180), 9.2% were insulin (n = 1,223), and 3.5% were a combination of metformin and a sulfonylurea (n = 468). Mean age ± SD was 63.6 ± 11.6 years, 3.5% were female (n = 467), and 13% (n = 1,727) were nonwhite. Mean A1C was 7.6 ± 1.9%, and mean BMI was 31.3 ± 6.2 kg/m2 at the time of the incident prescription.
From 2000 to 2005, metformin increased from 29 to 63% of the incident diabetes prescriptions, whereas sulfonylureas decreased from 57 to 21% (P < 0.0001, χ2 test). Insulin use increased from 8 to 11%, whereas the use of thiazolidinediones and metformin plus sulfonylurea as the incident diabetes regimen remained stable during this time.
For many years, sulfonylureas and insulin were the only treatment options for diabetes. Metformin has largely replaced sulfonylureas as the first-line agent in the treatment of type 2 diabetes in this VA population. Other less common incident diabetes prescriptions in this population included thiazolidinediones and dual metformin and sulfonylurea therapy.
We have demonstrated a significantly increased use of metformin as an incident glycemic control medication before the 2006 ADA recommendation. Further studies that compare the effectiveness of these regimens as first-line agents are needed to guide providers in the selection of the initial treatment of type 2 diabetes. These studies should include adherence, persistence, time to control, time in control, and other salutary or harmful features.
- DIABETES CARE