Office of the Assistant Secretary of Defense for Health Affairs/TRICARE Management Activity

1. A Comparison of Select Cardiovascular Outcomes by Anti-Diabetic Prescription Drug Classes Used to Treat Type 2 Diabetes among Military Health System (MHS) Beneficiaries, Fiscal Year 2003-2006 Office of the Assistant Secretary of Defense for Health Affairs/TRICARE Management Activity Michael R. Peterson, DVM, MPH, DrPH Director, Health Program Analysis and Evaluation Thomas Bacon, PharmD, MS LtCol, USAF, BSC Director, Pharmacy Utilization Management

2. Background • May 2007 article in NEJM by Nissen and Wolski reporting on association of Avandia use and risk of MI and death from cardiovascular causes • DoD interested in determining very quickly if our beneficiaries were at similar risk • Is there an increased incidence of select cardiovascular events (acute MI and CHF) among MHS beneficiaries who filled a script for Avandia compared to those who filled scripts for other anti-diabetic medications?

3. Methods • MHS has 9.1M beneficiaries • TRICARE Prime is an HMO-like benefit (65+ not eligible) – study limited to these beneficiaries • Cross-sectional analyses of data from fiscal years 2003-2006

4. Methods (cont.) • Data sources: • Defense Eligibility Enrollment System (DEERS) – gender/age/bencat/location • Inpatient/outpatient encounter claims – not clinical data – date of service/source of service/diagnoses (ICD-9)/procedure codes (DRG) • Pharmacy Data Transaction Service (PDTS) – real-time pharmacy fill record regardless of source of fill (military, mail order, civilian)

5. Methods (cont.) • Data sources linked by identifiers • Case definition for type 2 diabetics based on ICD-9 codes • Individual drugs grouped into therapeutic classes of anti-diabetic drugs • Drug categories not mutually exclusive (therefore unable to perform statistical comparisons) • Date of earliest diagnosis used to identify incident cases

8. Results • Table 1 • 232,000 Prime enrollees with dx of type 2 diabetes between fiscal year 2003 – 2006 • 54% female; 70% 45 – 64 years old; 8% active duty; 15% dependents of active duty; 54% retirees and their dependents; 21% missing beneficiary information

9. Results (cont.) • Table 2 • Annual incidence rates averaged over 4 years • AMI: range from 33/10,000 (biguanides) to 52/10,000 (insulin); 38/10,000 for Avandia, 35/10,000 for Actos, 37/10,000 for any TZD class of drugs, 49/10,000 for sulfonylureas

10. Results (cont.) • CHF: range from 68/10,000 (biguanides) to 114/10,000 (insulin); 86/10,000 for Avandia, 93/10,000 for Actos, 90/10,000 for any TZD class of drugs, 105/10,000 for sulfonylureas • Bottom line: No increased incidence of select cardiovascular events (acute MI and CHF) among MHS Prime beneficiaries who filled a script for Avandia compared to those who filled scripts for other anti-diabetic medications

11. Limitations • Cannot determine from these data if differences in average annual incidence rates of outcomes are due to anti-diabetic drug; disease progression; number, types and severity of comorbidities; or, other risk factors such as age • Study did not include those 65 and older • Claims data do not include socioeconomic status, comorbid conditions, current health status, medical history, drug dose, time on drug, concurrent medications, individual characteristics (e.g. BMI)

12. Limitations (cont.) • AMI/CHF attributed to anti-diabetic class if script filled any time prior to event (assumes cause-effect relationship) • Because drug categories were not mutually exclusive, unable to make statistical comparisons for significance

13. Future Efforts • Possible studies • Retrospective cohort study to include medical record review/time to event analysis • Prospective study of cohort of type 2 diabetics with reference cohort and appropriate uniform medical evaluation • DoD/FDA working on an MOU to permit surveillance of DoD pharmacy data to track sentinel events