Abstract 123: Replicating the Action to Control Cardiovascular Risk in Diabetes (ACCORD) Trial in Real-World Claims Data

Background: The ACCORD-Lipid Trial’s finding that statin-fibrate combination therapy (CT) provides no incremental cardiovascular risk reduction in type 2 diabetese over statin monotherapy (MT) prompted FDA to issue a public communication on 11/9/2011 stating that fenofibrate may not reduce risk of heart attack or stroke. Yet, fibrate use continues unabated with over $1 billion in sales in the US that raises concern regarding the inconsistency in diffusion of scientific evidence into clinical practice. Critics of ACCORD findings maintain that ACCORD trial adopted flexible thresholds for qualifying HDL and triglyceride levels and that it left unanswered whether the effects of non-ACCORD statins or fibrates or combinations thereof will be different. By replicating the ACCORD-Lipid trial as closely as possible using 16-year longitudinal claims database from a large national health plan, our study seeks to compare the following ACCORD outcomes between CT and MT cohorts: (i) primary composite outcome of nonfatal MI, nonfatal stroke and cardiovascular death; (ii) secondary outcomes of all-cause mortality, expanded macrovascular outcome, major CAD events and CHF. Methods (Research Design, Data Source and Data Analysis Methods) : Retrospective claims analysis that included patients enrolled between 1995 and 2010 using ACCORD inclusion/exclusion criteria including type 2 diabetes patients aged 40 to 79 with baseline A1C≥7.5 and on statin. Patients in the two study cohorts, CT and MT, were required to have minimum of 1-year baseline and 90-day follow-up periods. Propensity score (PS) matching was used to adjust for patient baseline characteristics. T- and Chi-squared tests were used to assess differences in continuous and categorical covariates and Cox proportional hazard model was used to assess the hazard of study events. Results: The study included 6765 patients (CT=954; MT=5811) with a mean follow-up of 2.4 years. An average patient in the sample was a White male aged 57 years from the South. The two study cohorts differed in demographics (age, female, ethnicity, income categories), baseline lipids (HDL, LDL, triglyceride and HbA1c), and in numerous comorbid conditions. After 1-to-1 PS matching, baseline LDL, triglycerides and total cholesterol were similar but HDL (HbA1c) was higher (lower) in CT than in MT cohort (n=943 in each cohort). Most other baseline covariates were balanced. Unadjusted results showed that compared to MT, CT cohort had higher primary composite outcomes (62 vs 45), all-cause deaths (113 vs 105), macrovascular events (16 vs 9), major CAD (84 vs 59), nonfatal stroke (35 vs 33) and CHF (60 vs 51). Adjusted results show no difference in the rate of primary composite outcome (hazard rate or HR=1.44, p=0.09) and in secondary outcomes of macrovascular events (HR=1.61), all-cause mortality (HR=1.22), major CAD (HR=1.45), CHF (HR=1.24) and non-fatal stroke (HR=0.98) [all p>.05] Conclusion: The study results appear to confirm the non-significance of CT over MT in cardiovascular risk reduction among type 2 diabetes patients in a large US commercial health plan.