The aim was to estimate the association between adherence to lipid-lowering drugs and LDL-cholesterol (LDL). The study included 2,973 patients with diabetes who met the following criteria during the 2003–2004 period: at least 2 separate fills for a lipid-lowering drug per year; 18 years or older; at least 1 laboratory test for LDL; and continuous enrollment in the health plan. A continuous measure of medication gaps (CMG) was constructed using claims data. Adherence was measured as %(1-CMG). In 2005, patients participated in a mixed-mode survey that included measures of selfreported adherence, mediators of adherence behavior, and self-reported less use of medication because of cost. 2,038 patients (69%) completed the survey. Patients with adherence 80% (claims-based) or 85% (selfreported) were classified as adherent. LDL outcomes were evaluated by the average LDL, the difference between the first and the last LDL test result available, and by LDL goal achievement (LDL < 100 mg/dl). GEE linear and logistic regression multivariable models included socio-demographic, clinical, and behavioral variables. Both self-reported and claims-based measures of adherence were significant (P < 0.05)predictors of LDL in multivariable linear models. Adherent patients were more likely to achieve LDL goal than nonadherent patients (odds ratio and 95% CI: 2.5 [1.5–4.1] for self-reported measures and 1.96 [1.5–2.6] for claims-based measures, respectively). Other significant predictors of goal achievement included race/ethnicity, age, gender, number of physician visits, number of lipidlowering drugs, number of LDL tests, and self-reported less use of medication because of cost (NIDDK # 5R01DK64695-3)
