Abstract
PURPOSE: Medication adherence plays an important role in glycemic control in type 2 diabetes mellitus (T2DM) self-management. To analyze the factors influencing medication adherence in T2DM patients and the effect of pharmacist-led interventions, we conducted a study in Beijing, China. PATIENTS AND METHODS: T2DM patients with hypoglycemic drugs for at least 6 months were enrolled. A pharmacist-led survey was conducted followed by individualized interventions for those non-adherent patients monthly within 3 months. FPG, HbA1c, and 2hPG were measured as the comprehensive glycemic control. Medication adherence was determined according to the patient's self-reported compliance with prescribed medication during the last 3 months. RESULTS: A total of 763 T2DM patients were included. The average age was 63.26±11.89 years, with 363 males. After pharmacist intervention, the patients with good adherence increased from 34.21% to 39.06%, while poor adherence decreased from 32.5% to 24.5% (p < 0.001). The average adherence score was a significant increase (p < 0.001) from 27.846±4.185 to 29.831±7.065. Furthermore, our study demonstrated that pharmacist-led interventions significantly increased glycemic control (FPG from 42.33% to 53.60%, p < 0.001; 2hPG from 41.68% to 48.75%, p = 0.005; HbA1c from 24.12% to 29.23%, p = 0.024). The results found that body mass index (OR 0.643, 95% CI 0.437-0.945), use of medications empirically (occasionally (OR=3.066, 95% CI 2.069-4.543); often (OR=2.984, 95% CI 1.107-8.044)), following the doctor's advice to visit (OR 2.129, 95% CI 1.079-4.202) and lifestyle compliance (OR 2.835, 95% CI 1.094-7.346) were the independent risk factors of non-adherence (p < 0.05), the area under the ROC curve was 0.716. CONCLUSION: Self-reported medication adherence and glycemic control in T2DM patients were poor which can be improved by pharmacist-led interventions. Interventions should focus on empirical medication behavior, non-adherence to lifestyle, and failure to follow the doctor's advice. The recall bias with self-reported results needs further objective data to verify.