Background: As part of a clinical decision support (CDS) and pharmacist outreach interventional study to improve medication adherence, we analyzed rates and clinical implications of poor adherence for diabetes (DM), hypertension (HTN) and lipid control. Methods: In two high performing care delivery systems in MN, 28 clinics were randomized to an adherence CDS intervention or control. The CDS uses web-based algorithms to assess clinical data and objectively capture poor adherence using an Epic based measurement of Proportion of Days Covered (ePDC) calculated using SureScripts dispense data with a moderate or high confidence rating. For 16 weeks in 2019, the CDS ran silently in all primary care encounters of patients age 18-75 to determine how many patients qualified for study eligibility using the following criteria:
1. DM cohort: A1c ≥8% with a non-insulin glycemic medication ePDC < 80%.
2. HTN cohort: BP ≥140/90 on two separate consecutive dates with a BP medication ePDC < 80%.
3. Lipid cohort: ACC/AHA criteria for statin use, on a statin with ePDC < 80%, excluding for LDL < 70 mg/dl.
Results: Of 142,879 patients, 138,252 remained after excluding for cancer, pregnancy, hospice or nursing home. A non-insulin DM medication was identified in 15,546 (11%). Of these, 3890 (25%) had A1C ≥ 8%, and of these 1467 (59%) had poor adherence. Among the DM cohort, the mean A1C was 9.6% and 30% had A1C ≥10%. Of patients identified with DM in the HTN cohort (494/1711, 29%), 60% had a systolic blood pressure ≥ 150 mm Hg. Of patients identified with DM in the lipid cohort (2298/4564, 50%), 19% had an LDL ≥130 mg/dl. Conclusion: Over 14 weeks, large numbers of DM patients were identified at primary care encounters with difficulty achieving A1C, BP and lipid goals in association with poor medication adherence, which was often unrecognized. Based on extremely high A1C, BP and LDL levels in these patients, there is a tremendous clinical opportunity to identify and address poor adherence using CDS and/or pharmacist care team involvement.