OBJECTIVES: To demonstrate how predictive models can improve the efficiency of medication adherence programs.

METHODS: This observational study includes 13,749 and 15,420 Medicare Part D patients from 30 randomly assigned test and control Walgreens pharmacies respectively. All patients had at least 1 script filled in any of the 3 CMS star drug classes (cholesterol, hypertension or diabetes) and 1 or more script that triggered late-to-refill (LTR) and/or new-to-therapy (NTT) pharmacist consultation calls from 1/1/2017 to 8/31/2017. Previous studies showed that both LTR and NTT significantly improved medication adherence. For the test group, patients predicted to be adherent or to have low probability of becoming adherent are placed on a no-call-list (NCL) and will not receive interventions. The predicted probabilities of medication adherence are derived from multinomial logistic regression which include age, gender, estimated socioeconomic status, number of comorbidities, prescription attributes (90-day supply, number of refills remaining, copay), recency of pharmacy visits, exposure period, and prior 1-year adherence to cholesterol, hypertension or diabetes medications. Medication adherence is compared between test and control groups in terms of proportion of days covered (PDC) and optimal adherence (PDC ≥ 80%), using Walgreens pharmacy claims data. Also estimated are suppressed calls due to model directed NCL and corresponding labor savings. Student’s t-tests and Chi-square tests are used to evaluate statistical significance of group differences.

RESULTS: Both average PDC and percent adherent patient are not statistically different between the test and control group for all three drug classes. The predictive model suppressed 40% of pharmacist calls and saved intervention resources with no negative impact on population-based adherence levels.

CONCLUSIONS: Predictive model directed interventions can improve the efficiency of established medication adherence programs with significant savings in labor resources, so the pharmacy staff can focus on other key clinical interventions to improve patient care.