Impact of Provider Mailings on Medication Adherence in Medicare Part D Patients
The Centers for Medicare and Medicaid (CMS) incentivize Medicare Part D plans to improve adherence by using a quality based payment system, the Medicare Star ratings. Adherence metrics account for 28% of a Part D plan’s overall Star rating. The objective of this study was to assess the impact of provider mailings on medication adherence among non-adherent patients enrolled in Part D prescription drug plans administered by a large national pharmacy benefits manager (PBM).
Data/Methods
Non-adherent members in the three Medicare Star rating adherence drug classes – oral diabetes medications (antidiabetics), cholesterol reducing medications (statins) and hypertension medications (renin angiotensin system or RAS antagonists) – were identified from over 600,000 continuously enrolled Medicare members’ prescriptions claims data based on whether the CMS adherence metric of the proportion of drugs covered (PDC) was less than 80% in the previous 12-month period (baseline). The PBM sent letters to the prescribing physicians of the non-adherent members requesting they discuss with their patients barriers to adherence. Letters were sent in the 4th quarter of 2011. The index date for each member was set at mailing date + 2 weeks. Baseline adherence (PDC) was assessed was from 10/1/2010 through 9/30/2011. The post-intervention PDC was calculated during the 12-month following the index date (follow-up). Analyzable members were: (i) age ≥ 18; (ii) had ≥1 drug claims in both baseline and follow-up
A historical control cohort was constructed from the PBM that satisfied the same eligibility criteria as applied to the intervention cohort with the timeline pushed back by 1 year (index date set at 10/01/2010)
Both a binary measure of adherence (whether PDC≥80%) and the continuous PDC were assessed as outcome measures. Multivariate logistic regression and a difference-in-difference (DID) regression were used for the binary and continuous adherence measures, respectively, both of which adjusted for potential baseline confounders including age, gender, pharmacy risk grouper score (proxy for severity), zip code level income, education and race variables, member cost share (copay). The analyses were conducted separately for each of the three drug classes.
Results
The final sample sizes for the three drug classes were 21,044 (Intervention=10,707; Control=10,337) for antidiabetics, 106,892 (Intervention=53,957; Control=52,872) for statins, and 73,560 (Intervention=36,706; Control=36,854) for RAS antagonists. The baseline PDCs between the intervention and control cohorts were not statistically different for all the three drug cohorts indicating that the intervention and control cohorts were similar in terms their baseline adherence to the study medications. Logistic regression results indicate that physician mailing was associated with 11%, 16%, and 7% improvement in adherence in members in the antidiabetic, statin and RAS antagonist cohorts compared with members from the pre-mailing period, all p<0.001. As indicated by the DID models, the mailing program was associated with average improvement of 1.27, 1.49 and 0.59 percentage points in continuous PDC, all p<0.001 for the antidiabetics, statin and RAS antagonist cohort, respectively.
Conclusion
The physician mailing program of a large U.S. PBM was associated with improved medication adherence in Medicare Part D enrollees in three drug classes: oral antidiabetics, statins and RAS antagonists.