Association of diabetes medication adherence quality measure performance with A1c control among Medicare Advantage...

The Centers for Medicare & Medicaid Services provide financial incentives for insurance plans demonstrating high-quality care to improve patient outcomes. Medicare Advantage (MA) plan performance is assessed by nearly 40 quality measures, with diabetes care heavily represented by 5 of those measures. Medication Adherence for Diabetes (MAD) and Hemoglobin A1c Control for Patients with Diabetes (HBD) are emphasized as triple-weighted measures that strongly impact the overall quality rating. The national average ratings for MA Prescription Drug (MA-PD) plans from 2022 to 2026 show a flat average pass rate 5 years in a row for the MAD measure but a steadily increasing pass rate each year for the HBD measure. Observation of this trend, along with rising 4- and 5-star cut points, raise questions about whether the MAD measure methodology currently contributes to the achievement of clinical outcomes in patients with diabetes and whether it guides health care resources to the population in need.

To identify the relationship between MAD performance and hemoglobin A1c (A1c) control. Secondary objectives were to evaluate the relationship between MAD and additional diabetes care quality measures and to evaluate factors associated with MAD measure performance.

This cross-sectional study included Aetna MA beneficiaries attributed to Prisma Health primary care providers participating in the inVio Health Network who indexed into the MAD measure during the 2023 measurement year. Data were collected from January 1, 2023, to December 31, 2023. The primary outcome was glycemic control, defined as achievement of A1c of less than 7%. Secondary outcomes included HBD, Eye Exam for Patients with Diabetes (EED), Statin Use in Persons with Diabetes (SUPD), and Kidney Health Evaluation for Patients with Diabetes (KED). Associations between MAD performance with A1c and diabetes quality measures were evaluated using descriptive statistics, χ² tests, and multivariable logistic regression. Independent variables included demographic characteristics, health care use measures, prior-year MAD status, day-supply category, and index medication class; these factors were evaluated as predictors of MAD performance and included as covariates in adjusted logistic regression models.

The final study population included 1,386 beneficiaries (mean age = 72.3 years; 50.5% male). Among beneficiaries with available A1c values (n = 1,278), mean A1c did not differ significantly between those who passed and those who failed the MAD measure (6.80 ± 0.98% vs 6.66 ± 1.16%; P = 0.114). In adjusted analyses, passing the MAD measure was not associated with achieving A1c less than 7% (adjusted odds ratio [aOR] = 0.87 [95% CI = 0.60-1.27]; P = 0.474). Similarly, passing the MAD measure was not associated with HBD or SUPD performance but was associated with higher odds of passing EED (aOR = 1.89 [95% CI = 1.12-3.19]; P = 0.017) and KED (aOR = 1.65 [95% CI = 1.10-2.47]; P = 0.015). Prior-year MAD pass (aOR = 4.04 [95% CI = 2.43-6.72]; P < 0.001) and longer medication day supply (≥84 vs 28-<56 days; aOR = 3.90 [95% CI = 2.49-6.12]; P < 0.001) were strongly associated with passing the MAD measure. In a predefined subgroup analysis evaluating polypharmacy, dual therapy (aOR = 2.30 [95% CI = 1.18-4.47]; P = 0.014) and triple-or-more therapy (aOR = 5.08 [95% CI = 2.23-11.56]; P < 0.001) were associated with higher odds of passing the MAD measure compared with monotherapy.

MAD performance was not independently associated with better glycemic control in our study population. The current proportion of days covered-based MAD measure methodology may not capture true medication adherence or its clinical impact and warrants reevaluation.