Abstract
Introduction
It is estimated that medication nonadherence is responsible for 10% of all hospitalizations and up to 16% of total health care expenditures annually in the United States. Therefore, payers and health care systems have implemented programs to improve medication adherence among patients with chronic diseases. One such intervention is the incorporation of medication adherence metric in the Star Rating System for Quality Bonus Program, which was established by the Centers for Medicare and Medicaid Services (CMS) for rating of Medicare Advantage (MA) and Medicare Part D plans (PDP), and adjusting their bonus and benchmark payments.
The Star Rating System grades the plans based on a number of process and outcome measures, and the results are published by the CMS annually on its website to allow the beneficiaries to make informed decision when selecting MA and PDP plans. If a plan achieves at least 4-star rating, the plan receives quality bonus payments (QBP), increase in benchmark payments, and is allowed to enroll new members year-round. This point of view article describes unintended consequences of the medication adherence metric utilized by CMS for the 5-star quality rating system.
How Is Medication Adherence Calculated?
CMS uses the Proportion of Days Covered (PDC) calculation as a proxy measure of adherence for the quality metric. PDC for medications in a therapeutic category is calculated by the total number of days supplied during an interval (in numerator), divided by the total number of days in that interval (denominator).
Refill data report for the calculation is obtained from pharmacy billing claims. Measures for medication adherence are triple weighted in the Star Rating System, and include renin angiotensin system antagonists, diabetes medications (excluding insulin), and statins. Patients enrolled under hospice programs, diagnosis of end-stage renal disease or dialysis coverage, and hospitalized and nursing home patients are excluded from the denominator.
The PDC calculation includes patients aged 18 years and older who have received at least 2 refills of any of the applicable medications within a year. Individuals are only included in the metric if their first refill is at least 91 days before the end of the enrollment period. The treatment period for the calculation begins with the date of first prescription claim, and ends on the last day of the year, or date of death or disenrollment, whichever occurs first. The quality measure is calculated yearly and is satisfied if the patient fills their prescriptions to cover 80% or more of the time during the treatment period. 1
Supportive Data Regarding the Medication Adherence Metric
Numerous studies have demonstrated that adherence to chronic disease medications calculated by claims-based measures is associated with improvement in clinical outcomes, including reduction in Emergency Department (ED) visits and hospitalizations, and resultant lower cost of care. 2 Research studies suggest racial disparities in adherence to medication classes included in QBP, with older Black and Hispanic individuals having a higher proportion of nonadherence compared with Whites. CMS data indicate that the QBP resulted in narrowing of disparities in PDC scores for MA and PDP beneficiaries from racial minorities and underserved communities during 2013–2018. 3 Since initiation of QBP, PDC claims data have improved for star rating-targeted and nontargeted medicines alike. 4
Limitations of the Medication Adherence Metric and Impact on Care
Following are notable limitations of the PDC metric.
Transitions of care or change in therapy
While the CMS star rating affects MA and PDP plans only, the plans contract with pharmacies that help them achieve the required metric goal. Pharmacies are incentivized by inclusion in the plans' networks and being charged lower network participation fees by the PDP and Pharmacy Benefit Managers. 5 Lacking an efficient tracking system, pharmacies are not aware that a medication has been discontinued by a clinician or that a patient is in a hospital or skilled nursing facility.
Studies indicate that 15%–46% of adult patients may be taking medications not noted in their medical records. 6 A study surveying Medicaid recipients who had more than a 30-day gap in medication refill identified by PDC claims data demonstrated that 28% of those patients labeled as nonadherent were actually adherent and were either receiving the medication through an alternative pharmacy or had appropriately discontinued a medication on the advice of their health care professional. 7
Impact on deprescribing efforts
As the US health care system moves toward value-based reimbursements, where value is measured and financial incentives are driven by quality metrics, this metric could potentially discourage health care professionals from deprescribing medications mid-year after a patient has received 2 refills, in the absence of a life-threatening side effect. While there is a dearth of data studying the impact of QBP on deprescribing tendencies in managed care practices, there is need to study the impact of the program on deprescribing initiatives of practices in QBP.
Flaws in calculation
There have been concerns about inaccuracy of the pharmacy claims data as most MA plans, even 2-star rated plans, report higher percentages of adherence to medications than reported in the peer-reviewed literature. 2 Healthy patients are also more likely to be adherent, resulting in an overestimate of savings in health care spending related to this metric by 20%. 8
Discount programs and other insurers
The MA insurance claims data do not include the use of free drug samples or discount programs such as GoodRx and may miss coverage through other insurance plans or the Department of Veterans Affairs (VA) pharmacies. Studies including older veterans utilizing both the VA and PDP benefits indicate that 30%–50% of patients obtained medications of similar drug classes from both VA and private pharmacies with overlapping days' supply. 9
How to Improve the Metric
The most critical flaw of the PDC calculation is the assumption that any gaps in refill result from suboptimal medication adherence, which results in underestimation of adherence, overprescribing, as well as resultant medication errors. Prieto-Merino et al recommend adjusting the denominator for the PDC for gaps larger than the duration of covered days, assuming that a large gap might be due to a medically justifiable treatment break instead of nonadherence. 10 Lester et al recommend replacing prescription adjudication date with pickup date in the PDC calculation to more accurately reflect medication adherence. 11
Supplemental Patient-Centered Medication-Related Metrics to Assess the Quality of MA and PDP Plans
Listed hereunder are some leading process barriers to medication adherence relevant to MA and PDP plans as demonstrated in research studies.
Out-of-pocket cost
The out-of-pocket cost has been associated with medication nonadherence and delay in seeking treatments, 12 and reduction in out-of-pocket costs of medications has demonstrated improvement in medication adherence and resultant clinical outcomes across a spectrum of medical conditions and in underserved communities. 13
Utilization management
Studies demonstrate that formulary restriction strategies implemented by insurance plans to reduce costs, including step therapy and prior authorizations, create systemic barriers to access medications, leading to lower adherence to medications, and higher Emergency Department and hospital admissions, and higher health care costs. 14
Discussion
Using a quality metric based on PDC calculation to estimate medication adherence may motivate health care plans to utilize value-based insurance design to reduce the out-of-pocket cost of medications included in the metric, and incentivize timely and longer duration refills, thus impacting key clinical outcomes. However, this metric has significant flaws that can potentially disincentivize deprescribing in a model of value-based reimbursement and can promote medication errors, 9 thus negating efforts toward developing age-friendly health care systems.
The calculation should be modified to allow for medically justified gaps and deprescribing. In addition, in line with the CMS initiatives to reduce health disparities, there is a need to add quality metrics that measure precipitants of disparities, such as prior authorization requirements for medications prescribed to treat common chronic conditions, to allow patients to make an informed decision when selecting MA and PDP plans.
Conclusions
Medication adherence is a key component of CMS's Star Rating System for grading MA and PDP plans. However, the currently utilized PDC calculation for the metric utilizing prescriptions claims data is suboptimal and may underestimate medication adherence 7 and potentially discourage deprescribing efforts.
Footnotes
Acknowledgments
The authors would like to thank the Society of General Internal Medicine's Leadership in Health Policy Program, Thomas O’ Staiger, MD (University of Washington Medical Center, Seattle, WA), and Jonathan Flacker, MD, AGSF (JenCare Senior Medical Center, Atlanta, GA) for review of an earlier version of this article.
Authors' Contributions
Conceptualization, investigation, and writing—original draft preparation by Q.S. Writing—reviewing and editing by S.S. and C.V.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
No funding was received for this article.