Dosing Frequency and Adherence in Patients With Nonvalvular Atrial Fibrillation
Published Online: April 15, 2014
Tanya Burton, PhD; Ying Fan, MS; and Winghan Jacqueline Kwong, PharmD, PhD
Higher medication adherence is widely believed to be associated with better clinical and economic outcomes.1-4 As highlighted by many published studies, medication adherence is gradually becoming an important part of healthcare quality assessment. Recently, the Centers for Medicare & Medicaid Services (CMS) incorporated medication adherence to antihypertensives, lipid-lowering agents, and oral antidiabetics in the Five-Star Ratings system for quality bonus payment to Medicare Advantage health plans.5 Yet, achieving high patient adherence is often a challenge, with variables affecting adherence covering multiple dimensions. According to the World Health Organization, treatment complexity, including dosing frequency, is one of many factors that contribute to medication nonadherence.6,7 Previous studies have shown that higher rates of adherence were observed among patients prescribed medications that dosed less frequently. 8-13 While many studies have examined the association between dosing frequency and medication adherence in various chronic conditions, few have examined this association in patients with nonvalvular atrial fibrillation (NVAF).
Atrial fibrillation (AF) is the most common arrhythmia disorder in the United States, affecting 2.4 to 5.9 million adults.14-18 It is associated with increased risk of stroke,19-21 heart failure,22 cognitive dysfunction,23,24 and premature death.25-27 In addition, AF patients are older and have a significantly greater chronic disease burden compared with the general population.28 It is estimated that more than half of patients with AF have hypertension26,29 and one-third have heart failure.30,31 NVAF is the most common form of AF, accounting for nearly 70% to 90% of all patients with AF.32,33 Due to their comorbidities, NVAF patients often require multiple chronic medications with different dosing schedules that may affect their medication adherence.
Previous studies examining the association of dosing frequency and medication adherence have evaluated multiple drug classes in aggregate. For example, medication adherence of patients receiving once-daily oral antidiabetic antihypertensive, or calcium channel blockers were studied as a group and compared with patients taking twice-daily regimens of the same therapeutic classes. While such a method may increase the sample size, differences in medications used by patients in the once-daily and twice-daily cohorts and their corresponding side effect profi les may introduce biases to study results.11,13 To minimize these types of heterogeneity biases within cohorts, we performed an analysis focused on 2 medications commonly prescribed to NVAF patients that have both once-daily (controlledrelease) and twice-daily (immediate-release) oral formulations available in the United States.
This was a retrospective cohort study using medical and pharmacy claims and eligibility information from an administrative claims database affi liated with a large US managed care health plan. The individuals covered by this health plan, approximately 14 million per year, are geographically diverse across the United States, with greatest representation in the US Census Southern and Midwest regions. The health plan provides full insurance coverage for physician, hospital, and prescription drug services for individuals with commercial and Medicare Advantage insurance. The data are linked longitudinally using an encrypted patient ID. All study data were de-identifi ed and accessed with protocols compliant with the Health Insurance Portability and Accountability Act. This study was exempt from both ethics approval from an institutional review board and subject informed consent because the research only involved the study of existing data, and the subjects could not be identifi ed directly or through identifi ers linked to the subjects.
Two study samples were created for the 2 study drugs: metoprolol and carvedilol. The metoprolol sample included adult patients (>18 years) with 2 or more pharmacy claims for metoprolol, and the carvedilol sample included adult patients with 2 or more pharmacy claims for carvedilol, on separate dates within a 365-day window between January 1, 2008, and December 31, 2010 (identifi cation period). Patients with qualifying pharmacy claims for both metoprolol and carvedilol were included in both study samples. The date of the first metoprolol or carvedilol claim during the identifi cation period was defi ned as the “index date.” National drug codes (NDC) were used to identify the dosing frequency of metoprolol and carvedilol on pharmacy claims. Within each study sample, patients were assigned to the once-daily or twice-daily cohort based on the dosing frequency on the index pharmacy claim. Patients with an index pharmacy claim for a once-daily medication were assigned to the once-daily cohort, and patients with an index pharmacy claim for a twice-daily medication were assigned to the twice-daily cohort. Patients were required to be continuously enrolled with both medical and pharmacy benefi ts for 12 months before and after the index date (baseline and follow-up periods, respectively).
Patients were also required to have at least 1 inpatient medical claim or 2 outpatient medical claims (on separate dates) with a primary or secondary AF diagnosis code (International Classification of Diseases, Ninth Revision, Clinical Modification code 427.31) during baseline. Patients in the metoprolol sample were required to have 1 or more medical claims for acute myocardial infarction, angina, heart failure, or hypertension, and patients in the carvedilol sample were required to have 1 or more medical claims for heart failure or hypertension during baseline. Evidence of these drug-specific medical claims during baseline was used to confi rm chronic usage of the study medication.
Patients who had 1 or more medical claims for valvular heart disease or hyperthyroidism during baseline, or who had prescription claims of more than 1 dosing frequency of the study medication during follow-up, were excluded from the analysis.
Adherence, discontinuation, and persistence to the study medication were assessed for all study patients during the 12 months after the index date (follow-up). In this study, adherence was assessed using a measure of the proportion of days covered (PDC). PDC represents the proportion of time over the course of treatment that the patient was in possession of the study medication. PDC was calculated by dividing the number of days during which the patient was theoretically in possession of the study medication (based on the total days supply on filled prescriptions) by 365 days. Previous literature has categorized patients with PDC 80% or greater as “adherent.”7,34,35
Discontinuation of the index study medication was defined by a gap of 30 or more (30+) days and 90 or more (90+) days between fills during follow-up. The date of discontinuation was defined as the run-out date of the last fill. Persistence was defined as time on treatment, which was measured by the number of days from the index medication fill date until discontinuation or the end of the 12-month follow-up period, whichever occurred first.
Patient demographics, including age, gender, type of insurance (commercial or Medicare Advantage), and geographic region of health plan (Northeast, Midwest, South, and West) were assessed on the index date. Baseline stroke risk was measured using the CHADS2 score.36 A summary score derived from the Elixhauser Comorbidity Index (ECI)37,38 was used to assess patients’ overall baseline comorbidity burden. Other comorbidities common in NVAF patients identified during baseline included: stroke and/or transient ischemic attack (stroke/TIA), diabetes, dyslipidemia, hypertension, heart failure, and acute myocardial infarction. Baseline utilization of warfarin, heart rate and heart rhythm control medications, and antihypertensives was captured, and total out-of-pocket (OOP) prescription drug cost was calculated during the 1-year baseline period.
Descriptive statistics were calculated for all study variables. Cohort comparisons were performed using t test for continuous variables and x2 test for categorical variables. Persistence with once-daily and twice-daily frequencies was analyzed using Kaplan-Meier curves and the log-rank test.
The likelihood of PDC 80% or greater with once-daily and twice-daily frequencies was estimated using logistic regression models for the metoprolol and carvedilol samples separately. Persistence with once-daily and twicedaily frequencies was compared using Cox proportional hazards models. Covariates included in the regression models included patient demographics, type of health insurance, geographic region of the health plan, baseline Elixhauser Comorbidity Index score, prior stroke/TIA, baseline use of study medication, warfarin, heart rate and heart rhythm control medications, the number of baseline antihypertensive medication classes, and OOP prescription drug cost for all baseline medications.
PDF is available on the last page.