Nonadherence to beta blockers, statins, or angiotensin-converting enzyme inhibitors is common (21%-29%) among patients with coronary artery disease (CAD). Patients who do not adhere to their medication regimens are at increased risk of mortality, cardiovascular hospitalizations, and revascularization procedures; thus, medication nonadherence should be a target for quality improvement interventions to maximize the outcomes of CAD patients.
Clinical trials have demonstrated the efficacy of beta blockers, angiotensin-converting enzyme (ACE) inhibitors, and statins for the secondary prevention of coronary artery disease (CAD).1 The American College of Cardiology/American Heart Association clinical practice guidelines have adopted these medications as class I recommendations for CAD patients to reduce recurrent cardiovascular events.1 Although there have been improvements in the prescription rates of secondary prevention medications for CAD patients, a gap persists between the benefits demonstrated with these medications in clinical trials and the effectiveness observed in clinical practice.2,3 One potential explanation for this discrepancy is suboptimal adherence to secondary prevention medications in practice compared with clinical trials, where adherence is often closely monitored.4-6
The objective of our study was to assess the association between medication adherence and a broad range of outcomes among patients with known CAD treated in a large managed care organization. Specifically, we evaluated the association between nonadherence to beta blockers, ACE inhibitors, and statins with outcomes of all-cause and cardiovascular mortality, hospitalization for acute myocardial infarction (MI) or heart failure, and need for coronary revascularization, including percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) surgery.
Kaiser Permanente of Colorado is an integrated, nonprofit managed care organization that provides medical services to more than 430,000 members in the Denver, Colorado, metropolitan area. A CAD registry was established in 1998.7 Patients were identified for potential inclusion in the registry using an algorithm applied to the Kaiser Permanente of Colorado automated databases consisting of hospitalization records and outpatient diagnoses. Once a potential patient was identified, the diagnosis of CAD was validated by chart review before inclusion in the registry.
We conducted a retrospective cohort study of patients in the CAD registry with prior MI, PCI, or CABG surgery. Patients were included if they were enrolled in the CAD registry on or after January 1, 2000, and had at least 6 months of continuous follow-up after entry into the registry. Applying these criteria, 15,767 patients met the inclusion criteria and were included in the study.
Medication adherence was calculated as the proportion of days covered (PDC) by dividing the total number of days supplied for each class of medication (ie, beta blockers, ACE inhibitors, angiotensin-receptor blockers, and statin medications) by the observation time interval. Patients were classified as “nonadherent” based on a PDC below 0.80, consistent with the dichotomization of medication adherence in the literature.4 The observation time interval was 180 days in the primary analysis, and a PDC was calculated for each 180-day interval during follow-up.
The primary outcomes were all-cause and cardiovascular mortality. The secondary outcomes included hospitalization for acute MI or heart failure and coronary revascularization (either PCI or CABG surgery after entry into the CAD registry). Follow-up and vital status information were available for 99% of subjects through December 31, 2005. The median follow-up was 4.1 years.
Multivariable Cox proportional hazards models with medication nonadherence as a time-varying covariate were used to assess the independent association between medication nonadherence and all-cause mortality, adjusting for patient demographics and both cardiac and noncardiac comorbidity variables. The hazard ratio (HR) estimate represents the risk of mortality associated with medication nonadherence from the most recent 180-day interval evaluated. The same analyses were repeated for the following end points: cardiovascular mortality, hospitalization for acute MI or heart failure, and revascularization procedures, PCI, or CABG surgery. To determine whether the association between adherence and outcomes resulted from a “healthy adherer” effect, we assessed the association between nonadherence to proton pump inhibitor or H2 antagonist regimens and outcomes in patients prescribed these medications (n = 6041).8 We hypothesized that a significant association between nonadherence with these medications and cardiac-specific outcomes would demonstrate the presence of a health adherer effect because neither of these treatments have an impact on cardiac outcomes.
The study was approved by the Kaiser Permanente Colorado Institutional Review Board. All analyses were performed using the SAS statistical package version 9.1.
Baseline characteristics of adherent and nonadherent patients for each class of medication are shown in Table 1. Among patients dispensed beta blockers (n = 11,865), 28.8% were nonadherent, as defined by a PDC below 0.80. For those dispensed ACE inhibitors or angiotensin-receptor blockers (n = 10,021), 21.6% were nonadherent. When it came to those taking statin medications (n = 13,596), 26.0% were nonadherent. Across the 3 classes of medications, nonadherent patients were more likely to be younger and to have chronic obstructive pulmonary disease and depression.
In multivariable analysis, the association between nonadherence and all-cause mortality remained significant for each class of medication (Table 2): beta blocker (HR, 1.50; 95% CI [confidence interval], 1.33-1.71), ACE inhibitor (HR, 1.74; 95% CI, 1.52-1.98), and statin (HR, 1.85; 95% CI, 1.63-2.09). Next, the independent association between nonadherence and increased risk for cardiovascular mortality was consistent for beta blockers (HR, 1.53; 95% CI, 1.16-2.01), ACE inhibitors (HR, 1.66; 95% CI, 1.26-2.60), and statins (HR, 1.62; 95% CI, 1.24-2.13). In addition, nonadherence to each class of medication was associated with a higher risk of cardiac hospitalizations (beta blockers: HR, 1.10 [95% CI, 0.99-1.23]; ACE inhibitors: HR, 1.40 [95% CI, 1.25-1.57]; statins: HR, 1.35 [95% CI, 1.21-1.50]) and revascularization procedures (beta blockers: HR, 1.15 [95% CI, 1.04-1.27]; ACE inhibitors: HR, 1.32 [95% CI, 1.18-1.48]; statins: HR, 1.11 [95% CI, 1.01-1.22]).
The risk associated with nonadherence was consistent when the observation time interval was extended to 360 days or the entire observation period was compared with 180-day results for the primary analysis. Finally, we evaluated for the presence of a healthy adherer effect by assessing the association between nonadherence to proton pump inhibitors or H2 antagonists and outcomes. Nonadherence to proton pump inhibitors or H2 antagonists was not associated with any of the cardiovascular outcomes, including all-cause mortality, cardiovascular mortality, cardiovascular hospitalizations, and revascularization procedures (Table 2).
The objective of our study was to assess the association between medication nonadherence and a broad range of outcomes among patients with known CAD. Over a median followup of 4.1 years, we found that medication nonadherence to statins, ACE inhibitors, and beta blockers was common, occurring in approximately 1 in 4 patients. Nonadherence was associated with increased risk for all-cause mortality, cardiovascular mortality, hospitalization for acute MI or heart failure, and coronary revascularization procedures. In contrast, nonadherence to proton pump inhibitors or H2 antagonists was not associated with adverse outcomes, suggesting that the lower risk of adverse outcomes associated with adherence in our study was due mainly to the direct benefits of these cardioprotective medications rather than a healthy adherer effect.
In contrast to prior studies, we evaluated the association between nonadherence and a range of cardiacspecific outcomes, including cardiovascular mortality, cardiovascular hospitalizations, and revascularization procedures. Previous studies have only focused on all-cause mortality, where the association between nonadherence and outcome have been attributed to the direct benefit of the medication as well as the “healthy adherer” effect.8-10 The findings of our study further support the direct benefits of cardioprotective medications in “real world” practice since there was no association between adherence with proton pump inhibitors or H2 antagonists and cardiac-specific outcomes. Our study also demonstrates that medication nonadherence is a common problem among outpatients with CAD, limiting the effectiveness of secondary prevention medications.
Prior studies suggest that nonadherence is modifiable. The FAME (Federal Study of Adherence to Medications) investigators used a multicomponent intervention comprised of standardized medical education, regular followup with pharmacists, and medications dispensed in specific packs to address nonadherence, and these measures resulted in improvements in medication adherence and blood pressure levels.11 In addition, several previous trials using telehealth technology have increased adherence to antihypertensive and oral hypoglycemic medications, resulting in improvements with intermediate outcomes of blood pressure and hemoglobin A1c levels.12,13 The composite of the literature to date suggests that medication nonadherence is important and should be routinely assessed in clinical practice. Once identified, clinicians should engage patients in a discussion of the barriers and suggest strategies to improve medication adherence.
Current quality improvement efforts for patients with CAD are focused on the appropriate prescription of indicated cardioprotective medications14; however, for patients to derive the maximal benefit of prescribed therapies, they must take their medications as directed. Our findings demonstrate that nonadherence to cardiac medications is common and suggests that quality improvement efforts should be expanded to include medication adherence as a key component of secondary prevention in addition to prescription of indicated medications. To achieve better outcomes, quality improvement interventions must be focused on the patient, not just hospitals and clinicians; this is consistent with the need for a more patient-centered health care system, as espoused by the Institute of Medicine.15
We found that nonadherence to cardioprotective medications was prevalent among outpatients with CAD and was associated with a broad range of adverse outcomes, including all-cause and cardiovascular mortality, cardiovascular hospitalizations, and the need for revascularization procedures. These findings suggest that current quality-of-care measures should be expanded to include medication adherence assessment, which may be an especially important quality metric for CAD patients. Our findings also show that there is a need for interventions to reduce medication nonadherence in clinical practice.
This work was supported by an award from the American Heart Association (0535086N). Dr Ho is supported by a VA Research & Development Career Development Award (05-026-2). Dr Peterson is supported by an award from the American Heart Association (0670017N).