Coronary revascularization does not affect the underlying disease process of coronary atherosclerosis, so medical therapy is routinely recommended after coronary artery bypass graft (CABG) surgery or percutaneous coronary intervention (PCI). Medications that reduce the rate of death or subsequent myocardial infarction (MI) in patients with coronary disease include antiplatelet drugs, beta-blockers, angiotensin converting enzyme inhibitors (ACEI), angiotensin II receptor blockers (ARBs), and lipid-lowering therapy, particularly statins in appropriate patients. Use of these agents is recommended as secondary prevention among patients with clinically evident coronary artery disease (1), including patients who have undergone CABG or PCI ((2),3).

Roughly 838,000 patients underwent a CABG or PCI in 2009 in the United States (4). Patients who have undergone coronary revascularization might not, however, fill an initial prescription for secondary preventive medications or use these medications consistently, especially after a CABG ((5),(6),(7),(8),(9),10). Patients who have undergone successful coronary revascularization might not use recommended medications consistently for several reasons, including a belief that cardiac medications are no longer important once a coronary obstruction has been treated mechanically. We hypothesized that patients who have undergone CABG would be less likely than patients who have undergone PCI to take secondary prevention medications. We tested this hypothesis in a large cohort of patients from an integrated health system that has emphasized use of cardiovascular preventive therapies and that charges minimal copayments to fill prescriptions at health plan pharmacies.

Kaiser Permanente Northern California is a large, integrated healthcare delivery system that provides care to more than 3.2 million individuals in the greater San Francisco Bay Area who are broadly representative of the local and statewide population, apart from slightly lower representation of the extremes of age and income (11). We constructed a cohort of patients with incident coronary artery disease who received CABG or PCI as an initial coronary revascularization strategy. We identified all health plan members 30 years of age or older who received an initial CABG or PCI procedure between January 1, 2000, through December 31, 2007 (International Classification of Diseases-Ninth Edition [ICD-9] codes 00.66, 36.0, 36.00, 36.01, 36.02, 36.05, 36.06, 36.07, 36.09, 36.10, 36.11, 36.12, 36.13, 36.14, 36.15, 36.16, 36.17, and 36.19; and Current Procedure Terminology–Fourth Edition codes 33510, 33511, 33512, 33513, 33514, 33516, 33517, 33518, 33519, 33521, 33522, 33523, 33530, 33533, 33534, 33535, 33536, 33572, 92973, 92980, 92981, 92982, 92984, 92995, and 92996). We assigned the index date as the date of the revascularization procedure. To capture complete data on demographic characteristics, comorbidities, and concurrent medication use, we restricted the analysis to patients with complete demographic data and at least 12 months of continuous membership and continuous pharmacy benefit before the index date. We excluded patients with a history of coronary artery disease, coronary revascularization, maintenance renal dialysis, organ transplantation, missing membership information after the index date, or unconfirmed index hospital stay.

This study was approved by the institutional review boards of the Kaiser Foundation Research Institute and of Stanford University. Waiver of informed consent was obtained because of the nature of the study.

We ascertained baseline and longitudinal medication use from pharmacy records with previously validated methods ((12),13). Medications of interest included statins, beta-blockers, ACEIs, ARBs, and clopidogrel. We did not assess use of aspirin, because it can be obtained without a prescription and, therefore, complete information on its use was not available. We computed the medication possession ratio (MPR) over the first 12 months after the index date, among those who received at least 1 prescription for the drug of interest after initial revascularization. Patients were assumed to have continuous coverage for a given medication from the date they filled the initial prescription until the end of the final prescription, as long as there were no gaps in coverage of 30 days or more. When a gap in coverage >30 days was noted, patients were considered uncovered from the last day of the final filled prescription until the date their next prescription was filled or the end of follow-up at 12 months after revascularization. For example, a patient with a single 2-week gap between prescription refills was assumed to have a 100% MPR, whereas a patient with a single 6-week gap during the year would have an MPR of 88% (46 of 52), and a patient who filled a single 30-day prescription during the year would have an MPR of 8% (30 of 365).

Because medication adherence might change over long-term follow-up, we analyzed MPR over the fixed time interval of 1 year after the index date. We required that each patient have at least 1 full year of follow-up to permit the MPR to be calculated over the same period of follow-up for all patients.

We obtained data on age, sex, self-reported race/ethnicity, and treatment facility from health plan databases. We identified comorbid conditions up to 4 years before the index date with previously validated approaches on the basis of ICD-9 diagnosis and procedure codes and Current Procedural Terminology–Fourth Edition procedure codes identified from health plan hospital stay, ambulatory, laboratory, and pharmacy databases ((14),(15),(16),(17),(18),(19),20). We ascertained the following covariates: smoking status and history of heart failure, ventricular fibrillation/tachycardia, ischemic stroke/transient ischemic attack, peripheral arterial disease, valvular heart disease, rheumatic heart disease, atrial fibrillation/flutter, diabetes mellitus, hypertension, dyslipidemia, arthritis, dementia, chronic liver disease, chronic lung disease, chronic kidney disease (CKD), diagnosed depression, hyperthyroidism, hypothyroidism, systemic cancer, or hospitalized bleed. The criteria for defining these covariates have been published previously ((14),(15),(16),(17),(18),(19),20). In brief, a diagnosis of heart failure required at least 2 outpatient visits with an ICD-9CM code for heart failure or an inpatient admission with a primary diagnosis of heart failure. CKD was defined as estimated glomerular filtration rate of 60 ml/min/1.73 m² or lower, on the basis of the Chronic Kidney Disease Epidemiology Collaboration estimating equation.

We used multivariable regression analysis to assess the relationship of CABG or PCI on MPR, while adjusting for baseline covariates. We analyzed non-use of medications (i.e., patients who never filled a prescription for the medication of interest) with logistic regression, and among patients who filled least 1 prescription, we analyzed MPR as a continuous variable with ordinary least-squares linear regression. Analyses were performed with the R statistical package (version 2.12.2, R Development Team, Vienna, Austria).

Between 2000 and 2007, a total of 28,477 patients with new-onset coronary disease underwent initial coronary revascularization in Kaiser Permanente Northern California. After excluding patients who did not have a full year of follow-up data available (1,486 who died, and 3,416 who did not have continuous prescription drug coverage) and 222 patients receiving chronic dialysis, the final study population consisted of 23,353 patients, 8,837 of whom received an initial CABG, and 14,516 of whom received an initial PCI. Patients who underwent CABG were somewhat older and less likely to have had a recent MI or diagnosed depression than patients who underwent PCI (Table 1). Patients who received CABG were also more likely to have a history of diabetes, hypertension, dyslipidemia, CKD, heart failure, and atrial fibrillation (Table 1). All subsequent analyses were adjusted for differences in baseline characteristics.

Most patients were treated with a statin in the first year after coronary revascularization, but significantly more CABG patients (7.1%) than PCI patients (4.8%) never filled a prescription for a statin (Table 2). Among patients who filled least 1 prescription for a statin, the MPR was significantly lower among CABG patients (82.7%) than among PCI patients (84.2%). The unadjusted difference of 1.5% in MPR was somewhat attenuated (to 1.0%) after adjustment for baseline characteristics but remained statistically significant (p = 0.001).

Most patients were treated with a beta-blocker after coronary revascularization, with a similar proportion of CABG patients (6.4%) and PCI patients (6.1%) who never filled a prescription (Table 2). After adjustment for baseline confounding factors, however, CABG patients were significantly less likely than PCI patients to have never filled a prescription for a beta-blocker. In contrast, among patients who filled at least 1 prescription for a beta-blocker, CABG patients had a significantly lower MPR (76.1%) than PCI patients (80.6%). The unadjusted difference of 4.5% in MPR was slightly attenuated (to 3.7%) but still significant (p < 0.0001) after adjusting for baseline characteristics.

The use of ACEI/ARBs after coronary revascularization was more variable than that of statins and beta-blockers (Table 2). A higher percentage of CABG patients (29.1%) than of PCI patients (22.4%) never filled a prescription for an ACEI or ARB in the year after revascularization (p < 0.0001). Among those who filled at least 1 prescription, the MPR among CABG patients (69.4%) was lower than for PCI patients (77.8%) over the first year. The unadjusted difference of 8.4% in MPR between CABG and PCI patients was essentially unchanged (8.2%) after multivariable adjustment for baseline characteristics and remained statistically significant (p < 0.0001).

We also examined use of clopidogrel between CABG or PCI patients, to provide a context for interpreting the other medication data. Most CABG patients (88.5%) did not fill a prescription for clopidogrel, whereas only 4.9% of PCI patients did not (p < 0.0001). Among patients who filled at least 1 prescription for clopidogrel, the MPR was much lower among CABG patients (40.1%) than PCI patients (54.5%). The unadjusted difference of 14.4% in MPR between CABG and PCI patients was somewhat attenuated (to 9.2%) but remained significant after adjusting for baseline characteristics (p < 0.0001).

The lower use of secondary preventive medications among patients who had CABG was generally consistent in key patient subgroups (Figure 18_gr1). Among patients with diabetes mellitus or heart failure, there were generally smaller gaps in use of secondary preventive medications, particularly for ACEI/ARBs, between CABG- and PCI-treated patients. The lower use of secondary medications among CABG patients was generally consistent among patients with or without CKD, lung disease, liver disease, and a prior MI (Figure 18_gr1).

Use of secondary preventive medications increased progressively between 2000 and 2007. The percentage of patients who never filled a prescription for an ACEI/ARB, beta-blocker, or statin declined progressively (Figure 18_gr2), and the gap between CABG and PCI narrowed over time for statins and beta-blockers but persisted for ACEI/ARBs. Similarly, the MPR among patients with at least 1 prescription rose steadily over time for all 3 medications; the gap between CABG and PCI narrowed slightly for beta-blockers, persisted for statins, and widened for ACEI/ARBs (Figure 18_gr3).

Use of medications for secondary prevention of coronary artery disease after coronary revascularization was quite high in this study. Nevertheless, patients who had undergone a CABG were generally less likely than patients who underwent a PCI to be started on statins and ACEI/ARBs and less likely to refill their medications consistently during the first year after revascularization. This gap in use of evidence-based therapies, although small in absolute terms, suggests there is an opportunity to further improve quality of care for patients after coronary revascularization.

Prescription of proven secondary preventive medications at the time of hospital discharge for a cardiovascular diagnosis has become an important focus of quality improvement efforts. Prescription at the time of discharge is under the control of physicians and health systems, and our data show steady improvements in these rates over time (Figure 18_gr2), with little to no gap between CABG and PCI patients for beta-blockers and statins in the final study year (2007). Subsequent adherence to initially prescribed medications is not measured commonly and depends on patients to refill and take their medications consistently. Use of secondary preventive medications among patients who filled an initial prescription also improved over time (Figure 18_gr3), yet MPRs were consistently lower after CABG than after PCI. The reasons for this persistent gap are uncertain but might be related to patterns of follow-up care, if patients are less likely to visit a cardiologist for subsequent care after a CABG than after a PCI and cardiologists are more likely to prescribe and encourage use of preventive medications ((21),22). Furthermore, patients might feel that a CABG was the definitive treatment of their coronary disease and hence that medications are no longer necessary, whereas patients after PCI are educated that medications, particularly antiplatelet agents, are important to maintain patency of the stented area. Surveys of patients after CABG found that they have concerns about the overuse and harms of prescription drugs and that poor medication adherence was correlated with social, economic, and demographic factors (23). Further research is needed to understand the reasons for discontinuation of medications after CABG or PCI and to develop appropriate interventions, either at the health system level (e.g., care protocols) or at the patient level (e.g., education on the importance of continued medical therapy).

This study was conducted in a large integrated health system that provides medication with minimal copayments and has pharmacies conveniently located in its medical centers. Medication use might also be higher in the Kaiser Permanente Northern California system because of its focus on quality improvement and uniform electronic health records. Other large, integrated delivery systems, such as the Veterans Administration, also have higher levels of medication use after CABG (24). Medication use in less well-integrated healthcare settings is likely to be much lower, particularly if care is fragmented or patients have to make large payments for prescription drugs. These differences might particularly affect long-term adherence and the frequency of prescription refills.

Prior studies have usually measured prescriptions for evidence-based medications at the time of hospital discharge ((25),26) but have not characterized long-term use of these therapies after coronary revascularization. Studies that did examine long-term medication use post-revascularization have usually done so in a cross-sectional manner, examining use at a single point in time, such as at 6 or 12 months. In 1 large national community-based claims analysis of medication use after acute coronary syndrome, Lee et al. (27) found rates of statin (73%), beta-blocker (73%), and ACEI/ARB (62%) use at 12 months after discharge, rates that were lower than those found in our study. Our study adds to the published data by measuring post-discharge medication use continuously over the first year after revascularization.

In spite of the large body of published data on use of medications for secondary prevention after revascularization, few studies have compared the rate of medication use after CABG with the rate after PCI. In 1 such study, Hiratzka et al. (28) examined the use of medications at discharge after CABG, PCI, or medically managed acute coronary syndrome in hospitals enrolled in the American Heart Association Get With the Guidelines program. They also found that patients who received CABG had lower rates of prescriptions at discharge than patients who received PCI: 57.3% versus 74.0% for ACEI; 90.8% versus 91.0% for beta-blockers, and 77.4% versus 89.2% for lipid-lowering medication (p < 0.0001 for all comparisons). The difference in use of ACEI and statins in their study remained significant after multivariable adjustment for clinical characteristics. They did not, however, have any data on medication use after discharge and were unable to measure adherence in subsequent follow-up.

First, we relied on pharmacy dispensing data to ascertain the timing of prescription refills and to estimate adherence. We do not have data on whether patients actually took their medications as prescribed. If some patients were advised to “split pills,” this could affect the calculated MPRs, because adherent patients would refill prescriptions less frequently. Nor do we have data on prescriptions filled at outside pharmacies, although prior surveys of health plan members show over 95% use of Kaiser Permanente pharmacies to obtain medications for chronic conditions. Second, we do not have access to detailed clinical data on reasons for not prescribing medications initially or for discontinuing medications during follow-up. Although we have no reason to believe that patient tolerance to medications would vary by revascularization method, unobserved differences between patients who underwent CABG or PCI could account for the observed differences in adherence rates. Third, we do not have information on some clinical characteristics, such as anemia or left ventricular ejection fraction, which might have affected initial prescription of medications or subsequent adherence. Fourth, we do not have data on patient socioeconomic status, the distance to a Kaiser Permanente pharmacy, or the specialty of physician who saw the patient in follow-up, all of which might have affected long-term adherence to medications. Finally, the last year of patient entry into this study was 2007, and medication use might have further improved since then.

Our study suggests that patients who receive CABG are less likely than patients who receive PCI to fill initial prescriptions for selected cardioprotective medications and to remain adherent to these medications during the first year after revascularization. Although the absolute differences in medication use in this study were small, our findings nevertheless suggest that there might be systematic factors after CABG and PCI that contribute to lower medication use after CABG. Addressing these differences might provide an opportunity to further improve the quality of care after coronary revascularization. Future research should examine the factors that predict poorer adherence to cardioprotective and should test interventions to improve both the initial prescription and long-term adherence to proven medications for secondary prevention (29).