Author + information
- Received March 17, 1998
- Revision received June 23, 1998
- Accepted July 9, 1998
- Published online November 1, 1998.
- Nicholas L Smith, PhD, MPH∗,†,* (, )
- Gayle E Reiber, PhD, MPH†,‡,¶,
- Bruce M Psaty, MD, PhD∗,†,‡,
- Susan R Heckbert, MD, PhD†,
- David S Siscovick, MD, MPH∗,†,
- James L Ritchie, MD, FACC∗∥,
- Nathan R Every, MD, MPH∗∥,¶ and
- Thomas D Koepsell, MD, MPH†,‡
- ↵*Address for correspondence: Dr. Nicholas L. Smith, Cardiovascular Health Research Unit, 1730 Minor Avenue, Suite 1360, Seattle, Washington 98101
Objective. We compared long-term health outcomes associated with beta-adrenergic blocking agents and diltiazem treatment for unstable angina.
Background. No long-term data have been published comparing these two antianginal treatments in this setting.
Methods. Eligible veterans were discharged from the Veterans Affairs Puget Sound Health Care System (VAPSHCS), Seattle Division, between October 1989 and September 1995 with an unstable angina diagnosis and were prescribed monotherapy beta-blocker or diltiazem treatment at discharge. Medication data were collected from medical records and computerized VAPSHCS outpatient pharmacy files. Follow-up death and coronary artery disease rehospitalization data were collected through 1996. Proportional hazards regression compared survival among diltiazem and beta-blocker users, controlling for patient characteristics with propensity scores.
Results. Two hundred forty-seven veterans (24% on beta-blockers, 76% on diltiazem) were included in this study. There were 54 (22%) deaths during an average follow-up of 51 months. After propensity score adjustment, there was no difference in risk of death comparing diltiazem to beta-blocker treatment (hazards ratios [HR] 1.1; 95% confidence interval [CI] 0.49 to 2.4). Among Washington residents (n = 207), there were 146 (71%) coronary artery disease rehospitalizations or deaths during follow-up. After adjustment, there was a nonsignificant increase in risk of rehospitalization or death associated with diltiazem use (HR 1.4; 95% CI 0.80 to 2.4). For both analyses, similar risks were found among veterans without relative contraindications to beta-blockers.
Conclusions. We found no survival benefit of diltiazem over beta-blocker treatment for unstable angina in this cohort of veterans.
The survival advantage conferred by beta-adrenergic blocking agents compared with placebo in patients who have experienced a myocardial infarction (MI) is clearly supported in the medical literature (1–5). Beta-blocker therapy has also been proved beneficial for the treatment of other forms of coronary disease, including unstable angina (6,7), an acute ischemic manifestation that can evolve to infarction. Diltiazem, a benzothiazepine calcium channel blocking agent, is frequently used in clinical practice to treat acute ischemic heart disease, although the literature supporting its long-term health benefits is lacking (8). Two short-term randomized clinical trials compared beta-blocker with diltiazem treatment in unstable angina patients and found that both agents produced similar hemodynamic and symptom control effects, but the studies were not designed to reach conclusions about long-term health effects (9,10). Nonetheless, diltiazem and other calcium channel blockers have been widely adopted in clinical practice to treat unstable angina and other cardiovascular conditions (11–13).
In an observational study of veterans initially hospitalized for unstable angina, we compared long-term health outcomes associated with beta-blocker and diltiazem treatment at hospital discharge. Although beta-blockers and diltiazem possess comparable antianginal and negative inotropic properties, beta-blockers are more potent antiarrhythmic and negative chronotropic agents that may also confer other advantages, such as endothelium stabilization and vagal tone improvement (14–21). Based on pharmacologic properties and clinical trial findings for post-MI patients, we hypothesized that beta-blocker treatment of unstable angina following hospitalization would be associated with better long-term health outcomes than treatment with diltiazem.
The Seattle Division of the Veterans Affairs Puget Sound Health Care System (VAPSHCS) was the site of the study, and the study protocol was approved by the Human Subjects Review Committee at the University of Washington.
Eligible veterans were discharged to the community from the Seattle Division of the VAPSHCS between October 1989 and September 1995 with a primary discharge diagnosis of unstable angina. Qualifying patients were identified using computerized inpatient summaries (patient treatment file, [PTF]) containing International Classification of Disease (ICD-9) codes for discharge diagnoses. Veterans who had an ICD-9 code of 411 (acute and subacute forms of ischemic heart disease) in the first coding position were potentially eligible for the study. Veterans who had either 413 (angina pectoris), 414 (other forms of chronic ischemic heart disease) or 786.5 (nonspecific chest pain) diagnostic codes in the first position, and who also had a cardiac catheterization procedure during their hospital stay, where: 1) the attending clinician characterized the chest pain as unstable angina and 2) coronary artery disease was confirmed by catheterization, were also eligible.
Hospitalization medical records were abstracted for all eligible veterans to confirm a diagnosis of unstable angina. A patient was considered to have unstable angina if 1) at hospital discharge, the physician stated that the hospital stay was attributable to coronary artery disease and 2) the chest pain was characterized as unstable angina either by patient report (unpredictable or at rest) or clinician report.
When a veteran had multiple admissions during the 6-year study interval, only the first admission was included. We excluded those veterans who had an MI or stroke during hospitalization according to hospitalization notes.
Discharge medication data were collected from discharge summaries in medical records and from computerized files of the VAPSHCS outpatient pharmacy. A veteran was considered to be using monotherapy of either a beta-blocker (atenolol, metoprolol or propranolol) or diltiazem, if the patient 1) was prescribed either agent at discharge according to medical records, 2) filled that prescription within 30 days of discharge at the VAPSHCS pharmacy and 3) did not fill a prescription for another antianginal drug other than nitroglycerin within 30 days of discharge.
Veterans were followed from the discharge date of their index hospitalization until December 1996. The primary end points of interest were all-cause mortality and a composite end point of all-cause mortality or coronary artery disease rehospitalizations. Mortality data were collected from two sources: 1) Beneficiary Identification and Record Locator Subsystem (BIRLS), which is a database maintained by the Department of Veterans Affairs (VA) to track benefit distribution and includes information on veterans’ vital status, and 2) Washington State electronic death certificate files that contain records of Washington State residents’ deaths. Cause of death was based upon the ICD-9 code for the underlying cause of death and was classified as follows: coronary artery disease (410 to 414), cerebrovascular disease (430 to 438), cancer (140 to 239) and other (all other underlying causes of death). The level of agreement between BIRLS and Washington State records for deaths was high (kappa = 0.84). Nonagreement was primarily attributed to veterans who were Washington State veterans at the time of index admission and who resided elsewhere at the time of their death.
Rehospitalization data were limited to Washington State hospital admissions and were collected from two sources: 1) PTF for hospitalizations that occurred at any Department of Veterans Affairs medical facility in Washington and 2) Washington State’s Comprehensive Hospital Abstract Reporting System (CHARS), which contains summary hospital stay information from all non-Federal hospitals in Washington. Coronary artery disease rehospitalizations were defined using the first three discharge diagnoses from PTF and CHARS records and included hospitalizations for MI (ICD-9 code of 410) and other forms of coronary artery disease (ICD-9 codes 411 to 414).
For veterans discharged with planned follow-up revascularization procedures according to discharge records (n = 16), we reassigned the cohort entry date as the discharge date of the subsequent revascularization hospitalization provided that 1) the revascularization was completed within 30 days and 2) discharge medications did not change. When condition 1 was not met, follow-up re-hospitalizations—if present—were counted as a study end point. If condition 2 was not met, the veteran’s exposure was reclassified according to postrevascularization discharge medications.
Hospitalization, demographic and medical history data
Computerized inpatient summaries contained information on each veteran’s hospital stay (procedures performed, discharge type and destination and length of stay) and demographic characteristics (age, sex, race, marital status and residential zip code). The VAPSHCS cardiac catheterization database provided information on the presence and extent of coronary artery disease and ejection fraction for those veterans who underwent a catheterization during their hospital stay.
Medical records provided information on the characteristics of the chest pain (precipitants of pain, duration of pain, onset of acute and chronic pain), electrocardiographic (ECG) results, current and past comorbidities and procedures (MI coronary artery bypass grafting [CABG], percutaneous transluminal coronary angioplasty, heart failure, left ventricular hypertrophy, valvular disease, arrhythmias, stroke, transient ischemic attacks, hypertension, hypotension, hypercholesterolemia, diabetes, asthma, chronic obstructive pulmonary disease [COPD], depression and psychoses), predischarge vital signs (blood pressure and pulse), smoking status and the provider’s assessment of the veteran at discharge (discharge diagnoses and recommended follow-up procedures).
Descriptive statistics were used to describe veterans’ demographic, hospital stay and clinical characteristics. Proportional hazards regression (22)was used to compare survival among diltiazem users compared with beta-blocker users, controlling for other patient characteristics. Follow-up time was defined as the time between the date of hospital discharge following unstable angina until the date of the follow-up event (death date for deaths or the admission date for coronary artery disease rehospitalizations) or until December 1996 for those veterans without events. For the composite death and/or coronary artery disease rehospitalization analysis, the event date was for the first event. Hazard ratios (HR) and 95% confidence intervals (CI) were derived from regression models.
For each end point, we constructed three proportional hazards regression models: 1) a crude model, 2) a multivariate model that adjusted for potential confounders using propensity score matching and 3) a multivariate model that excluded veterans with COPD, asthma, hypotension and heart block and adjusted for potential confounders using propensity score matching. The propensity scores reflected a veteran’s probability (from 0 to 1) of using diltiazem compared with using beta-blockers (23,24). Scores were stratified within 0.1 increments for each exposure and then matched. Unmatched strata were excluded from analyses. Propensity scores were derived from multivariate logistic regression models that adjusted for age, gender, marital status, race (white vs. other), Seattle residence (vs. non-Seattle residence), year of index hospitalization, chest pain characteristics, duration of pain episodes (<20 min vs. other), time of onset of acute (within 48 h of hospital presentation vs. other) and chronic pain (within 2 months of hospital admission vs. other), and ECG abnormalities (including any mention of ST-wave elevation or depression, T-wave inversion or left bundle branch block), inhospital revascularization procedures, systolic and diastolic blood pressure, current smoking status, a history of MI or stroke and past or present comorbidities mentioned in the medical record that included heart failure, left ventricular hypertrophy, valvular disease, arrhythmias, hypertension, hypotension, hypercholesterolemia, diabetes, asthma, COPD, depression and psychoses.
Blood pressure data were missing for five treated hypertensive patients, so the population’s mean diastolic and systolic blood pressure were imputed for these veterans. Six veterans who had missing ECG data and one veteran who had missing duration of chest pain information were excluded from the multivariate regression models.
Of the 1,462 veterans identified with an unstable angina discharge diagnosis between 1990 and 1995, 1,100 (75%) were discharged to the community. Among these, 878 (80%) had an outpatient pharmacy prescription filled at the VAPSHCS within 30 days of discharge and, 268 (31%) of whom were using beta-blocker or diltiazem monotherapy according to both medical records and the pharmacy database. Of the 268, we identified 3 (1%) who had an MI during hospitalization, 2 (<1%) who had a stroke, 11 (4%) who had no clinical evidence of coronary artery disease and 5 (2%) who did not have pain characteristics suggesting unstable angina. These exclusions left 247 (92%) veterans eligible for primary analysis.
Table 1lists the demographic and prehospitalization characteristics of the cohort. Ninety-eight percent of the veterans were men, the average age was 62 years, over one-third were smokers and most lived in Washington State. Two-thirds of the veterans were being treated with either beta-blockers or calcium channel blockers at admission, and, as expected, most veterans were using the same medication at admission and at discharge. Almost half had experienced an MI before these admissions for unstable angina, and approximately one-third had at least one previous cardiac revascularization procedure.
Hospitalization and clinical characteristics are listed in Table 2. Diltiazem users were more likely to have heart failure, COPD and higher systolic and diastolic blood pressure. Beta-blocker users were more likely to have ECG T-wave inversions, to undergo CABG during hospitalization and to have longer hospital stays. Apart from these differences, both beta-blocker and diltiazem users had similar hospitalization and clinical characteristics.
Among the 247 veterans, there were 55 (22%) deaths during an average follow-up of 51 months (Table 3). Twenty-four of these deaths (44%) listed coronary artery disease as the underlying cause. Twelve deaths (22%) had missing cause of death information and could not be classified. In unadjusted proportional hazards regression models (Table 4), diltiazem users had a nonsignificant 60% increase in the risk of all-cause mortality compared with beta-blocker users (HR 1.6; 95% CI 0.79 to 3.3). When we adjusted for potential confounders using a propensity score, diltiazem users no longer had an increased risk of death compared with beta-blocker users (HR 1.1; 95% CI 0.49 to 2.4). A similar result was found when we restricted analyses to veterans without asthma, COPD, hypotension or heart block (Table 4).
Among veterans who were Washington State residents at the time of discharge (n = 207), there were 146 deaths or rehospitalizations (71%) for coronary artery disease during follow-up (Table 5). Seven veterans (3%) died before a subsequent hospital discharge; 4 (2%) died during coronary artery disease hospitalization; 12 (6%) veterans sustained a nonfatal MI and were hospitalized; and 123 (59%) were hospitalized for other manifestations of coronary artery disease. In unadjusted proportional hazards regression models (Table 6), the use of diltiazem was associated with a 70% increase in coronary artery disease rehospitalization or all-cause mortality compared with the use of beta-blockers (HR 1.7; 95% CI 1.1 to 2.6). When we adjusted for propensity scores in a multivariate model, the association was decreased and was no longer statistically significant (HR 1.4; 95% CI 0.80 to 2.4). Similar nonsignificant HR were found when we restricted the analysis to veterans without asthma, COPD, hypotension or heart block (Table 6).
This study is the first to compare the long-term health outcomes of diltiazem treatment with beta-blocker treatment for unstable angina. Veterans treated with diltiazem for unstable angina at hospital discharge had no overall survival advantage compared with veterans treated with beta-blockers. The composite end point of rehospitalizations or death was associated with a nonsignificant, moderately elevated risk for diltiazem users compared with beta-blocker users. However, CIs were wide and included the possibility that diltiazem was associated with a reduced risk.
The only other published follow-up study comparing diltiazem to beta-blockers for unstable angina treatment followed patients for an average of 5 months and reported few end points (9). The authors concluded that hemodynamic and symptom control were similar in both groups and that events were too scarce to draw inferences. Other published trials compared verapamil, a diphenylalkylamine calcium antagonist, with beta-blocker therapy and found beta-blockers to be more effective at short-term symptom control for unstable angina (25)and to have equivalent long-term mortality outcomes among subjects with stable angina (26). Of note in this latter trial was a borderline, significantly increased risk of combined nonfatal cardiovascular events and deaths (HR 1.23: 95% CI 0.95 to 1.58) for metoprolol users compared with verapamil users. Other trials have compared dihydropyridine calcium antagonists with beta-blockers and have found either no advantage (27)or adverse outcomes (6)associated with dihydropyridine treatment.
Several limitations of this study need to be addressed. First, beta-blocker and diltiazem treatments were not randomly assigned to veterans: medications were chosen by the providers presumably based on the clinical profile and medical history of the veteran. We found evidence of preferential use of diltiazem in our data among veterans with COPD and heart failure, and less use among veterans who had undergone a revascularization procedure. Diltiazem users seemed to be less healthy than beta-blocker users based on these retrospective data. Observational studies of this nature may produce spurious associations if confounding by either drug indication or disease severity is not accounted for in the analysis. We attempted to minimize potential confounding by propensity score adjustments and by restriction to subgroups without relative contraindications to beta-blocker treatment. Empirically, both methods produced similar results. We cannot, however, rule out the possibility of confounding by factors that were not assessed in this study.
Unstable angina was defined primarily using ICD-9 codes for subacute and chronic ischemic heart disease. The retrospective cohort design of this study did not allow us to screen potential subjects at the time of their hospitalization, so we used discharge diagnoses and medical records to determine unstable angina status. Those veterans with unstable angina who entered the VAPSHCS during the study dates and did not receive the appropriate ICD-9 code would not have been included in this cohort. Results cannot be generalized to this group of veterans.
We collected hospitalization and mortality data from two sources to increase the sensitivity of identifying these events. Mortality data were collected from Washington State death certificate files and from VA benefits tracking. Deaths that occurred outside of Washington State that were not known to the VA benefits office would have been missed with this approach. Because veterans are entitled to burial funds, this VA benefit is claimed by 90% to 95% of veteran relatives (28). Rehospitalization data for Washington State residents were collected from the PTF and CHARS databases; non-Washington State residents were excluded to avoid an event ascertainment bias. Nonetheless, some events of Washington State residents may not have been identified. Out-of-state rehospitalizations at non-VA facilities for veterans who were Washington State residents at the time of their index event would not have been included in our data sources, although VA hospitalization data were uniformly available. The extent to which these potential omissions of deaths and rehospitalizations bias our findings depends on the degree to which the outcome is associated with discharge medication treatment and on the number of omissions, which is presumed to be small.
Our study was designed to include only veterans who were prescribed monotherapy at discharge and who filled such a prescription at the VAPSHCS pharmacy. We did not collect data on veterans who were discharged on more than one type of calcium channel blocker. We have no information about whether veterans were compliant with the medications or whether they filled prescriptions received from other care providers at other pharmacies, VA or otherwise. We did not include information on follow-up treatment because outpatient refill patterns at the VAPSHCS pharmacy were sporadic, presumably reflecting the increased use of non-VA pharmacies for outpatient medications.
Our pilot data for this study had estimated beta-blocker monotherapy to be more common than what we actually found over the 6 years of follow-up. The small numbers of beta-blocker users effectively increased the amount of variability in our risk estimates and produced wide CIs. More cohort members and/or longer follow-up would increase the precision of our risk estimates.
Despite scientific evidence supporting treatment effectiveness and that beta-blockers are well tolerated in most patients with ischemic heart disease, these agents are chronically underused in the clinical setting (11,12,29–31). In our cohort, diltiazem users were more likely to have a history of heart failure or COPD compared with beta-blocker users, and both clinical conditions were associated with an increased risk of death and, to a lesser extent, rehospitalizations. The addition of beta-blockers in the treatment of heart failure patients may be beneficial in some situations (32–37), and clinical guidelines emphasize the importance of using low-dose beta-blockade in the management of unstable angina patients, even those with COPD (38,39). On the other hand, clinical benefit has not been demonstrated for diltiazem in patients with heart failure, and available data suggest that it may be deleterious (40–42). In the absence of meaningful clinical data supporting the long-term use of diltiazem for the treatment of unstable angina, it may be prudent to follow AHCPR guidelines and prioritize the use of beta-blockers in this population.
☆ This study was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service, Seattle, Washington.
- Beneficiary Identification and Record Locator Subsystem
- coronary artery bypass grafting
- Comprehensive Hospital Abstract Reporting System
- chronic obstructive pulmonary disease
- International Classification of Disease, 9th Edition
- myocardial infarction
- patient treatment file
- Veterans Affairs Puget Sound Health Care System
- Received March 17, 1998.
- Revision received June 23, 1998.
- Accepted July 9, 1998.
- American College of Cardiology
- Report of the Holland Interuniversity Nifedipine/Metoprolol Trial (HINT) Research Group
- The Multicenter Diltiazem Postinfarction Trial Research Group
- Theroux P.,
- Taeymans Y.,
- Morisette D.,
- et al.
- Andre-Fouet X.,
- Usdin J.P.,
- Gayet C.,
- et al.
- Schreiber T.L.,
- Elkhatib A.,
- Grines C.L.,
- Rogers W.J.,
- Bowlby L.J.,
- Chandra N.C.,
- et al.
- Cruickshank J.M.
- Cook J.R.,
- Bigger J.T.,
- Kleiger R.E.,
- Fleiss J.L.,
- Steinman R.C.,
- Rolnitzky L.M.
- Drugs Facts and Comparisons. St. Louis: Fact and Comparisons, 1996.
- Cox D.R.,
- Oakes D.
- Rosenbaum P.R.,
- Rubin B.D.
- Rehnqvist N.,
- Hjemdahl P.,
- Billing E.,
- et al.
- Muller J.E.,
- Turi Z.G.,
- Pearle D.L.,
- et al.
- Pashos C.L.,
- Normand S.L.T.,
- Garfinkle J.B.,
- et al.
- Vantrimpont P.,
- Rouleau J.L.,
- Wun C.C.,
- et al.
- Szabo B.M.,
- Crijns H.J.G.M.,
- Wiesfeld A.C.P.,
- van Veldhuisen D.J.,
- Hillege H.L.,
- Lie K.I.
- Heidenreich P.A.,
- Lee T.T.,
- Massie B.M.
- Doughty R.N.,
- Rodgers A.,
- Sharpe N.,
- MacMahon S.
- ↵Clinical Practice Guideline. Unstable Angina: Diagnosis and Management. US Department of Health and Human Services. AHCPR Publication No. 94-0602, March 1994.
- Braunwald E.,
- Jones R.H.,
- Mark D.B.,
- et al.
- Goldstein R.E.,
- Boccuzzi S.J.,
- Cruess D.,
- Nattel S.