β-Blocker Dosages and Mortality After Myocardial Infarction: Data From a Large Health Maintenance Organization

Background  Although long-term β-blocker therapy has been found beneficial in patients after an acute myocardial infarction, these drugs are greatly underused by clinicians. Moreover, the dosages of β-blockers used in randomized controlled trials appear to be much larger than those routinely prescribed.

Objective  To determine whether an association exists between the dosage of β-blockers prescribed after a myocardial infarction and cardiac mortality.

Methods  We performed a retrospective cohort study of 1165 patients who survived an acute myocardial infarction from January 1, 1990, through December 31, 1992. These patients represent a subgroup of the 6851 patients hospitalized at northern California Kaiser Permanente hospitals.

Results  Of the 37.7% of patients prescribed β-blocker therapy, 48.1% were treated with dosages less than 50% of the dosage found to be effective in preventing cardiac death in large randomized clinical trials (lower-dosage therapy). Compared with patients not receiving β-blockers, those treated with lower-dosage therapy appeared to have a greater reduction in cardiovascular mortality (hazard ratio, 0.33; P=.009) than patients treated with a higher dosage (hazard ratio, 0.82; P=0.51), after adjustment for age, sex, race, disease severity, and comorbidities.

Conclusions  The dosages of β-blockers shown to be effective in randomized trials are not commonly used in clinical practice, and treatment with lower dosages of β-blockers was associated with at least as great a reduction in mortality as treatment with higher dosages. This suggests that physicians who are reluctant to prescribe β-blockers because of the relatively large dosages used in the large prospective clinical trials should be encouraged to prescribe smaller dosages.

DESPITE the fact that numerous large, randomized, placebo-controlled trials have demonstrated a beneficial effect of long-term β-blocker therapy in patients after an acute myocardial infarction,¹ these drugs are greatly underused by practicing clinicians.^2-5 This underuse appears to be more of a problem in the elderly.⁶ Moreover, the dosages of β-blockers used in these large, randomized, controlled trials (160 mg of propranolol hydrochloride,^7,8 200 mg of metoprolol tartrate per day,⁹ and 100 mg of atenolol per day¹⁰) appear to be much larger than those routinely prescribed.² In fact, Viskin et al² found that 89% of patients who were discharged from the hospital and prescribed a β-blocker regimen received dosages 50% or less of the dosages shown to be effective in randomized clinical trials. The effectiveness of these lower dosages in reducing morbidity and mortality is unknown. Furthermore, since it is unlikely that any randomized controlled trials will be designed to examine this question, we analyzed data from a large study of patients with acute myocardial infarction to determine the dosages of β-blockers commonly prescribed to infarct survivors and whether these clinically prescribed dosages of β-blockers are associated with the same reduction in cardiovascular mortality as the larger dosages used in randomized controlled trials.

This was a retrospective cohort study of patients who survived an acute myocardial infarction. The patient cohort included members of the Kaiser Permanente Medical Care Program of Northern California hospitalized from January 1, 1990, through December 31, 1992. More than 90% of program members who experience an acute myocardial infarction are hospitalized at 1 of the program's 16 hospitals. The remainder are transferred to a program hospital when they are stable. Patients transferred from nonprogram hospitals, regardless of the length of stay, were also included.

Between January 1, 1990, and December 31, 1992, 6851 patients aged 25 years or older were discharged with a principal diagnosis of acute myocardial infarction (International Classification of Diseases, Ninth Revision, Clinical Modification, code 410)¹¹ from among 16 health maintenance organization hospitals. An extensive chart review was performed in a subcohort of 1165 patients who were chosen from 3 hospitals with higher rates of angiography and 4 hospitals with lower rates. This study was initially undertaken to examine the relationship between revascularization rates at individual hospitals and clinical outcome.¹² The present study represents a secondary data analysis in this subcohort. Of the 1165 patients studied, 115 (9.9%) died during this index hospitalization. The remaining 1050 patients composed the study cohort.

For each of the 1050 patients, follow-up began on the day of hospital discharge and continued until December 31, 1993, or until date of death or disenrollment from Kaiser Permanente. The primary outcome variable in this study was cardiovascular death. Patients who died of a noncardiovascular cause (n=50) were censored at the time of death. Cardiovascular deaths (n=145) were identified by automated linkage of all cohort members to computerized State of California death records. Cardiovascular death was defined as any death coded with an International Classification of Diseases, Ninth Revision, Clinical Modification, code 410 to 429.

Comorbid conditions, medical treatment, and clinical outcomes were identified from secondary diagnoses listed for the index admission and from all diagnoses of other hospitalizations in the previous 5 years. A Charlson comorbidity index score¹³ was calculated and used to adjust regression analyses predicting clinical outcomes.

A congestive heart failure (CHF) score was developed to quantify the signs and symptoms of heart failure. The scoring was graded on a scale of 0 to 3: no mention of CHF (0), documentation of heart failure (1), documentation of functional class IV or intractable failure (2), and documentation of symptomatic hypotension or cardiogenic shock (3). This variable was treated as a categorical (rather than ordinal) variable in the regression analyses.

The severity of a myocardial infarction was assessed by means of the presence or absence of new Q-waves, infarct location, and the severity of CHF (as assessed by the CHF score, described above). Risk factors for cardiovascular disease and additional medical treatments (administration and timing of thrombolytic agents, in-hospital and discharge use of aspirin, β-blockers, calcium channel blockers, digoxin, and diuretics) were obtained from hospital records. The results of diagnostic tests (echocardiography, nuclear wall motion studies, Holter monitoring, and exercise, dipyridamole, and dobutamine hydrochloride stress tests) were recorded from both inpatient and outpatient records.

The type and daily dosage of the β-blocker prescribed at the time of hospital discharge was recorded. The term effective dosages was defined as the dosage of β-blockers shown to prolong survival in the large randomized clinical trials of β-blocker therapy.² Accordingly, we defined effective dosage as follows: propranolol hydrochloride, 160 mg/d^7,8; metoprolol tartrate, 200 mg/d⁹; or atenolol, 100 mg/d.¹⁰

A Cox proportional hazards model was used to identify independent predictors of cardiovascular mortality. The variables entered into the model were age, sex, race, infarct location, type of infarction, previous angina, previous infarction, heart failure score, Charlson index, history of smoking, thrombolytic use, mechanical revascularization within 3 months of discharge, aspirin and β-blocker treatment at the time of hospital discharge, and the dosages of β-blockers used (expressed as the percentage of the "effective" β-blocker dosage). The calculated percentage of effective dosage was dichotomized at 50%, as this was approximately the median dosage prescribed at discharge. Hazard ratios (HRs) were calculated as the exponentiation of coefficients from the proportional hazards model.

To check for the possibility that measured and unmeasured covariates may cluster by hospital, analyses were repeated by means of logistic regression and generalized estimating equations. These models treat the hospital as a clustering variable. For these analyses, the dependent variable was cardiac death within 2 years (yes or no). Noncardiac deaths that occurred during this period were treated as nonevents.

β-Blockers were prescribed to 396 (37.7%) of the 1050 patients included in this study. Information on the type and dosage of β-blocker prescribed at discharge was available on 365 (92%) of these patients. The mean length of follow-up in this study was 747 days. Patients who were treated with β-blocker therapy tended to be a lower-risk group, as they were younger, less likely to be female, and less likely to have had previous angina or a previous myocardial infarction compared with patients who were not treated with β-blockers. They also had a lower Charlson comorbidity index, had a lower CHF score, were more likely to have been treated with thrombolytic therapy, and were less likely to be treated with digoxin, calcium channel blockers, and diuretic therapy than patients not receiving β-blocker therapy (Table 1).

Approximately half of all patients treated with β-blockers received a dosage 50% or less of the dosage found to be "effective" in prolonging survival in large prospective β-blocker trials (Figure 1). The median β-blocker dosage prescribed at discharge was 53% of what was defined as an effective dosage. Patients treated with less than 50% of an effective dosage (low dosage) were similar to those treated with 50% or more of an effective dosage (higher dosage) with respect to sex, infarct location, type of infarct, Charlson index, and diuretic use (Table 1). However, patients who received lower dosages of β-blockers were older, had a greater CHF score, and were more likely to be prescribed digoxin at discharge. Despite appearing to be a higher-risk group of patients, these patients experienced a lower cardiovascular mortality rate than those treated with higher dosages of β-blockers (3.43% vs 6.88%) (Figure 2).

Stress testing was performed after acute myocardial infarction in 324 patients (31.8%). The use of stress testing was more common in patients who received higher-dosage therapy than in patients treated with lower-dosage therapy or patients not treated with any β-blocker (48.1%, 39.2%, and 25.1%, respectively). Of patients who underwent stress testing, patients treated with lower-dosage β-blockers were more likely to have had a positive test result and to have undergone a revascularization procedure within 3 months than patients prescribed higher-dosage therapy (Figure 3).

To determine the independent predictors of mortality in this cohort, a Cox proportional hazards model was constructed. This model demonstrated that greater age, higher Charlson comorbidity index, and CHF (CHF score >0) were predictive of worse survival (Table 2). Having undergone revascularization within 3 months of discharge, either percutaneously or by coronary artery bypass grafting, was associated with a 67% lower mortality (P<.001). Aspirin use was associated with a 21% reduction in mortality (P=.18). Receiving lower-dosage β-blockers was associated with a 67% reduction in cardiovascular mortality (P<.001), whereas receiving higher-dosage therapy was associated with an 18% reduction in cardiovascular mortality (P=.51) (Table 2). The difference in the associated reduction in mortality between the lower- and higher-dosage therapy (67% vs 18%) was statistically significant (P=.02). In a separate model, we determined that the overall effect of having received any dosage of a β-blocker was a 38% reduction in cardiovascular mortality, compared with patients not prescribed β-blockers at discharge (HR, 0.62; P=.06). The results did not change when a logistic regression model with generalized estimating equations was used.

Overall, only 37.7% of patients in this cohort were treated with β-blocker therapy, confirming previous reports that such therapy is underused in patients after a myocardial infarction.² Recommendations for the use of long-term β-blocker therapy in survivors of a myocardial infarction are based on large clinical trials demonstrating an improvement in survival. However, these studies invariably used relatively large dosages of β-blockers. In the β-Blocker Heart Attack Trial and the Norwegian trial of propranolol, propranolol hydrochloride was given in a dosage of 160 mg/d.^7,8 In the studies evaluating metoprolol and atenolol in acute myocardial infarction, the prescribed dosage of metoprolol tartrate was 200 mg/d⁹ and that of atenolol was 100 mg/d.¹⁰

In the present analysis of clinical practice in a large staff model health maintenance organization, we demonstrated that β-blocker use was associated with 38% lower cardiovascular mortality (HR, 0.62; 95% confidence interval, 0.38-1.01; P=.06) compared with patients who were not treated with β-blockers. This relative reduction in cardiovascular mortality is similar to what has been reported in randomized controlled trials.¹ In addition, we observed that the actual dosage of β-blockers prescribed at the time of discharge from the hospital was much lower than the dosages used in randomized controlled trials, confirming the observations made by Viskin et al.² The most important observation of this study was, however, that treatment with lower dosages of β-blockers (<50% of "effective" dosages) was associated with a significantly greater reduction in cardiovascular mortality (HR, 0.33; 95% confidence interval, 0.14-0.76; P=.009) than treatment with higher dosages (HR, 0.82; 95% confidence interval, 0.44-1.51; P=.51).

Before this study, the efficacy of long-term lower-dosage β-blocker therapy in patients who survive a myocardial infarction had not been examined. In a recent prospective randomized study, Aronow et al¹⁴ examined the effect of lower-dosage propranolol in asymptomatic elderly patients with complex ventricular arrhythmias, the majority of whom had had a previous Q-wave infarct. Interestingly, the mean dosage of propranolol hydrochloride was 85 mg/d, 53% of an "effective" dosage by our definition. In this study, there was a 37% reduction in total cardiac death and a 47% reduction in sudden cardiac death during a mean (±SD) follow-up of 29±14 months, suggesting that in elderly patients with asymptomatic ventricular arrhythmias, low-dosage propranolol is effective in reducing both sudden and total cardiac mortality.

Randomized controlled trials are needed to confirm our observation that lower-dosage β-blocker therapy is effective in reducing cardiovascular death after acute myocardial infarction and to explore the mechanism by which this reduction in mortality occurs. Whether lower-dosage therapy is more effective than higher-dosage therapy or whether patients treated with lower dosages of β-blockers are simply less likely to develop significant adverse effects and therefore less likely to discontinue therapy should be explored. Unfortunately, we did not collect information on whether patients continued to receive their β-blockers after discharge and therefore cannot comment on this possible mechanism.

The major limitation of this or any observational study of treatment effects is that patients were not randomly assigned to receive high- or low-dosage β-blocker therapy. Thus, it is possible that "healthier" patients could have been preferentially treated with low-dosage therapy. If such a selection bias occurred, then the association between patients receiving low-dosage therapy and reduced mortality would be confounded. However, if anything, it appears that those patients treated with lower-dosage β-blockers were at slightly higher risk. Patients treated with lower-dosage β-blockers were older and had a higher CHF score, similar Charlson score, similar rates of diuretic use, and similar rates of revascularization compared with patients treated with higher-dosage β-blocker therapy. In addition, the 2 groups did not appear to differ in the severity of their coronary artery disease, as similar proportions of patients in each group had previous angina and previous myocardial infarction. Furthermore, we reviewed the results of all of the noninvasive stress tests performed. The percentage of patients who underwent stress testing who developed a positive stress test was 43%, 45%, and 31% for patients who did not receive any β-blocker, lower-dosage β-blocker, and higher-dosage β-blocker therapy at discharge, respectively (Figure 3). Thus, patients treated with higher-dosage therapy were slightly less likely to have inducible ischemia than were patients treated with lower dosages of β-blockers.

β-Blocker therapy has been clearly shown to save lives when administered after a myocardial infarction. Despite this, such therapy is underused. One possible explanation for this underuse is that physicians believe that some patients will not tolerate the high dosages that have been shown to be effective in randomized clinical trials. Physicians should be encouraged to use β-blocker therapy after a myocardial infarction in all patients who do not have contraindications. On the basis of our findings, if a patient cannot tolerate the medication at dosages that have been shown to be effective, then lower dosages should be considered. While our data do suggest that the reduction in mortality associated with lower-dosage β-blocker therapy is at least as great as that of the higher dosages, confirmation of this finding, in a randomized controlled trial, is needed.

Accepted for publication August 20, 1997.

This study was supported by The Permanente Medical Group, Oakland, Calif.

Reprints: Hal V. Barron, MD, 460 Point San Bruno Blvd, South San Francisco, CA 94080-4990 (e-mail: barron@ep4.ucsf.edu).