NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Helfand M, Peterson K, Christensen V, et al. Drug Class Review: Beta Adrenergic Blockers: Final Report Update 4 [Internet]. Portland (OR): Oregon Health & Science University; 2009 Jul.

Cover of Drug Class Review: Beta Adrenergic Blockers

Drug Class Review: Beta Adrenergic Blockers: Final Report Update 4 [Internet].

Show details

Results

Overview

Searches identified 6325 citations, with 606 new in Update 4. The results of study selection are outlined in Figure 1. Dossiers were received for Update 4 from the manufacturers of carvedilol, carvedilol controlled release, and nebivolol. Studies excluded at the full text level are listed in Appendix D.

Figure 1

Figure 1

Results of study selection

Summary of Findings

Efficacy/effectiveness

Hypertension

  • Direct comparisons
    • There were no head-to-head trials of different beta blockers on long-term (≥6 months) health or quality-of-life outcomes.
    • No consistent differences between beta blockers in quality-of-life outcomes were found in shorter-term, head-to-head trials of beta blockers.
  • Placebo-controlled trials
    • Long-term placebo-controlled trials of propranolol and atenolol were found, however no reliable indirect comparisons can be made from them.
  • Gaps: Long-term effectiveness; quality of life

Angina

  • Direct comparisons
    • There were no significant differences in exercise tolerance or attack frequency in 6 head-to-head trials of carvedilol compared with metoprolol, pindolol compared with propranolol, betaxolol and propranolol, or betaxolol compared with metoprolol in patients with stable angina.
    • Atenolol and bisoprolol were equivalent in angina patients with chronic obstructive pulmonary disease.
    • Atenolol and labetalol (when combined with chlorthalidone) were equivalent in angina patients with hypertension.
  • Placebo-controlled trials
    • One short-term, placebo-controlled trial of propranolol did not add any meaningful evidence of comparative efficacy in attack frequency or exercise parameters.

After coronary artery bypass graft

  • Direct comparisons
    • There were no head-to-head trials of beta blockers in adults following coronary artery bypass graft.
  • Placebo-controlled trials
    • Two placebo-controlled trials suggested that long-term use of a beta blocker after coronary artery bypass graft does not improve mortality or other outcomes. For example, the MACB Study Group conducted a fair-quality trial that randomized 967 patients (85.5% male, median age 64 years) to metoprolol 200 mg once daily or placebo within 5 to 21 days following coronary artery bypass graft and measured the effects of treatment on death and cardiac events. No differences between metoprolol and placebo were found in mortality (3.3% compared with 1.8%; P=0.16) or in ischemic events (myocardial infarction, unstable angina, need for additional coronary artery bypass graft, or percutaneous transluminal coronary angioplasty).
  • Gaps: long-term direct comparisons

Recent myocardial infarction

  • Direct comparisons
    • One fair-quality head-to-head trial found no differences in mortality after 1 year between atenolol and propranolol, but this was a relatively small trial.
    • One fair-quality head-to-head trial found no differences in time to serious cardiovascular events between carvedilol and atenolol.
    • One fair-quality head-to-head trial found no differences in time to first cardiac adverse event or all-cause mortality between carvedilol and metoprolol tartrate.
  • Placebo-controlled trials
    • In placebo-controlled trials, similar mortality reductions were reported for acebutolol, metoprolol tartrate, propranolol, and timolol for patients following myocardial infarction without other complications. Similar reductions in sudden death and reinfarction were reported for metoprolol tartrate and timolol and in sudden death for propranolol. Carvedilol is the only beta blocker shown to reduce mortality in post-myocardial infarction patients who are already taking an ACE (angiotensin-converting enzyme) inhibitor. No studies of carvedilol phosphate (extended-release carvedilol) in patients with recent myocardial infarction were identified. Carvedilol reduced mortality and reinfarction in 1 placebo-controlled trial of patients with a mean left ventricular ejection fraction of greater than 32.8% (Carvedilol Post-Infarct Survival Control in Left Ventricular Dysfunction [CAPRICORN] trial).

Heart failure

  • Direct comparisons
    • There were no direct comparator trials comparing 2 or more of the drugs proven to reduce mortality (bisoprolol, carvedilol, and sustained release metoprolol succinate).
    • In the Carvedilol or Metoprolol European Trial (COMET) trial, carvedilol was superior to metoprolol tartrate reducing all-cause mortality (number needed to treat, 18) after a mean follow-up of 58 months in patients with mild to moderate heart failure.
    • No differences were found between carvedilol and metoprolol tartrate in improving symptoms (quality of life; New York Heart Association classification) or exercise capacity in 4 smaller head-to-head trials.
    • Improvements in New York Heart Association function class and on walking distance (6-minute walk test) were similarly slight for both carvedilol and nebivolol.
  • Placebo-controlled trials
    • Bisoprolol, metoprolol succinate, and carvedilol have each reduced total mortality, as a planned primary endpoint, by approximately 35%.
    • Based on findings from the COPERNICUS trial (N=2289), carvedilol is designated as the beta blocker with the most direct, strongest evidence of having a mortality benefit in patients with severe heart failure. In a post-hoc subgroup analysis of 795 patients from the good-quality MERIT-HF trial, metoprolol succinate has also demonstrated a mortality reduction relative to placebo similar to that for carvedilol in patients who had a similar mortality risk.
    • In the SENIORS trial (N=2128), which involved patients who were, overall, older (mean age of 76 years) and healthier than in the prior major trials (higher mean left ventricular ejection factor, lower annual placebo mortality rate), nebivolol was superior to placebo in reducing the risk of the primary composite outcome of all-cause mortality or cardiovascular hospital admission (31.1% compared with 35.3%; hazard ratio, 0.86; 95% CI, 0.74 to 0.99). When components of the primary outcome were examined individually as secondary outcome measures, differences between nebivolol and placebo were no longer statistically significant.
    • We found no trials that directly evaluated the effects of carvedilol phosphate, the long acting form of carvedilol, on mortality in adults with heart failure. Approval of the heart failure indication for carvedilol phosphate was based on “equivalence of pharmacokinetic and pharmacodynamic parameters between carvedilol phosphate and conventional carvedilol tablets.”

Atrial arrhythmia

  • Direct comparisons
    • There were no differences between bisoprolol 10 mg and carvedilol 50 mg in preventing relapse of atrial fibrillation in patients subjected to cardioversion of persistent atrial fibrillation (>7 days).
  • Placebo-controlled trials
    • Atenolol, nadolol, and pindolol, but not labetalol, were effective in controlling the ventricular rate, while labetalol was no more efficacious than placebo based on findings from a good-quality systematic review examining 12 studies of rate control in patients with chronic atrial fibrillation.
    • One placebo-controlled trial found that metoprolol CR/XL 100 to 200 mg was effective in preventing relapse of atrial fibrillation/flutter after cardioversion. Over 6 months, atrial fibrillation or flutter relapse rates were significantly lower in patients taking metoprolol CR/XL. Death rates were similar. The study was not powered to examine mortality.
    • A study examining the effects of carvedilol in managing patients with concomitant atrial fibrillation and heart failure found that when added to digoxin, carvedilol significantly improved mean left ventricular ejection fraction scores and reduced severity of symptoms/functional capacity when compared to digoxin alone. There were no differences between monotherapies of carvedilol or digoxin.

Migraine

  • Direct comparisons
    • Head-to-head trials showed no difference in efficacy in reduction of attack frequency, severity, headache days, or acute tablet consumption, or in improvement in any subjective or composite index in any of the comparisons made (atenolol, metoprolol durules, metoprolol, or timolol compared with propranolol or nebivolol compared with metoprolol).
  • Placebo-controlled trials

Bleeding esophageal varices

  • Direct comparisons
    • One small head-to-head trial showed no difference between atenolol 100 mg and propranolol 40 to 160 mg in rates of non-fatal/fatal rebleeding and all-cause mortality.
  • Placebo-controlled trials
    • Results of 1 trial of nadolol and 8 small placebo-controlled trials of immediate-release and 2 formulations of extended-release propranolol for the secondary prevention of bleeding esophageal varices secondary to cirrhosis and schistosomiasis did not provide any additional indirect evidence of the comparative efficacy across beta blockers in these clinical outcomes. The somewhat mixed results across the placebo-controlled trials of propranolol suggest that treatment initiation interval may have an effect on rebleeding rates.

Harms

  • There were no consistent significant differences between beta blockers in head-to-head trials in overall adverse events, withdrawals due to adverse events, or individual adverse events.

Subgroups

  • A meta-analysis (see Table 16) suggested that beta blockers are equally effective in reducing mortality in subpopulations stratified by gender and race.
  • There is insufficient evidence to draw conclusions about the effects of beta blockers on perinatal mortality or preterm birth.
Table 16

Table 16

Results of Shekelle (2003) meta-analysis by gender, race, and diabetics

Key Question 1. Do beta blocker drugs differ in efficacy or effectiveness?
Key Question 1a. For adult patients with hypertension, do beta blockers differ in efficacy or effectiveness?

Summary

Beta blockers are equally efficacious in controlling blood pressure in patients with hypertension. No beta blocker has been demonstrated to be more efficacious or to result in better quality of life than other beta blockers, either as initial therapy or when added to a diuretic, ACE inhibitor, or angiotensin receptor blocker. Evidence from long-term trials is mixed; overall, beta blockers are generally less effective than diuretics, and are usually no better than placebo, in reducing cardiovascular events. The exception was 1 large trial in which treatment with metoprolol resulted in lower all-cause mortality than treatment with a thiazide diuretic.

Detailed Assessment

Primary or initial therapy

Beta blockers have been used as initial therapy in patients with hypertension and as additional therapy in patients whose blood pressure is not well controlled with a diuretic. In several head-to-head trials, beta blockers have similar effects on blood pressure control.3–11 No trials have examined whether beta blockers have different effects on all-cause mortality, cardiovascular mortality, or cardiovascular events among patients with hypertension.

By the time beta blockers became available, diuretics had already been shown to prevent cardiovascular events, primarily strokes. It was considered unethical to compare a beta blocker to placebo in patients who were likely to benefit from a diuretic. For this reason, most large, long-term trials of beta blocker therapy for hypertension used a comparison group taking a diuretic rather than a placebo. Unlike diuretics, then, beta blockers have not been clearly demonstrated to be more effective than placebo in reducing cardiovascular events when used as initial therapy in the general population of patients with hypertension.

The Medical Research Council trials, the International Prospective Primary Prevention Study in Hypertension, the Heart Attack Primary Prevention in Hypertension study, and the Metoprolol Atherosclerosis Prevention in Hypertensives study compared a beta blocker to a thiazide diuretic. Of these trials, only the 2 Medical Research Council trials compared a beta blocker to placebo. In 1 Medical Research Council trial, atenolol 50 mg daily was not better than placebo and less effective than a diuretic in adults ages 65 to 74 who had baseline blood pressures of 160/115 mm Hg or higher.12 In the other Medical Research Council trial, which recruited 17 361 patients with mild diastolic hypertension (90 to 109 mm Hg), beta blocker therapy (atenolol) reduced the odds for stroke, but only in nonsmokers and to a smaller degree than a low dose of a thiazide diuretic (bendrofluazide).13

Of the trials that compared a beta blocker with a diuretic, only 1 (Metoprolol Atherosclerosis Prevention in Hypertensives study) had any suggestion that the beta blocker was more effective. In that trial, deaths from heart attacks and strokes as well as total mortality were lower in the metoprolol treated group than in those treated with a diuretic (hydrochlorothiazide or bendroflumethiazide).14 The trial continues to be cited as strong evidence that beta blockers reduce mortality when used as primary treatment for hypertension. However, it must be weighed against the mixed results of the Medical Research Council trials and other trials of beta blockers compared with diuretics. In a good-quality meta-analysis of 10 trials published in 1998 or earlier, beta blockers were ineffective, or less effective than comparator drugs, in preventing coronary heart disease, cardiovascular mortality, and all-cause mortality (odds ratios 1.01, 0.98, and 1.05, respectively).15

Secondary treatment

The Systolic Hypertension in the Elderly Program (SHEP) trial examined a stepped approach for treating isolated systolic hypertension in the elderly.16 Chlorthalidone was the first step. Atenolol was prescribed if the blood pressure goal could not be achieved with chlorthalidone 25 mg daily. Compared to placebo, stepped treatment prevented 55 cardiovascular events per 1000 patients over 5 years. The contribution of beta blocker therapy with atenolol to the overall benefit is not clear; most of the benefit was attributed to chlorthalidone.

The Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (2002) did not include a beta blocker arm.17 Based on the results of this trial, the Joint National Committee on the Prevention, Detection, Evaluation and Treatment of High Blood Pressure (JNC-7) recommends a diuretic as the first-line treatment for most patients who have Stage 1 hypertension without compelling indications.18

Quality of life

There was no definitive evidence that 1 beta blocker yields a better quality of life than another for patients who have hypertension. Eight trials directly compared different beta blockers19 on changes of quality of life-associated measures. We excluded 2 trials of atenolol compared with propranolol based on poor-quality ratings.7, 20 The methods described in these publications were insufficient to rule out the possibilities that results were biased by inadequate randomization procedures (methods weren’t described and baseline characteristics weren’t reported) and/or by mishandling of missing data (attrition reasons not described and proportion of patients included in analyses not reported). Table 4 below summarizes the results of the fair-quality trials.

Table 4

Table 4

Quality-of-life outcomes in head-to-head trials of hypertensives

The strongest evidence of any differences between beta blockers came from a 4 week trial of captopril, enalapril, propranolol, and atenolol that used a larger sample size (N=360) and a parallel design.8 This was the only trial that is clearly industry-funded. Patients were all men that were “at least 21 years of age, employed or retired, educated at high-school level or equivalent, and married or living with a significant other.” Self-ratings of improvements were greater for atenolol than propranolol in Psychologic General Well-Being-measured self-control, distress overall and that caused by obsessions and hostility symptoms (Symptom Check List-90-R), and on global and social satisfaction indices from the Life Satisfaction Index. It remains unclear, however, as to whether these short-term results in men can be generalized to a broader population over a longer period of time.

The strength of the evidence from the remaining trials was limited by smaller sample sizes and, in the crossover trials, results that were averaged across treatment periods.5, 19, 21–23 Improvement in self-rated sexual interest (Minor Symptom Evaluation profile) was greater for atenolol than metoprolol CR in 1 trial of 60 patients (mean age 58 years; 43.3% male).5

Two trials of metoprolol succinate compared to nebivolol examined quality of life measures. One trial was conducted in Germany and compared nebivolol 5 mg to metoprolol succinate 95 mg. After 12 weeks of treatment, 48 men (ages 40 to 55) with newly diagnosed hypertension experienced decreased sexual function on metoprolol 95 mg, but not nebivolol 5 mg.23 However, the article provides insufficient detail to determine how the metoprolol succinate 95 mg product compares to the metoprolol succinate product available in the United States and Canada. In another trial, after 6 weeks of treatment of 46 adults with mild hypertension, sleep quality, as measured by scores on the Pittsburgh Sleep Quality Index, was improved by treatment with nebivolol 5 mg, but declined following treatment with metoprolol CR 100 mg.19

Two placebo-controlled trials reported the effect of long-term beta blocker therapy on quality of life in otherwise healthy patients who have hypertension (Evidence Tables 1 and 2). The Trial of Antihypertensive Interventions and Management24–26 had a serious flaw: only patients who were available for the 6-month blood pressure readings (79.4%) were included in the quality-of-life analysis. After 6 months, atenolol and placebo were similar on several dimensions from the Life Satisfaction Scale, the Physical Complaints Inventory, and the Symptoms Checklist, including summary (“total physical problems”, “overall psychological functioning”, “overall life satisfaction”), distress (“sexual physical problems”, “depression”, “anxiety”, “sleep disturbances”, “fatigue”), and well-being (“satisfaction with physical health”, “sexual satisfaction”). In the second trial,27 there were no differences between propranolol and placebo in cognitive or psychological measures after 1 year of treatment.

Key Question 1b. For adult patients with angina, do beta blockers differ in efficacy or effectiveness?

Summary

There were no differences in exercise tolerance or attack frequency in head-to-head trials of carvedilol compared with metoprolol, pindolol compared with propranolol, betaxolol compared with propranolol, and betaxolol compared with metoprolol tartrate in patients with chronic stable angina. Atenolol and bisoprolol were equivalent in angina patients with chronic obstructive pulmonary disease. Atenolol and labetalol (when combined with chlorthalidone) were equivalent in angina patients with hypertension.

Beta blockers that had intrinsic sympathomimetic activity reduced the resting heart rate less than other beta blockers, a potential disadvantage in patients suffering from angina pectoris. For this reason, experts recommend against using beta blockers with intrinsic sympathomimetic activity in patients with angina.

Detailed Assessment

In 1966 the first beta blocker, propranolol, was shown in a multicenter controlled trial to improve symptoms in patients with angina pectoris.28 Several other beta blockers (acebutolol, atenolol, metoprolol tartrate, metoprolol succinate, nadolol, propranolol, and propranolol long-acting) have been demonstrated to reduce symptoms of angina in placebo-controlled trials.

Most head-to-head trials of beta blockers in patients with angina pectoris observe patients for only 2 to 4 weeks of treatment.29–36 In these trials, exercise tolerance, attack frequency, or nitroglycerin use were generally similar at comparable doses.

Six fair-quality head-to-head trials evaluated angina symptoms after 2 or more months of treatment with beta blockers (Table 5, Evidence Tables 3 and 4). Mean ages ranged from 55 to 61.5 years and most subjects were men (71.5% to 100%), with the exception of 1 study, which included 40% male subjects.37 Exercise parameters were measured using bicycle ergometric testing in all but 2 trials,38, 39 which used a treadmill. One study, however, did not include exercise parameters in its study design.37 There were no significant differences in exercise tolerance or attack frequency. No significant differences were found between betaxolol 20 mg and metoprolol tartrate 100 mg on 5 of 6 health-related quality-of-life parameters. Compared with metoprolol tartrate (15%), however, significantly greater numbers of patients on betaxolol improved on the ‘Physical Function’ parameter (43%; P<0.01).37

Table 5

Table 5

Results of head-to-head trials in patients with angina

Over the long term, beta blockers may differ in their ability to prevent or reduce the severity of anginal attacks. In 1 fair-quality 2-year multicenter European trial, propranolol was better than placebo after 8 weeks but not after 24 weeks of treatment.40 Specifically, after 8 weeks propranolol 60 to 240 mg reduced the proportion of patients using nitroglycerin (57% compared with 73% in the placebo group; P=0.04) and increased the mean total work time by 48% compared with 13% (P=0.04). These effects were transient, however, and propranolol was equivalent to placebo on those parameters after 24 weeks of treatment. Propranolol and placebo had similar effects on the number of weekly angina attacks, the number of attack-free days, maximum workload, and exercise duration at 8- and 24-week endpoints. The relevance of this trial was limited because since the time it was conducted, the rate of progression of angina may have been altered by advances in treatment of atherosclerosis (for example statin therapy).

A good-quality meta-analysis identified 72 randomized controlled trials of a beta blocker compared with a calcium channel blocker and 6 trials comparing a beta blocker to a nitrate.41 This meta-analysis found that, in general, beta blockers had similar efficacy but fewer discontinuations due to adverse events than calcium channel blockers, but the authors did not report results for each beta blocker separately.

Key Question 1c. For adult patients who have undergone coronary artery bypass grafting, do beta blockers differ in efficacy or effectiveness?

We did not examine the short-term (4 to 10 days) use of beta blockers to prevent or control atrial tachyarrhythmias after coronary artery bypass graft.42–46 In addition to the beta blockers included in our review, esmolol, a very short-acting, intravenous beta blocker, is used postoperatively to control tachyarrhythmias.

In 7 trials, long-term use of a beta blocker after coronary artery bypass graft did not improve mortality or other outcomes (Evidence Tables 5 and 6). For example, the MACB Study Group conducted a fair-quality trial47 that randomized 967 patients (85.5% male, median age 64 years) to metoprolol 200 mg once daily or placebo within 5 to 21 days following coronary artery bypass graft and measured the effects of treatment on death and cardiac events. No differences between metoprolol and placebo were found in mortality (3.3% compared with 1.8%; P=0.16) or in ischemic events (myocardial infarction, unstable angina, need for additional coronary artery bypass graft or percutaneous transluminal coronary angioplasty).

Key Question 1d. For adult patients with recent myocardial infarction, do beta blockers differ in efficacy or effectiveness?

Summary

Table 6 summarizes evidence from meta-analyses and major trials of beta blockers in patients with recent myocardial infarction. Timolol was the first beta blocker shown to reduce total mortality, sudden death, and reinfarction outcomes in the Norwegian Multicenter Study.48 Subsequently, similar total mortality reductions were reported across trials of acebutolol,49 metoprolol tartrate (Goteborg), and propranolol (Beta Blocker Heart Attack Trial) in comparable populations. In addition, similar benefits in sudden death were reported for propranolol50 and metoprolol tartrate51, 52 and in reinfarction for metoprolol tartrate.52

Table 6

Table 6

Comparison of outcomes of mortality-reducing beta blockers in patients following myocardial infarction

Carvedilol reduced reinfarction rates in the Carvedilol Post-Infarct Survival Control in Left Ventricular Dysfunction (CAPRICORN) trial, which recruited stable inpatients with recent myocardial infarction and a left ventricular ejection fraction of 40% or less. Carvedilol is the only beta blocker shown to reduce mortality in post-myocardial infarction patients who are already taking an ACE inhibitor. An extended-release form of carvedilol (carvedilol phosphate) was approved by the US Food and Drug Administration in October 2006. No studies of carvedilol phosphate in patients following myocardial infarction were identified through literature searches. Approval of the left ventricular dysfunction following myocardial infarction indication for carvedilol phosphate was based on pharmacokinetic and pharmacodynamic data that demonstrated bioequivalence with carvedilol.

Indirect comparisons of beta blockers across these trials must be done with caution because the study populations differed in duration, the presence or absence of left ventricular dysfunction, the dose and timing of therapy, and the use of other medications.

Head-to-Head Trials

No consistent differences between beta blockers were found in 3 head-to-head trials in post-myocardial infarction patients.53–55 A 6-week trial comparing atenolol 100 mg to propranolol 120 mg had inconclusive results.53 The second trial, an open-label study with a median follow-up of 1.6 years, compared carvedilol to atenolol. Patients in this study had mean left ventricular ejection fraction 53.9% at baseline. The primary outcome of the study was the change in left ventricular ejection fraction at 1 year; time to first serious cardiovascular event was a secondary endpoint. No significant difference was found between the 2 interventions in either change in left ventricular ejection fraction (P=NR) or time to occurrence of a serious cardiovascular event (P=0.524), which remained when controlling for use of diuretics (P=0.990).56 However, these results are not conclusive, as the study’s authors acknowledge that the study was underpowered to detect such a difference for this secondary outcome. A study of 313 patients comparing metoprolol tartrate 100 mg twice daily to carvedilol 25 mg twice daily for a mean of 13.4 months found no differences in time to first composite cardiac adverse event (all-cause death, postinfarction angina, heart failure, rehospitalization, and revascularization) or time to composite hard event (cardiovascular death and nonfatal reinfarction).55 There were statistically significant differences in 5 of 8 health-related quality-of-life domains measured using the Short Form-36 questionnaire (adjusted for age and baseline differences) favoring the carvedilol group.55

Placebo-controlled Trials

Because there are so few comparative trials, inferences about the comparative effectiveness of beta blockers in post-myocardial infarction patients must be made on other grounds. The criteria for making these comparisons might include:

  1. Demonstration of reduced mortality in large, multicenter placebo-controlled trials
  2. Degree of mortality reduction compared with other beta blockers
  3. Improvements in other outcomes
  4. Effectiveness studies and applicability of efficacy studies to current practice.

Mortality

Three systematic reviews have analyzed over 60 trials of beta blockers after myocardial infarction.57–59 The first (Yusuf, 1985) analyzed 22 long-term trials of beta blockers in acute myocardial infarction. Overall beta blockers reduced mortality by 23%, from an average of 10% to 8%. The second (Hjalmarson, 1997) found an average 20% mortality reduction in 24 trials of a total of 25000 patients.

A more recent review (Freemantle, 1999) used meta-regression to examine the relationship of characteristics of different beta blockers with the outcome of treatment.59 In their analysis of 24 long-term trials, cardioselectivity had no effect, but there was a near significant trend towards decreased benefit in drugs with intrinsic sympathomimetic activity. Individually, acebutolol (0.49; 0.25–0.93), metoprolol tartrate (0.80; 0.66–0.96), propranolol (0.71; 0.59–0.85), and timolol (0.59; 0.46–0.77) significantly reduced mortality, but there was insufficient data to distinguish among them. The analysis included just 1 trial of carvedilol, a pilot study in 151 post-myocardial infarction patients.60

Table 7 summarizes placebo-controlled trials that enrolled over 100 patients, had long-term follow-up (greater than 6 weeks), and met our other inclusion criteria. All of these trials were analyzed in the 1999 systematic review except for CAPRICORN, which was conducted from 1997 to 2000 at 163 sites in 17 countries and published in 2001.61 Unlike the other trials, CAPRICORN included only patients who had reduced left ventricular function (≤ 40%) after acute myocardial infarction as determined by echocardiography or cardiac catheterization. Patients with uncontrolled heart failure, such as those requiring intravenous diuretics, were excluded. Of 1959 subjects randomized to either carvedilol or placebo at an average of 10 days following a confirmed myocardial infarction, 1289 had no clinical signs of heart failure (Killip Class I), 593 had Killip Class II heart failure, and 65 had Killip Class III failure. The mean ejection fraction was 32.8%.

Table 7

Table 7

Summary of results from placebo-controlled trials of beta blocker therapy following myocardial infarction

The original primary endpoint was all-cause mortality. Subsequently, following a masked interim analysis in which the data and safety monitoring board found that overall mortality rates were lower than predicted, the CAPRICORN steering committee decided to adopt the co-primary endpoints of all-cause mortality together with all-cause mortality plus cardiovascular hospital admissions. There was no difference between carvedilol and placebo for the primary endpoint of mortality plus cardiovascular admissions (35% compared with 37% for placebo over 1.3 years, P=0.299). However, carvedilol reduced the original primary endpoint of total mortality in the first 30 days (19% compared with 33%; hazard ratio, 0.58; 95% CI, 0.33 to 1.02) 62 and over 1.3 years (12% compared with 15% for placebo over 1.3 years; number needed to treat, 30 or number needed to treat for 1 year, 43). The P value was 0.03, which, although nominally significant, did not meet the higher level of significance specified when the combined primary outcome measure was adopted.

CAPRICORN was the only trial to demonstrate the added benefit of a beta blocker in post-myocardial infarction patients taking ACE inhibitors or having undergone thrombolytic therapy or angioplasty. It was also the only trial specifically designed to evaluate a beta blocker in post-myocardial infarction patients who have asymptomatic left ventricular dysfunction. Based on CAPRICORN, the United States Food and Drug Administration gave carvedilol an indication to reduce mortality in “left ventricular failure after a myocardial infarction.”

The use of ACE inhibitors, thrombolytics, and angioplasty support the relevance of CAPRICORN to current care in the United States and Canada. However, the case for relevance could be strengthened if data were available to compare other practices and the quality of care between sites that recruited successfully and those that did not. Additional information about the recruitment of patients and the centers at which the CAPRICORN was conducted might provide additional insight into its relevance to current practice in the United States and Canada. Of the 1949 subjects in the trial, 83 were enrolled in the United States and 5 were from Canada. Five of the 6 top recruiting sites were in Russia, which enrolled the most subjects of any country (600). Of the 163 study sites, 24 enrolled only 1 subject. In their Lancet paper, the authors of CAPRICORN noted that “recruitment was slow in some countries where it was widely perceived that the case for beta blockers in all patients with myocardial infarction was proven.” The statement leaves open the possibility that, in North America, the subjects in CAPRICORN would already have been taking beta blockers.

Is the mortality reduction in CAPRICORN different from what would be expected from older trials of beta blockers in post-myocardial infarction patients or in patients with heart failure? The authors of the Lancet paper raised this question, noting that the 23% mortality reduction in CAPRICORN is identical to that found in meta-analyses of the older beta blocker trials.

Mortality was higher in CAPRICORN than in previous trials of beta blockers in post-myocardial infarction patients. The likeliest explanation is that many earlier trials included a broader mix of patients, including many who had normal left ventricular function and a better prognosis. Unlike many major trials, the CAPRICORN publication did not say how many patients with myocardial infarction were seen at the participating centers during the period of recruitment. It was also not clear what proportion of potentially eligible patients were excluded because they had an ejection fraction greater than 40%. These statistics would be useful in comparing the CAPRICORN subjects to the subjects of previous trials of beta blockers in post-myocardial infarction patients.

There was no direct evidence that other beta blockers shown to reduce mortality in post-myocardial infarction patients or in patients with heart failure worked as well as carvedilol in post-myocardial infarction patients with decreased left ventricular function and few or no symptoms of heart failure. While the older trials undoubtedly included some subjects with left ventricular dysfunction, it is difficult to determine how many, or how this subset did compared with post-myocardial infarction patients with normal left ventricular function.

Indirect evidence came from a good-quality meta-analysis.63 This analysis examined the relationship between the mortality reduction reported in each trial and the proportion of patients in the trial who had heart failure. There were few data on the effects of beta blockers after myocardial infarction in patients with documented left ventricular systolic dysfunction, but some studies included subjects with clinical findings of heart failure and reported the proportion of subjects that had these findings. As expected, studies that included patients with heart failure had higher mortality rates. The relative benefit of beta blockers on mortality after a myocardial infarction was similar in the presence or absence of heart failure.

Two retrospective subgroup analyses in heart failure patients from individual trials included in this meta analysis provided additional details supporting this hypothesis. One is from the Beta Blocker Heart Attack Trial (BHAT), a large, 3-month trial of propranolol published in 1980. In BHAT, 710 of 1916 subjects had a history of congestive heart failure prior to randomization. Propranolol lowered total mortality from 18.4% to 13.3% (a 27% reduction) in patients with a history of heart failure and from 7.8% to 5.9% (25% reduction) in patients who did not have a history of heart failure.64

The other retrospective subgroup analysis was from a 1980 placebo-controlled trial of metoprolol. At the time of randomization, 262 (19%) of the 1395 subjects had signs or symptoms of mild heart failure.65 Metoprolol or placebo was administered intravenously once, followed by oral metoprolol or placebo for 3 months, followed by open treatment with metoprolol for up to 2 years in all patients who had signs of ischemia. For patients with heart failure, mortality during the first year of the study was 28%, compared with 10% in subjects without signs of heart failure (P<0.0001). Among the subjects with heart failure at the time of randomization, metoprolol reduced mortality during the 3-month double-blind phase of the trial (14% compared with 27%, P<0.0009, number needed to treat=8).

Sudden death

Significant reductions in sudden death were reported in 2 of 3 trials of metoprolol tartrate,51, 52 1 trial of propranolol,50 and in 1 trial of timolol.48

Reinfarction

Significant reductions in reinfarction rates were reported in 1 of 2 trials of metoprolol tartrate52 and in 1 trial of timolol.48 Carvedilol was also associated with significantly reduced reinfarction rates in the CAPRICORN trial.

Arrhythmias

Evidence on the effect of beta blockers on post-myocardial infarction arrhythmias is unclear based on the available evidence. No significant difference in occurrence of post-myocardial infarction arrhythmia (defined as cardiac arrhythmia, fibrillation, or tachycardia) was found in placebo-controlled trials of acebutolol (1 trial)66 or propranolol (1 trial),50 while 1 placebo-controlled trial of propranolol found a small, but significantly higher, percentage of withdrawals due to serious ventricular arrhythmia in the placebo group (0.3% propanolol compared with 1.0% placebo; P<0.025).67 One trial of timolol found a significantly higher proportion of patients experiencing ventricular tachycardia with placebo use (20% placebo compared with 8.5% timolol; P=0.05) while the number of episodes of ventricular tachycardia (55 placebo compared with 10 timolol) was not statistically significant (data not provided).68

Two publications comparing carvedilol to placebo presented mixed results. One older trial found no significant difference between the 2 drugs in the rate of cardiac arrhythmias among all enrolled patients.60 In a subgroup analysis of patients (N=49/151; 32%) with baseline left ventricular ejection fraction <45%, carvedilol was associated with a significant decrease in serious cardiac events, a combined endpoint that included death, reinfarction, unstable angina, congestive heart failure, and ventricular tachycardia (P=0.04). The second publication, a post-hoc analysis of data from the CAPRICORN trial, compared rates of atrial and ventricular arrhythmias.69 As stated above, patients enrolled in the CAPRICORN trial had baseline left ventricular ejection fraction ≤40%. Atrial and ventricular arrhythmias were found to be less common with carvedilol use relative to placebo (hazard ratio, 0.48; 95% CI, 0.30 to 0.76; P=0.0015 and hazard ratio, 0.37; 95% CI, 0.24 to 0.58; P<0.0001, respectively. These values remained significant when controlling for history of arrhythmias. Carvedilol was also found to reduce the risk of all analyzed combinations of death and arrhythmia outcomes.

Withdrawals

Among the major trials, rates of withdrawal ranged from 9.3% to 36.6%, probably indicating differences in patient characteristics. Within studies, rates of withdrawal were generally similar for the beta blocker and placebo groups, with 3 exceptions. Rates of withdrawal were greater for metoprolol tartrate in 170 of 5 trials, pindolol in 1 trial,71 and propranolol in 1 trial.67

Key Question 1e. For adult patients with heart failure, do beta blockers differ in efficacy or effectiveness?

Summary

The United States Food and Drug Administration approval of metoprolol succinate for mild to moderate heart failure (New York Heart Association Class II or III) is based on MERIT-HF. United States Food and Drug Administration approval of carvedilol for severe heart failure is based on COPERNICUS. Its approval for mild to moderate heart failure is based on 5 other trials, 4 of which constitute the United States Carvedilol Study plus the Australian-New Zealand Heart failure study (see Table 10). Heart failure is not a United States Food and Drug Administration-approved indication for nebivolol or bisoprolol, which is a generic drug.

Table 10

Table 10

Patient characteristics and annualized mortality rates adjusted for active drug run-in periods in trials of beta blockers for heart failure

The main findings from placebo-controlled trials in patients with mild to moderate heart failure are summarized in Table 8. Reductions in mortality, sudden death, cardiovascular deaths, and death due to heart failure were similar for bisoprolol, metoprolol succinate, and carvedilol. Because several carvedilol trials performed in the United States had significant mortality reductions, the evidence for carvedilol may be more relevant to a United States population. When titrated gradually in stable patients, there is no difference in tolerability among these drugs.

Table 8

Table 8

Main findings in placebo-controlled trials of patients with mild to moderate heart failure

No studies of carvedilol phosphate (extended-release carvedilol) in patients with heart failure were identified through literature searches. Approval of the heart failure indication for carvedilol phosphate was based on pharmacokinetic and pharmacodynamic data that demonstrated bioequivalence with carvedilol.

In 2289 patients with severe heart failure (COPERNICUS), carvedilol clearly reduced mortality and the combined endpoint of mortality and hospitalizations. Carvedilol had the most direct, strongest evidence. In a post-hoc subgroup analysis of 795 patients from the good-quality MERIT-HF trial, metoprolol succinate demonstrated a mortality reduction relative to placebo similar to that for carvedilol in patients who had a similar mortality risk. This was a weaker level of evidence than that for carvedilol, but the lack of a direct comparator and the difficulty of comparing subjects from the different trials makes it uncertain whether one of these drugs is superior in patients with the various degrees of heart failure.

In the Carvedilol or Metoprolol European Trial (COMET) trial, a head-to-head trial conducted in patients with mild to moderate failure, carvedilol reduced mortality compared with metoprolol tartrate, the immediate-release form of metoprolol. In previous trials, however, metoprolol tartrate had not been proven to reduce mortality. The COMET trial does not resolve the question of whether carvedilol is superior to metoprolol succinate or bisoprolol, the preparations that have been shown to reduce mortality.

Detailed Assessment

Placebo-controlled trials

Mortality

Eight meta-analyses of placebo-controlled trials of various beta blockers in heart failure were published in the mid-1990’s through 2000 (Evidence Tables 9 and 10).73–80 In general, these meta-analyses found that beta blockers reduce mortality by about 30%, preventing 3.8 deaths per 100 patients in the first year of treatment. Nevertheless, the authors of the meta-analyses agreed that larger trials were needed before beta blockers could be recommended routinely for patients with heart failure.

The mortality benefits of seven beta blockers (atenolol, bisoprolol, bucindolol, carvedilol, metoprolol tartrate, metoprolol succinate, and nebivolol) have been evaluated in placebo-controlled trials in adults with heart failure. Five of these beta blockers (bisoprolol, bucindolol, carvedilol, metoprolol succinate, and nebivolol) have been evaluated in major trials that enrolled 1000 to almost 4000 patients (Table 9). Bisoprolol, in the Cardiac Insufficiency Bisoprolol Study II trial (CIBIS-II); carvedilol, in the Carvedilol Prospective Randomized Cumulative Survival trial COPERNICUS; and metoprolol succinate, in the Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure trial (MERIT-HF); but not bucindolol, in the BEST trial, reduced total mortality (as planned primary endpoint) by approximately 35%. The nonsignificant result for bucindolol suggest that individual beta blockers may differ in their effectiveness to reduce mortality in heart failure patients (bucindolol is not available in the United States, but is included in Table 9 for comparison).

Table 9

Table 9

Comparison of major beta blocker trials in heart failure

Two trials evaluated nebivolol in relation to all-cause mortality or cardiovascular hospitalization, New York Heart Association class reduction, and quality of life.72, 98 Mortality was included as a secondary outcome measure in both of these trials. The SENIORS study of 2128 elderly patients included patients with a history of heart failure (hospital admission for heart failure during the past 12 months or an ejection fraction of ≤35%). The mean age of patients was 76 and the mean ejection fraction was 36%. SENIORS included some patients who were similar to those included in the other trials, but a majority of patients who were older, had little or no left ventricular dysfunction, and had a lower risk of death. Thirty-five percent had an ejection fraction of >35%, and the annualized placebo mortality rate was 10%. When compared with placebo, nebivolol reduced the composite risk of all-cause mortality or cardiovascular hospital admission (31.1% compared with 35.3%; hazard ratio, 0.86; 95% CI, 0.74 to 0.99)72 but had nonsignificant effects on the individual variables examined as secondary outcomes. A subgroup analysis demonstrated that the risk of mortality or hospitalization for patients with a left ventricular ejection fraction of either ≤35 or >35% was not significantly different (P=0.42). In a post-hoc analysis, researchers identified the subgroup of patients most similar to the other major outcome trials. In this subgroup, defined as patients of less than 75.2 years with an ejection fraction ≤35% (n=342 for nebivolol and n=342 for placebo), findings were similar to that seen with metoprolol-controlled release, bisoprolol, and carvedilol (hazard ratio for the primary composite variable was 0.73; 95% CI, 0.56 to 0.96). For all-cause mortality alone, the hazard ratio was 0.62 (95% CI, 0.43 to 0.89). It should be noted, however, that the older and healthier patients (those with less severe left ventricular distinction) in the SENIORS trial were not evaluated in a subgroup analysis, and therefore it is unknown as to whether nebivolol would be effective in this population.

In the ENECA trial, nebivolol was examined for 8 months as an add on therapy in 260 elderly patients with chronic heart failure.98 Total mortality, included as a secondary outcome measure, was not significant when compared to placebo (survival rate 67.47% compared with 62.89; P=0.696). Results of the ENECA study are discussed below in relation to the study's primary outcome measures of New York Heart Association class reduction and quality of life.

Table 10 summarizes 16 placebo-controlled trials (including those in Table 9) that enrolled over 100 patients and met our other inclusion criteria (Evidence Tables 9 and 10). These trials evaluated atenolol 50 to 100 mg,81 bisoprolol 5 to 10 mg,82, 83 carvedilol 50 to 100 mg,84–93 metoprolol tartrate 100 to 150 mg,94, 95 metoprolol succinate (CR) 12.5 to 25 mg,96, 97 and nebivolol 10 mg.72, 98

Relation of mortality reduction to severity of heart failure

The trials in Table 9 leave no doubt that, in certain patients, bisoprolol, carvedilol, and metoprolol succinate reduce mortality. The main unresolved questions are 1) whether any of these agents is superior to the others in patients with mild to moderate failure, and 2) whether, in patients with severe failure, bisoprolol or metoprolol succinate are equivalent to carvedilol, which is the only drug that has a United States Food and Drug Administration indication in this group.

Many authors have used the placebo group mortality rates to make inferences about the baseline severity of patients in the various trials. However, several factors, including New York Heart Association Class, ejection fraction, blood pressure, lifestyle, and the quality of medical care influence mortality in patients with heart failure. For this reason it has proven difficult to judge the relative severity of illness among the major trials listed in Table 9.

MERIT-HF provides interesting data about the relationship of New York Heart Association class and ejection fraction:

MERIT-HF Subgroups EF<25%EF>25%
New York Heart Association Class II707 (“A”)928
New York Heart Association Class III–IV7951561 (“D”)

The large number of Class II patients with “severe” left ventricular dysfunction (ejection fraction <25%) illustrates the hazards of inferring functional class from ejection fraction. Conversely, a significant proportion of patients with “moderate to severe” heart failure (Class III and IV) had an ejection fraction >25%. As one would expect, the subgroup with New York Heart Association Class III–IV and ejection fraction <25% had the highest mortality. It would be impossible to distinguish between patients in cells “A” and “D” based on mortality rates and entry criteria.

The 4 United States Carvedilol trials and the Australian-New Zealand trial demonstrated that in patients with New York Heart Association Class II to IV heart failure, carvedilol reduced mortality. As shown in Table 10, the severity of heart failure of patients in these trials varied substantially, suggesting that carvedilol was effective across a broad spectrum of heart failure patients. These trials used an active drug run-in period during which patients who could not tolerate a small dose of carvedilol, were noncompliant, or died were excluded prior to randomization. For this reason, the mortality reductions and rates of withdrawal and adverse events are not comparable to those of other trials. In Table 10 we summarize mortality results of these and other trials after adjusting the number of deaths in the carvedilol group by adding in deaths that occurred during the run-in period.

COPERNICUSwas a well-designed, well-conduct ed placebo-controlled trial of carvedilol conducted in 334 Centers. Of 2289 subjects randomized, 627 were recruited from the United States and Canada; the rest were recruited in Europe (including Russia), the United States, Canada, Israel, Australia, South Africa, Argentina, and Mexico. It is difficult to compare the COPERNICUS subjects to those of other trials because COPERNICUS did not report New York Heart Association Class or exercise capacity, which were inclusion criteria in the other trials. COPERNICUS was intended to recruit a more severely ill population than the United States carvedilol trials. COPERNICUS subjects had higher mortality than 3 of the 4 trials that make up the United States Carvedilol Trial.

The mortality effect in COPERNICUS was consistent for sex, age, and other subgroups. The effect was lower, but not significantly so, for patients who had an ejection fraction <20% compared with those who had ejection fraction >20% and for those recruited in Europe, Australia, and the Middle East compared with North and South America.

MERIT-HF, conducted in the United States and Europe, recruited stable subjects with mild to severe heart failure. Although it had a significant proportion of subjects with New York Heart Association Class II symptoms, the mean ejection fraction was similar to that of CIBIS-II. MERIT-HF was well-designed and well-conducted and had clear-cut overall reductions in overall mortality, death from cardiac causes, sudden death, and heart transplantation, as well as a reduction in all-cause hospitalization (RR, 0.84; CI, 0.76–0.95).

The MERIT-HF investigators defined a “high risk” group consisting of the 795 patients who had New York Heart Association class III–IV and ejection fraction <25%. This subgroup had a mean ejection fraction (19%) and placebo group mortality (18.2%) close to that of COPERNICUS.

The applicability of the results of any trial to a United States population is a major issue in all of these trials, because heart failure survival depends on other aspects of care. The United States Food and Drug Administration review of the MERIT-HF trial found “a strong suggestion of a treatment-by-region (United States compared with Europe) interaction with respect to mortality.” MERIT-HF had 1071 United States subjects and 2920 European subjects. The placebo group mortality was higher in Europe (168/1462; 11.5%) than in the United States (49/539; 9.1%). Metoprolol succinate reduced all-cause mortality in Europe (hazard ratio, 0.55; P=0.0001) but not in the United States subgroup (hazard ratio, 1.05; P=0.7961). The lack of any trend toward reduced mortality in the United States subgroup is of concern.

For carvedilol, relevance to the United States population is not a concern, because the United States Carvedilol Trials were performed in the United States. Rather, the concern is what COPERNICUS adds to what was already known from the United States Carvedilol Trials. About 1 in 5 patients in COPERNICUS were from the United States; the hazard ratio was 0.80 in the United States patients and 0.60 in the rest of the world. Statistically, this difference is not meaningful, but that is not the whole story, for 2 reasons. First, the “rest of the world” is not homogeneous. Second, the proportion of United States patients in COPERNICUS was much lower than in MERIT-HF, so it is not surprising that the United States subgroup (n=482) was not a statistical outlier in COPERNICUS. Next to the United States, Russia (n=309) and Poland (n=299) recruited the most patients in COPERNICUS, and carvedilol had larger mortality reductions in these 2 countries than in 9 of 13 others.

CIBIS-II was a well-conducted multicenter European study designed to recruit stable subjects with moderate to severe heart failure (New York Heart Association Class III–IV).83 Most patients were New York Heart Association Class III. The annual placebo mortality rate was 13%, which is higher than the rate projected by the CIBIS-II investigators based on the results of CIBIS-I. Nevertheless, this mortality rate and the average ejection fraction of 27% are closer to those of MERIT-HF, which recruited mostly Class II and III patients, than to those of COPERNICUS, which is thought to have recruited New York Heart Association Class III and IV patients.

In CIBIS-II, 752 subjects were New York Heart Association Class III or IV and had an ejection fraction less than 25%, but the results in this subgroup have not been reported completely, although the hazard ratio was said to be 0.78 (0.56 to 1.07). For the Class III patients, annual placebo group mortality was about 13%; over the entire study (averaging 1.3 years of followup), the number needed to treat to prevent 1 death was about 19. For the Class IV patients, the annual placebo mortality was about 18%, and the number needed to treat to prevent 1 death over 1.3 years was about 15. The mortality reduction for Class IV patients was of borderline statistical significance; when measured as a difference of probabilities, the confidence interval was 0.0005 to 0.127 (from that is, from 0 to 12.7 lives saved for every 100 patients).

In addition to all-cause mortality, sudden death, and cardiovascular mortality, endpoints in beta blocker trials include symptoms, progression of disease, need for hospitalization, and need for (or time to) transplantation. The major placebo-controlled trials and many smaller trials evaluated these outcomes in Table 11.

Table 11

Table 11

Outcomes in placebo-controlled trials of beta blockers for heart failure

New York Heart Association class

The effect on New York Heart Association class rating was inconsistently reported. The CIBIS trial found that significantly more patients taking bisoprolol improved by at least 1 New York Heart Association class (21% compared with 15%; P=0.03) but there was no differences in patients that deteriorated by at least 1 class (13% compared with 11%). Results were mixed for carvedilol. Three trials suggest carvedilol is superior to placebo in improving the overall New York Heart Association class distribution.85, 86, 91 This includes the MUCHA trial of Japanese patients with heart failure.91 In 3 other trials, including a subset of dialysis patients with heart failure,92 carvedilol had no effect.84, 88, 92 Metoprolol tartrate did not significantly improve the New York Heart Association class in either of 2 trials. In the MERIT-HF trial, metoprolol CR increased the proportion of patients that improved by at least 1 New York Heart Association class overall (28.6% compared with 25.8%; P=0.003). A post-hoc analysis found the same effect in a subgroup of patients with baseline New York Heart Association class III–IV and left ventricular ejection fraction < 25% (46.2% compared with 36.7%; P=0.0031).99 By contrast, carvedilol did not reduce progression of heart failure in COPERNICUS. In the ENECA study of 260 patients with chronic heart failure treated with nebivolol as an add on therapy, compared with placebo (27%), slightly fewer elderly patients (≥65 years) with heart failure taking nebivolol at an average dose of 7.4 mg improved by at least 1 New York Heart Association class overall (26%).98

Exercise capacity

The carvedilol trials84–86, 88 were consistent in showing equivalency to placebo in exercise capacity improvement as measured by both the 6-minute walk and 9-minute treadmill tests. Results of treadmill testing (modified Naughton protocol) were mixed in 2 placebo-controlled trials of metoprolol.

Quality of life

Quality of life in heart failure patients was most commonly assessed using the Minnesota Living with Heart Failure Questionnaire. Overall, placebo-controlled trials provided limited evidence that beta blockers significantly improve quality of life in heart failure patients. Carvedilol was consistently associated with nonsignificant improvements in quality of life in patients with mild to moderate84–86 or severe87 heart failure.

In the MDC trial, patients taking immediate release metoprolol experienced significantly greater improvements in quality of life than those taking placebo, however, no data were provided and it is unclear as to which measurement instrument was used. For controlled-release metoprolol, results of quality-of-life assessments were mixed across 2 trials.97, 100 In the ENECA study, reductions in Minnesota Living with Heart Failure Questionnaire scores were similar for nebivolol compared with placebo.98

Head-to-head trials

There are no direct comparator trials comparing 2 or more of the drugs proven to reduce mortality (bisoprolol, carvedilol, and sustained release metoprolol succinate). We are aware of 1 trial in process that compares the tolerance of bisoprolol and carvedilol in elderly patients (≥65 years) with systolic or diastolic chronic heart failure.101

Otherwise, we found 6 fair-quality, head-to-head trials comparing immediate-release metoprolol tartrate to carvedilol in patients with heart failure and 1 trial that compared nebivolol to carvedilol (see Evidence Tables 11 and 12 for characteristics and quality assessments and Evidence Table 13 for outcomes).102–107 These trials recruited stable patients with Class II–IV (mainly II and III) heart failure, most of whom took ACE inhibitors and diuretics.

Only 1 trial (COMET) was adequately powered to evaluate mortality and cardiovascular events (N=3029). The target dose of carvedilol was 25 mg twice a day and the target for metoprolol tartrate was 50 mg twice a day. The patients were mostly (79.8%) men, with a mean age of 62 years and a mean ejection fraction of 26% on optimal treatment with ACE inhibitors and diuretics for New York Heart Association class II–IV heart failure.

When COMET was designed, extended-release metoprolol was not yet available, and immediate-release metoprolol was a logical comparator because in the MDC trial metoprolol tartrate was clearly effective, even though it did not change mortality. Specifically, metoprolol tartrate improved ejection fraction, left ventricular end diastolic pressure, and exercise time and prevented clinical deterioration, reducing the need for transplantation by almost 90% during the followup period.94

Mortality

In COMET, after a mean followup of 58 months (nearly 5 years), the intention-to-treat analysis showed an all-cause mortality reduction in favor of carvedilol (34% compared with 40%; number needed to treat, 18; P<0.0017). The annual mortality rate was 10% for metoprolol tartrate and 8.3% for carvedilol. For comparison, the rates were for metoprolol succinate in MERIT-HF (7.2%) and bisoprolol in CIBIS-II (8.8%). There was no difference between carvedilol and metoprolol in the combined endpoint of deaths plus all-cause admissions (74% compared with 76%).

COMET demonstrates unequivocally that carvedilol 25 mg twice a day was better than immediate-release metoprolol (metoprolol tartrate) twice a day. There is disagreement, however, about the relevance of the result, because immediate-release metoprolol had not been shown to reduce mortality in previous trials. Several years ago, after metoprolol tartrate failed to reduce mortality in the Metoprolol in Dilated Cardiomyopathy (MDC) trial, it was hypothesized that the patients who received it were subjected to daily variations in the degree of beta blockade. In COMET, the mean dose of metoprolol tartrate was less than that used in the MDC trial (85 mg daily compared with 108 mg daily), and the mean decrease in heart rate was also less (11.7 compared with 15 beats per minute). Subsequently, extended-release metoprolol (metoprolol succinate) was proven to reduce mortality in heart failure patients in the MERIT-HF trial. In MERIT-HF, the mean dose of metoprolol succinate was 159 mg daily and the mean reduction in heart rate was 14 beats per minute.

Other outcomes
Carvedilol compared with metoprolol

Evidence on numerous secondary outcomes from the COMET trial have been published.108, 109, 110 Carvedilol was superior to immediate-release metoprolol in reducing rates of cardiovascular death, sudden death, stroke, cardiovascular events, and unstable angina, and similar to immediate-release metoprolol in reducing death due to circulatory failure and other cardiovascular deaths, as well as in reducing days lost due to impaired well being.108, 109

Greater reductions in rates of first hospitalization due to potential complication of heart failure treatment were more associated with immediate-release metoprolol than with carvedilol. Both interventions had similar effects on rates of overall hospitalization and cause-specific hospitalizations, with 1 exception.108, 109 Rates of non-cardiovascular death, worsening heart failure, change in New York Heart Association classification, and medication withdrawal were similar for carvedilol and immediate release metoprolol.108

With regard to combined endpoints, carvedilol was superior in reducing rates of fatal or nonfatal myocardial infarction and the combination of cardiovascular death, heart transplantation, hospitalization for nonfatal acute myocardial infarction, or worsening heart failure and was similar to immediate-release metoprolol in reducing the combined rate of all-cause mortality and cardiovascular hospitalizations.108 Another combined endpoint of days of life lost due to death, hospitalization, impaired well-being, or need to increase diuretic use (deemed the ‘patient journey’) found carvedilol to be superior to metoprolol over 4 years when compared to baseline composite scores (P=0.0068).109 It is important to note however, that this combined endpoint considered all factors to be equal; days lost due to death were considered equivalent to days lost due to hospitalization.

In the older trials, there was a nonsignificant trend favoring carvedilol over immediate-release metoprolol. Carvedilol and immediate release metoprolol (124+/−55 mg daily) had similar effects on quality of life, but metoprolol improved exercise capacity more. There were no differences between the carvedilol and metoprolol groups in quality of life.

Nebivolol compared with carvedilol

One trial of 70 patients with heart failure, a left ventricular ejection fraction of 40% or lower, and a New York Heart Association functional class of II or III, compared treatment with mean doses of carvedilol 44 mg and a lower than recommended target dose of nebivolol (4.4 mg) over 6 months. Compared with baseline, carvedilol and nebivolol demonstrated slight improvements in New York Heart Association functional class and the 6-minute walk test.111

Key Question 1f. For adult patients with atrial arrhythmia, do beta blockers differ in efficacy or effectiveness?

Several beta blockers have been used to reduce the heart rate in patients with atrial tachyarrhythmias and to prevent relapse into atrial fibrillation or flutter. A recent good-quality systematic review examined 12 studies of rate control in patients with chronic atrial fibrillation.112 Atenolol, nadolol, and pindolol were effective in controlling the ventricular rate, while labetalol was no more efficacious than placebo.

We found 1 head-to-head trial comparing bisoprolol 10 mg and carvedilol 50 mg in patients subjected to cardioversion of persistent atrial fibrillation (> 7 days).113 This fair-quality, 12-month trial enrolled 90 patients (mean age, 65.5; 82% male) (Evidence Tables 14 and 15). Similar proportions of patients relapsed into atrial fibrillation during follow-up in the bisoprolol and carvedilol groups (53.4% compared with 43.6%; P=NS).

Two placebo-controlled trials evaluated beta blockers in patients with persistent atrial fibrillation.114–116 One placebo-controlled trial found that metoprolol CR/XL 100 to 200 mg was effective in preventing relapse of atrial fibrillation/flutter after cardioversion (Evidence Table 14).114, 115 This fair-quality trial was conducted in Germany and enrolled 433 patients after cardioversion of persistent atrial fibrillation that were 70% male, with a mean age of 60. Over 6 months, atrial fibrillation or flutter relapse rates were significantly lower in patients taking metoprolol CR/XL (48.7% compared with 59.9%; P=0.005). This trial was not powered to detect differences in rates of mortality as a primary endpoint. Death was reported as an adverse event and rates were not significantly different for the metoprolol CR/XL and placebo groups (3.1% compared with 0).

The other study examined the effects of carvedilol in managing patients with concomitant atrial fibrillation and heart failure.116 This study was divided into 2 phases. The first phase involved a 4-month comparison of digoxin alone to the combination of digoxin and carvedilol and the second phase involved a 6-month comparison of digoxin alone to carvedilol alone. Forty-seven patients (mean age, 68.5; 61.7% male) with atrial fibrillation (mean duration, 131.5 weeks) and heart failure (predominantly New York Heart Association class II–III; mean left ventricular ejection fraction, 24.1%) were enrolled in this fair-quality study. When added to digoxin, carvedilol significantly lowered the 24-hour ventricular rate (65.2 compared with 74.9 bpm; P≤0.0001) and improved mean left ventricular ejection fraction scores (30.6% compared with 26%; P=0.048) and severity of symptoms/functional capacity on a 33-point scale (6 compared with 8; P=0.039). There were no differences between monotherapies with either carvedilol or digoxin in the second phase, however.

Key Question 1g. For adult patients with migraine, do beta blockers differ in efficacy or effectiveness?

Summary

Six head-to-head trials show no difference in efficacy in reduction of attack frequency, severity, headache days or acute tablet consumption, or in improvement in any subjective or composite index in any of the comparisons made (atenolol or metoprolol durules or metoprolol or timolol compared with propranolol or nebivolol compared with metoprolol). Results from placebo-controlled trials on similar outcome measures generally supports those for atenolol, metoprolol durules, and propranolol seen in head-to-head trials. Placebo-controlled trial results also show that bisoprolol had a significant effect on attack frequency reduction and that pindolol had no appreciable effects.

Detailed Assessment

Head-to-head trials

We found 6 fair-quality117–122 head-to-head trials of beta blockers for the treatment of migraine (Table 12). One study comparing bisoprolol and metoprolol appears to have been published twice.123, 124 This trial was rated poor quality due to inadequate descriptions of methods of randomization and allocation concealment, lack of use of an intention to treat principle, and a high rate of attrition (37.6%).

Table 12

Table 12

Outcomes in head-to-head trials of migraine patients

The 6 included trials compared propranolol 160 mg to atenolol 100 mg,120 slow release metoprolol (durules) 200 mg daily,118 immediate release metoprolol 200 mg daily,117 timolol 20 mg,121, 122 propranolol 80 mg to metoprolol 100 mg daily,119 and nebivolol 5 mg to metoprolol 142.5 mg.125 All 6 trials were conducted outside of the United States, were relatively short-term in duration (12 to 20 weeks), and were small (30 to 96 patients). Most patients had common migraine per Ad Hoc Committee and World Federation of Neurology Research Group guidelines (83 to 93%) and migraine without aura per International Headache Society (92.8%). These patients have mean ages of 33.8 to 42.3, are 68.6% to 88.9% female, and have a history of migraine frequency of greater than 3 attacks per month. Use of concomitant analgesics and ergotamines was allowed for abortive migraine treatment. Headache frequency, intensity, severity, duration, and abortive treatment tablet usage efficacy parameters were analyzed using patient diary data.

The methods used to assess treatment effects differed across studies. Some of the common outcome results are summarized in Table 13 below. Analysis of variance was used to assess comparative efficacy of metoprolol 200 mg and propranolol 160 mg in 1 trial.117

Table 13

Table 13

Variceal rebleeding rates

Attack frequency

Metoprolol durules 200 mg, metoprolol tartrate 200 mg, and timolol 20 mg all were similar to propranolol 160 mg in decreasing 4-week attack frequency rates.117–119, 121, 122 A recent, well-conducted systematic review comparing propanol to other beta blockers found that there was little difference between propanol and the comparators (metoprolol, nadolol, timolol) in reducing attack frequency (pooled standard mean difference, −0.01; 95% CI, −0.24 to +0.22) based on data from 4 crossover trials.126 In a study comparing nebivolol to metoprolol there were no statistically significant differences in attack frequency, although nebivolol fared better with regards to tolerability.125

Migraine days

There were differences across trials in methods of assessment of this parameter. When the total number of headache days recorded over 42 days across all 28 patients analyzed was considered in the Stensrud trial, no difference between atenolol and propranolol treatment was found. Metoprolol durules and metoprolol tartrate reduced number of migraine days at rates similar to propranolol across 3 trials.117–119 In a comparison of nebivolol to metoprolol over an 18-week period, no differences were found.125

Severity

Severity rating methods differed across trials. Metoprolol durules, metoprolol tartrate, and timolol all were similar to propranolol at comparable doses in decreasing attack severity.118, 119, 121, 122 As measured using a 100-mm visual analog scale, headache severity at endpoint was similar for nebivolol and metoprolol (50 compared with 54 points).125

Tablet consumption

There were no differences in reduction of acute medication (analgesics, ergots) for metoprolol durules or metoprolol tartrate and propranolol.118, 119, 121, 122 Moreover, the number of patients using pain medication at endpoint were similar between nebivolol and metoprolol.125

Subjective assessment

Patients in 2 trials118, 119 were asked to make a subjective assessment of therapeutic improvement using descriptions of marked, moderate, slight, and unchanged or worse. There were no differences found between slow release metoprolol (durules) and propranolol (76% compared with 63%) or between low doses of immediate release metoprolol or propranolol (63% compared with 64%) in rates of decreased frequency of mean or median attacks per month.

Miscellaneous

Two trials120–122 measured treatment efficacy using a composite score (attack frequency x severity x duration) and found no differences between atenolol or timolol and propranolol. The Gerber et al. trial included an analysis of duration of migraine in hours and didn’t find any difference between metoprolol and propranolol in percent of patients qualifying as responder type A or B for decrease on this variable.

Placebo-controlled trials

We found 19 fair-quality, placebo-controlled trials (see Evidence Tables 16 and 17) of atenolol 100 mg,127 bisoprolol 5 mg or 10 mg,128 metoprolol slow release (durules) 200 mg,129, 130 131 pindolol 7.5 mg to 15 mg,132, 133 propranolol immediate release 80 mg to 240 mg,134–142 and long-acting propranolol 160 mg.143, 144 One trial145 did not report propranolol dosage and will be discussed separately.

All but 2136, 145 of these trials were conducted outside of the United States A crossover design was used in 12 trials, while the other 6 compared parallel groups. All but 2 trials reported allowing the use of various concomitant medications to abort migraine pain including common analgesics, ergotamines, and narcotics. These trials ranged in duration from 8 to 52 weeks, generally enrolling patients with a 1 to 2 year history of common or classic migraine (Ad Hoc Committee), generally occurring at an average frequency of 3 per week. One trial included only patients with classic migraine.130 Patient characteristics reflected the target migraine population, with mean ages in the range of 37 to 39 and gender predominantly female (>75%). Sample sizes ranged from 24 to 259 patients enrolled. Assessment of attack frequency, duration, severity, and use of acute medication variables was made using patient diary card data.

Placebo-controlled trial data was consistent with head-to-head trial data for atenolol 100 mg and slow-release metoprolol (durules) 200 mg, but added no additional evidence that is not reported in the head-to-head trials. Propranolol 80 mg and 160 mg, as discussed above, added information regarding efficacy of bisoprolol and pindolol. An exception was found in 1 of the 10 fair-quality trials of propranolol137 where a dosage of 120 mg was not significantly superior to placebo in increasing the proportion of patients that had at least a 50% reduction of migraine attacks in the last 4 weeks of treatment (42.3% compared with 30.9%) or in reducing the mean duration of migraine in hours per month (34.4% compared with 13.7%).

Bisoprolol

The results of 1 placebo-controlled trial of 12 week’s duration and involving 226 patients128 indicated that both bisoprolol 5 and 10 mg daily had a significant (P<0.05) effect in reducing attack frequency (39% for both bisoprolol doses compared with 22% for placebo). Neither dose of bisoprolol showed any obvious influence on reducing attack duration or severity.

Pindolol

The results of 2 placebo-controlled trials of pindolol 7.5 to 15 mg daily132, 133 in a total of 58 patients with predominantly common migraine showed no obvious advantage of this nonselective beta blocker in reducing averages per 4 weeks in headache frequency, headache index, or duration of attacks.

Twelve other placebo-controlled trials of beta blockers were found.121, 122, 146–155 These were rated poor quality due to insufficient detail in reporting randomization and allocation concealment methods, failure to perform efficacy analyses using an intention to treat principle, and rates of attrition ranging from 24% to 48.1%, which were not discussed here.

We found 1 meta-analysis156 that evaluated the effects of propranolol in 2403 migraine patients across a combination of 53 head-to-head, active- and placebo-controlled trials published through 1991. This review was rated poor quality due to failure to report critical assessment of internal validity and will not be discussed here. We independently assessed and included 3 head-to-head and 12 placebo-controlled trials from this meta-analysis in our report.

Key Question 1h. For adult patients with bleeding esophageal varices, do beta blockers differ in efficacy or effectiveness?

Summary

One small head-to-head trial showed no difference between atenolol and propranolol in rates of non-fatal/fatal rebleeding and all-cause mortality. Results of 1 trial of nadolol and 8 small placebo-controlled trials of immediate release and 2 formulations of extended release propranolol do not provide any additional indirect evidence of the comparative efficacy across beta blockers in these clinical outcomes. The somewhat mixed results across the placebo-controlled trials of propranolol suggest that treatment initiation interval may have an effect on rebleeding rates.

Detailed Assessment

Head-to-head trials

We found 1 head-to-head trial of beta blockers for the treatment of bleeding esophageal varices.157 This trial compared the efficacy of propranolol 40 to 160 mg daily, a nonselective beta blocker, atenolol 100 mg daily, a selective beta blocker, and placebo in cirrhotic patients. The results of this trial are summarized in Evidence Tables 18 and 19. This trial was rated fair quality. This trial, conducted in Italy, was designed to measure rebleeding and death and had a mean follow-up of 357 days. The patient population enrolled was typical for esophageal variceal bleeding, with a mean age of 53, 80.8% male and 81.9% alcoholic patients. This study also enrolled a small proportion of patients in which the prior hemorrhage was of a gastric erosion (12.8%) or unknown (inconclusive endoscopy) (6.4%) origin. Concomitant use of ranitidine, oral antacids, spironolactone, saluretics, lactulose, and nonabsorbable antibiotics was allowed.

No significant differences were found between propranolol and atenolol at 1 year for percentage of patients with fatal/nonfatal rebleeding episodes (2.4% compared with 3.1%) or total deaths (12% compared with 10%) or deaths due to rebleeding (3.1% compared with 3.1%), liver failure (6.2% compared with 3.1%) or other unrelated causes (3.1% compared with 3.1). Results of a multivariate analysis of parameters hypothesized to have had an influence on rebleeding were also reported. Drinking habits after enrollment was found to have significant effect on rebleeding, in that patients continuing to drink had higher incidences of rebleeding in both the propranolol (drinkers 50% compared with abstainers 0%) and atenolol (drinkers 43% compared with abstainers 27%) groups. Results of the analyses of the other parameters (severity of prior bleed, randomization time, number of bleeds prior to enrollment, treatment center, interval between index bleed, and endoscopy) were insignificant.

Other-controlled trials

We found numerous fair-quality, placebo-controlled trials of nadolol158 and propranolol159–166 for the secondary prevention of bleeding esophageal varices secondary to cirrhosis and schistosomiasis.167 Results are summarized in Evidence Tables 18 and 19. These trials were all conducted outside of the United States, enrolled samples of 12 to 84 patients, and ranged from 3 months to 2 years in duration. Mean ages ranged from 43 to 60 for the cirrhotic and 35.8 for non-cirrhotic patients. Populations were predominantly male with alcoholism as the most common etiology for cirrhosis. Treatment was initiated earlier, within 72 hours of the index bleeding episode, in only 3 of the trials.159, 162, 166

Variceal rebleeding rates

As shown in Table 13 below, compared to placebo, no differences in effect on variceal rebleeding rates were shown for immediate release propranolol in 2 early treatment trials. 159, 166 A significant difference between the effects of slow release propranolol and placebo was found in a third early treatment trial (20% compared with 75%; P<0.05).162 For trials of later (≥14 days)161, 163, 164, 168 and unspecified160, 169 treatment initiation, atenolol was equivalent to placebo (31% compared with 24%), nadolol was superior (25% compared with 71%; P<0.05), results of immediate release propranolol trials were mixed, and long-acting propranolol was superior (2% compared with 20%; P<0.02).

Deaths due to variceal rebleedingwere reported by 7 comparisons to placebo across 6 trials.159–161, 163, 166, 168 Results are summarized in Table 14 below and in Evidence Tables 18 and 19. In 1 trial of atenolol and 5 trials of propranolol, no differences from placebo in effect on death due to variceal rebleeding were established regardless of treatment initiation interval. In 1 trial of patients with portal hypertension secondary to schistosomiasis,169 however, significantly more patients (17%) experienced death due to variceal rebleeding on placebo than after late intervention (2 weeks) with propranolol (0%).

Table 14

Table 14

Death due to variceal rebleeding

All-cause mortality

No trial of patients with bleeding esophageal varices involved large enough sample sizes to measure all-cause mortality with sufficient power. Although crude trends suggest numerically smaller numbers of patients taking atenolol, nadolol and propranolol experienced deaths due to any cause in all but 1 trial of propranolol,159 no significant differences between beta blockers and placebo were found (Table 15).

Table 15

Table 15

All-cause mortality in patients with bleeding esophageal varices

Key Question 2. Do beta blocker drugs differ in safety or adverse effects?

Summary

Side effects are common among patients taking beta blockers. In longer-term trials (12 to 58 months) directly comparing beta blockers in patients with hypertension (atenolol compared with bisoprolol compared with propranolol), heart failure (carvedilol compared with metoprolol), bleeding esophageal varices (atenolol compared with propranolol), or atrial fibrillation (bisoprolol compared with carvedilol), a few differences in specific adverse events were noted. But, overall, no particular beta blocker stood out from the others as being consistently associated with a significantly less favorable adverse effect profile.

In everyday practice, weight gain, fatigue, dizziness, and dyspnea are the most common side effects in patients with heart failure. About 1 in 5 patients require discontinuation of the initial beta blocker choice. In a retrospective review of 1 series of 268 patients seen in a United States heart failure clinic, 54% were started on carvedilol and 46% on metoprolol succinate or metoprolol tartrate.170 Overall, about 1 in 5 patients (51 total) could not tolerate the initial choice of treatment. Forty of the 51 patients who could not tolerate the initial choice were switched to another beta blocker. Twenty-two of these 40 patients tolerated the second choice, with equal proportions tolerating a switch to carvedilol from metoprolol and to metoprolol from carvedilol.

A higher rate of beta blocker intolerance was reported in another trial that enrolled 90 consecutive patients in a heart failure clinic in Denmark.171 This trial compared bisoprolol and carvedilol and was designed to measure treatment failure rates under conditions that mimic daily clinical practice. The eligibility criteria were lax and the dosing regimen was flexible. Overall, 40% of patients (35 of 87) did not tolerate beta blocker therapy. Intolerance rates were similar in the bisoprolol and carvedilol groups (39% compared with 40%). This trial had some important methodological flaws, however. The trial used an inadequate method of randomization. Between-group differences at baseline confirm the inadequacy of the randomization method. The bisoprolol group was comprised of a significantly higher proportion of females (31% compared with 17%) and a numerically lower proportion of patients with a left ventricular ejection fraction < 25% (27% compared with 43%). Further, the team that treated and assessed the patients was not blinded to beta blocker assignment and the analysis excluded 3 patients that died prior to completing 2 months of follow-up. Group assignment of the 3 excluded patients was not reported. For these reasons, we rated this trial as poor quality and recommend a cautious interpretation of these potentially unreliable results.

Detailed Assessment

Adverse events of beta blockers most commonly reported in randomized controlled trials include cardiovascular symptoms of bradycardia and hypotension and central nervous system symptoms of dizziness. Relatively low rates of withdrawal due to these adverse events suggest that they were mild to moderate in severity. Other adverse events associated with beta blockers that were less commonly reported include sexual dysfunction and various dermatologic and gastrointestinal symptoms.

Head-to-head safety analyses were provided by 9 trials of patients with hypertension (Evidence Table 1),5, 8–11, 21, 22 19, 23 4 trials of patients with angina (Evidence Table 3),37–39, 172 5 trials of patients with heart failure (Evidence Table 11),95, 103, 106, 173, 111 7 trials of migraine patients (Evidence Table 16),117–120, 122, 174, 125 1 trial of patients with bleeding esophageal varices (Evidence Table 18),157 3 trials of patients post-myocardial infarction (Evidence Table 7),53, 55, 56 and 1 trial of patients with atrial fibrillation (Evidence Table 14).113 Trial characteristics have been described in detail previously and can also be found in the cited evidence tables. In general trials ranged in duration from 4 weeks to 58 months. Sample sizes ranged from 28 to 3029 patients. All but 2117, 125 of the head-to-head trials in patients with migraine used crossover designs, only reporting results of the combined intervention periods. Furthermore, in a hypertension study examining nebivolol and metoprolol,23 authors reported “no critical” adverse events were found, but did not supply data nor did they define “critical” adverse events.

Only 1 trial9 of atenolol 100 mg and pindolol SR 20 mg in 107 essential hypertensive patients was designed specifically for adverse event assessment and was rated good quality. Safety assessment in the remaining 21 head-to-head trials was fair to poor quality due to a lack of descriptive information regarding evaluation techniques. Events analyzed were generally not specified or defined. There was much heterogeneity across the trials in specific adverse events reported. All safety data reported can be found in the evidence tables cited above. The safety data that was most consistently reported (overall adverse event rate, incidence of bradycardia, dizziness, and hypotension, and withdrawals due to adverse events) across a more limited number of trials are summarized in Evidence Table 11.

Overall adverse events

Overall adverse event incidencewas reported in 17 head-to-head trials. 5, 8, 10, 21, 22, 38, 39, 106, 118, 119, 122, 123, 172 19, 37, 111, 125 Rates varied across the trials. For example, rates for carvedilol and metoprolol in a 3-month trial of 368 angina patients were 30% and 25%, respectively, as compared to 96% and 94% in a 58 month trial of 3029 patients with heart failure. No significant differences between the beta blocker comparisons were found, with 1 exception. In one 8-week trial of 40 angina patients,38 adverse events were more frequent in the propranolol group (94.4%) than in the pindolol group (17.4%; P<0.0001). Specific adverse events seen more frequently in the propranolol group include fatigue (44.4% compared with 0; P<0.0005) and mild hypotension (27.8% compared with 0; P=0.0114). The difference in safety favoring pindolol should be interpreted with caution due to variation between groups in illness severity at baseline. The mean 2-week angina attack rate was higher in the propranolol group during run-in [28.5 (95% CI, 26.4 to 30.6) compared with 18.4 (95% CI, 17.4 to19.4)]. This suggests problems with the randomization methods.

Withdrawals due to adverse events were reported by 13 head-to-head trials.5, 8, 11, 21, 22, 95, 113, 122, 123, 157 37, 111, 125 No significant differences were found in any of the comparisons.

Specific adverse events

Bradycardia

Rates of bradycardia were reported in short-term hypertension trials, in longer-term heart failure trials, a 2-month angina trial,3, 6, 17, 18, 937, 106, 111 and in a long-term trial for treatment of migraine.125 Overall, no significant differences between beta blockers were reported, with the exception of the 1 trial, which found a difference of bradycardia/electrocardiogram pauses >2.5 seconds for carvedilol 3 (9%) and 1 (3%) for nebivolol.111

Dizziness

Eight head-to-head trials reported dizziness incidence.21, 56, 103, 111, 120, 122, 123, 172 All but 1 reported no significant differences between beta blockers.103 Carvedilol was associated with higher rates of dizziness than metoprolol in a 44-month trial of 122 patients with heart failure (14.7% compared with 1.3%; P=0.0046).103 This significant difference was not seen in another shorter trial [3 months in 368 patients with angina (4.8% compared with 5.0%)],172 nor was there a significant difference in rates of dizziness in a head-to-head trial of carvedilol compared with atenolol in patients with recent myocardial infarction (36.4% compared with 27.2%; P=0.131).56 Reasons for this inconsistency may include differences in definition of dizziness and evaluation techniques between the 2 trials. This assumption cannot be verified, however, as the methods were not provided. Indirect comparison of the inconsistent head-to-head trial results to available fair- to good-quality placebo-controlled trials safety data did not offer any additional information as dizziness rates in metoprolol trials were not reported.

Hypotension

Rates of hypotension were similar for carvedilol and metoprolol across 2 longer-term trials of patients with heart failure.103, 106 Only 2.7% of patients from either treatment group experienced hypotension in the smaller (N=122), 44-month trial. After 58 months in the COMET trial (N=3029), 14% of patients taking carvedilol and 11% of patients taking metoprolol had hypotensive events. A study of left ventricular dysfunction after acute myocardial infarction (carvedilol compared with metoprolol), reported incidence of hypotension leading to withdrawal, but did not report the incidence for each study arm.55 In a 6-month heart failure study, no differences were found between nebivolol and carvedilol.111 A 30-week trial of treatment for migraine found similar rates between metoprolol compared with nebivolol.125

New-onset diabetes

Direct comparisons between beta blockers on risk of new-onset diabetes were only available from 1 retrospective analysis of data from the COMET trial, which compared metoprolol tartrate and carvedilol in adults with heart failure.173 New-onset diabetes was identified post-hoc among a cohort of 2298 patients without diabetes at baseline. The endpoint of new-onset diabetes was based on patient reporting and notes in hospital files and was considered present when there was documentation of a diagnosis of diabetes mellitus or diabetic coma, patients started antidiabetic treatment during the trial, or if patients had 2 or more random blood glucose readings above 11.1 mmol/l. The main finding of this analysis was that more patients receiving metoprolol tartrate developed new-onset diabetes than those receiving carvedilol (10.1% compared with 8.7%; hazard ratio, 0.78; 95% CI, 0.61 to 0.997). Although noteworthy, this finding should be interpreted with caution, keeping in mind that it is based on a post-hoc analysis and relies on a clinical, rather than guideline-based definition of diabetes.

Otherwise the only evidence we found came from a meta-analysis that pooled data from 12 trials (94 492 patients) of beta blockers compared with placebo, diuretics, ACE inhibitors, and calcium channel blockers which generated combined estimates of risk of new-onset of diabetes for each beta blocker.175 Pooled estimates based on a random effects model found that when compared to other comparators (placebo, diuretics, ACE inhibitors, calcium channel blockers) there is an increased the risk of new-onset DM for atenolol (pooled RR, 1.30; 95% CI, 1.11 to 1.52) and metoprolol (pooled RR, 1.34; 95% CI, 1.04 to 1.73), but not for propranolol (pooled RR, 0.77; 95% CI, 0.37 to 1.60).175 It should be noted that had a fixed effects model been used, only atenolol would have resulted in a statistically significant finding. The results of this meta-analysis should be interpreted with caution, as it did not evaluate the potential effects of variation among trials in internal validity factors.

Key Question 3. Are there subgroups of patients based on demographics (age, racial groups, gender), other medications, or co-morbidities for which one beta blocker is more effective or associated with fewer adverse effects?

Summary

There is no data that suggests that any beta blocker is superior in any subgroup of patients based on demographics, other medications, or comorbidities.

Detailed Assessment

Head-to-head trials

None of the 14 fair-quality head-to-head trials included in our efficacy analyses across all indications provided any subgroup analyses that differentiated one beta blocker from another based on demographics, concomitant medications, or comorbidities.

Placebo-controlled trials

We are aware of 1 placebo-controlled trial that examined the efficacy and tolerability of nebivolol in hypertensive African American patients.176 This study, however, did not meet our inclusion criteria as its focus was on blood pressure lowering and it did not report long-term health outcomes.

Meta-analyses

A recent systematic review conducted by the Cochrane Collaboration compared beta blockers to placebo in reducing the risk of severe hypertension and need for additional antihypertensives during pregnancy.177 Studies of acebutolol, atenolol, metoprolol, pindolol, and propranolol were included in this review, but no evidence of comparative effectiveness is provided. Rather, the focus of the review is on comparing beta blockers as a class to placebo. The review found that there was insufficient evidence to draw conclusions about the effects of beta blockers on perinatal mortality or preterm birth.

The Beta Blocker Pooling Project178 analyzed mortality in post-infarction patients relative to subgroup risk factors from trials of propranolol,50, 67, 179 pindolol,67 and other beta blockers not available in the United States. This analysis found that none of the age, gender, heart failure, or prior diabetes mellitus baseline characteristics interacted significantly with the effect on mortality. This analysis also does not offer any meaningful information about the comparative efficacy of beta blockers in these subgroups.

A 2003 meta-analysis180 analyzed the effects of bisoprolol (CIBIS-II), carvedilol (US Carvedilol, COPERNICUS), and controlled release metoprolol (MERIT-HF) on mortality in heart failure patients stratified by gender, race, and diabetics. Results are summarized in Table 16 below and suggest that beta blockers are equally effective in reducing mortality in subpopulations stratified by gender and race.

Observational analyses

A 12-month observational study comparing the tolerability of carvedilol (target dose 25 mg daily) in patients (ages ≥70) with and without diabetes mellitus found the rates of withdrawal due to adverse events (bradycardia, bronchospasm) were low in both the diabetes and nondiabetes subgroups (6% compared with 3%).181

Subgroup analyses and prescribing information

Atenolol

The SHEP trial assessed the use of chlorthalidone compared with placebo in controlling hypertension. Once desired blood pressure was reached, participants were further randomized to receive atenolol or reserpine. A subgroup analysis of long-term data (median 14.3 years) found that adding atenolol to chlorthalidone did not significantly affect mortality relative to placebo in diabetic patients, including both patients who were diabetic at baseline and those who developed diabetes during time on trial.182

Carvedilol

Prescribing information for carvedilol (http://us.gsk.com/products/assets/us_coreg.pdf) reports that effects on efficacy and adverse events were equivalent regardless of age (48% were ≥ 65 years; 11% were ≥75 years) in patients with left ventricular dysfunction following myocardial infarction in the CAPRICORN trial.61 We found no other source of publication of results from this subgroup analysis.

A number of additional meta-analyses have been published that evaluate the effects of carvedilol in subgroups of patients based on demographics and/or comorbidities. The United States Carvedilol Heart Failure Study Group published an analysis183 of the pooled results from a stratified set of 3 fair-quality and 1 poor-quality concurrently conducted protocols,84–87 discussed in detail above, that showed no significant interaction between race and carvedilol treatment in patients with mild to moderate heart failure. More recent analyses from the COPERNICUS trial89 show that carvedilol had similar effects regardless of age and gender in patients with severe heart failure.

The most recent and largest manufacturer-funded meta-analysis (N=5757) of published and unpublished data from 7 clinical trials focused on evaluating the effects of carvedilol in patients with heart failure, with and without comorbid diabetes.184 Consistent with previous analyses, the main findings confirmed that similar reductions in risk of all-cause mortality were seen in heart failure patients, regardless of diabetes status. The relative risk reduction in the subgroup of patients with diabetes was 28% (95% CI, 3 to 46) and was 37% (95% CI, 22 to 48) in the non-diabetic patients.

Labetolol

Product information for labetalol (http://www.prometheuslabs.com/pi/TrandateTab.pdf) suggests that required maintenance doses may be lower in geriatric patients due to a reduced rate of elimination. However, we did not find any evidence of differential efficacy of labetalol relative to age.

Metoprolol

A fair-quality review185 that pooled results from 5 placebo-controlled trials of metoprolol (Amsterdam, Belfast, Goteborg, Stockholm, Lopressor Intervention Trial) found that neither age nor gender had a significant influence on mortality. When considered individually, results from the Goteborg Metoprolol Trial186 show a nonsignificant trend that patients aged 65 to 74 years had a more marked reduction in mortality at 3 months post-myocardial infarction (45%) than did all patients aged 40 to 74 (36%). Results from the MERIT-HF trial also reported that neither age nor gender had any influence on the effects of metoprolol CR in patients with mild to moderate heart failure.

A subgroup analysis of the MERIT-HF trial evaluated the influence of comorbid diabetes on the effects of metoprolol CR.187 This analysis found higher rates of all-cause mortality in the placebo group when compared to metoprolol (12.7% compared with 10.1% per patient year; risk reduction, 18%; 95% CI, +44 to −19). Metoprolol CR also significantly reduced risks of hospitalizations for worsening heart failure (including those patients identified as having severe heart failure) regardless of diabetic status.

Propranolol

The fair-quality, placebo-controlled Beta Blocker Heart Attack Trial67 comprised of 3837 patients found that the protective of propranolol on mortality 25 months (average follow-up) following myocardial infarction was equivalent regardless of age or gender.

Nebivolol

Subgroup analysis of the SENIORS trial found no significant differences in the effect of nebivolol on subpopulations of gender, ejection fraction, age, diabetes, and prior myocardial infarction.72

Copyright © 2009, Oregon Health & Science University, Portland, Oregon.
Bookshelf ID: NBK47170
PubReader format: click here to try

Views

Other titles in this collection

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...