β blockers, as a class, provided credible mortality benefits in comparison with placebo or standard treatment in patients with heart failure, and the beneficial effects on mortality persisted irrespective of the duration of treatment. However, we found no obvious differences when comparing the different β blockers head to head for the risk of death, sudden cardiac death, death due to pump failure, or drug discontinuation by way of a network meta-analysis or mixed treatment comparison. Accordingly, improvements in left ventricular ejection fraction were also similar irrespective of the individual study drug.
β blockers have long been shown to have unequivocal mortality benefits in chronic heart failure, especially with long term follow-up, although this is slightly controversial as most of the landmark randomized controlled trials of β blockers have been stopped prematurely.35
Several large trials and meta-analyses have assessed the mortality benefit of β blockers in chronic heart failure with reduced ejection fraction and have shown a mortality benefit of β blocker use in chronic heart failure as well as in ischemic and non-ischemic cardiomyopathy.36
Various studies have looked at important reduction in clinically relevant endpoints such as sudden death, pump failure, and hospital readmission rates with β blockers in heart failure.37
Retrospective analyses have compared different β blockers and their effect on clinical endpoints in chronic heart failure, and no superiority was found for one individual class over another in reducing mortality or readmissions.40
When the surrogate endpoint of improvement in ejection fraction has been investigated, superiority of any one class of β blockers could not be definitely established.42
With this robust evidence in favor of β blockers in chronic heart failure, we decided to assess the efficacy of one β blocker over another by comparing them using a Bayesian network meta-analysis, in view of paucity of head to head trial data, except for the COMET study.10
Also, to our knowledge, the question of tolerability and premature discontinuation of β blockers in chronic heart failure has not been investigated, except in isolated trials, so we decided to pool the data from different trials to assess the important question of tolerability and the discontinuation rates of different β blockers in heart failure with reduced ejection fraction.43
Our analysis reassuringly shows that overall, the tolerability, as indicated by discontinuation rates in different randomized trials, of β blockers as a class of drugs is generally good and comparable to a range of pharmacologically active and inactive comparators.
Our analysis shows that among the different β blockers in current use, bisoprolol, carvedilol, and metoprolol all had significant mortality benefit compared with placebo/standard heart failure treatment. However, our data additionally indicate that carvedilol was not only superior to placebo/standard heart failure treatment for all cause mortality but also had the lowest cardiac mortality numerically among all β blockers tested, although this was not statistically significant, indicating a benefit for its use in patients with cardiovascular comorbidities as an empiric initial treatment of choice. This finding was congruent in our analysis with the results and subgroup analyses from COMET,10
possibly as a result of a beneficial effect of carvedilol on endothelial function, its stimulatory effects on β arresting signaling, and its anti-oxidant properties.45
Carvedilol has also been found to be superior to metoprolol in maintaining a favorable glycemic profile in patients with diabetes, improved insulin sensitivity, and decreased progression to microalbuminuria, all of which have been shown to have cardioprotective effects.48
Considering the lack of differences in improvement of mortality with use of individual β blockers in chronic heart failure, as shown in our data, we consider it pragmatic to infer that, although the data may be limited, we have three agents (bisoprolol, sustained release metoprolol succinate, and carvedilol) that have been tested more extensively and have been shown to be superior to placebo in our analysis (95% credible interval did not include 1). Hence, one of these three agents should be chosen as the empiric drug of choice in treatment of chronic heart failure with reduced ejection fraction.
Our analysis indicates that the cost of individual β blocker treatment for patients can possibly be reduced by choosing the least expensive agent. Another significant finding was that tolerability, as measured by premature discontinuation rates in trials, was also numerically the best with carvedilol (although again the difference did not reach statistical significance), making a case for its increased use among sicker patients needing a β blocker. This could be a crucial point in patients taking many different drugs, as non-adherence to treatment is the most frequent reason for hospital admission for acute heart failure, with important implications for clinical management and potentially for prognosis.49
In harmony with the current American College of Cardiology/American Heart Association recommendations, and a previous Bayesian analysis,50
our analysis shows mortality benefit with use of β blockers in chronic heart failure—especially for bisoprolol, carvedilol, and sustained release metoprolol succinate—and their use should be recommended for all stable patients with current or previous symptoms of heart failure and reduced left ventricular ejection fraction, unless contraindicated.
Limitations of study
Our analysis was limited by the data in the included studies and the structure of the reported data. Meta-analysis of rare events is known to produce erroneous results. This is further compounded in network meta-analysis if few trials per comparison are available, as was the case with a few of the β blockers. Therefore, extra caution should be exercised when interpreting the treatment rankings for the primary outcome (mortality). Furthermore, outcomes were not consistently reported across the ensemble of trials. An individual participant meta-analysis might overcome this limitation, even if this might reduce the sample size. Although we feel reasonably confident of our search strategy, some trials, such as potentially some non-English language trials, may not have been included. The results may also be limited by the modeling assumptions. Because of the multitude of doses used for many trials, we did a weighted meta-regression analysis by target and achieved doses of individual drugs and their effects on the natural log of the calculated odds ratio, which did not alter the outcomes.
Another potential limitation of our study stems from the interesting analysis in a recent paper,52
which suggested that β blockade was associated with a lower magnitude of survival benefit in the United States than was seen in the rest of the world. Non-availability of patient level data prevents us from commenting definitively on the effect of geographic distribution on the outcomes seen with β blockers; however, our preliminary analysis using trial level data does refute the phenomenon. In the absence of patient level data, an adequately rigorous analysis to assess the effect of geographic location with use of β blockers is extremely difficult to do, and future efforts should focus on explaining or refuting this interesting phenomenon.
Comparison with other studies
A previous retrospective study attempted to compare different β blockers in clinical use by using data from an administrative database.53
The authors concluded that atenolol and acebutolol were superior to metoprolol in reducing mortality in patients, whereas carvedilol and bisoprolol were not superior to metoprolol in improving survival. The population studied in that analysis was quite different from the population of randomized trials assessed in our study—the mean age of the population of that study was 77 years, whereas the mean age of the population in our analysis was 61 years, and the percentage of male patients was 49% in the study, whereas male patients comprised 76% of the population in our analysis. Also, ejection fraction was not accounted for in the analysis of the retrospective study, whereas the mean ejection fraction of the population in our analysis was 26%. Hence, the difference from our study, which pooled data from randomized trials only, can likely be explained by differences in the population studied and differences in the analytic methods.
Our analysis, using one of the largest sample sizes of β blockers in chronic heart failure with reduced ejection fraction and a novel analytic method of comparing individual β blockers by using Bayesian network meta-analysis, suggests that β blockers are safe and tolerable for patients with chronic heart failure and that the mortality benefit and other clinically relevant effects, as well as the improvement in ejection fraction, are a class effect of β blockers and superiority of one over another cannot be determined with the currently available data. More head to head comparisons of individual β blockers are needed to definitively answer the question of whether one agent is clearly superior.
What is already known on this topic
- β blockers are useful in preventing deaths in patients with heart failure
- Carvedilol, bucindolol, and metoprolol succinate are the only β blockers shown in randomized trials to be of benefit in lowering mortality in patients with heart failure
- A previous study has contended that β blockers may not be as effective in preventing mortality in the US as in rest of the world
What this study adds
- This study, using novel methods in the framework of a Bayesian analysis, computes the relative mortality benefit of β blockers in patients with heart failure as 20-35%
- The mortality benefits observed with the individual β blockers seems to be largely due to a “class effect,” with no evidence of a clearly superior agent among individual β blockers
- The extent of reduction in mortality seen in trials conducted solely in the US was comparable to that seen in trials conducted in the rest of the world