Eighteen published trials involving 2,986 patients were included (duration of treatment 1 to 23 months; weighted average 13 months). Complete data, by allocated treatment, were available for mortality (10 trials; 2,841 patients), for hospitalisation for heart failure (5 trials; 1,514 patients) and for heart transplantation (6 trials; 2,330 patients).
Data on left ventricular ejection fraction were available in 11 trials (941 patients) and complete data on exercise end points were available in 9 trials (855 patients). Metoprolol was evaluated in 7 trials (562 patients), bucindolol in 4 trials (209 patients), carvedilol in 2 trials (1,509 patients), nebivolol in 2 trials (36 patients), bisoprolol in 1 trial (641 patients), acebutolol in 1 trial (17 patients), and labetalol in 1 trial (12 patients).
There were 7 trials of patients with an idiopathic dilated etiology, 2 trials of those with an ischaemic etiology, 5 trials of those with a combination of idiopathic and ischaemic etiology, and 4 trials of those with a combination of idiopathic, ischaemic, hypertensive and valvular etiologies.
There was a lower rate of death in the active treatment group, 8.2% versus 12.6% in the control group (OR 0.72, 99% CI: 0.51, 1.00). In addition, there was a lower rate of hospitalisation for heart failure, 18.1% versus 28.8% (OR 0.54, 99% CI: 0.39, 0.74) and a trend towards a lower proportion of patients receiving heart transplantation, 1.1% versus 2.7% (OR 0.45, 99% CI: 0.20, 1.03). Overall, the absolute change in the left ventricular ejection fraction was 6.3% higher in the active treatment group (99% CI: 1.1, 11.54), and there was a non significant absolute decrease of 14.4 seconds in the exercise time for the treatment group, compared with the placebo (99% CI: -202.1, +175.2). There was no significant heterogeneity among the results of the trials investigating any of the outcomes.
Thirteen trials reported information on adverse effects (84% of all patients randomly assigned). There was an excess of the following in the treatment group when compared with the control group: dizziness, 16.2% versus 7.5% (OR 2.2, 95% CI: 1.7, 2.8); fatigue, 9.0% versus 2.6% (OR 1.10, 95% CI: 0.83, 1.46); bradycardia, 5.1% versus 0.6% (OR 4.29, 95% CI: 2.75, 6.70); and hypotension, 4.7% versus 1.7% (OR 2.21, 95% CI: 1.41, 3.46). In contrast, there was a significant reduction in the proportion of patients reporting worsening heart failure in the treatment group, 16.7% versus 25.1% in the control group (OR 0.59, 95% CI: 0.49, 0.71).
In evaluating the reliability of the available data, the optimal information size was calculated on the basis of a control mortality of 15%. These calculations indicated that 7,720 and 14,044 patients, based on a relative risk reduction of 20 and 15%, respectively, are the minimum sample sizes needed to reliably detect plausible treatment effects. The corresponding mortality end point indicated that the available evidence was not sufficiently convincing in the context of a cumulative meta-analysis.