The evidence about the effect of CRT on mortality is less clear, however, since none of the individual clinical trials found a statistically significant reduction. McAlister and associates used meta-analysis to combine the results from 9 trials of CRT and reported a statistically significant 21% reduction in mortality (2). While this result is suggestive, it is not definitive. Meta-analyses based only on small trials often yield lower relative risk reductions than definitive large trials (4), perhaps because small positive studies are more likely to be published than small negative studies. All of the randomized trials used in McAlister and associates' meta-analysis have flaws that limit inferences about the effect of CRT on mortality. Eight of the 9 trials required successful device implantation before patients were randomly assigned to have the CRT function turned on or off. This design caused the results to overstate the effect of CRT on mortality by omitting all of the implantation failures (10% or more) and procedural deaths (0.4% or more) while exposing all of the controls to the adverse effects of an implanted device. A more serious limitation is that CRT was combined with an ICD in some studies but not in others, making it tricky to isolate the effect of CRT on mortality. The Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure (COMPANION) trial (5) poses particular problems because of its unbalanced, 3-arm design (medical therapy, CRT, and CRT combined with an ICD) and the weight it was given in the meta-analysis because of its large enrollment and high mortality rates. We think that McAlister and associates should have omitted the CRT-ICD arm of the COMPANION trial, since including it inappropriately credited CRT with the mortality reduction that was due to the ICD. With omission of this unpaired arm of the COMPANION trial, the estimated effect of CRT on mortality would certainly be much less favorable and, if so, no longer statistically significant. Thus, while CRT improves functional status and quality of life, there is not yet significant evidence that it reduces mortality.