Compared with optimal medical treatment, biventricular cardiac resynchronization therapy (CRT), with or without an implantable cardioverter defibrillator (ICD), can reduce all-cause death and all-cause hospitalizations in patients with moderate or severe heart failure, according to preliminary results from the Comparison of Medical Therapy, Pacing, and Defibrillation in Chronic Heart Failure (COMPANION) trial. The COMPANION trial, presented at the American College of Cardiology 2003 Annual Scientific Sessions by Drs. Arthur M Feldman and Michael R. Bristow, was halted prematurely in November 2002 when the trial s data and safety monitoring board (DSMB) projected that the trial would reach its primary endpoint ahead of schedule. In his presentation, Bristow emphasized that full and final COMPANION results will require adjudication of the trial endpoints as well as repeat consent from patients who withdrew from the study, a step that will require approval from the DSMB. As such the data presented must be considered preliminary. However, we do believe the results presented today accurately reflect what the ultimate outcome will be in COMPANION, Bristow said. This is a very positive study, Bristow commented later. The results speak for themselves. COMPANION was a parallel, randomized clinical trial of 1,600 patients with moderate or severe heart failure with QRS > 120 ms and P-R interval > 150 ms. Entry criteria stipulated that patients had to have been hospitalized at least once in the past year for heart failure management, to have had an outpatient visit in which inotropes or a vasoactive infusion were administered continuously for at least 4 hours, or to have had an emergency room visit of at least 12 hours during which intravenous heart failure medications were administered. Patients were randomized in a 1:2:2 fashion to optimal medical therapy (including beta blockers, diuretics, ACE inhibitors or ARBs, spiralactone, and/or digoxen), optimal drug therapy plus CRT, or optimal drugs plus a CRT with ICD (CRT-D). Bristow reported that the primary endpoint of the study, a combination of all-cause death and all-cause hospitalizations over 12 months, was significantly reduced by 19% for both the CRT and CRT-D arms of the study. CRT alone was also associated with a nonsignificant trend toward a 23.9% reduction in all-cause mortality, a secondary endpoint of the study, while the addition of an ICD to CRT increased the mortality reduction to 43.4%, a highly significant result. Hospitalizations related to device implantations were not considered for the purposes of the comparative analysis. When results were examined according to heart failure etiology, the investigators found no significant differences in treatment effects between patients with ischemic and nonischemic cardiomyopathy. Reduction in the combined endpoints of death plus all-cause, cardiovascular, or heart failure hospitalizations was likely due to CRT, since CRT and CRT-D resulted in similar effect sizes, Bristow noted. He added, Based on a reduction in the primary endpoint that included all hospitalizations, CRT or CRT-D would likely be cost-effective in as much as heart failure hospitalizations are a major determinant of heart failure care costs, although this would need to be confirmed in further studies. Commenting on the COMPANION trial, Dr. Milton Packer (Columbia University, New York City) pointed out that the COMPANION results apply only to a small proportion of heart failure patients. The patient population studied in COMPANION represents a pretty small slice of the overall population with heart failure, because we're talking about class III/IV patients with a wide QRS and LV dysfunction. My guess is we're probably talking about 250,000 patients in the US, out of 5 million heart failure patients. Packer was particularly encouraged by the CRT-D results, noting that patients treated with a CRT-D in COMPANION enjoyed similar benefit, regardless of whether their underlying cardiomyopathy was ischemic in origin. In the past we ve had only defibrillator data in ischemic patients, so this was the first clue that we might in fact be seeing an effect of defibrillators in nonischemic cardiomyopathies as well. By contrast, the CRT data were more confusing, Packer said, primarily because CRT was credited with the reduction in all-cause hospitalizations. The problem with this, Packer said, is that hospitalizations related to device implantation were not included in the comparison. This is postrandomization censoring, and we know from other trials that postrandomization censoring causes biases, and it s particularly worrisome because, if a patient were hospitalized for a device and went home in 24 hours, did the investigators count that? I asked the investigators and they said no. Was it counted if a patient were hospitalized for a device and had a complication and stayed in the hospitalized for days and weeks? They said no. I think if you censor events in the first week for the device group, you have to censor the control group for the same period of time, and it doesn't appear that they did that.