Author + information
- Martin H. Ruwald, MD∗ ( and )
- Arthur J. Moss, MD
- ↵∗University of Rochester Medical Center, Heart Research Follow-up Program, 265 Crittenden Boulevard, Rochester, New York 14642
We read with interest the comments on our study (1) by Dr. Ghali. Dr. Ghali felt we did not emphasize the differences in baseline characteristics between patients on metoprolol and carvedilol. Because our study was not a randomized trial to evaluate a comparison between metoprolol and carvedilol, we carried out a fully adjusted multivariate Cox regression model and an additional propensity scored-adjusted analysis. As we explained in the article, we adjusted the multivariate model for baseline characteristics that had a significant impact on the endpoint of hospitalization for heart failure or death (Table 2 in our article). This approach does not eliminate unmeasured confounding, as we pointed out in the limitations section. Therefore, the differences in the baseline characteristics between patients on metoprolol and carvedilol as mentioned by Dr. Ghali are taken into account in our analysis.
Dr. Ghali felt that the 12% of patients on metoprolol tartrate should be excluded from the analysis since it has not been documented to reduce survival or hospitalization in a randomized trial. Because some of the patients changed from tartrate to succinate (or to carvedilol or came off the drug) throughout the study we wanted to include and keep these patients in the analysis. Because we used time-dependent analysis, thus always taking into account the risk time the patient was on a particular beta-blocker with a specific dose, we kept these patients in the analysis. If we exclude the patients on metoprolol tartrate at baseline, the results on the endpoint of heart failure or death were, however, similar and remained statistically significant.
Regarding question 3, we used time-dependent analysis. This means that throughout follow-up, we are taking into account the risk time the patient is on a specific drug (at a specific dose) and its contribution to the overall risk of the endpoint. Thus, if a patient decided to go off the drug or change from metoprolol to carvedilol or the opposite, the risk time contributed on either drug is taken into account in the analysis. Therefore, we think it would wrong to exclude these patients from the analysis. We do not have precise information on the cause of why patients changed type of beta-blocker.
Finally, we believe that the synergistic effect seen in patients treated with carvedilol who had a left bundle branch block may be related to favorable reverse remodeling associated with the use of resynchronization therapy in combination with carvedilol.
Please note: Dr. Ruwald has received unrestricted funding grants from the Danish Heart Association, the Lundbeck Foundation, Helsefonden, Arvid Nilssons Fond, and Knud Hoejgaard Fonden. Dr. Moss has received grant support from Boston Scientific and lecture fees from Boston Scientific, Medtronic, and St. Jude Medical.
- American College of Cardiology Foundation