Author + information
- Brandi J. Witt, MD,
- Steven J. Jacobsen, MD, PhD and
- Véronique L. Roger, MD, MPH⁎ ()
- ↵⁎Division of Cardiovascular Diseases, Mayo Clinic, 200 First Street SW, Rochester, MN 55905
We appreciate the interest of Dr. West in our work (1). We respectfully take issue with the statement that the use of propensity-score methodology is unusual. Indeed, the use of propensity score is a commonly used, well-accepted method of statistical adjustment (2,3). It is considered by many to be preferable to conventional regression analysis to adjust for differences in baseline characteristics and control for confounding by indication. As in any observational study, however, we cannot rule out residual confounding related to unmeasured characteristics. This point, which was emphasized in our report, is important to underscore in the interpretation of our data.
As underscored by Dr. West, and as stated in our study, randomized controlled trials constitute the methodological gold standard to test the effect of an intervention. Dr. West quotes one meta-analysis of four trials (4) and one multicenter randomized trial (4). Both of these are published only in abstract format, and neither one provides sufficient information to interpret the findings. For example, the trial inclusion criteria or components of the rehabilitation programs may be substantially different from what is reported in our community-based myocardial incidence cohort (1). These differences could, in turn, explain the observed differences in survival. More importantly, the duration of follow-up in the randomized trial is only 12 months (5), shorter than in our published follow-up of 6.6 years (1). Finally, the apparent age and gender disparities in the delivery of care noted in our analysis could only be addressed in observational studies such as ours, through the analysis of clinical practice.
- American College of Cardiology Foundation
- Witt B.J.,
- Jacobsen S.J.,
- Weston S.A.,
- et al.
- Miettinen O.S.
- West R.R.,
- Beswick A.D.