Author + information
- Jonathan Afilalo, MD, MSc⁎ (, )
- Mark J. Eisenberg, MD, MPH,
- Howard Bergman, MD,
- Johanne Monette, MD, MSc,
- Jean-Francois Morin, MD,
- Yves Langlois, MD,
- Nicolas Noiseux, MD,
- Louis P. Perrault, MD, PhD and
- Karen P. Alexander, MD
- ↵⁎Division of Cardiology, Sir Mortimer B. Davis Jewish General Hospital, McGill University, 3755 Cote Ste Catherine, Montreal, QC H3T 1E2, Canada
We appreciate the interest of Dr. Thum and colleagues in our prospective study of gait speed in elderly patients undergoing cardiac surgery (1). Two points are raised: the first concerns the validity of the gait speed protocol used, and the second concerns the selection of the covariates evaluated in the multivariable model.
First, the gait speed protocol used in this study was prespecified to be a 5-m distance with a cutoff fit to optimally predict the occurrence of mortality or major morbidity in our patient population (which was 6 s [0.83 m/s] as determined by receiver-operating characteristic analysis). Dr. Thum and colleagues suggest that this distance and cutoff are not consistent with consensus definitions. We respectfully disagree and point to the recent task force position statement on gait speed that highlights a variety of distances and cutoffs appropriately used in the medical literature (2). Accordingly, the most common distances were between 4 and 6 m, and the most common cutoffs were between 0.6 and 1.0 m/s, depending on the patient population being evaluated and the outcome being predicted. The task force authors go on to state that “the use of gait speed at usual pace as a predictor makes the course-distance of less importance.” In keeping with this, Graham et al. (3) demonstrated that course distance was not a significant determinant of mean gait speed. Therefore, although a 4-m, 0.8-m/s protocol is endorsed by some, there remains justified variability and the 5-m, 6-s protocol used in this study is entirely within evidence-based standards. Moreover, we find it important to correct that short-distance gait speed is not intended to be a marker of exercise capacity nor of sarcopenia and cachexia, as suggested, but rather a marker of frailty, which is regarded as a distinct entity.
Second, the covariates evaluated in the multivariable model were prespecified to be the Society of Thoracic Surgeons predicted risk of mortality or major morbidity and a group of 7 individual risk factors that had been shown to account for the bulk of the outcomes observed. It is undoubtedly true that other risk factors exist; however, the benefit of adding covariates in a statistical model must be weighed against the risk of overfitting and detecting spurious associations (4). In light of this, we, like many others, opted for a parsimonious model containing core risk factors rather than exhaustive ones.
Good clinical practice dictates that the incorporation of new tests and treatments should be based on sound evidence, ideally from more than a single study. As stated in our paper, we wholeheartedly endorse and look forward to future efforts to study the optimal cutoff for slow gait speed and to validate the role of gait speed as a prognostic marker in patients undergoing cardiac surgery.
- American College of Cardiology Foundation