Author + information
- Andreas W. Schoenenberger, MD∗ (, )
- André Moser, PhD and
- Andreas E. Stuck, MD
- ↵∗Bern University Hospital, University Department of Geriatrics, Inselspital, Bern 3010, Switzerland
We thank Dr. Afilalo and colleagues and Dr. Lantelme and colleagues for their interest in our paper (1).
Dr. Afilalo and colleagues refer to the results from the multicentric FRAILTY-AVR (Frailty Assessment Before Cardiac Surgery & Transcatheter Interventions) study (2), which they published after submission of the manuscript of our study. Their study investigated 7 frailty scales, including our own scale, to predict mortality and disability in patients undergoing transcatheter aortic valve replacement. We agree with Dr. Afilalo and colleagues that the huge variety of different frailty scales so far has impeded the implementation of frailty in clinical practice. Undoubtedly, the study from Afilalo et al. (2) is very comprehensive and helps clarify the situation.
To respond to the comment of Dr. Lantelme and colleagues about potential overoptimism and overfitting in our models, we re-analyzed the C-statistics of our main models using 1,000 bootstrap replications (Table 1). We conclude on the basis of these results that overfitting did not affect our conclusions.
Dr. Lantelme and colleagues also asked about clarification of our choice of confounders and predictors. We a priori chose a maximum of 4 variables in the modeling based on the recommended 10 events-per-variable (EPV) approach. Given the observed 52 deaths within 1 year of follow-up, we obtained an EPV of 17 (3 variables) and 13 (4 variables), which is even more than the recommended 10 EPV. Thus, based on our a priori variable selection approach and the observed EPV, we a priori minimized model overfitting and misspecification.
However, we agree with Dr. Lantelme and colleagues that our findings are limited to our observed data and study population, which need further validation. However, we adequately mentioned in the limitations section of our paper that external validation will be necessary to generalize our results.
Please note: The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2018 American College of Cardiology Foundation
- Schoenenberger A.W.,
- Moser A.,
- Bertschi D.,
- et al.
- Afilalo J.,
- Lauck S.,
- Kim D.H.,
- et al.