We have read the Letter to the Editor and thank you for the opportunity to respond to the concerns regarding our methodology/statistical analyses . We appreciate the work of these colleagues and would like to comment on these considerations.
The authors argue that the Cox proportional hazard regression model is problematic in terms of the integration of significant variables into the final multivariate model. This is not incorrect from a statistical point of view, and we also performed a forward and backward analysis before these variables were taken into account in this final analysis if a P -value of <0.2 was obtained. In light of this, we believe that this is a correct way to identify variables of value for the patient-specific outcomes as described. This is the same method that the authors describe in their Letter.
In addition, a Cox analysis needs correct and corresponding data. The problem mentioned regarding the use of our chosen Cox regression model is a highly statistical problem and has been a subject of discussion by statisticians for a long time. We agree that every statistical model has advantages and disadvantages, but, with regard to good scientific practice, we provided the methodology used in this paper and came to the results presented. We believe that in a peer-reviewed international paper (with a statistical editor reviewing all manuscripts), the careful description of methods, which were obviously reproducible to the authors of the Letter, imply an intent in terms of the methods and results used. We thank the authors for their critical point of view, but believe that there are some more problems than statistical criticism on a well described and established method, which is reproducible for every reader of this Journal.
This article was not funded by any organization.