A commentary on “Retrospective cohort development and internal validation of a prediction model for 30-day mortality after TEVAR in Type B aortic dissection”

Dear Editor, We read with great interest the article by Ming et al, which aimed to develop and internally validate a multivariable prediction model for 30-day mortality after thoracic endovascular aortic repair[1]. The study identified several factors associated with 30-day mortality; however, several methodological and presentational aspects require clarification to ensure the robustness and interpretability of the findings. First, regarding variable selection, the abstract indicates that variables with P<0.05 in univariate analysis were included in the multivariate model, whereas the methods section states that variables with P<0.1 were selected. This inconsistency in the significance threshold could substantially influence the final model composition and its performance. A detailed explanation of the variable inclusion criteria would be appreciated. Second, there is ambiguity concerning the dataset used for model development and validation. Standard practice involves screening variables and constructing the model in a training set[2,3], followed by validation in an independent internal or external set. In this study, the methods section notes that “the full cohort was randomly divided into training (70%, n = 708) and validation (30%, n = 303) sets to develop and internally validate the prediction model.” However, the results section reports model performance metrics, such as an AUC of 0.92 (95% CI 0.88–0.96), for the full cohort rather than separately for the training and validation sets. Clarification on which dataset was used for variable selection and model evaluation would greatly enhance reader understanding. Finally, we noted discrepancies between certain results described in the text and the corresponding figures. Specifically, the data presented in Figure 2I, Figure 4A, and Figure 4D appear inconsistent with the descriptions provided in the results section. Further verification and, if necessary, correction of these figures or their captions would ensure the accuracy of the reported findings. We thank the authors for their valuable work and look forward to their clarifications on these points.