Predicting postoperative hypoxemia risk factors in the patients after triple‐branched stent graft implantation surgery with acute type A aortic dissection: A retrospective study

Abstract

Objective To evaluate the risk factors of postoperative hypoxemia in patients after triple‐branched stent graft implantation surgery with acute type A aortic dissection by conducting a nomogram. Methods We evaluated 97 patients with acute type A aortic dissection (2020–2021), who underwent triple‐branched stent graft implantation surgery. The independent risk factors were screened using univariate and multivariate logistic regression. We integrated significant factors as well as potential interference factors to build the nomogram model. The accuracy of the nomogram model was determined by using receiver operating characteristic curves (ROC), decision curve analyses (DCA), and calibration plots. Internal verification was evaluated using bootstrap validation. Results Logistic regression analysis illustrated that the postoperative lactic acid, postoperative creatinine, and intraoperative aortic occlusion time were all independent risk factors for hypoxemia. Age, sex, and body mass index (BMI) were clinically relevant for predicting postoperative hypoxemia. We established a nomogram based on these six risk factors. The ROC (area under the curve [AUC] = 0.765), DCA, and calibration confirmed good discriminatory applicability and accuracy of the nomogram. Bootstrap validation (AUC = 0.76) verified the applicability of the nomogram. Conclusions The nomogram model could serve as a tool for the prediction of postoperative hypoxemia in patients after modified triple‐branched stent graft implantation surgery with acute type A aortic dissection.

The catastrophe of hypoxia in complex aortic surgery

“…As postoperative hypoxemia is the focus of this study, 6 it is expected that the variables collected should pertain to the precise definition of hypoxia. However, Wang et al 6 refer to hypoxemia in the context of ARDS with a PaO 2 /FiO 2 ratio of less than or equal to 200 mmHg, not coinciding with the Berlin Definition of a PaO 2 /FiO 2 ratio of equal to or less than 300 mmHg. 6,9 Additionally, there was no mention of the different severities of hypoxia, as hypoxemia is classified into mild hypoxemia for PaO 2 /FiO 2 ratios between 300 and 201 mmHg, moderate hypoxemia for PaO 2 /FiO 2 ratios between 200 and 101 mmHg, and severe hypoxemia for PaO 2 /FiO 2 ratios below or equal to 100 mmHg.…”

“…A population of 97 patients were included in this study, all of whom underwent TBSG implantation at Fujian Union Hospital in the Fujian Province of China within a 3‐month window. The predictive nomogram was based on the result of their study, being that postoperative lactic acid, creatinine, intraoperative, and aortic occlusion time were all independent risk factors for hypoxemia, and that age, sex, and body mass index were clinically relevant for predicting postoperative hypoxemia 6 . The methodology that Wang et al 6 utilize to finalize their nomogram, although practical in approach, is not without its limitations.…”

“…When involving patients in their sample population, Wang et al 6 failed to report their individual demographics. In patients with aortic dissections, certain demographic features could be considered independent risk factors for morbidity and mortality.…”

“…As postoperative hypoxemia is the focus of this study, 6 it is expected that the variables collected should pertain to the precise definition of hypoxia. However, Wang et al 6 refer to hypoxemia in the context of ARDS with a PaO 2 /FiO 2 ratio of less than or equal to 200 mmHg, not coinciding with the Berlin Definition of a PaO 2 /FiO 2 ratio of equal to or less than 300 mmHg 6,9 .…”

“…For example, in the United States of America, acute type B aortic dissections occur more frequently in the black population 7 . As a result of this, applying the results of the predictive nomogram developed by Wang et al 6 to an international population will not be possible. To be reliable, the application of such models should be demonstrated on varied validation data sets in the same setting.…”