Construction and validation of early warning model for acute aortic dissection
10.3760/cma.j.issn.1671-0282.2023.07.005
- VernacularTitle:急性主动脉夹层早期预警模型的构建及验证
- Author:
Fengqing LIAO
1
;
Chenling YAO
;
Guorong GU
;
Yao YU
;
Dongxu CHEN
;
Yannan ZHOU
;
Canguang CAI
;
Humaerbieke ALIMA·
;
Chen CHEN
;
Siying ZHOU
;
Zhenju SONG
;
Chaoyang TONG
Author Information
1. 复旦大学附属中山医院急诊科，上海 200032
- Keywords:
Aortic dissection;
Chest pain;
Risk factors;
Diagnostic model;
Nomogram
- From:
Chinese Journal of Emergency Medicine
2023;32(7):874-880
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the clinical characteristics of patients with acute aortic dissection (AAD) through a retrospective and observational study, and to construct an early warning model of AAD that could be used in the emergency room.Methods:The data of 11 583 patients in the Emergency Chest Pain Center from January to December 2019 were retrospectively collected from the Chest Pain Database of Zhongshan Hospital Affiliated to Fudan University. Inclusion criteria: patients with chest pain who attended the Emergency Chest Pain Center between January and December 2019. Exclusion criteria were 1) younger than 18 years, 2) no chest/back pain, 3) patients with incomplete clinical information, and 4) patients with a previous definite diagnosis of aortic dissection who had or had not undergone surgery. The clinical data of 9668 patients with acute chest/back pain were finally collected, excluding 53 patients with previous definite diagnosis of AAD and/or without surgical aortic dissection. A total of 9 615 patients were enrolled as the modeling cohort for early diagnosis of AAD. The patients were divided into the AAD group and non-AAD group according to whether AAD was diagnosed. Risk factors were screened by univariate and multivariate logistic regression, the best fitting model was selected for inclusion in the study, and the early warning model was constructed and visualized based on the nomogram function in R software. The model performance was evaluated by accuracy, specificity, sensitivity, positive likelihood ratio and negative likelihood ratio. The model was validated by a validation cohort of 4808 patients who met the inclusion/exclusion criteria from January 2020 to June 2020 in the Emergency Chest Pain Center of the hospital. The effect of early diagnosis and early warning model was evaluated by calibration curve.Results:After multivariate analysis, the risk factors for AAD were male sex ( OR=0.241, P<0.001), cutting/tear-like pain ( OR=38.309, P<0.001), hypertension ( OR=1.943, P=0.007), high-risk medical history ( OR=12.773, P<0.001), high-risk signs ( OR=7.383, P=0.007), and the first D-dimer value ( OR=1.165, P<0.001), Protective factors include diabetes( OR=0.329, P=0.027) and coronary heart disease ( OR=0.121, P<0.001). The area under the ROC curve (AUC) of the early diagnosis and warning model constructed by combining the risk factors was 0.939(95 CI:0.909-0.969). Preliminary validation results showed that the AUC of the early diagnosis and warning model was 0.910(95 CI:0.870-0.949). Conclusions:Sex, cutting/tear-like pain, hypertension, high-risk medical history, high-risk signs, and first D-dimer value are independent risk factors for early diagnosis of AAD. The model constructed by these risk factors has a good effect on the early diagnosis and warning of AAD, which is helpful for the early clinical identification of AAD patients.
