摘要： 目的 分析急性敌草快中毒患者死亡危险因素并构建列线图预测模型。方法 回顾性分析2018年2月至2023年2月中国医科大学附属第一医院急诊科收治的123例急性敌草快中毒患者的临床资料，以是否28 d内死亡为终点事件，分为存活组和死亡组。其中，存活组61例，死亡组62例。记录2组患者性别、年龄、口服敌草快剂量、来诊时是否低血压、是否在本院进行灌流、入院首次外周血白细胞计数（WBC）、血红蛋白（Hb）、外周血血小板计数（PLT）、谷丙转氨酶（ALT）、碱性磷酸酶（ALP）、总胆红素（TBil）、直接胆红素（DBil）、血清白蛋白（ALB）、钾离子（K⁺）、血清尿素氮（BUN）、血清肌酐（Cr）、葡萄糖（Glu）、肌钙蛋白Ⅰ （TnⅠ）、脑钠肽（BNP）、pH、PaO₂、PaCO₂、乳酸（Lac）。采用LASSO回归分析筛选可能影响急性敌草快中毒患者死亡的危险因素并进行多因素logistic回归分析，根据多因素分析结果建立列线图模型，绘制受试者工作特征曲线（ROC）进行内部验证。结果 LASSO回归分析选出14个潜在危险因素，分别为敌草快剂量、低血压、WBC、Hb、PLT、ALT、TBil、DBil、K⁺、Cr、Glu、BNP、PaO₂、Lac；多因素logistic回归分析结果显示敌草快剂量、低血压、WBC、Lac为急性敌草快中毒死亡的独立危险因素［敌草快剂量（OR=12.059，95% CI：1.822~79.822）、低血压（OR=85.182，95% CI：2.2~3 365.3）、WBC （OR=14.96，95% CI：2.03~110.12）、Lac （OR=4.026，95% CI：1.878~6.174）；均P<0.05 ］。列线图模型预测急性敌草快患者死亡风险的曲线下面积（AUC）为0.978 （95% CI：0.958~0.997），用留一法交叉验证的AUC为0.963 （95% CI：0.935~0.990）；校准曲线分析结果显示列线图预测与实际观测之间具有最佳的一致性。结论 本研究成功构建了急性敌草快中毒患者死亡风险列线图预测模型；该列线图模型可准确分析急性敌草快中毒患者的死亡风险，能够帮助临床医生正确决策并可准确预测患者结局。

急性敌草快中毒患者死亡风险列线图预测模型的构建和验证

李蕙伊, 董雪松

中国医科大学附属第一医院急诊科, 沈阳 110001

收稿日期:2023-03-27出版日期:2023-08-30发布日期:2023-08-07
通讯作者:董雪松E-mail:xsdong@cmu.edu.cn
作者简介:李蕙伊(1992-),女,医师,硕士.
基金资助:国家自然科学基金（81971821）


关键词: 敌草快, 中毒, 死亡风险, 列线图, 预测模型
Abstract: Objective To analyze the risk factors of death in patients with diquat (DQ) poisoning in the emergency department and to construct a nomogram prediction model. Methods A total of 123 patients with DQ poisoning admitted to the emergency Department of the First Hospital of China Medical University between February,2018,and February,2023,were retrospectively investigated. The patients were categorized into two groups:the survival group,consisting of 61 cases,and the non-survival group,comprising 62 cases where death occurred within 28 days as the defined end event. Clinical parameters,such as gender,age,DQ dose,hypotension at admission,hospital perfusion history and blood parameters at admission,such as peripheral blood white blood cell count (WBC),hemoglobin (Hb),peripheral blood platelet count (PLT),alanine aminotransferase (ALT),alkaline phosphatase (ALP),total bilirubin (TBil),direct bilirubin (DBil),serum albumin,potassium ion (K⁺),serum urea nitrogen,serum creatinine (Cr),glucose (Glu),troponinⅠ,brain natriuretic peptide (BNP),pH,PaO₂,PaCO₂,and lactic acid (Lac) were recorded. LASSO regression was used to screen risk factors affecting mortality in patients with acute DQ poisoning,and multivariate logistic regression was performed. A nomogram model was developed based on the results of the multivariate logistic regression analysis,and the receiver operating characteristic (ROC) curves were drawn for internal validation. Results A total of 14 potential risk factors were selected by LASSO regression analysis,which were DQ dose,hypotension,WBC,Hb,PLT,ALT,TBil,DBil,K⁺,Cr,Glu,BNP,PaO₂ and Lac. Methods of the multivariate logistic regression analysis showed that DQ dose,hypotension,WBC and Lac were independent risk factors of death in patients with acute DQ poisoning[DQ dose (OR=12.059,95%CI:1.822-79.822),hypotension (OR=85.182,95%CI:2.2-3 365.3),WBC (OR=14.96,95%CI:2.03-110.12),Lac (OR=4.026,95%CI:1.878-6.174),all P < 0.05]. Area under the curve (AUC) for predicting mortality risk in patients with acute DQ poisoning was 0.978 (95%CI:0.958-0.997) according to the nomogram model and 0.963 (95%CI:0.935-30.990) based on the leave-one-out cross-validation. Calibration curve analysis showed that the predicted risk probability of acute DQ poisoning based on the nomogram model was consistent with the actual probability. Conclusion A nomogram model for predicting mortality risk in patients with acute DQ poisoning was successfully constructed. The nomogram model can accurately analyze the risk of death in patients with acute DQ poisoning and can help with clinical decision making and accurately predicting patient outcomes.
Key words: diquat, poisoning, risk of death, nomogram, prediction model
PDF全文下载地址:

https://journal.cmu.edu.cn/CN/article/downloadArticleFile.do?attachType=PDF&id=3258