检索历史 清除

目的:构建急诊重症监护病房(EICU)住院患者ARDS发生风险列线图模型，并验证模型的预测能力。方法:本研究为观察性研究，采用便利抽样法选取2021年6月至2023年6月徐州医科大学附属连云港医院EICU收治的398例患者作为研究对象，根据建模组在院期间是否发生ARDS分为ARDS组( n＝52)和正常组( n＝226)，验证组120例。通过多因素logistic回归分析EICU患者ARDS发生的影响因素，建立风险预测评分模型，采用受试者操作特征(ROC)曲线下面积检验模型的预测能力，校准图评价模型的校准度，利用Bootstrap法进行内部验证。 结果:ARDS组52例，男27例，女25例，年龄(57.36±8.34)岁；正常组226例，男129例，女97例，年龄(55.42±7.66)岁。2组的性别、年龄差异无统计学意义。与正常组相比，ARDS组肺部感染率高，26.9%(14/52)比12.8%(29/226)；误吸率高，38.5%(20/25)比11.9%(27/226)；既往呼吸系统疾病者多，44.2%(23/52)比16.8%(38/226)；全身感染率高，63.5%(33/52)比13.3%(30/226)；多发性骨折者多，25.0%(13/52)比13.3%(30/226)；严重脑外伤者多，25.0%(13/52)比13.3%(30/226)；休克者多，53.8%(28/52)比20.4%(46/226)；药物中毒者多，30.8%(16/52)比15.0%(34/226)；严重电解质紊乱者多，42.3%(22/52)比18.6%(42/226)(均 P<0.05)。多元logistic回归结果显示：肺部感染、误吸、既往呼吸系统疾病、全身感染、多发性骨折及严重电解质紊乱是EICU住院患者发生ARDS的独立危险因素。列线图结果显示，每个危险因素相加得到的总分明确地表达了ARDS发生的概率。患者既往有呼吸系统疾病(60分)、存在肺部感染(45分)、误吸(85分)、全身感染(65分)、多发性骨折(55分)、严重电解质紊乱(50分)，发生ARDS的概率超过80%，需要临床给予重视。预测模型ROC曲线下面积0.853，敏感度为89.6%，特异度为66.3%。模型预测中校准曲线接近理想曲线，提示该模型有很好的预测效果。 结论:该模型预测效果良好，在临床上能够快速预测EICU住院患者ARDS的发生风险，为医务人员及时提供干预措施提供依据。

Objective:The risk nomogram model of acute respiratory distress syndrome (ARDS) in emergency intensive care unit (EICU) inpatients was constructed, and the predictive ability of the model was verified.Methods:This was a observational study.A total of 398 patients admitted to the EICU of Lianyungang Hospital Affiliated to Xuzhou Medical University from June 2021 to June 2023 were selected by convenience sampling method as the research subjects.According to whether ARDS occurred during the hospital stay, they were divided into ARDS group ( n=52) and normal group ( n=226), validation group ( n=120).Multivariate logistic regression was used to analyze the influencing factors of ARDS occurrence in EICU patients, and a risk prediction score model was established, and the area under the receiver operating characteristic (ROC) curve was used to test the predictive ability of the model.The calibration chart was used to evaluate the calibration degree of the model, and Bootstrap method was used for internal verification. Results:There were 52 cases in ARDS group, including 27 males and 25 females, aged (57.36±8.34) years, and 226 cases in normal group, including 129 males and 97 females, aged (55.42±7.66) years.There was no significant difference in gender and age between the two groups.Compared with the normal group, there was higher proportion of patients in the ARDS group in the following items: pulmonary infection, 26.9% (14/52) vs 12.8% (29/226); aspiration, 38.5% (20/52) vs 11.9% (27/226); previous history of respiratory diseases, 44.2% (23/52) vs 16.8% (38/226); systemic infection, 63.5% (33/52) vs 13.3% (30/226); multiple fractures, 25.0% (13/52) vs 10.2% (23/226); severe brain trauma, 25.0% (13/52) vs 13.3% (30/226); shock, 53.8% (28/52) vs 20.4% (46/226); drug poisoning, 30.8% (16/52) vs 15.0% (34/226), and severe electrolyte disorder, 42.3% (22/52) vs 18.6%(42/226)(all P<0.05).The results of multivariate logistic regression showed that pulmonary infection, aspiration, previous respiratory diseases, systemic infection, multiple fractures, and severe electrolyte disorder were the influencing factors of ARDS in EICU inpatients.The results of the nomogram showed that the total score obtained by adding each risk factor clearly expresses the probability of ARDS occurrence.If patients have respiratory diseases (60 points), pulmonary infection (45 points), aspiration (85 points), systemic infection (65 points), multiple fractures (55 points), and severe electrolyte disorders (50 points), the probability of ARDS is more than 80%, which requires clinical attention.The area under ROC curve of the prediction model was 0.853, the sensitivity was 89.6%, and the specificity was 66.3%.The calibration curve is close to the ideal curve, indicating that the model has a good prediction effect. Conclusions:The model has a good predictive effect, and can quickly predict the risk of ARDS in EICU inpatients in clinical practice, and provide a basis for medical staff to provide timely intervention measures.