ICU患者耐碳青霉烯类铜绿假单胞菌感染风险预测模型的构建

Construction of a predictive model for risks of carbapenem-resistant pseudomonas aeruginosa infection in ICU patients

目的构建ICU患者耐碳青霉烯类铜绿假单胞菌(CR-PA)感染的预测模型并验证其效能。方法选择2021年1月至2022年12月在苏州市立医院ICU发生铜绿假单胞菌(PA)感染的846例患者作为研究对象,并将其分为模型组(n=507)和验证组(n=339)。收集患者的临床资料,使用单因素分析和多因素Logistic回归分析其发生CR-PA感染的影响因素,并通过R统计软件建立列线图预测模型,采用受试者工作特征(ROC)曲线、校准曲线、临床决策曲线(DCA)评价模型并通过验证组验证模型。结果多因素Logistic回归分析结果显示,住院时间≥14 d[OR=5.01(95%CI:2.97~8.46)]、碳青霉烯类药物使用时间≥7 d[OR=6.52(95%CI:3.39~12.52)]、使用机械通气[OR=4.35(95%CI:2.20~8.60)]、合并下呼吸道感染[OR=4.56(95%CI:2.42~8.58)]、检出前使用抗菌药物[OR=1.88(95%CI:1.11~3.20)]和清蛋白≤30 g/L[OR=2.09(95%CI:1.19~3.69)]是ICU患者发生CR-PA感染的独立危险因素(P<0.05),据此构建风险预测模型。建模组和验证组验证该模型的ROC曲线下面积(AUC)分别为0.880(95%CI:0.850~0.910)和0.878(95%CI:0.838~0.918)。校准曲线显示该模型具有较好的校准度。DCA显示该模型有较好的临床效用。结论该研究构建的模型预测效果良好,可以帮助临床医生识别高风险患者并采取相应的预防和干预措施,从而减少CR-PA感染的发生率。

Objective To construct a predictive model for carbapenem-resistant Pseudomonas aeruginosa(CR-PA)infection in ICU patients and validate its efficacy.Methods A total of 846 patients with PA infection in the ICU from January 2021 to December 2022 in Suzhou Municipal Hospital were selected as the study subjects and divided into a modeling group(n=507)and a validation group(n=339).Clinical data of the patients were collected,and the influencing factors for CR-PA infection were analyzed by univariate and multivariate Logistic regression.A nomogram predictive model was established by R software,and the model was evaluated by receiver operating characteristic(ROC)curve,calibration curve and decision curve analysis(DCA),which were then validated in the validation group.Results Multivariate Logistic aggression analysis results showed that the duration of hospitalization≥14 days[OR=5.01(95%CI:2.97-8.46)],the duration of carbapenem usage≥7 d[OR=6.52(95%CI:3.39-12.52)],the usage of mechanical ventilation[OR=4.35(95%CI:2.20-8.60)],combined with lower respiratory tract infections[OR=4.56(95%CI:2.42-8.58)],prior usage of antibiotics[OR=1.88(95%CI:1.11-3.20)]and albumin≤30 g/L[OR=2.09(95%CI:1.19-3.69)]were independent risk factors for CR-PA infection in ICU patients(P<0.05),the risk prediction model was constructed based on the above indicators.The area under the ROC curve(AUC)for validating the risk prediction mode in modeling group and validation group were 0.880(95%CI:0.850-0.910)and 0.878(95%CI:0.838-0.918),respectively.The calibration curve indicated good calibration of the model,and the DCA demonstrated good clinical utility.Conclusion The model constructed in this study has a good predictive effect,which can help clinicians to identify high-risk patients and take appropriate preventive and intervention measures,thus reducing the incidence of CR-PA infection.