利用回归模型构建呼吸机相关肺炎发病风险预测评分系统

Establishment of a risk assessment scoring system for ventilator-associated pneumonia

目的构建呼吸机相关肺炎(VAP)发病风险预测评分系统。方法选取温州市人民医院2017年1月至2018年12月行机械通气的595例住院患者为研究对象。基于赤池信息量准则构建预测模型,取75%的研究对象为建模组(445例),剩余研究对象为验证组(150例)。利用logistic回归模型对机械通气患者(建模组)发生VAP的影响因素进行分析,建立风险预测评分系统,再利用ROC曲线验证该评分系统的预测效果。结果595例患者VAP发生率为7.73%。年龄≥65岁、连续发热≥2 d、手术、曾检出耐碳青霉烯类肠杆菌科(CRE)、曾检出耐碳青霉烯类鲍曼不动杆菌(CRAB)、曾检出耐碳青霉烯类铜绿假单胞菌(CRPA)、脑血管意外史、血降钙素原升高是机械通气患者(建模组)发生VAP的独立影响因素(均P<0.05),其对应的风险分值为年龄≥65岁(1分)、连续发热≥2 d(2分)、手术(-1分)、曾检出CRE(2分)、曾检出CRAB(2分)、曾检出CRPA(2分)、脑血管意外史(1分)、血降钙素原升高(1分);≥3分为高危人群。利用风险预测评分系统评估机械通气患者发生VAP的效能,结果显示建模组AUC、灵敏度、特异度、约登指数分别为0.938、1.000、0.694、0.694,验证组分别为0.855、0.769、0.891、0.660。结论本研究建立并验证的ICU住院患者VAP发病风险预测评分系统有助于对高危患者进行监测治疗。

Objective To establish a risk assessment scoring system for ventilator-associated pneumonia(VAP).Methods A total of 595 patients receiving mechanical ventilation in our hospital from January 2017 to December 2018 were enrolled in the study.A prediction model was constructed according to Akaike information criterion,75%of the research objects were assigned as derivation group(445 cases),and the remaining 150 cases were assigned as the validation group.The influencing factors of ventilator-associated pneumonia were analyzed with Logistic regression model and the risk assessment scoring system was established;the prediction effect of the model was assessed with ROC curve.Results The ventilator-associated pneumonia occurred in 46 patients with an incidence rate of 7.73%(46/595).Logistic regression analysis showed that≥65 years old,fever≥2 d,surgery,CRE,CRAB,CRPA,history of cerebrovascular accident and elevated PCT were independent risk factors for VAP.The risk assessment scoring system was:≥65 years old(1 score),fever≥2 d(2 scores),surgery(-1 score),CRE(2 scores),CRAB(2 scores),CRPA(2 scores),history of cerebrovascular accident(1 score)and elevated PCT(1 score).Patients with the risk assessment score≥3 points were at high risk of VAP.The area under ROC curve(AUC)of the scoring system in derivation group and validation group was 0.938 and 0.855,sensitivity was 1.000 and 0.769,specificity was 0.694 and 0.891,respectively.Conclusion Risk assessment scoring system for VAP has been established and validated.It may be used for targeted surveillance to high risk patients.