基于床旁简易指标急性胰腺炎护理预警模型的构建  

Construction of Nursing Early Warning Model for Acute Pancreatitis Based on Simple Bedside Indicators

作　　者：李备 林俊 翟惠敏[1] LI Bei;LIN Jun;ZHAI Hui-min(School of Nursing,Southern Medical University,Guangzhou 510515,China;Affiliated Xiaolan Hospital,Southern Medical University,Zhongshan 528415,China)

机构地区：[1]南方医科大学护理学院,广东广州510515 [2]南方医科大学附属小榄医院,广东中山528415

出　　处：《护理学报》2022年第12期6-10,共5页Journal of Nursing（China）

基　　金：中山市医学科研项目(2020J242)。

摘　　要：目的基于床旁简易指标,构建急性胰腺炎护理预警模型并进行验证,探讨其在急性胰腺炎患者病情变化时的预测价值。方法回顾性分析广东省某三级甲等医院的急性胰腺炎病例378例,将建模集(n=244)的重症急性胰腺炎组和非重症组各项危险因素进行比较,采用Logistic回归构建模型,应用ROC曲线下面积检验模型区分度,利用Hosmer-Lemeshow拟合优度检验评价模型校准度;将模型带入验证集(n=134),并与BISAP评分、APACHE Ⅱ评分进行预测效能对比。结果最终纳入模型的预警指标有心率(OR=1.040)、呼吸(OR=1.269)、血氧饱和度(OR=0.824)、肠鸣音微弱(OR=2.383)、肠鸣音消失(OR=3.227)、肺部听诊音(OR=2.185)和腹膜刺激征(OR=2.097)。Hosmer-Lemeshow拟合优度检验P=0.278,ROC曲线下面积为0.840(95%CI:0.788~0.892,P<0.001),最大约登指数为0.541,灵敏度为68.3%,特异度为85.8%。模型验证结果:验证集ROC曲线下面积0.802(95%CI:0.714~0.891,P<0.001),最大约登指数0.605;灵敏度70.5%,低于BISAP评分的84.1%和APACHE II评分的79.5%;特异度90.0%,高于两种评分的84.4%和83.3%;准确率78.4%,低于两种评分的85.1%和82.1%。结论本模型易得易测、简便快捷,预测急性胰腺炎严重程度与公认的BISAP和APACHE II评分相近,凸显了护士床旁开展快速评估的优势,可作为急性胰腺炎病情评价体系的一种补充。Objective To construct a nursing early warning model for acute pancreatitis based on simple bedside indicators,to make validation and to discuss its predictive value when the disease condition changes.Methods We conducted a retrospective study with data of 378 acute pancreatitis cases from a tertiary grade-A Hospital in Guangdong.Cases were categorized into the training set(n=244)and further subdivided into severe acute pancreatitis cases and non-severe cases;the remaining cases were categorized into the validation set(n=134).Risk factors in the validation set and two non-severe groups in the training set were comparatively analyzed.Logistic regression analysis was adopted to construct a warning model.The area under the receiver operating characteristic curve(ROC-AUC)was calculated to test the discrimination of the model,and Hosmer-Lemeshow goodness-of-fit test was performed to test the accuracy.The validation set was used to test the warning performance of the model,which was compared with the BISAP and APACHE Ⅱ scoring methods.Results Seven indicators were identified and included in the warning model,including heart rate(OR=1.040),respiratory(OR=1.269),pulse oxygen saturation(OR=0.824),weak bowel sounds(OR=2.383),disappearance of bowel sounds(OR=3.227),pulmonary auscultation(OR=2.185)and peritoneal irritation(OR=2.097).the p-value in Hosmer-Lemeshow goodness-of-fit test was 0.278.ROC-AUC was 0.840(95%CI:0.788~0.892,P<0.001);maximized Youden index 0.541,sensitivity 68.3%and specificity 85.8%.In the validation set,the ROC-AUC was 0.802(95%CI:0.714~0.891,P<0.001),and maximized Youden index 0.605.The sensitivity(70.5%vs 84.1%vs 79.5%)and accuracy(78.4%vs 85.1%vs 82.1%)were lower than those by BISAP and APACHE II scoring methods,while the specificity was higher(90.0%vs 84.4%vs 83.3%).Conclusion The model in the study is easy to establish and convenient and exhibits good performance equivalent to the recognized BISAP and APACHE II scores in predicting acute pancreatitis severity.It also shows advantages in rapid be

关 键 词：急性胰腺炎 预警模型 床旁 护理 

分 类 号：R473[医药卫生—护理学]