老年患者髋关节置换术后谵妄风险预测模型构建及验证

Construction and validation of delirium risk prediction model in elderly patients after hip replacement

目的:分析老年髋关节置换(HR)术患者术后谵妄的风险因素,并建立列线图模型及进行验证。方法:回顾性分析2019年6月~2021年6月合肥市第二人民医院收治的156例行HR术老年患者的临床资料,根据其术后谵妄发生情况分为发生组(n=35)及未发生组(n=121)。采用多因素Logistic回归法分析术后谵妄的危险因素,并在此基础上建立风险列线图预测模型,采用受试者工作特征曲线(ROC)对该模型的预测效能进行评价,另外绘制校准曲线、通过Hosmer-Lemeshow拟合优度检验对其准确度进行测评。结果:多因素Logistic回归向前逐步分析法结果显示年龄、手术时间长、全麻麻醉方式、复苏时间≥1 h、术后疼痛评分≥4分、术后低氧血症和术后4 h C反应蛋白(CRP)高水平均是老年患者HR术后发生谵妄的危险因素(P<0.05);据此建立风险预测模型,ROC曲线分析显示风险模型的Youden指数为0.618,曲线下面积(AUC)为0.820(95%CI:0.758~0.872),灵敏度、特异度分别为83.51%、78.26%;校准曲线显示列线图模型预测术后谵妄的发生和实际谵妄的发生拟合度良好,结果显示该模型预测老年HR术后谵妄风险具有良好准确度(χ^(2)=8.760,P=0.363)。结论:高龄、手术时间长、全麻、复苏时间≥1 h、术后疼痛评分≥4分、术后低氧血症和术后4 h CRP高水平均是老年HR术患者术后谵妄的危险因素,以此结果为基础建立的列线图模型对评估其术后谵妄风险有一定的指导作用,可协助临床工作者筛选高风险人群。

Objective:To analyze the risk factors of postoperative delirium in elderly patients after hip replacement(HR),and to establish a nomogram model and verify it.Methods:The clinical data were retrospectively analyzed in 156 elderly patients undergone HR in our hospital between June of 2019 and 2021.The cases were divided into occurrence group(n=35)and non-occurrence group(n=121)according to the relapse of postoperative delirium.The risk factors of postoperative delirium were analyzed by multivariate logistic regression,and a risk nomogram prediction model was established on this basis,and the prediction efficiency of the model was evaluated by the receiver operating characteristic(ROC)curve.In addition,the calibration curve was drawn and the accuracy was evaluated by the Hosmer lemeshow goodness of fit test.Results:Multivariate logistic regression forward stepwise analysis showed that age,long operation time,general anesthesia,resuscitation time≥1h,postoperative pain score≥4 points,postoperative hypoxemia and high C-reactive protein(CRP)level at 4h after operation were the risk factors for delirium after HR in elderly patients(P<0.05).The risk prediction model was established based on the model.ROC curve analysis indicated that Youden index of the risk model was 0.618,and the area under the curve(AUC)was 0.820(95%CI 0.758-0.872).The sensitivity and specificity were 83.51%and 78.26%,respectively.The calibration curve showed that the nomogram model had a good fit between the prediction of postoperative delirium and the actual delirium,and the result showed that the model had a good accuracy in predicting the risk of delirium after HR in the elderly(X 2=8.760,P=0.363).Conclusion:Age,longer operation time,general anesthesia,resuscitation time≥1h,postoperative pain score≥4 points,postoperative hypoxemia and high CRP at 4h after operation are the risk factors for postoperative delirium in elderly patients after HR,and the nomogram model established based on the results has a certain guiding role in evaluating the risk of postoperative delirium,which can assist clinical workers in screening high-risks for this group of patients.