术前血糖达标的老年糖尿病骨折患者术中血糖变异性与麻醉术后谵妄的相关性

Association between intraoperative glucose variability and postoperative delirium in elderly patients with diabetic fractures who meet preoperative blood glucose targets

目的探讨术前血糖达标的老年糖尿病骨折患者术中血糖变异性与麻醉术后并发谵妄(POD)的相关性,分析麻醉术后并发谵妄的独立影响因素并构建和验证风险预测模型。方法回顾性分析浙江省绍兴市柯桥区中医医院2019年8月~2023年4月进行骨折手术的老年糖尿病患者120例的相关资料,并按照7:3分为模型集(84例)和验证集(36例),术后1~7天内对患者采用谵妄诊断量表(The Confusion Assessment Method,CAM)评估POD发生情况并按其结果分为谵妄组和非谵妄组。收集患者一般资料和术中血糖变异性相关指标[血糖变异系数(CV)、平均血糖值(GluAve)、血糖标准差(GluSD)和血糖不稳定指数(GLI)]的临床资料,通过单因素分析和多因素Logistic回归分析筛选患者术后并发谵妄的独立影响因素,并构建术后并发谵妄的风险预测模型。运用R绘制出Nomogram图(列线图)将风险预测模型可视化,通过验证校准图判断实际发生率和预测发生率的一致性。以ROC曲线评估风险预测模型的预测价值,并采用Hosmer-Lemeshow检验判断模型的拟合优度。结果资料显示,84例模型集患者中有24例术后并发谵妄,另60例患者术后1~7天内未发生谵妄,谵妄发生率28.57%。单因素分析结果显示:患者年龄、术前认知状况(MoCA评定量表)、麻醉时间、血糖变异性相关指标(CV、GluAve、GluSD和GLI)、住院时间等在组间比较具有统计学意义(均P<0.05)。进一步多因素Logistic回归分析得出:患者年龄、麻醉时间、CV、GluAve、GluSD和GLI均为术后并发谵妄的独立影响因素。风险预测模型为:Logit(P)=23.954-1.106×(麻醉时间)-0.696×(年龄)-1.216×(CV)+1.276×(GluAve)-2.977×(GluSD)-0.677×(GLI)。ROC曲线下面积(AUC)为0.964(95%CI:0.928—1.000),敏感度92.3%,特异度89.7%,绘制建模集的预测模型校准图当中校准曲线贴近于标准曲线,提示该模型一致性较好,Hosmer-Lemeshow拟合优度检验结果显示Х^(2)=5.608,P=0.870;外部验证AUC为0.956(95%CI:0.915—0.997),其敏感度88.5%,特异度93.1%。结论患者年龄、术前认知状况、麻醉时间、血糖变异性(CV、GluAve、GluSD和GLI)均为术后并发谵妄的独立影响因素,基于上述因素构建出的风险预测模型具有很好的临床评估效能和参考价值,可以帮助及时评估患者病情并制定临床救治方案,降低老年糖尿病骨折患者术后并发谵妄的风险。

Objective To explore the correlation between intraoperative glucose variability and postoperative delirium(POD)in elderly patients with diabetic fractures who met the preoperative blood glucose standard,analyze the independent influencing factors of postoperative delirium after anesthesia,and construct and verify the risk prediction model.Methods A retrospective analysis was performed on 120 elderly diabetic patients who underwent fracture surgery in Shaoxing Keqiao Distract of Traditional Chinese Medicine Hospital,Zhejiang Province from August 2019 to April 2023.According to the 7:3 scale,the model set(84 cases)and the verification set(36 cases)were divided,and the Confusion Assessment Method(CAM)was used to evaluate the occurrence of POD within 1~7 days after surgery,and the patients were divided into delirium group and non-delirium group according to their results.The general data of patients and the relevant indicators of intraoperative blood glucose variability[glycemic coefficient of variation(CV),mean blood glucose value(GluAve),blood glucose standard deviation(GluSD)and glycemic instability index(GLI)]were collected,and the independent influencing factors of postoperative delirium were screened by univariate analysis and further multivariate logistic regression analysis,and a risk prediction model of postoperative delirium was constructed.A Nomogram plot(nomogram plot)is drawn using R to visualize the risk prediction model,and the consistency between the actual incidence rate and the predicted incidence rate is judged by verifying the calibration chart.The predicted value of the risk prediction model was evaluated by the receiver operating characteristic curve(ROC),and the Hosmer-Lemeshow test was used to judge the goodness-of-fit of the model.Results The data showed that 24 of the 84 patients in the model set had postoperative delirium,and the other 60 patients did not have delirium within 1~7 days after surgery,and the incidence of postoperative delirium was 28.57%.Univariate analysis showed that the age,preoperative cognitive status(MoCA rating scale),duration of anesthesia,blood glucose variability(CV,GluAve,GluSD and GLI),and length of hospital stay were statistically significant between groups(all P<0.05).Further multivariate logistic regression analysis showed that the patient’s age,duration of anesthesia,CV,GluAve,GluSD and GLI were independent influencing factors of postoperative delirium.The risk prediction model is:Logit(P)=23.954-1.106×(duration of anesthesia)-0.696×(Age)-1.216×(CV)+1.276×(GluAve)-2.977×(GluSD)-0.677×(GLI).The area under the ROC curve(AUC)is 0.964(95%CI:0.928—1.000),it sensitivity is 92.3%,the specificity is 89.7%,the calibration curve in the calibration chart of the predictive model of the modeling set is close to the standard curve,indicating that the model consistency is good,the Hosmer-Lemeshow goodness-of-fit test results showХ^(2)=5.608,P=0.870;the external verification AUC is 0.956(95%CI:0.915—0.997),it sensitivity was 88.5%,and it specificity was 93.1%.Conclusion The risk prediction model constructed based on the above factors has good clinical evaluation efficiency and reference value,which can help timely assess the patient’s condition and formulate clinical treatment plan,and reduce the risk of postoperative delirium in elderly diabetic fracture patients.