Predictive Model Construction And Effect Verification Of Postoperative Delirium In ICU Patients With Multiple Injuries

ObjectiveIdentify the risk factors of postoperative delirium in ICU patients with multiple injuries,establish and verify delirium prediction models.In order to provide guidance for medical staff to identify high-risk groups of delirium in time,and provide reference for medical staff to take preventive measures,so as to effectively prevent and control the occurrence of delirium.MethodsThis study used a retrospective cohort study to investigate the relevant medical records of patients with multiple injuries who were admitted to the ICU of a tertiary hospital in Zhengzhou from July 2018 to December 2019.The literature related to postoperative delirium in patients with multiple injuries in ICU was searched through Chinese database.Wanfang,and Weipu,and foreign language databases such as Web of science,PubMed,and Cochrane Library,"ICU delirium","postoperative delirium","risk factors","risk factors" as the search terms,and the English is "multiple injury patients","ICU delirium","postoperative delirium","risk factors","dangerous""factors" is the search term.Referring to clinical practice guidelines at home and abroad,systematic reviews,expert consensus,consult ICU and surgical experts to select risk factors that may be related to the occurrence of postoperative delirium in ICU patients with multiple injuries,use the hospital’s electronic medical record system to collect relevant data,and use Excel for data Data is sorted out,SPSS26.0 software was used for statistical analysis.The APACHE Ⅱ data was scored by ±standard deviation and compared between groups by t-test.The ratio was expressed as(%)and compared between groups by χ2 test.Divide the collected medical records into a modeling group and a verification group in chronological order,and select 208 patients from July 1,2018 to June 30,2019 as the modeling group,from July 1,2019 to 2019 On December 31,96 patients were used as the validation group.The single factor analysis method was used to screen out the risk factors that entered the model,multivariate data were mined by Logistic regression and β value was determined by analysis,and the regression equation was established.Using the Hosmer-Lemeshow(H.L)chi-square test model to test the degree of agreement,it is statistically significant,it shows that the prediction results of the model have significant differences and the actual incidence in ICU patients with multiple injuries,and the degree of agreement is poor.On the contrary,it shows that the consistency between the two is better;use the validation group data to evaluate the predictive performance of the model,the model can accurately distinguish whether patients have postoperative delirium,which is reflected by the area under the AUC curve.The closer the AUC is to 1,the stronger the distinguishing ability is.ResultsCollected the data of 208 patients in the modeling group,and performed univariate analysis.The results showed that age,history of alcoholism,high blood pressure,diabetes,history of smoking,type of surgery,education level,method of anesthesia,intraoperative blood infusion,mechanical ventilation10 There are statistically significant differences in these clinical variables.Through binary logistic regression analysis,the results showed:age OR=2.217,95%CI(1.821～4.824);hypertension OR=5.325,95%CI(1.782～25.773);type of surgery 2 OR=2.204,95%CI(1.922～5.857),type of surgery 3 OR=2.910,95%CI(1.664～5.044),type of surgery 4 OR=2.049,95%CI(1.507～6.326);intraoperative allogeneic blood infusion OR=2.276,95%CI(1.624～5.542);the 4 risk factors in the postoperative delirium group and non-delirium in ICU patients with multiple injuries The difference between the groups was statistically significant,therefore,these risk factors are the four independent risk factors for postoperative delirium in ICU patients with multiple injuries.The constructed prediction model is eZ/(1+eZ)×100%,and e is an exponential function;Z=-15.734+(0.191 ×age)+(1.669×assignment of hypertension)+(1.592×assignment of operation type 2)+(0.940×assignment of operation type 3)+(3.021 ×assignment of operation type 4)+(1.459×assignment of intraoperative allogeneic blood infusion).The area under the ROC curve of the model is 0.830,and the 95%CI is(0.747～0.913).When the model threshold is set to 5.65%,the corresponding combination of sensitivity and specificity is optimal.After the prediction effect test,it is found that the sensitivity of this model is 65.38%,the specificity is 81.43%,and the accuracy rate is 77.08%.ConclusionsDelirium is a common complication in ICU and is related to many factors.Through this study,it was found that the patients age,hypertension,type of surgery,and intraoperative blood transfusion were independent risk factors for delirium.And through the prediction model of delirium to verify the risk factors,in order to provide guidance for medical personnel to identify the high-risk groups of delirium in time,for medical personnel to take preventive measures to provide reference,so as to effectively prevent and control the occurrence of delirium in ICU patients with multiple injuries.