Construction And Validation Of Risk Assessment Tool For Stress Hyperglycemia In ICU Patients Based On Meta-analysis

Objectives:Meta-analysis was used to obtain the combined effect size of the influencing factors of stress hyperglycemia in ICU patients,the risk prediction model for early prediction of stress hyperglycemia in ICU patients was established,and the risk assessment chart of ICU patients with stress hyperglycemia is drawn,so as to facilitate the early prediction of the risk level of ICU patients with stress hyperglycemia.Targeted prevention and intervention measures are taken to alleviate the pain of patients,reduce hospitalization costs,reduce the workload of medical staff,and prevent the occurrence of stress hyperglycemia.Methods:This study involved three parts:（1）Meta-analysis of the influencing factors of stress hyperglycemia in ICU patients:developing retrieval strategies and retrieving data,the eligible literatures,based on the determined inclusion and exclusion criteria,were screened out,the influencing factors and related data were extracted,and the combined effects were calculated by meta-analysis;（2）Construction and evaluation of stress hyperglycemia risk prediction model in ICU patients:The combined effect size corresponding to each influencing factor was logarithmically transformed as the model coefficient,and logistic regression analysis was carried out.ROC curve was drawn by using the collected clinical data to evaluate the performance of the model;（3）Drawing and calculation of risk assessment chart of ICU patients with stress hyperglycemia:draw the logistic regression analysis curve according to the 10 000simulated samples generated by the R Studio software,paint the risk assessment chart according to the divided risk level nodes,draw the ROC curve of risk assessment chart using the collected clinical data,and calculate the positive predictive value and negative predictive value of risk assessment chart.Results:（1）Meta-analysis results:Finally,7 qualified literatures were screened out from 10databases,and 6 influencing factors were extracted,namely age>65,female,lung disease,abnormal renal function,abnormal blood pressure and smoking history,with OR values of 2.65、1.23、1.71、1.35、1.40 and 1.62 respectively;（2）The risk prediction model of stress hyperglycemia in ICU patients:Logit（P）=α+0.975X₁+0.207X₂+0.536X₃+0.3X₄+0.336X₅+0.482X₆According to the clinical data of 86 patients meeting the criteria for inclusion and exclusion,α=-0.426 was obtained.The ROC curve was drawn using SPSS21.0,AUC=0.793,the 95%CI was（0.685,0.900）,P=0.000,the sensitivity was 67.6%,and the specificity was 94.2%;（3）The determination of the incidence risk level nodes and verification results ofthe risk assessment chart:according to the trend of scattered point map of logistic regression model and repeated verification,0.552 and 0.750 were selected as the low,medium and high risk risk classification nodes.The risk assessment chart was used to evaluate the risk level of 86 patients,and the ROC curve was drawn with SPSS21.0,AUC=0.776,95%CI was（0.669,0.883）,P=0.000,sensitivity was 70.6%,and specificity was 84.6%.Conclusions:In this study,Meta-analysis was used to screen the influencing factors of stress hyperglycemia in ICU patients,and a risk prediction model of stress hyperglycemia in ICU patients was established.On the basis of the model,a risk assessment chart was drawn.The validation results showed that the prediction is effective,which has reference significance for early identification and targeted early intervention of high-risk patients with stress hyperglycemia in ICU.