Research On The Formation Of Hierarchical Management Modules Under The ICU Delirium Risk Prediction Model

ObjectiveTo analyze and explore the risk factors of delirium in ICU patients and build a delirium risk prediction model.To understand the needs of ICU nurses for delirium management.To build the module content of the ICU patient delirium hierarchical management platform,providing a basis for future ICU patient delirium management based on mobile medicine.Method(1)In the first part,a questionnaire for data collection of delirium risk factors was formed on the basis of literature review and group discussion.According to the established inclusion and exclusion criteria,the researchers collected a total of 367 ICU patients in a tertiary hospital in Hengyang from September 2019 to January 2020 and May to October2020,and analyzed the risk factors of ICU delirium,establishing a risk prediction model and verify the effectiveness of the model.(2)The second part adopted the purpose sampling method to select ICU nurses from a tertiary hospital in Hengyang from May to August2020 to conduct semi-structured interviews.Through in-depth interviews,we can understand the knowledge and management needs of ICU nurses about ICU delirium.(3)The third part was based on the previous two parts.Through literature review and group discussion,the content of the platform module was initially drafted,and the final module content of the ICU delirium grading management platform was determined through the expert meeting method.The statistical analysis of the data used SPSS23.0 software.Measurement data and count data were described by means±standard deviation and frequency and composition ratio respectively.The measurement data that obey the normal distribution adopted the t test of two independent samples,otherwise the Mann-Whitney non-parametric test was used.The enumeration data were compared using the chi-square test.The risk factors of delirium in ICU were analyzed by binary logistic regression analysis,and P<0.05 indicated that the difference was statistically significant.The predictive ability of ICU delirium risk prediction model was analyzed by receiver operating curve(ROC curve).Result(1)73 of 367 ICU patients(19.89%)developed delirium.After binary logistic regression analysis,ICU patients’age,APACHEⅡscore,infection,Surgery,mechanical ventilation and Aspartate aminotransfer--ase were the main risk factors for ICU delirium.The final prediction model formula were R=e~Y/(e~Y+1),e is the exponential function,Y=-8.029+0.614*age(=1,2,3 or 4)+1.292*APACHEⅡ(=1 or 2)+1.175*是否感染(0 or 1)+1.085*Surgery(=0 or 1)+0.855*whether mechanical ventilation(=0 or 1)+1.048*Aspartate aminotransferase(=1 or 2).Using the Hosmer-Lemeshow model fit test to getχ~2=8.671,P=0.371,showing that the model fited well.(2)Used the prediction model formula R as the test variable,and drawed the ROC curve with the delirium outcome as the state variable.Finally,the area under the ROC curve(AUC)was 0.823,the 95%CI was(0.769,0.877),and the standard error was 0.028,P=0.000,indicating that the constructed delirium prediction model could predict patients with delirium and non-delirium,and the predictive ability was moderate(AUC between 0.70-0.90).According to the model’s maximum Youden index(0.55)and model risk threshold(24.24%),patients were divided into low risk(R≤25%),moderate risk(R=25%-55%),and high risk(R≥55%).The difference in the incidence of ICU delirium among the three levels was statistically significant(P<0.05).The sensitivity of the model was0.740,and the specificity was 0.810.(3)In-depth interviews with ICU nurses extracted 5 themes:limited knowledge of delirium(unclear/incomplete);the majority of negative feelings during delirium nursing(high physical and mental stress/worries/helpless);delirium in ICU patients There are limitations in management(ICU working environment/unregulated requirements of departments);ICU delirium management needs clear(knowledge training/management tools and platforms);understanding of the combination of medical information platforms and delirium(high recognition and affirmation/clear module requirements).(4)Based on the qualitative research and the constructed delirium risk prediction model,the content of the platform modules was determined through the expert meeting method,and the expert’s authority coefficient was 0.8625,which was highly authoritative.Finally,the ICU delirium grading management platform formed through the expert meeting method The module content includes 3 ports and 15 modules.Conclusion(1)The risk factors for delirium in ICU patients were age,history of alcohol abuse,APACHE II score,infection,and mechanical ventilation.A delirium risk prediction model was constructed through Logistic regression analysis.The area under the ROC curve verified by the model was 0.823,which has moderate predictive ability.Based on this,clinical medical staff can quantify the risk of delirium in ICU patients,prevent it early,reduce the occurrence of delirium and reduce the harm.(2)ICU nurses had insufficient grasp of the basic knowledge and assessment methods of delirium,and had certain needs in this regard.Therefore,relevant departments should strengthen delirium knowledge training and provide convenient platforms for medical staff to learn and use,so as to improve their management of delirium ability.(3)Based on the analysis of ICU delirium risk factors,the construction of risk prediction models,qualitative research,guidelines and literature review,and group discussions,this study initially constructed the content of the ICU delirium grading management platform module,and adopted the expert meeting method to evaluat and revisie,the final content of the platform module had more authoritative scientificity and reliability,which can provide a reference for the construction of a mobile medical-based ICU patient delirium management platform in the future.