Construction And Application Of The Risk Prediction Model On 30-Day Readmission In Older Patients With Chronic Heart Failure

Aim1.To construct "The risk factors assessment tool of readmission in older patients with chronic heart failure" based on Andersen’s Behavioral Model.2.To investigate the influencing factors of readmission in older patients with chronic heart failure within 30 days after discharge.3.To build the risk prediction model of 30-day readmission in older patients with chronic heart failure and to test the prediction ability through internal verification.4.To apply the risk prediction model in clinical environment and evaluate the accuracy.Method1.Firstly,all influencing factors that might affect the readmission of older patients with chronic heart failure were collected through literature review.Secondly,older patients with chronic heart failure who readmitted within 30 days after discharge were interviewed to explore the influencing factors of readmission."The risk factors assessment tool of readmission in older patients with chronic heart failure" was constructed after focus group discussion.Finally,pre-research was conducted to evaluate the feasibility and applicability of the assessment tool.2.Patients with heart failure in three tertiary hospitals in Tianjin from March2019 to June 2020 were enrolled.A prospective nested case-control study design was adopted and patients were followed up for 30 days after discharge."The risk factors assessment tool of readmission in older patients with chronic heart failure" was used to investigate.3.The data from the second part was used to construct the prediction model.The readmission within 30 days after discharge was taken as the dependent variable and the influencing factors with statistical significance in univariate analysis were taken as the independent variables.The Logistic regression,Cox proportional hazards regression and BP neural network were used to establish the risk prediction model.4.The patients with heart failure in three tertiary hospitals in Tianjin from July to October 2020 were enrolled in applied research.A prospective nested case-control study design was used to collect the independent variables with statistical significance in univariate analysis from the second part.The accuracy of prediction model was evaluated by the area under the ROC curve.Results1.Based on Andersen’s Behavioral Model,the methods of literature review,semi-structured interview,focus group discussion and pre-research were used and "The risk factors assessment tool of readmission in older patients with chronic heart failure " was constructed which included 4 dimensions and 41 indicators.2.A total of 381 patients with heart failure were included,of which 55(14.4%)readmitted within 30 days after discharge.There were 22 factors in univariate analysis statistically significant: social support,objective support,subjective support,utilization of support,health literacy,information acquisition ability,communication and interaction ability,willingness to improve health,willingness to financial support,anxiety level,anxiety grade,depression level,depression grade,history of operation,history of major events,changing the type of medicine by oneself,return visit on time,quality of life,physical field,emotional field,other field,self-care behavior.3.After the multiple collinearity test,collinearity were found between the following variables: anxiety level and anxiety grade,depression level and depression grade,history of operation and history of major events.It was difficult to determine which three variables should be removed to get an effective prediction model.Therefore,Logistic regression,Cox proportional hazard regression and BP neural network were used to construct the prediction model with four combined independent variables.The four Logistic regression models and the second,the third BP neural network models had good internal verification effect.4.A total of 150 patients with heart failure were enrolled in applied research.The results showed that the first Logistic regression model constructed in this study had the biggest AUC value which meant the best predictive effect.Conclusion1.Through the methods of literature review,semi-structured interview,focus group discussion and pre-research,"The risk factors assessment tool of readmission in older patients with chronic heart failure" based on Andersen’s Behavioral Model was constructed.2.In this study,the readmission rate of older patients with heart failure within 30 days after discharge was 14.4%.The 30-day readmission risk was higher in following patients: lower social support,lower health literacy,higher anxiety level,higher depression level,no history of operation,no history of major events,no changing drug types by oneself,failure to return visit on time,lower quality of life.,More attention should be paid for patients with these characteristics.Early nursing intervention should be carried out to reduce the 30-day readmission risk for these patients.3.The four Logistic regression models and two BP neural network models had good internal verification effect,while the four Cox proportional hazards regression models and other two BP artificial neural network models had poor internal verification effect.4.The first Logistic regression model had the best predictive effect in applied research.This model provides a reliable method for clinical nurses to evaluate 30-day readmission risk in older patients with heart failure.