Construction And Validation Of An Early Warning Model For Readmission Within 90 Days In Patients With Chronic Obstructive Pulmonary Disease And Frailty

Objective:To explore the risk factors of readmission within 90 days in patients with chronic obstructive pulmonary disease and frailty,and to construct a clinically applicable early warning model.Then,combined with external verification to test the prediction accuracy of the model.Methods:This study selected COPD patients with frailty who were hospitalized in the Department of Respiratory and Critical Care Medicine of Yixing Hospital Affiliated to Jiangsu University from January 1,2020 to September 30,2021 as the development cohort study subjects.The subjects were divided into readmission group and control group according to the readmission within 90 days after discharge.The general data of the two groups were collected for univariate analysis,and the variables with statistically significant differences were included into multivariate logistic regression analysis to identify the risk factors for readmission within90 days in patients with COPD and frailty.On the basis of clear risk factors,a risk quantitative warning model for readmission within 90 days of COPD patients with frailty was established.Then,external verification was carried out to evaluate the predictive efficacy of the model.The external validation cohort selected COPD patients with frailty who were hospitalized in the Department of Respiratory and Critical Care Medicine of Yixing Hospital Affiliated to Jiangsu University from October 1,2021 to June 30,2022 as subjects.We substituted the data of the external validation cohort into the constructed early warning model,calculated the prediction probability and draw the receiver operating characteristic curve accordingly.Then we compared the area under the ROC curve(AUC)of the development cohort and the external validation cohort and used the Hosmer-Lemeshow goodness of fit test to evaluate the effectiveness of the warning model.Results:1.A total of 303 COPD patients were included in the development cohort.104 patients were readmitted within 90 days after discharge,and the readmission rate within 90 days was34.32 %.2.Compared with the control group,the BMI of the readmission group was lower,the proportion of patients Hospitalized for acute exacerbation ≥ 2 times in the past year was higher,the scores of m MRC,CCI and REFS were higher,and 4MGS was slower(P<0.05).3.BMI,Hospitalized for acute exacerbation ≥ 2 times in the past year,CCI score,REFS score and 4MGS were independent risk factors for readmission within 90 days in patients with COPD and frailty.The OR(95%CI)were 0.847(0.737-0.973),2.635(1.422-4.884),1.304(1.069-1.589),1.500(1.224-1.839),0.040(0.003-0.564)(P<0.05).4.Construction of risk early warning model(PRE_D): the probability of readmission within90 days of COPD patients with frailty P=1/(1+e-X);X=-0.927 +(-0.166×BMI)+(0.969×Hospitalized for acute exacerbation ≥ 2 times in the past year)+(0.265×CCI score)+(0.405×REFS score)+(-3.209×4MGS).5.The AUC of the early warning model(PRE_D)was 0.744 [95% CI(0.687-0.801)].The sensitivity and specificity were 70.2% and 68.3%,respectively.The optimal critical value of the early warning model(PRE_D)is P=0.335,which can be used as a stratification standard for the risk of readmission within 90 days in patients with COPD and frailty.6.The early warning model(PRE_D)contained five predictors: BMI,Hospitalized for acute exacerbation ≥ 2 times in the past year,CCI score,REFS score and 4MGS.The AUC was0.603[95%CI(0.540-0.666)],0.598[95%CI(0.528-0.667)],0.602[95%CI(0.535-0.669)],0.648[95%CI(0.585-0.710)],0.577[95%CI(0.511-0.643)].The predictive value of the early warning model(PRE_D)for the risk of readmission within 90 days in patients with COPD and frailty was better than that of a single variable.7.The 303 subjects were divided into high-risk group(n=136)and low-risk group(n=167)by using the optimal critical value P=0.335 of the early warning model(PRE_D).The readmission rates of high-risk group and low-risk group were 53.68% and 18.56%,respectively.The median readmission time in the high-risk group was 62.213 days(95%CI:56.869-67.558),which was significantly shorter than 80.503 days(95%CI:77.103-83.903)in the low-risk group(χ2=42.210,P<0.01).8.External validation of the early warning model(PRE_D): A total of 124 COPD patients with frailty were included in the external validation cohort,and the readmission rate within90 days was 32.26%.The AUC of the external validation cohort was 0.737[95%CI(0.648-0.826)],the sensitivity was 70%,and the specificity was 69%.The difference between the AUC of the development queue and the external validation queue is 0.007,which proves that the prediction efficiency of the early warning model(PRE_D)for the development queue and the external validation queue is similar.9.The results of Hosmer-Lemeshow goodness of fit test: χ2=11.112,P=0.195>0.05,indicating that the early warning model(PRE_D)has good calibration.Conclusion:BMI,Hospitalized for acute exacerbation ≥ 2 times in the past year,CCI score,REFS score and 4MGS were independent risk factors for readmission within 90 days in COPD patients with frailty.The early warning model(PRE_D)constructed by combining these five predictors has a moderate predictive value for assessing the 90-day readmission risk of COPD patients with frailty.The Prognostic Early Warning Model(PRE_D)can help clinicians identify potential populations at high risk of readmission as early as possible and reduce the risk of readmission through early intervention and treatment.