Development And Clinical Empirical Validation Of Prognosis Prediction Model Of Chronic Critical Illness Patients

ObjectiveMore and more patients suffered from chronic critical illness(CCI)and died of delayed organ failure in Intensive Care Unit(ICU).This study is to develop and validate a prediction model which could be used for accurate,timely,simple and objective identification of the critical degree of the patients’ condition,as well as provide reference for clinical decisions and communication with patients’ families.Methods1.Development of prediction model of chronic critical illness(PPCCI model):Retrospective case-control study.Adopt the definition of CCI proposed by Research Triangle Institute(RTI)in 2014.From January 2012 to December 2017,344 CCI patients admitted to the Intensive Care Unit(ICU)of the First Affiliated Hospital of Fujian Medical University were collected as the development cohort.There were 172 cases in case group and 172 cases in control group.The control(survival)group was randomly sampled from the eligible survival cases at a ratio of 1: 1 according to the sample size of the case(death)group.The risk factors of P < 0.1 were screened out by Univariate analysis of study variables in case group and control group.Univariate analysis(P < 0.1)and clinically significant variables were gradually eliminated by Binary Logistic regressive analysis to determine independent predictors.PPCCI model is developed by using the corresponding assignment and Beta coefficient of each prediction factor.Area Under Curve(AUC)of Receiver Operating Characteristic Curve(ROC Curve)was used to compare the prediction efficiency of the PPCCI model,Acute Physiology and Chronic Health Evaluation(APACHE)Ⅱ,Modified Early Warning Score(MEWS)and Sequential Organ Failure Assessment(SOFA).Artificial Neural Network(ANN)was used to test the accuracy and stability of the predicted variables in PPCCI model.2.Clinical verification of PPCCI model: Prospective cohort study.From January2018 to March 2019,88 CCI patients admitted to the ICU of the First Affiliated Hospital of Fujian Medical University were collected prospectively as the validation cohort to test the clinical applicability of PPCCI model.There were 44 cases in case group and 44 cases in control group.The control(survival)group was randomly sampled from the eligible survival cases at a ratio of 1: 1 according to the sample size of the case(death)group.The chi-square test was used to compare the adverse outcomes of each risk stratification between the development cohort and the validation cohort.AUC was used to compare the predictive performance of PPCCI model,APACHE Ⅱ,MEWS and SOFA in the validation cohort.Result1.The score of PPCCI model = age assignment ×0.5+ prolonged mechanical ventilation time ×1.5+ sepsis or other serious infections ×1.1+ Glasgow Coma Scale×0.8+ mean arterial pressure ×1.5+ heart rate ×0.9+ respiratory rate ×1.2+ oxygenation index ×0.6+ active bleeding ×1.12.ROC curve analysis showed that the AUC of PPCCI model was 0.934,95% CI was 0.908-0.960,the Youden index was 0.755,the sensitivity was 0.849,and the specificity was 0.906.ANN was used to test the accuracy and stability of the variables in PPCCI model.The AUC calculated by ANN was 0.957 in both the case group and the control group.PPCCI model scores range from 0 to 20.8,and risk stratification is carried out by Quartile : <4.7 is the low-risk group,4.7-6.6 is the medium risk group,6.7-9.2 is the high-risk group,and ≥ 9.3 is the extremely high-risk group.3.Clinical verification of PPCCI model: Chi-square test was used to compare the incidence of adverse outcomes in each risk stratification between the development cohort and the validation cohort: P value of low risk group,medium risk group and high risk group were all 1.000,P value of high risk group was 0.133.The AUC of PPCCI model was 0.965,95% CI was 0.931-0.999,Youden index was 0.819,sensitivity was0.864,and specificity was 0.955.Conclusions1.Age,Prolonged Mechanical Ventilation(PMV),sepsis or other serious infection,Glasgow Coma Scale(GCS),Mean Artery Pressure(MAP),Heart Rate(HR),Respiration Rate(RR),Oxygenation Index(OI)and active bleeding were 9 independent predictors of PPCCI model.2.The PPCCI model developed in this study has a score range of 0-20.8.The higher the score,the higher the incidence of adverse outcomes.3.The PPCCI model has been proved to have high sensitivity and specificity by preliminary clinical verification.And it has high accuracy in predicting the change of CCI condition,which has certain clinical guiding significance.