Development Of An Early Predictive Model For Delirium In The Cardiac Surgical ICU And Preliminary Study Of The Gut Microbiome In The Mechanism Of Delirium

AimsTo establish an early predictive model for cardiac surgical ICU(CSICU)delirium,and to preliminarily explore the mechanism of the CSICU delirium induced by gut microbiome and gut barrier function.Methods1.Validation of the E-PRE-DELIRIC model:Data were extracted from the electronic records in a tertiary teaching hospital in Zhejiang,China during 2018.Predictors were retrieved and delirium was assessed using the Confusion Assessment Method for the ICU(CAM-ICU).The score of the E-PRE-DELIRIC model was calculated and the model was validated.2.Development and validation of a model for early prediction of delirium:Data of patients admitted to the CSICU after cardiac surgery between January and September in2019 were extracted.Candidate predictor variables were selected from patient characteristics and perioperative information available at CSICU admission via logistic regression.The nomogram was then developed in the development set and further validated in the validation set.3.Gut barrier function in the CSICU delirium patients:Thirty-five delirious subjects were matched according to age ranges,sex,type of surgery and postoperative days with 35 non-delirious subjects.Blood samples were collected for quantification of plasma levels of gut barrier function biomarkers(diamine oxidase,D-lactate and endotoxin).The difference between groups and the association between delirium predictors and gut barrier were analyzed.4.Gut microbiome in the CSICU delirium patients:Thirty delirious subjects were matched with 30 non-delirious subjects.Stool samples for bacterial 16S r RNA gene sequencing were collected.Composition of gut microbiome was compared to identify gut microbiota biomarker,and the model of delirium risk-gut barrier-gut microbiome was developed.Results1.A total of 725 patients were included,of whom 120(16.6%)developed delirium.The AUROC was 0.538(95%CI 0.483-0.593)and the AUPRC was 0.175(95%CI0.121-0.201).There was a significant difference between predicted probability and delirium occurrence(x~2=17.326,P=0.027)and also a chance of overestimated delirium.2.A total of 595 patients were included and randomly divided into development set(N=396)and validation set(N=199).The early predictive nomogram consisted of six factors:age,drinking,cognitive impairment,preoperative atrial fibrillation,surgery duration and the European System for Cardiac Operative Risk Evaluation score.The specificity and sensitivity of the model was 0.776 and 0.711,respectively.The AUROC was 0.833(95%CI,0.782-0.884)in the development set and 0.786(95%CI,0.703-0.869)in the validation set.There was no significant difference between the predicted delirium and observed delirium(x~2=7.848,P=0.448),while the performance of the new-developed model was better than that of the E-PRE-DELIRIC model(P<0.001).3.Significant difference in abnormal D-lactate was observed between groups(x~2=6.293,P=0.012).Surgery duration and drinking was associated with elevated D-lactate.The level of D-lactate(OR=1.150,95%CI 1.075-1.355)or abnormal D-lactate(OR=3.120,95%CI 1.091-6.769)were associated with delirium development.4.Theαdiversity was significantly lower in the delirium group than the control group(P<0.05),while only a trend of difference inβdiversity(P=0.06)was observed.The genus Bacteroides and family Bacteroidaceae were identified as biomarkers in the delirium group.The abundance of class Clostridia was lower in the delirium group(14.5%vs 19.9%,P<0.05),while the abundance of family Bacteroidaceae was much higher(44.5%vs 32.2%,P<0.05).The early predicted risk of delirium development was associated with gut barrier and their relationship was mediated by gut microbiome.Conclusion1.The performance of the E-PRE-DELIRIC model in the CSICU was not satisfactory and the early predictive model developed with perioperative information could effectively predict delirium development in the CSICU.2.Patients with delirium in the CSICU have certain gut microbiome dysbiosis and gut barrier dysfunction,and those with a higher risk of delirium development are susceptible to gut microbiota dysbiosis.