Development Of 28-day Mortality Risk Prediction Model For Patients With Sepsis Induced Coagulopathy Based On MIMIC-Ⅲ Database

Objective 1.To confirm the incidence,general characteristics,comorbidities and clinical or lab test item of sepsis induced coagulopathy patients in MIMIC-Ⅲ database.2.Extracting target population data from the MIMIC-Ⅲ database,then develop a 28-day mortality risk prediction model based on the independent risk factors screened by multivariate Logistic regression analysis.Methods Target population data in MIMIC-Ⅲ database were obtained using SQL statement codes,and then screened by the following inclusion and exclusion criteria.Inclusion criteria:（1）Patients with confirmed or suspected infection（Patients receiving antibiotics within 72 hours of admission to the intensive care unit）;（2）Patients with SOFA score ≥2.Exclusion criteria:（1）Juvenile patients（age<18 years old）;（2）Patients with primary blood system diseases;（3）Pregnant or birth-giving women;（4）Cancer patients or patients on chemotherapy drugs;（5）Patients admitted to intensive care unit for less than 24 hours;（6）Patients who received heparin anticoagulant therapy within 24 hours after admission to ICU and warfarin therapy for long duration of disease.The data of sepsis patients obtained by the above criteria were further integrated and screened and interpolated with missing values by Stata.The patients were divided into SIC group and non-SIC group according to SIC score,and the general situation,comorbidities and laboratory indicators of the obtained patients were statistically described.According to whether the SIC patients died at 28 days,the patients were divided into survival group and death group.Univariate and multivariate Logistic regression analysis was used to determine the risk factors of death in SIC patients.The prognostic model was developed based on the results of the multivariate Logistic regression analysis.Results 1.A total of 9186 eligible sepsis patients were included,among which 3034 patients met the diagnosis of SIC.About 60% of SIC patients are male,people aged 40~60 years and 60~80 years were the high-risk population of SIC.A total of 1058 SIC patients died within 28 days（35%）.There were significant differences between patients who died within 28 days and those who survived in basic complications,laboratory indicators,organ function,disease severity score,and clinical outcome.2.Univariate Logistic regression analysis showed that basic diabetes,basic liver diseases,renal replacement therapy,antiplatelet aggregation therapy,AGE,HR,MAP,RR,TEMP,INR,PT,APTT,PLT,WBC,MCH,MCHC,MCV,RDW,PH,PO2,SO2,GLU,Lac,K+,Na+,Cl-,AG,AB,creatinine,urea nitrogen and the results of pathogen culture were significantly related to the 28-day mortality of SIC patients.3.Multivariate Logistic regression analysis showed that base with liver disease,using of norepinephrine,older age,low mean arterial pressure,pyknocardia,shortness of breath frequency,hypothermia,the reduce of spo2,elevated levels of blood lactic acid,chloride ion concentration anomaly,low platelet count,increased RDW,blood urea nitrogen increased are independent risk factors for 28-day mortality of SIC patients,Long-term treatment with antiplatelet aggregation,definite urinary tract infection and definite catheter-associated infection were protective factors for SIC patients.The prediction model of SIC patients was established based on the results of Logistic regression analysis and backward stepwise regression analysis The equation=1/（1+exp-（13.53+Age×0.04+Basic complicated liver disease× 0.50+Use of norepinephrine×0.75-Received antiplatelet aggregation therapy×0.82+ Heart rate×0.02-Mean arterial pressure×0.02+Respiratory rate×0.05-Body temperat-ure×0.30-SO2×0.06-Chloride ion concentration×0.03+ Blood lactic acid level×0.08-Platelet count×0.1+RDW×0.18+ Urea nitrogen×0.01-Identify urinary tract infection×0.47-Identify catheter-related infections×1.34））.The AUROC was 0.81.4.The continuous variables in the independent risk factors screened by multivariate Logistic regression analysis were transformed into categorical variables according to certain criteria to further improve the multivariate Logistic regression and stepwise regression analysis.Then a 28-day mortality risk scale was developed for SIC patients.The area under ROC curve of the scale was 0.78 based on the model group dataset and 0.75 based on the validation group dataset.Compared with the existing scoring system,the area under ROC curve drawn based on model group data set is significantly larger than SOFA,LODS,SAPSII and SIC scores.The area under ROC curve drawn based on validation group data set is significantly larger than SOFA and SIC scores.It is also larger than SAPSII and LOD scores,However,the distinction does not prove a statistical significance.But,compared with LODS and SAPSII score,the 28-day mortality risk scale scoring rules are simpler.Conclusion 1.SIC has a high morbidity and mortality,middle-aged and elderly men are the high incidence of SIC,thrombocytopenia is the most sensitive diagnostic and prognostic index of SIC.2.Advanced age,basic complicated liver disease,shock,renal insufficiency,large heterogeneity of erythrocyte volume,and low platelet count predict poor prognosis of SIC patients 3.This study based on patients’ general condition and clinical,laboratory indicators established a 28-day mortality risk prediction model for patients with SIC.Compared with the existing prognostic scoring system,it has certain advantages in predictive performance,and easy to be operated.Besides it has a certain reference significance for the prognosis assessment of clinical diagnosis and treatment and the population inclusion criteria of subsequent related studies.