Prediction model for poor prognosis of patients with severe pulmonary infection-related sepsis and its predictive efficiency
OBJECTIVE To establish the prediction model for poor prognosis of the patients with severe pulmonary infection-related sepsis and assess the predictive efficiency of the model.METHODS The clinical data were collect-ed from 165 patients with severe pulmonary infection-related sepsis who were treated in Sanmenxia Central Hospi-tal from Apr 2018 to Feb 2021 and were divided into the favorable prognosis group with 58 cases and the poor prognosis group with 107 cases according to the survival status after the treatment for 28 days.The prediction model for poor prognosis was established by logistics regression analysis,the fitting degree of model was evaluated by Hosmer-Lemeshow test,and the predictive efficiency of the model was assessed by means of receiver operating characteristic(ROC)curves.RESULTS The proportion of patients complicated with acute kidney injury was signifi-cantly higher in the poor prognosis group than in the favorable prognosis group(P＜0.05).The white blood cell counts,urea nitrogen,serum creatinine,total bilirubin,aspartate aminotransferase,alanine aminotransferase,C-reactive protein,procalcitonin,acute physiology and chronic health evaluationⅡ(APACHE Ⅱ)score and sepsis-related organ failure assess-ment(SOFA)score were significantly elevated(P＜0.05);the eradication rates of prealbumin and 24-hour lactic acid were reduced(P＜0.05).The result of logistic regression analysis showed that complication with acute kidney injury,A-PACHEⅡ score and SOFA score were the independent risk factors for the poor prognosis of the patients with severe pul-monary infection-related sepsis(P＜0.05),and the eradication rate of 24-hour lactic acid was the independent protective factor(P＜0.05).The prediction model for poor prognosis of the patients with severe pulmonary infection-related sepsis was acquired by multivariate logistic regression equation,and the Hosmer-Lemeshow test showed that the fitting degree of the model was good(x2=12.691,P=0.124).ROC curve analysis indicated that the model showed high predictive effi-ciency,with the area under curve(AUC)0.881,95％CI:0.814-0.949.CONCLUSION The prediction model for poor prognosis of the patients with severe pulmonary infection-related sepsis that is established based on logistics re-gression analysis shows high prediction efficiency,the fitting degree of the model is satisfying,and it can provide data for guidance of diagnosis and treatment of sepsis.
Severe pulmonary infectionSepsisPrognosisPrediction modelLogistic regression analysisPre-dictive efficiency
NETL
NSTL
万方数据
