Construction of risk model for healthcareassociated infection with multidrugresistant organisms in general intensive care unit

Abstract:

ObjectiveTo construct the risk model for healthcareassociated infection (HAI) with multidrugresistant organisms(MDROs) in intensive care unit (ICU).Methods836 patients who were admitted to ICU for more than 48 hours between October 2012 and September 2015 were analyzed retrospectively, logistic regression model of HAI was constructed, the model was conducted goodness of fit tests and the area under ROC curve analysis.ResultsAmong 836 patients, incidence of HAI with MDROs was 14.23%(n=119). 15 variables that were statistically significant in univariate analysis were included in logistic multivariate analysis, the results showed that the following variables entered into logistic regression equation: length of ICU stay(OR,2.493 ［95%CI,1.816-3.494］）, underlying diseases(OR,1.536 ［95%CI, 1.243-1.898］), hypoproteinemia(OR,87.211 ［95%CI,36.165-210.304］), ventilator days (OR,1.723 ［95%CI,1.399-2.121］), fever（OR,20.639 ［95%CI,3.462-123.043］）, and primary pulmonary infection (OR,0.295 ［ 95%CI,0.133-0.664］). Evaluation of model effect: sensitivity 95%, specificity 87.9%, the area under ROC curve 0.973. ConclusionLogistic regression model has a high goodness of fit in predicting HAI among ICU patients.