Multiple predictor models for diagnosis of sepsis in trauma patients / 中华创伤杂志

ABSTRACT

Objective To set up a warning diagnostic model by using the commonly used clinical indicators in an attempt to provide a basis for the early,fast and accurate diagnosis of posttraumatic sepsis.Methods Based on the presence of sepsis,165 patients were grouped into sepsis group (n =45) and non-sepsis group (n =120).Body temperature,respiration,heart rate,C-reactive protein(CRP),white blood cell,blood platelet count(PLT),activated partial thromboplastin time (APTT),acute physiology and chronic health evaluation Ⅱ (APACHE Ⅱ) score and sepsis-related organ failure assessment(SOFA) score were tested to identify the independent predictors of sepsis.Warning diagnostic models of unweighted score (unwScore) and weighted score (wScore) for posttraumatic sepsis were constructed by combining the independent variables.Receiver operation characteristic curve (ROC) was used to evaluate the independent predictor and warning diagnostic models for posttraumatic sepsis.Results Body temperature,respiration,heart rate,CRP,APACHE Ⅱ score and SOFA score were significantly different between the two groups(P ＜ 0.05).Multiple analysis showed body temperature,CRP and APACHE Ⅱ score were independently associated with sepsis.With the ROC analysis,areal under the curve (AUC),sensitivity,specificity,positive predictive value and negative predictive value of unwScore (0.915,0.87,0.85,69.64％ and 94.50％) and wScore (0.931,0.96,0.78,63.24％ and 97.85％) were better than these of body temperature (0.855,0.84,0.78,59.38％ and 93.07％),CRP (0.761,0.64,0.80,55.77％ and 85.84％) and APACHE Ⅱ (0.884,0.84,0.82,64.41％ and 93.40％).Conclusions Body temperature,CRP and APACHE Ⅱ score are independent predictors of sepsis.Models combining body temperature,CRP and APACHE Ⅱ score demonstrate high performance in diagnosing sepsis in trauma patients.