Logistic regression analysis of the risk factors of patients with sepsis in intensive care unit

Abstract

To investigate the correlation of the data of patients with sepsis in intensive care unit (ICU) and their prognosis. Clinical data of 450 cases with sepsis in ICU from January 2004 to December 2008 were retrospectively studied. Twenty-four possible prognostic factors including clinical, chemical examinations and scores contributing to the survival of these patients were analyzed with univariate and multivariate logistic regression analysis. Univariate logistic regression analysis: eight factors were non-risk factors for sepsis, including days of hospitalization [odds ratio (OR)=0.951, P<0.001], Glasgow coma score (GCS, OR=0.858 , P<0.001), mean arterial pressure (OR=0.991, P=0.041), platelet count (OR=0.997, P<0.001), total protein (OR=0.975,P=0.003), albumin (OR=0.955, P<0.001), arterial pH (OR=0.077,P=0.001), arterial oxygen pressure (OR=0.996, P=0.014). The risk factors for sepsis included acute physiology and chronic health evaluation II (APACHE II) score (OR=1.115, P<0.001), body temperature (OR=1.215, P=0.041), prothrombin time (OR=1.057, P=0.007), aspartate aminotransferase (OR=1.001, P=0.011), serum creatinine (OR=1.002, P=0.002), blood potassium level (OR=1.391, P=0.003), blood glucose (OR=1.051, P=0.002), and C-reactive protein (OR=1.245, P=0.001). Multivariate logistic regression analysis showed that risk factors of sepsis included APACHE II score (OR=1.094, P<0.001), days of hospitalization (OR=0.946, P<0.001), platelet count (OR=0.976, P=0.004), and albumin level (OR=0.957, P=0.017). APACHE II score, days of hospitalization, platelet count and albumin could predict the prognosis of sepsis.