Hypoalbuminemia, Low Base Excess Values, and Tachypnea Predict 28-Day Mortality in Severe Sepsis and Septic Shock Patients in the Emergency Department

Abstract

Purpose: The objective of this study was to develop a new nomogram that can predict 28-day mortality in severe sepsis and/or septic shock patients using a combination of several biomarkers that are inexpensive and readily available in most emergency departments, with and without scoring systems.

Materials and methods: We enrolled 561 patients who were admitted to an emergency department (ED) and received early goal-directed therapy for severe sepsis or septic shock. We collected demographic data, initial vital signs, and laboratory data sampled at the time of ED admission. Patients were randomly assigned to a training set or validation set. For the training set, we generated models using independent variables associated with 28-day mortality by multivariate analysis, and developed a new nomogram for the prediction of 28-day mortality. Thereafter, the diagnostic accuracy of the nomogram was tested using the validation set.

Results: The prediction model that included albumin, base excess, and respiratory rate demonstrated the largest area under the receiver operating characteristic curve (AUC) value of 0.8173 [95% confidence interval (CI), 0.7605-0.8741]. The logistic analysis revealed that a conventional scoring system was not associated with 28-day mortality. In the validation set, the discrimination of a newly developed nomogram was also good, with an AUC value of 0.7537 (95% CI, 0.6563-0.8512).

Conclusion: Our new nomogram is valuable in predicting the 28-day mortality of patients with severe sepsis and/or septic shock in the emergency department. Moreover, our readily available nomogram is superior to conventional scoring systems in predicting mortality.

Keywords: Severe sepsis; mortality; nomograms; septic shock.

Fig. 1. Flow diagram of the study subjects. ED, emergency department; SIRS, systemic inflammatory response syndrome.

Fig. 2. Comparisons of APACHE II, NEWS, and SOFA scores versus model 4 in predicting 28-day mortality. Model 4 was composed of albumin, BE, and RR as predictive factors, and showed an AUC value of 0.8173 (95% CI, 0.7605–0.8741). The AUCs of the APACHE II, NEWS, and SOFA scores were 0.6177 (95% CI, 0.5423–0.6931), 0.5940 (95% CI, 0.5137–0.6743), and 0.6005 (95% CI, 0.5256–0.6754), respectively. Model 4 demonstrated a significantly higher AUC value than those of conventional scoring systems (p<0.001) by the Delong test for comparisons of receiver operating characteristic curves. APACHE II, Acute Physiology and Chronic Health Evaluation II; NEWS, National Early Warning Score; SOFA, Sepsis Organ Failure Assessment; BE, base excess; RR, respiratory rate; AUC, area under the curve; CI, confidence interval.

Fig. 3. The newly developed nomogram and external validation. (A) A nomogram for predicting 28-day mortality among patients with severe sepsis and/or septic shock using the training set. (B and C) External validation of the nomograms using the validation set. The discriminative ability of the nomogram was good, with an AUC value of 0.7537 (95% CI, 0.6563–0.8512) (B). Calibration plots (dotted line) show close approximations to the logistic calibration (solid line), indicating good agreement between the predicted and observed probabilities of 28-day mortality (C). BE, base excess; RR, respiratory rate; AUC, area under the curve; CI, confidence interval.