Abstract
Introduction. Early detection of patients with a high risk of death in sepsis makes it possible to optimize intensive care. The objective of the study was to perform a comparative assessment of the prognostic values of various laboratory indicators and their combinations in predicting mortality in patients with sepsis. Materials and methods. We analyzed the medical records of 280 patients diagnosed with sepsis. Levels of more than 50 laboratory parameters were assessed upon admission in groups of patients who survived and those who died during treatment. The discriminatory abilities of the identified laboratory predictors were assessed using ROC analysis. The Kaplan–Meier method was used to assess survival. Results. Among the studied patients, the mortality rate was 50%. Differences between the groups of surviving and deceased patients were identified according to levels of urea, total protein, albumin, lactate, pH, platelet and neutrophil count. The lowest level α (p < 0.001) was found for the ratio of urea to creatinine and lactate to albumin. Based on this, an integral biomarker consisting of the product of these two parameters was calculated and analyzed. Its values also differed between groups and amounted to 3.9 (2.9–6.8) and 9.8 (4.8–18.8) in the group of surviving and deceased patients, respectively (p < 0.001). When performing ROC analysis, the optimal cut-off point for our predictor, separating survivors from deceased patients, was 6.6 with an AUC of 0.758 (95% CI 0.693–0.813), with a sensitivity of 67% and specificity of 75%, exceeding similar indicators for the APACHE II scale. We also found that patients with low levels of our proposed predictor had a threefold increase in median survival compared to the subgroup of patients with high levels (23 days and 8 days, respectively; p < 0.001). Conclusion. An integrated indicator based on plasma levels of urea, creatinine, lactate, and albumin is a simple and accessible biomarker for early prediction of mortality in patients with sepsis.

This publication has 32 references indexed in Scilit: