Analysis of risk factors for death in patients with sepsis

Abstract

Objective To construct a death risk prediction model of sepsis, in order to provide reference for improving the early warning of death risk and improving the outcome of sepsis patients. Methods The data of 286 sepsis patients who met the diagnostic criteria in the comprehensive ICU of Tongji Hospital, Tongji Medical College of Huazhong University of Science and Technology from January 2019 to May 2022 were retrospectively collected. The t test, chi-square test and Mann-Whitney Utest were used to conduct univariate analysis on the death risk of sepsis patients, and then multivariate analysis was carried out, and a prediction model of sepsis death risk was established. Results The survival status of 28 days after admission to ICU was calculated. Of 286 patients with sepsis, 165 (57.69%) survived and 121 (42.31%) died. The initial infection sites were lung, abdominal cavity, skin and soft tissue, urinary system, the time between diagnosis of sepsis and admission, time from onset to first cluster therapy, age, lactic acid level and APACHEⅡ score at admission to ICU, occurrence of hypothermia, continuous renal replacement therapy (CRRT), first symptoms of fever at onset, length of stay in ICU, the levels of creatinine, procalcitonin and fibrinogen at ICU admission were independent influencing factors of death in sepsis patients (P＜0.05). The area under the ROC curve of the death risk prediction model for sepsis patients was 0.970, with a sensitivity of 0.893, a specificity of 0.933, and a maximum Youden index of 0.826. Conclusion The prevention focus of death risk in sepsis patients needs to be moved forward, the risk factors of pre-hospital death of patients should be paied attention to, the changes in the condition before confirmatory treatment should be grasped, in order to reduce and avoid adverse outcomes. In addition, the preliminarily constructed death risk prediction model for sepsis patients in this study has good predictive ability, and can provide certain reference value in clinical work.

Keywords

• sepsis， death， risk prediction model， continuous renal replacement therapy， procalcitonin， fibrinogen