Risk factors and prediction for 28-day mortality of patients with negative systemic inflammatory response syndrome scores at the early stages of sepsis: A retrospective study based on MIMIC and eICU databases

Abstract

Objective: Analyze mortality and risk factors in septic patients with negative systemic inflammatory response syndrome (SIRS) scores and assess the predictive value of various critical illness scoring systems for mortality in this patient population. Methods: A retrospective cohort study using Medical Information Mart for Intensive Care IV and eICU databases focused on early-stage septic patients without mechanical ventilation or vasopressor therapy at intensive care unit admission. Patients were divided into SIRS-positive and SIRS-negative groups based on SIRS scores at admission. Mortality rates were compared between groups, and SIRS-negative patients were categorized by 28-day survival. Logistic regression identified 28-day mortality risk factors in SIRS-negative patients, and the predictive performance of Acute Physiology Score III (APS-3), Sequential Organ Failure Assessment (SOFA), National Early Warning Score (NEWS), and quick SOFA (qSOFA) scores was assessed using receiver operating characteristic curves. Results: The study included 36,794 septic patients, with 34,371 in the SIRS-positive group (7108 mortalities) and 2423 in the SIRS-negative group (166 deaths). The SIRS-negative group had a significantly lower mortality rate (6.85% vs. 20.68%, p < 0.001). Nonsurvivors in the SIRS-negative group were older, had higher Charlson Comorbidity Index (CCI), and had higher lactate and APS-3 scores (all p < 0.05). Logistic regression identified CCI, age, lactate, and APS-3 score as independent mortality predictors. The APS-3 score's predictive power for 28-day mortality was similar to SOFA and superior to NEWS and qSOFA. Conclusions: SIRS-negative septic patients have a lower mortality rate. Risk factors for mortality include high acute physiological scores, elevated lactate, advanced age, and comorbidities. NEWS and qSOFA have limited predictive power for mortality in these patients.