Analysis of clinical features and risk factors for mortality in patients with culture-negative sepsis: a single-center retrospective cohort study based on MIMIC-Ⅳ / 中华危重病急救医学

ABSTRACT

Objective:To compare the characteristics and outcomes of culture-positive sepsis (CPS) with culture-negative sepsis (CNS) patients in order to understand the impact of CNS on prognosis and explore the possible risk factors for mortality.Methods:A retrospective cohort study was conducted. Patients with sepsis were identified from the Medical Information Mart for Intensive Care database-Ⅳ v0.4 (MIMIC-Ⅳ v0.4). Patients were divided into CPS and CNS groups according to the culture results within 24 hours before and after the diagnosis of sepsis. General information, baseline characteristics, and medical operation data between CNS and CPS groups were compared. Logistic regression analysis was used to calculate the relationship between CNS and in-hospital mortality under three regression models. Chi-square analysis and mediation analysis were used to analyze the effect of initial antibiotic and prior antibiotic use within 90 days on the in-hospital mortality of CNS. Results:A total of 8 587 patients with sepsis were enrolled in the final analysis, including 5 483 patients in the CPS group and 3 104 patients in the CNS group. Compared with the CPS group, the patients in the CNS group were younger [years old 68 (56, 79) vs. 70 (58, 81)], had higher sequential organ failure assessment (SOFA) score and higher proportion of using mechanical ventilation, renal replacement therapy and vasopressin within 24 hours after intensive care unit (ICU) admission [SOFA score 3 (2, 5) vs. 3 (2, 4), mechanical ventilation 48.61% (1 509/3 104) vs. 39.25% (2 152/5 483), renal replacement therapy 13.69% (425/3 104) vs. 9.68% (531/5 483), vasopressin 15.79% (490/3 104) vs. 13.44% (737/5 483)], longer length of ICU stay [days 5 (3, 10) vs. 3 (2, 6)] and higher in-hospital mortality [25.00% (776/3 104) vs. 18.53% (1 016/5 483)], with significant differences (all P < 0.01). However, there was no significant difference in gender, ICU type, simplified acute physiology score Ⅱ (SAPS Ⅱ), and Charlson comorbidity index (CCI) score between the two groups. After adjustment for multiple confounding factors, CNS was still a risk factor for in-hospital mortality [odds ratio ( OR) = 1.441, 95% confidence interval (95% CI) was 1.273-1.630, P < 0.001]. The results of Chi-square analysis and mediation analysis showed that the initial antibiotic had no significant effect on the higher in-hospital mortality of CNS, while the prior use of antibiotics within 90 days was related to higher in-hospital mortality of CNS ( OR = 1.683, 95% CI was 1.328-2.134, P < 0.05). The mediating effect of CNS in prior antibiotic use within 90 days and in-hospital death was significant ( Z = 5.302, P < 0.001), accounting for 7.58%. Conclusions:Compared with CPS, CNS was more severe and had a worse prognosis. Prior use of antibiotics within 90 days may be related to the higher in-hospital mortality of CNS patients, but it could not fully explain the high mortality of CNS.