| Objective:In this study,relevant clinical Information was extracted from the Medical Information Mart for Intensive Care(MIMIC)Ⅳ database,a large open database of intensive care in the United States.The clinical features and 28-day mortality of sepsis patients infected with gram-positive bacteria,gram-negative bacteria,and fungi were compared,and the influence of different pathogens on the prognosis of sepsis patients was further determined.At the same time,we also analyzed the effects of different infection sites and whether multidrug-resistant bacteria were involved on the mortality of sepsis patients.Methods:We extracted clinical data of septic patients with a single pathogen from the online critical care medicine database-MIMIC-Ⅳ.According to the results of microbial culture within 48 hours before and after admission to ICU,the patients were divided into gram-negative bacteria group,gram-positive bacteria group,and fungal group.The clinical characteristics of patients with sepsis of different infection types were analyzed.Then,we analyzed the clinical characteristics of sepsis patients with different infections.The primary outcome was 28-day mortality.The secondary outcomes were in-hospital mortality,7-day Mortality,length of hospital stay,length of ICU stay,and ventilation duration.In addition,Kaplan-Meier analysis was used for the 28-day cumulative survival rate of patients with sepsis.We then performed univariate and multivariate Logistic regression analyses for 28-day mortality to identify independent risk factors in sepsis patients.Finally,we created and validated a nomogram for predicting 28-day mortality.Results:A total of 6584 patients with sepsis conforming to the exclusion criteria were extracted from the MIMIC-Ⅳ database(n=6584).According to different microbial test results,they were divided into the gram-negative bacteria group(n=2142),the gram-positive bacteria group(n=3438),and the fungal group(n=1004).The overall 28-day mortality rate was 19.1%.Among these,the 28-day mortality rate was 31.8%in the fungal group,which was significantly higher than that in the gram-negative bacteria group(15.4%,p<0.05)and gram-positive bacteria group(17.7%,p<0.05).In a univariate regression model,the risk of outcome events was 1.22times(95%CI:1.05-1.42,p=0.009)and 2.48 times(95%CI:2.07-2.97,p<0.001)in the gram-positive bacteria and fungal group respectively,with the gram-negative bacteria group as the reference group.After adjustment for multivariate regression models,gram-positive bacterial and fungal infections were still significant risk factors for 28-day mortality,1.50 times(95%CI:1.24-1.81)and 2.16 times(95%CI:1.72-2.71,p<0.001)as compared with the gram-negative group,respectively.The in-hospital mortality,7-day in-hospital mortality,and mechanical ventilation duration in the gram-negative bacteria group,gram-positive bacteria group,and fimgus group were 15.8%,18.0%,32.6%,10.6%,12.1%,19.6%,40.0 ± 89.3 days,40.9 ± 82.7 days,77.2 ± 110.4 days,respectively.The fungal group was significantly higher than the gram-negative group and gram-positive bacteria group,with statistical significance(p<0.001).The length of hospital stay in the gram-negative bacteria group,gram-positive bacteria group,and the fungal group was 13.2 days(IQR,5.6-15.6),9.5 days(IQR,5.9-16.6)and 13 days(IQR,7.4-21.5),respectively,with statistical significance.Survival curve analysis showed that the prognosis of the fungal group was poor,with log-rank test p<0.001.Furthermore,in subgroup analysis,we found that 28-day mortality from bloodstream infection was higher than that from non-bloodstream infection in both the Gram-negative and fungal groups(p=0.002),while no difference was found in the overall and gram-positive bacteria groups.In the gram-positive bacteria group,the prognosis of patients with multi-drug-resistant bacteria worse than that of non-drug-resistant bacteria(p=0.002),but there was no statistical difference in the overall and gram-negative bacteria groups.Conclusion:Organism type of infection is associated with mortality of sepsis,and early identification of the microbiological type of a patient with sepsis will provide an understanding of the patient’s condition and guide treatment. |