Predictive Accuracy of Sepsis-3 Definitions for Mortality Among Adult Critically Ill Patients with Suspected Infection
Introduction
Sepsis, a life-threatening condition caused by the body’s dysregulated response to infection, has been a significant challenge in critical care medicine. Over the years, the definitions of sepsis have evolved to improve diagnostic accuracy and patient outcomes. The first consensus definitions, known as Sepsis-1, were established in 1991 by the American College of Chest Physicians and the Society of Critical Care Medicine. These definitions characterized sepsis as systemic inflammatory response syndrome (SIRS) resulting from infection. Despite widespread adoption, the Sepsis-1 definitions were criticized for their lack of specificity and inability to accurately stratify mortality risk.
In 2016, the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) were introduced, marking a significant shift in the clinical criteria for sepsis. Sepsis-3 defines sepsis as life-threatening organ dysfunction caused by a dysregulated host response to infection, using the Sequential Organ Failure Assessment (SOFA) score as a diagnostic tool. Septic shock is identified by the need for vasopressors to maintain a mean arterial pressure (MAP) above 65 mmHg and a serum lactate level greater than 2 mmol/L despite adequate fluid resuscitation. While Sepsis-3 has been validated in large databases, its diagnostic value remains controversial, particularly in critically ill patients with suspected infection.
This study aims to assess the predictive accuracy of Sepsis-3 definitions compared to Sepsis-1 definitions in stratifying mortality among adult critically ill patients with suspected infection. By evaluating the performance of these definitions in a multicenter, prospective cohort study, we seek to provide insights into their clinical utility and potential impact on patient outcomes.
Methods
The study was conducted as a multicenter, prospective cohort study across five Intensive Care Units (ICUs) in four teaching hospitals in Jiangsu Province, China. The study period spanned from November 10, 2017, to October 10, 2018. Ethical approval was obtained from the Institutional Review Board of Northern Jiangsu People’s Hospital, and written informed consent was secured from each participant or their guardian prior to enrollment.
Patients admitted to the ICUs with suspected infection were eligible for inclusion. Suspected infection was defined based on clinical presentation, radiological or laboratory examination, and subsequent administration of antimicrobials. Exclusion criteria included patients under 18 years of age, those with a stay in the ICU of less than 24 hours, missing information, and repeated admissions.
The study population was stratified according to both Sepsis-1 and Sepsis-3 definitions. Sepsis-1 defined sepsis as documented or suspected infection with two or more SIRS criteria, while septic shock was defined as sepsis-induced hypotension requiring vasopressors despite adequate fluid resuscitation. Sepsis-3 defined sepsis as an acute change in the total SOFA score of two or more points resulting from infection, with septic shock identified by the need for vasopressors and elevated lactate levels.
Data collection included demographic information, vital signs, APACHE II scores, SOFA scores, infection sources, ICU length of stay, organ support measures, and initial lactate levels. The primary endpoint was all-cause 30-day mortality, assessed through telephone interviews 30 days after enrollment.
Statistical analysis was performed using SPSS version 16.0. Continuous variables were presented as mean ± standard deviation (SD) and compared using independent sample t-tests. Categorical variables were compared using Chi-square or Fisher exact tests. Logistic regression analysis was conducted to evaluate the predictive accuracy of Sepsis-1 and Sepsis-3 definitions, followed by measurement of the area under the receiver operating characteristic curve (AUROC) for 30-day mortality rates. AUROCs were compared using the DeLong test, with a P-value of less than 0.05 considered statistically significant.
Results
Of the 993 ICU patients with suspected infection during the study period, 244 were excluded, leaving 749 patients enrolled in the study. The mean age of the enrolled patients was 63.4 years, with 74.8% being male. The mean APACHE II score was 21.4, and the most common infection sources were respiratory infections (55.8%), followed by urinary tract infections (20.9%) and abdominal infections (8.7%).
According to the Sepsis-1 definitions, 644 patients (85.9%) were diagnosed with sepsis, of whom 362 (48.3%) had septic shock. In contrast, the Sepsis-3 definitions identified 483 patients (64.5%) with sepsis, including 299 (39.9%) with septic shock. Among the 749 patients, 170 (22.6%) met only the Sepsis-1 definitions, while 9 (1.2%) met only the Sepsis-3 definitions. A total of 474 patients (63.3%) met both definitions, and 96 (12.8%) met neither.
The 30-day mortality rate for the entire cohort was 29.1%. Patients who met both Sepsis-1 and Sepsis-3 definitions had a significantly higher mortality rate (38.2%) compared to those who met only the Sepsis-1 definitions (15.3%) or neither criteria (9.3%). Sepsis-3 definitions stratified mortality more effectively, with sepsis and septic shock patients having a higher 30-day mortality rate (41.8%) compared to Sepsis-1 definitions (31.8%).
Logistic regression analysis revealed that sepsis and septic shock, as defined by both Sepsis-1 and Sepsis-3, were independent risk factors for 30-day mortality. The AUROC for the Sepsis-3 model (0.746) was significantly higher than that of the Sepsis-1 model (0.620), indicating superior predictive accuracy. Additionally, the SOFA score (AUROC: 0.828) outperformed both the SIRS score (AUROC: 0.609) and the qSOFA score (AUROC: 0.694) in predicting mortality.
Discussion
This study demonstrates that the Sepsis-3 definitions are more accurate than the Sepsis-1 definitions in stratifying mortality among adult critically ill patients with suspected infection. The findings support the use of Sepsis-3 criteria to identify high-risk patients and guide clinical decision-making.
The Sepsis-3 definitions identified a smaller proportion of patients with sepsis compared to Sepsis-1, but with higher mortality rates. This suggests that Sepsis-3 criteria are more specific in identifying critically ill patients with a higher risk of death. The inclusion of the SOFA score in the Sepsis-3 definitions provides a more comprehensive assessment of organ dysfunction, which is a key determinant of mortality in sepsis patients.
The superior predictive accuracy of the Sepsis-3 model, as evidenced by the higher AUROC, highlights the clinical utility of these definitions. The SOFA score, in particular, was found to be an excellent tool for predicting mortality, outperforming both the SIRS and qSOFA scores. This reinforces the importance of assessing organ dysfunction in the diagnosis and prognosis of sepsis.
Despite these strengths, the study has limitations. The 30-day follow-up period may not capture long-term outcomes, and the findings may not be generalizable to non-ICU settings. Additionally, the study population was limited to patients in China, and further research is needed to validate these findings in diverse populations.
Conclusion
In conclusion, the Sepsis-3 definitions are more accurate than the Sepsis-1 definitions in stratifying mortality among adult critically ill patients with suspected infection. The use of the SOFA score in Sepsis-3 provides a more comprehensive assessment of organ dysfunction, leading to better identification of high-risk patients. These findings support the adoption of Sepsis-3 criteria in clinical practice to improve patient outcomes and guide therapeutic interventions. Future research should focus on validating these findings in broader populations and exploring long-term outcomes.
doi.org/10.1097/CM9.0000000000000166
Was this helpful?
0 / 0