MORTALITY ATTRIBUTION IN CHILDREN WITH COMMUNITY ACQUIRED SEPSIS AND IMPLICATIONS FOR RISK STRATIFICATION: A MULTI-CENTRE COHORT STUDY

Background: Pediatric sepsis is a leading cause of childhood mortality worldwide. Risk stratification tools are widely used to predict mortality in children with suspected sepsis, assuming mortality is directly attributable to sepsis. However, this assumption may not be valid, particularly in medically complex children. Objectives: to describe mortality attribution in children with community acquired sepsis and implications for risk stratification using the Phoenix sepsis criteria. Design: this prospective observational study used data from the multi-centre multi-country Sepsis Epidemiology in Australian and New Zealand Emergency Departments (SENTINEL) study, collected from 2021 to 2023. Population: participants were children aged 0 to <18 years with suspected community acquired sepsis admitted to the hospital through 11 emergency departments in Australia and New Zealand. Main outcomes: we describe mortality attribution and the test characteristics of the Phoenix sepsis criteria for predicting sepsis attributable deaths. Results: A total of 6232 patients were eligible and included in the analysis, with a median age of 2.1 years, IQR 0.3 to 7.1 years, in whom 55 died in-hospital within 30 days (0.9%). Phoenix sepsis criteria were met in 306 (4.9%), in whom 32 (10.4%) died in-hospital within 30 days. Thirty day in-hospital mortality was attributable to sepsis in 26/55 patients overall (47.3%). Sepsis attributable deaths were not associated with patient age, sex, ethnicity, or presence of comorbidities. Death due to a known chronic condition was the most observed alternative cause of death (13/29; 44.8%). Of the 32 children who met the Phoenix sepsis criteria and died in hospital within 30 days, 19 (59.4%) deaths were attributable to sepsis. The worst Phoenix sepsis score calculated over the first 24 hours of hospitalisation had an area under the receiver operating characteristic curve (AUROC) of 0.84 (95% CI 0.76-0.93) for predicting 30-day mortality, with an area under the precision–recall curve (AUPRC) of 0.048 (95% CI 0.025-0.076). Conclusion: In-hospital mortality in children with suspected community acquired sepsis is not attributable to sepsis in more than half of cases. Known chronic disease was the most observed alternative cause of death. This challenges the validity of using all-cause mortality as an endpoint in sepsis research and underscores the need for sepsis-specific mortality attribution and early risk identification strategies. The Phoenix Sepsis Criteria had high discriminative ability for predicting sepsis attributable mortality; however, the low AUPRC limits their ability to precisely identify patients who will die from sepsis.

Background: Pediatric sepsis is a leading cause of childhood mortality worldwide. Risk stratification tools are widely used to predict mortality in children with suspected sepsis, assuming mortality is directly attributable to sepsis. However, this assumption may not be valid, particularly in medically complex children. Objectives: to describe mortality attribution in children with community acquired sepsis and implications for risk stratification using the Phoenix sepsis criteria. Design: this prospective observational study used data from the multi-centre multi-country Sepsis Epidemiology in Australian and New Zealand Emergency Departments (SENTINEL) study, collected from 2021 to 2023. Population: participants were children aged 0 to <18 years with suspected community acquired sepsis admitted to the hospital through 11 emergency departments in Australia and New Zealand. Main outcomes: we describe mortality attribution and the test characteristics of the Phoenix sepsis criteria for predicting sepsis attributable deaths. Results: A total of 6232 patients were eligible and included in the analysis, with a median age of 2.1 years, IQR 0.3 to 7.1 years, in whom 55 died in-hospital within 30 days (0.9%). Phoenix sepsis criteria were met in 306 (4.9%), in whom 32 (10.4%) died in-hospital within 30 days. Thirty day in-hospital mortality was attributable to sepsis in 26/55 patients overall (47.3%). Sepsis attributable deaths were not associated with patient age, sex, ethnicity, or presence of comorbidities. Death due to a known chronic condition was the most observed alternative cause of death (13/29; 44.8%). Of the 32 children who met the Phoenix sepsis criteria and died in hospital within 30 days, 19 (59.4%) deaths were attributable to sepsis. The worst Phoenix sepsis score calculated over the first 24 hours of hospitalisation had an area under the receiver operating characteristic curve (AUROC) of 0.84 (95% CI 0.76-0.93) for predicting 30-day mortality, with an area under the precision–recall curve (AUPRC) of 0.48 (95% CI 0.25-0.76). Conclusion: In-hospital mortality in children with suspected community acquired sepsis is not attributable to sepsis in more than half of cases. Known chronic disease was the most observed alternative cause of death. This challenges the validity of using all-cause mortality as an endpoint in sepsis research and underscores the need for sepsis-specific mortality attribution and early risk identification strategies. The Phoenix Sepsis Criteria had high discriminative ability for predicting sepsis attributable mortality; however, the low AUPRC limits their ability to precisely identify patients who will die from sepsis.