Objective:To explore the differences between the Phoenix sepsis scoring system including Phoenix sepsis score (PSS) and Phoenix-8 organ dysfunction score (hereinafter referred to as Phoenix-8) and the commonly used pediatric sepsis scores in evaluating clinical characteristics and prognostic analysis of pediatric patients with severe sepsis diagnosed under traditional standards, namely the diagnostic criteria from the 2005 International Pediatric Sepsis Consensus Conference.Methods:This study was a retrospective observational study. From December 2020 to March 2023, 202 pediatric patients with severe sepsis meeting the inclusion criteria were admitted to the Children's Hospital of Nanjing Medical University. Based on the sepsis diagnostic criteria outlined in the International Consensus Criteria for Pediatric Sepsis and Septic Shock (2024), the pediatric patients were categorized into a sepsis group and a non-sepsis group. Sepsis group was further subdivided into a death subgroup and a survival subgroup based on the outcomes. The age, hospitalization costs, disease outcome indicators (e.g., mortality rate and incidence of septic shock), major organ (e.g., heart, liver, lungs, and kidneys) damage and their correlations, as well as PSS, Phoenix-8 and commonly used pediatric sepsis scores (e.g., pediatric sequential organ failure assessment (pSOFA), pediatric risk of mortality score Ⅲ (PRISM Ⅲ), pediatric logistic organ dysfunction-2 score (PELOD-2), pediatric multiple organ dysfunction score (P-MODS), pediatric critical illness score (PCIS), and pediatric early warning score (PEWS)) were collected and compared. Receiver operating characteristic (ROC) curve and precision-recall curve were plotted to evaluate the predictive ability of PSS, Phoenix-8, and commonly used pediatric sepsis scores for mortality risk in pediatric patients with severe sepsis under traditional standards. Predictive performance was quantified using the area under the ROC curve (AUROC). Univariate logistic regression analysis was employed to quantify the odds ratios of PSS and Phoenix-8 for predicting mortality risk. Patients with severe sepsis under traditional standards were further stratified into subgroups based on complications and comorbidities, including central nervous system (CNS) diseases, multiple infections, cardiovascular system diseases, shock, and malignancies. The Hosmer-Lemeshow goodness-of-fit test was used to assess calibration of PSS and Phoenix-8, and the DeLong test was used to compare whether there were statistically significant differences in the AUROC of PSS and Phoenix-8 for predicting mortality risk among different subgroups of pediatric patients. Results:Compared with those in non-sepsis group, pediatric patients in sepsis group were significantly older ( Z=-2.92, P<0.05) with higher incidences of septic shock and mortality, hospitalization costs, PRISM Ⅲ, PEWS, pSOFA, PELOD-2, PSS, and Phoenix-8 (with χ2 values of 21.28 and 13.64, respectively, Z values of -1.99, -5.33, -5.10, -8.55, -6.91, -10.98, and -9.93, respectively, P<0.05), and lower PCIS ( Z=-3.34, P<0.05). Compared with those in survival subgroup, hospitalization costs, PSS, Phoenix-8, PRISM Ⅲ, PEWS, pSOFA, PELOD-2, and P-MODS of pediatric patients in death subgroup was significantly higher (with Z values of -2.50, -3.50, -2.47, -5.11, -3.84, -2.94, -3.61, and -3.04, respectively, P<0.05). Compared with those in survival subgroup, the incidences of lung damage and liver damage of pediatric patients in death subgroup were also significantly higher (with χ2 values of 6.20 and 10.94, respectively, P<0.05), and 64.7% (97/150) of patients exhibited two or more concurrent organ damage. For predicting mortality risk in pediatric patients with severe sepsis under traditional standards, the AUROC values for PRISM Ⅲ, PCIS, PEWS, pSOFA, PELOD-2, P-MODS, PSS, and Phoenix-8 were approximately 0.70, with optimal cutoff values of 17.5, 91.0, 5.5, 4.5, 2.5, 4.5, 3.5, and 4.5, respectively; PELOD-2 demonstrated the highest sensitivity (0.83); while PRISM Ⅲ, PSS, and Phoenix-8 showed high specificity (>0.80). Univariate logistic regression analysis showed that for every 1-point increase in the PSS within 24 hours of pediatric intensive care unit admission, the relative risk of mortality increased by 63.7% (with odds ratio of 1.64, 95% confidence interval of 1.34-1.99, P<0.05). Similarly, for every 1-point increase in the Phoenix-8, the relative risk of mortality increased by 37.5% (with odds ratio of 1.38, 95% confidence interval of 1.18-1.60, P<0.05). The AUROC values (around 0.80) of PSS and Phoenix-8 for predicting mortality risk in pediatric patients with severe sepsis combined with CNS diseases, multiple infections, and cardiovascular system diseases were relatively high. In contrast, the AUROC values (0.60-0.80) for predicting mortality risk in pediatric patients with severe sepsis combined with shock or malignant tumors were moderate. All models passed the Hosmer-Lemeshow goodness-of-fit test ( P>0.05). The DeLong test indicated no statistically significant differences in predictive ability between PSS and Phoenix-8 across subgroups of pediatric patients ( P>0.05). Conclusions:PSS and Phoenix-8 exhibited higher specificity than most of the commonly used pediatric sepsis scores in predicting mortality risk under traditional standards. Both scores performed much better in predicting the mortality risk in pediatric patients with severe sepsis combined with CNS diseases, multiple infections, and cardiovascular system diseases.

Objective:To explore the differences between the Phoenix sepsis scoring system including Phoenix sepsis score (PSS) and Phoenix-8 organ dysfunction score (hereinafter referred to as Phoenix-8) and the commonly used pediatric sepsis scores in evaluating clinical characteristics and prognostic analysis of pediatric patients with severe sepsis diagnosed under traditional standards, namely the diagnostic criteria from the 2005 International Pediatric Sepsis Consensus Conference.Methods:This study was a retrospective observational study. From December 2020 to March 2023, 202 pediatric patients with severe sepsis meeting the inclusion criteria were admitted to the Children's Hospital of Nanjing Medical University. Based on the sepsis diagnostic criteria outlined in the International Consensus Criteria for Pediatric Sepsis and Septic Shock (2024), the pediatric patients were categorized into a sepsis group and a non-sepsis group. Sepsis group was further subdivided into a death subgroup and a survival subgroup based on the outcomes. The age, hospitalization costs, disease outcome indicators (e.g., mortality rate and incidence of septic shock), major organ (e.g., heart, liver, lungs, and kidneys) damage and their correlations, as well as PSS, Phoenix-8 and commonly used pediatric sepsis scores (e.g., pediatric sequential organ failure assessment (pSOFA), pediatric risk of mortality score Ⅲ (PRISM Ⅲ), pediatric logistic organ dysfunction-2 score (PELOD-2), pediatric multiple organ dysfunction score (P-MODS), pediatric critical illness score (PCIS), and pediatric early warning score (PEWS)) were collected and compared. Receiver operating characteristic (ROC) curve and precision-recall curve were plotted to evaluate the predictive ability of PSS, Phoenix-8, and commonly used pediatric sepsis scores for mortality risk in pediatric patients with severe sepsis under traditional standards. Predictive performance was quantified using the area under the ROC curve (AUROC). Univariate logistic regression analysis was employed to quantify the odds ratios of PSS and Phoenix-8 for predicting mortality risk. Patients with severe sepsis under traditional standards were further stratified into subgroups based on complications and comorbidities, including central nervous system (CNS) diseases, multiple infections, cardiovascular system diseases, shock, and malignancies. The Hosmer-Lemeshow goodness-of-fit test was used to assess calibration of PSS and Phoenix-8, and the DeLong test was used to compare whether there were statistically significant differences in the AUROC of PSS and Phoenix-8 for predicting mortality risk among different subgroups of pediatric patients. Results:Compared with those in non-sepsis group, pediatric patients in sepsis group were significantly older ( Z=-2.92, P<0.05) with higher incidences of septic shock and mortality, hospitalization costs, PRISM Ⅲ, PEWS, pSOFA, PELOD-2, PSS, and Phoenix-8 (with χ2 values of 21.28 and 13.64, respectively, Z values of -1.99, -5.33, -5.10, -8.55, -6.91, -10.98, and -9.93, respectively, P<0.05), and lower PCIS ( Z=-3.34, P<0.05). Compared with those in survival subgroup, hospitalization costs, PSS, Phoenix-8, PRISM Ⅲ, PEWS, pSOFA, PELOD-2, and P-MODS of pediatric patients in death subgroup was significantly higher (with Z values of -2.50, -3.50, -2.47, -5.11, -3.84, -2.94, -3.61, and -3.04, respectively, P<0.05). Compared with those in survival subgroup, the incidences of lung damage and liver damage of pediatric patients in death subgroup were also significantly higher (with χ2 values of 6.20 and 10.94, respectively, P<0.05), and 64.7% (97/150) of patients exhibited two or more concurrent organ damage. For predicting mortality risk in pediatric patients with severe sepsis under traditional standards, the AUROC values for PRISM Ⅲ, PCIS, PEWS, pSOFA, PELOD-2, P-MODS, PSS, and Phoenix-8 were approximately 0.70, with optimal cutoff values of 17.5, 91.0, 5.5, 4.5, 2.5, 4.5, 3.5, and 4.5, respectively; PELOD-2 demonstrated the highest sensitivity (0.83); while PRISM Ⅲ, PSS, and Phoenix-8 showed high specificity (>0.80). Univariate logistic regression analysis showed that for every 1-point increase in the PSS within 24 hours of pediatric intensive care unit admission, the relative risk of mortality increased by 63.7% (with odds ratio of 1.64, 95% confidence interval of 1.34-1.99, P<0.05). Similarly, for every 1-point increase in the Phoenix-8, the relative risk of mortality increased by 37.5% (with odds ratio of 1.38, 95% confidence interval of 1.18-1.60, P<0.05). The AUROC values (around 0.80) of PSS and Phoenix-8 for predicting mortality risk in pediatric patients with severe sepsis combined with CNS diseases, multiple infections, and cardiovascular system diseases were relatively high. In contrast, the AUROC values (0.60-0.80) for predicting mortality risk in pediatric patients with severe sepsis combined with shock or malignant tumors were moderate. All models passed the Hosmer-Lemeshow goodness-of-fit test ( P>0.05). The DeLong test indicated no statistically significant differences in predictive ability between PSS and Phoenix-8 across subgroups of pediatric patients ( P>0.05). Conclusions:PSS and Phoenix-8 exhibited higher specificity than most of the commonly used pediatric sepsis scores in predicting mortality risk under traditional standards. Both scores performed much better in predicting the mortality risk in pediatric patients with severe sepsis combined with CNS diseases, multiple infections, and cardiovascular system diseases.