Phoenix Sepsis Score: Predictive Accuracy for Clinical Deterioration in Pediatric Critical Care
Hadzhieva-Hristova, Adriana; Darina, Krumova; Halilova, Mergyul; Tsochev, Kaloyan; Karamfilova, Teodora; Deyanova, Yana; Zlateva, Tanya; Bazdarska, Yuliya; Stoeva, Temenuga; Iotova, Violeta
- Publisher
- Zenodo
- Language
- en
Abstract
Background/Objectives: Early identification of pediatric sepsis complications in intensive careis challenging and requires improved diagnostic tools. This study aimed to compare the PhoenixSepsis Score (PSS), pSOFA, PELOD-2, and PRISM III in assessing clinical complexity inchildren with septic and critical conditions in the PICU and to identify the most suitable scale forthis patient cohort. Methods: Data were collected prospectively from 53 children between June2022 and January 2024. Patients were categorized into septic (n=42) and non-infectious SIRS(n=11) and further classified by outcome—with/without complications (n=23/30). The predictiveaccuracy of the scoring systems was evaluated by discrimination and calibration and byrecalibration for the PSS for improved performance. Predictive power was assessed using theArea Under the Precision-Recall Curve (AUPRC) to evaluate the ability to identify patients withcomplications. Results: Respiratory (18.8%) and neurological complications (9.4%) were themost common adverse events. Clinical deterioration was observed in 43.4% of cases, includingone fatality. Patients with complications stayed longer in the PICU (14 ± 10 days). In the patientswith complications, all scoring systems had higher median values. Only PSS showed asignificant difference (p=0.0023). PSS demonstrated the highest overall predictive accuracy(76.2%) outperforming PRISM III (62.3%) and PELOD-2 (58.5%). The pSOFA scale showedhigh accuracy (88.0%) in identifying patients without complications. Phoenix Sepsis Score иPRISM III demonstrated the highest precision in identifying patients with complications(AUPRC=0.62). The strongest association was between chronic disease (hazard ratioExp(B)=1.718) and deteriorations, while mechanical ventilation suggested a reduced risk ofcomplications (Exp(B) = 0.509). Conclusions: PSS showed superior predictive accuracy(76.2%) and discriminative capacity (AUPRC=0.62) for deteriorations in pediatric patients withsuspected infection and proved adaptable for further validation in larger populations.
Full text
Phoenix Sepsis Score: Predictive Accuracy for Clinical Deterioration in Pediatric Critical Care Paper Number: 28 Adriana Hadzhieva-Hristova¹˒³, Darina Krumova¹˒³, Mergyul Halilova¹˒³, Kaloyan Tsotchev¹˒³, Teodora Karamfilova¹˒³, Yana Deyanova¹˒³, Tanya Zlateva¹˒³, Yuliya Bazdarska¹˒³, Temenuga Stoeva²˒⁴, Violeta Iotova¹˒³ ¹ Department of Pediatrics, Medical University of Varna, Bulgaria ² Department of Microbiology and Virology, Medical University of Varna, Bulgaria ³ First Clinic and PICU, University Hospital St. Marina, Varna, Bulgaria ⁴ Microbiology Laboratory, University Hospital St. Marina, Varna, Bulgaria Created with BioRender Poster Builder INTRODUCTION Early recognition of complications in pediatric sepsis is a major challenge in intensive care. Reliable scoring systems are essential to improve diagnostic accuracy and guide risk stratification. This study aimed to compare the Phoenix Sepsis Score (PSS), pSOFA, PELOD-2, and PRISM III in predicting complications and assessing clinical severity in children admitted to the PICU. METHODS ●Period: June 2022 – January 2024 ●Cohort: 53 children (42 with sepsis; 11 with noninfectious SIRS) ●Groups: with complications (n=23) vs. without complications (n=30) ●Analysis: discrimination, calibration, recalibration (for PSS), and predictive performance (AUPRC) RESULTS Most common complications: respiratory (18.8%), neurological (9.4%) • Clinical deterioration: 43.4% (incl. 1 fatality) • PICU stay: 14 ± 10 days (with complications) • Significant difference observed only with PSS (p=0.0023) Suppl. Table Full-text article ●Predictive accuracy – PSS: 76.2% – PRISM III: 62.3% – PELOD-2: 58.5% – pSOFA: 88.0% (no complications) ●AUPRC – PSS & PRISM III: 0.62 ●Risk factors – Chronic disease – highest risk factor (hazard ratio Exp(B)=1.718) – Mechanical ventilation – lower risk (Exp(B) = 0.509) CONCLUSION The Phoenix Sepsis Score demonstrated superior predictive accuracy and discriminative capacity in identifying complications among pediatric patients with suspected infection. These findings support its potential as a reliable instrument for the early detection of clinical deterioration, meriting further validation in larger and more diverse cohorts.
