External validation of the Phoenix Sepsis Score in a paediatric intensive care unit in Saudi Arabia

Abstract

Abstract We externally validated the Phoenix Sepsis Score (PSS) for predicting n-hospital mortality in a Paediatric Intensive Care Unit (PICU) setting in Saudi Arabia and compared its performance with those of the International Paediatric Sepsis Consensus Conference (IPSCC) criteria and Paediatric Logistic Organ Dysfunction-2 (PELOD-2) score. This retrospective cohort study of a prospectively maintained paediatric sepsis registry included 0–14-year-old PICU-admitted children with suspected sepsis between 2015 and 2023. Of 431 admissions, 281 (65.2%) met the PSS criteria (PSS ≥ 2), of whom 197/281 patients (70.1%) met the septic shock criteria. The in-hospital mortality rates were 21.8% for the entire cohort, 12% for infection, and 27% for sepsis. A PSS ≥ 2 yielded a sensitivity of 80.9% (95% confidence interval [CI]: 71.7–87.5) and positive predictive value of 27.0% (95%CI: 22.2–32.5). The area under the precision-recall curve for the PSS was 0.45 (95% CI: 0.35–0.56), which was higher than that for the IPSCC criteria (0.38, 95%CI: 0.29–0.48) and comparable to that for the PELOD-2 score (0.48, 95%CI: 0.37–0.58). PSS discriminative performance, assessed by the area under the receiver-operating characteristic curve, was 0.67 (95% CI: 0.61–0.74), similar to that of the IPSCC criteria (0.67, 95%CI: 0.60–0.73) and PELOD-2 score (0.72, 95%CI: 0.65–0.77). The PSS demonstrated favourable precision-recall performance and fair discrimination for in-hospital mortality in this critical care cohort. Its precision-recall performance was superior to that of the IPSCC criteria and comparable to that of the PELOD-2 score. Notably, the PSS relies on fewer variables across four organ systems, which may enhance its practicality for risk stratification in comparable high-acuity critically ill children.