Phoenix Sepsis Criteria Outperform Procalcitonin in Predicting Child Sepsis Deaths

Sepsis remains the leading cause of childhood mortality worldwide, claiming more young lives than any other single condition, and clinicians have long struggled to identify which children with the syndrome are most likely to die. A new retrospective study from Indonesia now offers a striking answer: a simple bedside scoring system, the Phoenix Sepsis Criteria, predicts death in children with sepsis far more accurately than procalcitonin, the expensive blood biomarker many hospitals rely on. The findings, published in BMC Pediatrics by researchers at Universitas Padjadjaran and Dr. Hasan Sadikin General Hospital in Bandung, suggest that the free clinical criteria could transform how pediatric intensive care units in resource-limited settings triage their sickest patients.

The study analyzed 192 pediatric patients with sepsis, aged one month to 18 years, admitted to the Pediatric Intensive Care Unit of Dr. Hasan Sadikin General Hospital beginning in August 2023. The researchers collected data on patient characteristics, Phoenix sepsis scores, and procalcitonin levels, then compared how well each tool discriminated between children who survived and those who died within 28 days. The mortality rate in the cohort was a sobering 54.7 percent, a figure that underscores the severity of sepsis presentations reaching this tertiary referral center and the high stakes of accurate mortality prediction.

The Phoenix Sepsis Criteria, introduced in 2024 as part of a global redefinition of sepsis and septic shock, evaluate four organ systems using routinely collected clinical data: the respiratory system, the cardiovascular system, the coagulation system, and the neurological system. Each system is scored from zero upward based on the degree of dysfunction, and the total score places children on a spectrum from no organ dysfunction to septic shock. Because the criteria depend entirely on bedside measurements such as oxygen saturation, blood pressure, platelet count, and consciousness level, they require no laboratory tests beyond those already performed in any seriously ill child, making them essentially free to apply.

Procalcitonin, by contrast, is a peptide precursor of the hormone calcitonin that rises dramatically during bacterial infections and has been widely promoted as a biomarker of sepsis severity. Its blood levels correlate with the intensity of the inflammatory response, and many clinicians use it to gauge infection severity and guide antibiotic decisions. But the new study found that while procalcitonin levels were indeed significantly higher in children who died than in those who survived, with both the Phoenix score and procalcitonin showing p values below 0.001 for this difference, the biomarker’s overall discriminatory power was considerably weaker than the clinical criteria.

The technical comparison centered on the area under the receiver operating characteristic curve, or AUC, a statistical measure of how well a test separates those who experience an outcome from those who do not. An AUC of 0.5 indicates performance no better than chance, while 1.0 indicates perfect discrimination. The Phoenix Sepsis Criteria achieved an AUC of 0.848, placing them in the range of good discriminatory performance, whereas procalcitonin managed only 0.689, a value that falls into the weak-to-moderate range and implies substantial overlap in biomarker levels between survivors and nonsurvivors. In practical terms, the clinical criteria misclassified far fewer children at any given threshold than the blood test did.

The researchers also calculated odds ratios at specific cutoff values to quantify how strongly each tool predicted death. A Phoenix sepsis score greater than 4 was associated with an odds ratio of 9.30, meaning children above this threshold had more than nine times the odds of dying compared with those below it. Procalcitonin levels above 13.42 micrograms per liter carried an odds ratio of only 3.00, a threefold increase in odds that, while statistically meaningful, pales next to the clinical criteria. The cutoff of 13.42 micrograms per liter itself emerged from the study’s statistical analysis of the cohort, illustrating how biomarker thresholds often require local calibration rather than universal application.

Perhaps the most intriguing technical finding came from the linear regression analysis, which revealed a multiplicative relationship between the two measures: for every one-point increase in the Phoenix sepsis score, procalcitonin levels increased approximately threefold. This exponential relationship suggests that the clinical score and the biomarker are tracking related but distinct dimensions of the septic process, with the score capturing the cumulative degree of organ dysfunction and the biomarker reflecting the intensity of the underlying inflammatory response. The fact that the score outperformed the biomarker despite this coupling indicates that organ dysfunction, rather than inflammation alone, is the dominant driver of mortality in pediatric sepsis.

The implications for low- and middle-income countries, where the vast majority of childhood sepsis deaths occur, are considerable. Procalcitonin assays are costly, require laboratory infrastructure, and often take hours to result, whereas the Phoenix criteria can be calculated immediately from data already available at the bedside. In settings like the Bandung PICU, where families may face severe financial constraints and laboratory capacity is limited, a free scoring system that identifies high-risk children more accurately than an expensive blood test could redirect scarce resources toward timely intervention for those who need it most. The authors argue that implementing the Phoenix criteria in clinical practice may improve timely and effective identification of high-risk patients.

The study does carry limitations that temper its conclusions. As a retrospective analysis of a single-center cohort of 192 patients, its findings require validation in larger, prospective, multicenter studies before the specific cutoff values can be generalized. The extremely high mortality rate of 54.7 percent reflects the case mix of a tertiary referral hospital in Indonesia and may not represent the mortality seen in community hospitals or in high-income settings, where earlier presentation and different standards of care could shift both mortality and predictive performance. The authors themselves note that the Phoenix criteria showed stronger statistical performance for predicting mortality in their cohort, framing the result as evidence for implementation rather than a final verdict.

Nevertheless, the study adds to a growing body of evidence that the Phoenix Sepsis Criteria, developed through a Delphi consensus process involving hundreds of global experts, are living up to their promise as a practical, accurate tool for pediatric sepsis. For the world’s pediatric intensive care units, particularly those serving the populations where sepsis kills the most children, the message is clear: the most powerful predictor of death may already be written in the clinical chart, waiting to be summed, at no cost, before any blood test returns from the laboratory.

Subject of Research: Comparative performance of the phoenix sepsis criteria and procalcitonin for mortality prediction in pediatric sepsis

Article Title: Comparative performance of the phoenix sepsis criteria and procalcitonin for mortality prediction in pediatric sepsis

Article References: Sirait, A. E., Hakim, D. D. L., Rachmadi, D., Peryoga, S. U., Adrizain, R., & Rahayuningsih, S. E. (2026). Comparative performance of the phoenix sepsis criteria and procalcitonin for mortality prediction in pediatric sepsis. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07409-x

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07409-x

Keywords: Comparative, performance, phoenix, sepsis, criteria, procalcitonin, mortality, prediction, pediatric, scientific research