New PSS+ pediatric sepsis score significantly outperforms international standards in predicting hospital mortality for critically ill children

A study in the Chinese Medical Journal reveals the PSS+ score outperforms international standards in predicting mortality for children with suspected infection.

By: AXL Media

Published: Mar 31, 2026, 12:24 PM EDT

Source: Information for this report was sourced from Chinese Medical Journals Publishing House Co., Ltd.

The Challenge of Universal Sepsis Scoring

The Phoenix Sepsis Score (PSS) was recently introduced by an international task force to provide a global standard for identifying organ dysfunction in children with suspected infections. However, the effectiveness of such universal tools often fluctuates when applied to diverse healthcare systems with varying patient demographics and resource levels. A new multicenter study involving five hospitals across four Chinese provinces suggests that while the PSS is a vital framework for sepsis identification, its ability to predict actual mortality outcomes requires local refinement to be truly effective in a pediatric intensive care unit (ICU) setting.

Evaluating the Original Phoenix Framework

To test the international standard, researchers assembled a retrospective cohort of over 9,200 pediatric ICU encounters recorded between 2012 and 2023. The study specifically looked at patients with suspected infections, a group that faced an in-hospital mortality rate of 13.4%. Upon evaluation, the original PSS demonstrated only moderate discriminative performance, with accuracy scores hovering around 0.60. These results indicate that the standard PSS may have a limited ability to distinguish between high-risk and low-risk patients in this specific population, highlighting a gap in its prognostic utility outside of the Western environments where it was first derived.

The Development of the PSS+ Modification

In response to these limitations, the research team sought to enhance the existing framework without sacrificing its ease of use at the bedside. They utilized advanced machine learning, specifically extreme gradient boosting (XGBoost), to analyze candidate predictors of mortality. However, the researchers did not rely solely on algorithmic output; they only incorporated variables that were clinically meaningful and routinely available to frontline doctors. This process led to the creation of the PSS+, a modified score that blends the original organ dysfunction metrics with specific demographic factors, pre-existing comorbidities, and real-time vital signs.