Diagnosis and Treatment of Paediatric Septic Shock: A Critical Narrative Review of Recent Advances, Persistent Uncertainties and the Translation Gap

Abstract

Paediatric septic shock remains one of the largest contributors to childhood mortality worldwide, yet the field has changed more in how the condition is defined and detected than in how it is treated. The publication of the Phoenix sepsis criteria in 2024 replaced two decades of reliance on the systemic inflammatory response syndrome with a data-derived, organ-dysfunction-based construct, and the 2026 update of the international paediatric sepsis guidelines consolidated a management framework that nonetheless rests overwhelmingly on low-certainty evidence. This critical narrative review examines what has genuinely advanced, what remains contested, and why diagnostic refinement has not yet translated into measurable therapeutic gain. Literature was identified through structured searching of biomedical databases, scholarly indexes and citation tracking, with critical appraisal directed at study design, setting, outcome selection and generalisability rather than citation frequency. Four arguments are developed. First, the Phoenix criteria improve prognostic discrimination and cross-setting comparability, but they were derived to identify children already experiencing life-threatening organ dysfunction and perform substantially less well when repurposed as frontline screening instruments, particularly in intensive care populations where specificity falls sharply. Second, machine-learning and electronic screening systems demonstrate strong discrimination in retrospective evaluation yet have almost never been tested as interventions, leaving their clinical value unestablished. Third, host-response biomarkers, transcriptomic endotypes and latent-profile phenotypes have produced biologically coherent subgroups that remain unvalidated as treatment-selection tools, since no paediatric trial has yet randomised participants according to biological subclass. Fourth, the therapeutic evidence base has produced informative neutral and negative results, including a large pragmatic crystalloid trial showing no advantage of balanced fluid over saline, while the two adequately powered corticosteroid trials remain unreported. Across all domains, evidence is concentrated in high-resource settings, whereas mortality is concentrated elsewhere. Closing this gap requires trials designed around biological subclassification, functional outcomes and context-adapted delivery rather than further refinement of diagnostic labels alone.