Serum biomarkers of delirium in critical illness: a systematic review of mechanistic and diagnostic evidence

Abstract Delirium is a frequent and serious complication of critical illness, yet its pathophysiological mechanisms remain incompletely understood. Serum biomarkers offer a potential avenue for improved diagnosis, risk stratification, and mechanistic insight. This systematic review synthesises evidence from 28 studies evaluating 54 serum biomarkers in relation to delirium among critically ill adult patients. Biomarkers were categorised by mechanistic pathway, including central nervous system (CNS) injury, immune activation, hormonal dysregulation, neurotransmission, coagulation, and amino acid metabolism. Among CNS injury markers, S100β and neurofilament light chain (NfL) demonstrated the most consistent associations with delirium presence and severity, supporting a role for astrocytic and axonal injury in delirium pathogenesis. Inflammatory markers such as interleukin-6 (IL-6), C-reactive protein (CRP), and tumour necrosis factor-alpha (TNF-α) were frequently studied but showed variable associations, reflecting the complex and non-specific nature of systemic inflammation. Hormonal biomarkers, including cortisol and prolactin, showed preliminary promise, while neurotransmitter-related biomarkers yielded inconsistent results, challenging canonical hypotheses. A major limitation in the literature was the lack of standardisation in delirium assessment, sampling timelines, and adjustment for confounding variables. Only a minority of studies incorporated temporal profiling or longitudinal outcomes, and replication across cohorts was limited. Heterogeneity in ICU populations further reduced generalisability. This review proposes a new conceptual framework of mechanistic endotyping, integrating multimodal biomarker profiling with clinical phenotyping to define biologically distinct subtypes of delirium. Such an approach may support precision medicine strategies by aligning therapeutic interventions with underlying pathophysiology. Future biomarker research should prioritise longitudinal sampling, harmonised protocols, and integration with EEG, imaging, and cognitive outcomes. Despite early promise, serum biomarkers for ICU delirium remain investigational and require further validation before clinical application.