Identifying subphenotypes of patients undergoing post‐operative delirium assessment

INTRODUCTION: Delirium has heterogeneous etiologies and clinical presentations and is often associated with poor outcomes. Its pathophysiological mechanisms remain largely hypothetical and without targeted pharmacological treatment. This work investigates subphenotypes of patients undergoing delirium assessment based on clinical features and fluid biomarkers. METHODS: We performed latent class analysis of an observational cohort of older adults undergoing elective surgery. RESULTS: Two classes were identified, both containing individuals experiencing delirium symptoms, with a higher number in Class 1 (p < 0.001). Class 1 were older, less educated, and had more depression (p < 0.001). They performed worse in all pre-operative cognitive assessments (p < 0.001) and had more markers of central nervous system damage: cerebrospinal fluid glial fibrillary acidic protein, neurofilament light chain, and soluble triggering receptor expressed on myeloid cells 2 (p < 0.001); plasma phosphorylated tau (p = 0.024); and amyloid beta 42/40 ratio (p < 0.001). Class 2 experienced more pain (p = 0.006) and received more morphine equivalents (p = 0.018). DISCUSSION: Delirium and neighboring phenotypes should be investigated thoroughly in the newly dawning era of precision medicine, to establish novel treatments. HIGHLIGHTS: Latent class analysis identified two subphenotypes of patients. Both groups contained patients with delirium or its individual symptoms. Groups differed by age, education, depression, independent living, and pain levels. Groups differed by pre-operative and post-operative cognition. Groups differed by biomarker levels of neurodegeneration and neuronal injury.