Investigating subphenotypes of delirium using existing research data

Univariate analyses explored inflammatory and neurodegenerative biomarkers from blood and CSF. A combination of biomarkers, individual symptoms and relevant covariates were investigated in latent class analyses (LCA) of populations at risk for delirium.<br><br>Results<br>The 78 delirium symptoms described by systematic review and DSI recommendations support the generation of large datasets of standardised delirium symptoms, for validation and refinement of delirium and its related phenotypes.

Associations of CSF A42, sPDGFR, GFAP, NfL, and sTREM2, and plasma pTau181, IL-6, IL-8, GFAP, NfL, and TNF- varied with delirium, individual symptoms, resilience, and coma. LCA identified two subphenotypes in a surgical population at risk for delirium, which were associated with differential outcomes in living situation, cognition, and death at 8-year follow up. Four delirium subphenotypes were identified in a critical care population of delirious patients, which were associated with varying number of days in coma during ICU stay. <br><br>Discussion<br>This thesis provides evidence to support parsing the delirium syndrome whilst ensuring standardisation and operationalisation of its individual features.

Individual symptoms of inattention and altered level of consciousness displayed associations with raised markers of neurological damage, BBB dysfunction and inflammation, regardless of delirium status. Rigorous clinical observation and biospecimen collection in big-data cohorts are required to continue investigation of delirium phenotypes. The previously unobserved latent classes must be validated to provide insight into potential targets for treatment. <br><br><i>Thesis is embargoed until 31 July 2027</i>.<br><br>