To explore associations between neuro-inflammation and delirium in patients with hemorrhagic stroke.

Neuro-inflammation has been implicated as a potential mechanism in delirium pathogenesis. While there is limited direct evidence in non-neurologic populations, patients with neurosurgical drains may provide insights due to frequent measurements of cerebrospinal fluid (CSF) protein, a common inflammatory marker.

We performed a retrospective cohort study on consecutive patients from 2018-2019 with intracerebral hemorrhage (ICH) or subarachnoid hemorrhage (SAH) and available CSF data. CSF protein measurements were transformed using rank-based inverse normal transformation due to non-normality. Delirium was diagnosed according to DSM-5 criteria and categorized as persistent or resolved by hospital discharge. We compared CSF protein measurements and their trajectory based on delirium status on day of measurement and at hospital discharge. Analyses accounted for repeated measurements by patient-level using linear mixed-effects models incorporating age, stroke severity, and measurement day.

We identified 60 patients (mean age 56.5 [SD 15.5], 57% male) with 116 CSF measurements; 55% had SAH and 45% had ICH (median [IQR] Hunt-Hess 3 [3-4], ICH score 2 [2-3]). Delirium was present on 74% of all measurement days (86/116), while 33/57 patients with delirium had resolution by hospital discharge. In univariate analyses, CSF protein was higher in patients with delirium vs. without delirium on day of measurement (median 145 [77-282] vs. 117 [76-214] mg/microliter, p=0.29), as well as in patients whose delirium was persistent vs. resolved by discharge (median 168 [91-322] vs. 125 [80-246] mg/microliter, p=0.18). In our mixed-effects models, we were underpowered to detect a significant difference in CSF protein based on delirium status on the day of measurement (p=0.17) and at discharge (p=0.08).

CSF protein may have potential utility as a biomarker for delirium in hemorrhagic stroke, and this novel preliminary study provides estimates of effect size and trajectory.