To explore associations between acute and chronic neural substrates of delirium in patients with intracerebral hemorrhage (ICH).

Delirium occurs frequently in patients hospitalized with both neurologic and non-neurologic diseases. In non-neurologic patients, brain volume, white matter hyperintensity volume (WMHV), and other neuroimaging surrogates of brain reserve have been implicated in delirium pathophysiology. However, it is unclear if pre-existing neural substrates play a similar role in the context of an acute neurologic injury.

Using data from a single-center registry, we identified consecutive non-comatose patients with acute non-traumatic ICH and available magnetic resonance imaging (MRI). Delirium was diagnosed according to DSM-5 criteria. T1 and T2-FLAIR MRIs were processed and analyzed using semi-automated software, with volumetric measurements of each acute hematoma along with WMHV and segmented brain volumes from the contralateral hemisphere. We tested associations between WMHV and incident delirium using multivariable regression models, then determined the predictive accuracy of these neuroimaging models via area under the curve (AUC) analysis.

Of 139 ICH patients, 58 (42%) experienced delirium (mean [SD] age 67.3 [17.3], 53% male). Compared to non-delirious patients, patients with delirium had greater ICH volume (mean [SD] 29.7 [30.3] vs. 11.4 [16.3] cm³, p<0.001) and WMHV (mean [SD] 15.0 [14.3] vs. 10.9 [11.2] cm³, p=0.06) on univariate analysis. After adjusting for ICH features, WMHV was significantly associated with delirium (OR 1.44 per 10 cm³, 95% CI 1.06-1.95), and this association was strengthened (OR 1.78 per 10 cm³, 95% CI 1.20-1.67) after further adjustment for segmented brain volume in patients with high-resolution scans. Neuroimaging-based models predicted delirium with high accuracy (AUC 0.79), especially in patients with smaller ICH (AUC 0.89).

Pre-existing neural substrates are independently associated with delirium in patients with acute ICH, and neuroimaging biomarkers may have utility in predicting delirium occurrence.