# Hemoglobin and Perihematomal Edema After Intracerebral Hemorrhage: A Post Hoc Analysis of the i-DEF Trial

## TL;DR

This study explores the link between blood hemoglobin levels and brain swelling after intracerebral hemorrhage, suggesting a potential connection that needs further research.

## Contribution

The study identifies a potential association between baseline hemoglobin levels and perihematomal edema after intracerebral hemorrhage.

## Key findings

- Hemoglobin levels were weakly inversely correlated with baseline and follow-up edema extension distance.
- Anemia was present in 15% of participants, but no strong correlation was found between hemoglobin and interscan edema change.
- Absolute and relative perihematomal edema showed no clear correlation with hemoglobin levels.

## Abstract

Anemia is common after intracerebral hemorrhage (ICH). It has been attributed to inflammation and is associated with poor outcomes. We investigated whether this could be related to the effects of hemoglobin (Hb) on perihematomal edema (PHE).

We performed an exploratory post hoc analysis of the Intracerebral Hemorrhage Deferoxamine (i-DEF) randomized controlled trial. We included participants with primary supratentorial ICH, available baseline Hb levels, and computed tomography scans at baseline and follow-up after 72–96 h. We investigated the association of Hb and anemia (as continuous and dichotomous exposures, respectively) with edema extension distance (EED) as the main continuous outcome at baseline and follow-up and as its interscan change using Spearman correlation and unadjusted and adjusted linear models. We examined absolute and relative PHE in ancillary analyses.

We analyzed data from 276 of 293 (94%) i-DEF participants. The median age was 61 (interquartile range [IQR] 52–70) years, and 39% of participants were female. The median Hb level was 14.1 (IQR 13–15.2) g/dL, and 41 participants (15%) were anemic. The median EED was 4.4 (IQR 3.5–5.3) mm at baseline and 6.4 (IQR 5.3–7.3) mm at follow-up. Hb was weakly inversely correlated with baseline (ρ =  − 0.12, p = 0.05) and follow-up EED (ρ =  − 0.11, p = 0.07) but not with interscan EED change (ρ =  − 0.01, p = 0.89). Linear models showed similar relationships of Hb with baseline and particularly follow-up EED but not with EED change. In ancillary analyses, absolute and relative PHE showed no clear correlation with Hb but maintained similar relationships with Hb in linear models as in the main analysis.

We identified signals for an association of baseline Hb with PHE after ICH. These findings may warrant further exploration in larger cohorts.

ClinicalTrials.gov identifier: NCT02175225.

The online version contains supplementary material available at 10.1007/s12028-025-02284-3.