Risk factors and development of a prediction model for hematoma expansion in elderly patients with spontaneous...

Hematoma expansion (HE) is a major determinant of early neurological deterioration after spontaneous intracerebral hemorrhage (sICH) in older adults. We aimed to identify independent risk factors for HE and to develop a predictive model for elderly patients with sICH.

This retrospective observational study consecutively enrolled patients aged ≥60 years with sICH admitted between May 2023 and May 2024. Baseline non-contrast computed tomography (NCCT) and a follow-up NCCT within 48 h were required. HE was defined as an absolute increase in hematoma volume >6 mL or a relative increase ≥33%, with volumes quantified using the ABC/2 method. Candidate clinical, laboratory, and NCCT variables were screened with univariable logistic regression, then entered into multivariable logistic regression, and the variance inflation factor was assessed. Internal validation was performed using 1,000 bootstrap resamples.

Among 194 patients, 56 (28.9%) developed HE. Independent predictors of HE included age (adjusted odds ratio [aOR] 1.13 per year, 95% confidence interval [CI] 1.02-1.26; p = 0.025), National Institutes of Health Stroke Scale (NIHSS) score (aOR 1.21 per point, 95% CI 1.08-1.35; p < 0.001), neutrophil count (aOR 1.91 per 1 × 10⁹/L, 95% CI 1.34-2.73; p < 0.001), international normalized ratio (INR; aOR 1.70 per 0.1, 95% CI 1.08-2.69; p = 0.023), baseline hematoma volume (aOR 2.01 per 10 mL, 95% CI 1.34-3.02; p < 0.001), black hole sign (aOR 7.34, 95% CI 1.87-28.77; p = 0.004), and irregular shape (aOR 5.57, 95% CI 1.67-18.63; p = 0.005), whereas higher serum sodium was protective (aOR 0.76 per mmol/L, 95% CI 0.64-0.91; p = 0.003). The model showed an apparent area under the receiver operating characteristic (ROC) curve of 0.916 and an optimism-corrected AUC of 0.891.

In elderly patients with sICH, HE might be associated with neurological severity, inflammatory burden, coagulation status, baseline hematoma volume, and adverse NCCT markers. The proposed model could provide strong discriminative performance and might support early risk stratification.