Risk factors for early poor outcomes in intracerebral hemorrhage with mild-to-moderate volume: A retrospective cohort study

Compared to intracerebral hemorrhage (ICH) with large hemorrhage volume, early poor outcomes in mild-to-moderate hemorrhage volume ICH (MMV-ICH) remains poorly understood. We aimed to analyze risk factors in patients with MMV-ICH and establish a nomogram model to predict early poor outcomes in such cases. We conducted a retrospective analysis of data from 681 patients with MMV-ICH treated at our hospital (May 2019–May 2024). Baseline demographic information, clinical characteristics, and outcomes were collected. Early poor outcomes were defined as in-hospital death or discharge against medical advice. Least absolute shrinkage and selection operator (LASSO) and stepwise logistic regression analyses were used to screen risk factors for early poor outcomes and construct the prediction model, respectively. The model was visualized using a nomogram; its predictive performance was compared with the original ICH score. The nomogram prediction model was internally validated using 10-fold cross-validation combined with Bootstrap resampling. Of 681 patients, 71 (10.4%) experienced early poor outcomes during hospitalization. Univariate analysis revealed 16 risk factors. LASSO and stepwise logistic regression analyses identified four independent predictors: admission Glasgow Coma Scale (GCS) score, hematoma volume, hematoma expansion, and ventricular enlargement. The nomogram based on these factors demonstrated good calibration (mean absolute error 0.011) and superior predictive performance compared with the ICH score (area under the receiver operating characteristic curve: 0.895 vs. 0.801, p<0.001). Decision curve analysis further demonstrated that the model had greater net benefit than the ICH score. Internal validation confirmed the model’s stability and generalizability (10-fold cross-validation mean: 0.8893; bootstrap-corrected C-index: 0.8899). Patients with MMV-ICH with lower admission GCS, larger hematoma, ventricular enlargement, and hematoma expansion post-admission exhibit higher risk of early poor outcomes. The nomogram incorporating these factors potentially identifies early poor outcomes during hospitalization in this population, aiding neurologists in early clinical-decision making.