Predictive value of the homocysteine-to-albumin ratio for 90-day functional outcome in patients with spontaneous intracerebral hemorrhage

Objective This study aimed to evaluate the association between the homocysteine-to-albumin ratio (Hcy/Alb) and 90-day functional outcome in patients with spontaneous supratentorial intracerebral hemorrhage, and to assess its incremental prognostic value beyond established clinical and imaging predictors. Methods This single-center retrospective cohort study was conducted within a prognostic model updating framework and followed the TRIPOD reporting guideline. We consecutively included 356 patients with spontaneous supratentorial intracerebral hemorrhage admitted between January 2023 and December 2025. The primary outcome was unfavorable functional outcome at 90 days, defined as a modified Rankin Scale score of 4–6. A baseline model was developed using age, admission NIHSS score, hematoma volume, hemorrhage location, intraventricular hemorrhage, and midline shift ≥5 mm. Hcy/Alb was then added to construct a combined model. Multivariable logistic regression was used to assess the independent association between Hcy/Alb and outcome. Model performance and internal stability were evaluated using receiver operating characteristic curve analysis, calibration assessment, Brier score, decision curve analysis, bootstrap resampling, and 10-fold cross-validation. Results Among 356 patients, 218 (61.2%) had an unfavorable functional outcome at 90 days. In multivariable logistic regression, Hcy/Alb was independently associated with unfavorable outcome when analyzed as a continuous variable (OR = 1.25, 95% CI: 1.12–1.39, p < 0.001). Patients with Hcy/Alb above the median value of 0.42 also had a higher risk of unfavorable outcome than those with lower Hcy/Alb (OR = 2.65, 95% CI: 1.58–4.45, p < 0.001). Hcy/Alb alone showed moderate discrimination for unfavorable outcome, with an AUC of 0.78. The AUC increased from 0.82 in the baseline model to 0.86 in the combined model, and the difference was statistically significant by DeLong test ( p = 0.018). Internal validation showed a bootstrap-corrected AUC of 0.86 and a 10-fold cross-validation AUC of 0.85 for the combined model. Calibration indices, Brier score, and decision curve analysis indicated modest improvements in predictive agreement and relative net benefit after adding Hcy/Alb, although the absolute increase in discrimination was modest (ΔAUC = 0.04), indicating limited incremental predictive value beyond the established clinical and imaging predictors. Conclusion Hcy/Alb was independently associated with 90-day unfavorable functional outcome in patients with spontaneous supratentorial intracerebral hemorrhage. Adding Hcy/Alb to established clinical and imaging predictors provided statistically significant but limited incremental prognostic information (ΔAUC = 0.04). Therefore, Hcy/Alb should be interpreted as an adjunctive biomarker rather than a standalone prediction tool. Because this model was developed in a single-center retrospective cohort and evaluated only through internal validation, external validation in independent multicenter prospective cohorts is required before broader clinical application.