Risk factors for intracerebral hemorrhage in patients with end-stage renal disease on maintenance hemodialysis and establishment of a predictive model.

BACKGROUND AND

OBJECTIVE

Patients with end-stage renal disease (ESRD) undergoing maintenance hemodialysis (MHD) are at a significantly increased risk of intracerebral hemorrhage (ICH), a devastating complication associated with high mortality and disability. However, reliable tools for early risk assessment remain limited. This study aimed to identify independent risk factors for ICH in this population and to develop a clinically applicable predictive model.

METHODS

This is a retrospective, two-center study, 294 ESRD patients undergoing MHD who met the inclusion and exclusion criteria at our hospital and Yubei District People's Hospital between January 2019 and January 2025 were retrospectively selected as the research objects. Patients were randomly divided into a modeling group (n = 206) and a validation group (n = 88) in a 7:3 ratio. The modeling group patients were further categorized into a cerebral-hemorrhage group (n = 42) and a non-cerebral-hemorrhage group (n = 164). Univariate and binary Logistics regression analyses were used to identify the influencing factors for intracerebral hemorrhage in ESRD MHD patients. A predictive model was constructed using SPSS, and R language was used to assess the model's application value through receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA).

RESULTS

The calibration curve slopes in both the modeling group and validation group were close to 1, indicating good consistency between predicted and actual risks. The ROC analysis results revealed an area under the curve (AUC) of 0.96 (95% CI: 0.873-0.979) in the modeling group, while the ROC analysis results revealed an AUC of 0.88(95% CI: 0.793-0.917) in the validation group, demonstrating the model's effectiveness and clinical benefit in both groups.

CONCLUSION

This study successfully identified smoking, stroke history, hyperlipidemia, and longer dialysis duration as independent risk factors for ICH in ESRD patients on MHD. The developed prediction model demonstrated good discriminative ability, calibration, and clinical utility in both internal and validation cohorts, providing clinicians with a practical tool for early risk stratification and individualized preventive management. These findings directly address the study's primary research question and support the model's potential for clinical implementation pending further external validation.