To develop a new model and validate its prognostic predictive capacity for HBV-ACLF patients.
Approach:
Study Design: 1170 HBV-ACLF cases were enrolled from 5854 liver failure patients, divided into development (n=822) and validation (n=348) cohorts. Cox regression analysis identified predictive factors for 28-day and 90-day mortality, leading to the development of new progno…
Statistical Analysis: The new scores were compared with existing models using ROC curves, calibration curves, and DCA curves. Survival analysis was conducted for risk stratification.
Key Findings:
The AHF-MELD and AHN-MELD scores demonstrated stronger predictive capacity than MELD, MELD-sodium, CLIF-C OF, and AARC scores.
The new scores were comparable to CLIF-C ACLF, COSSH ACLF, and COSSH ACLF II scores.
There was no significant difference between observed and predicted mortality probabilities for the new scores.
The new scores provided positive net benefits over a wide range of threshold probabilities.
Interpretation:
Limitations:
The study was conducted at a single center, which may limit generalizability.
The complexity of existing models may still pose challenges for routine bedside use.
