Clinical Report: Evaluation of Prognostic Models for Mortality in ACLF
Overview
This systematic review evaluates 75 prognostic models for acute-on-chronic liver failure (ACLF), highlighting their variable performance and high risk of bias. While ACLF-specific models like CLIF-C ACLF show better discrimination, the overall evidence is insufficient for precise clinical decision-making.
Background
ACLF is a critical condition characterized by rapid deterioration and high mortality rates, necessitating accurate prognostic assessments to guide treatment. Current management strategies rely on identifying patients' prognostic expectations to tailor interventions, ranging from aggressive treatment to palliative care. The need for reliable prognostic models is underscored by the high short- to medium-term mortality rates associated with ACLF.
Data Highlights
A total of 185 studies covering 241 external validation cohorts were included, evaluating 75 distinct prognostic models.
Key Findings
99.51% of analysis units had a high risk of bias due to insufficient outcome events and inadequate model performance assessment.
24 models were included in the meta-analysis, with c-statistics ranging from 0.58 to 0.84.
ACLF-specific models demonstrated better discrimination than general models.
CLIF-C ACLF showed stability across subgroups, but further validation is necessary.
Model discrimination declined with longer prediction horizons, increasing estimation uncertainty.
Clinical Implications
Clinicians should be cautious when using existing prognostic models for ACLF due to the high risk of bias and limited calibration reporting. The findings suggest a need for further validation of ACLF-specific models to enhance their reliability in clinical settings. Systematic risk assessment is essential for optimizing treatment strategies in ACLF patients.
Conclusion
The review underscores the necessity for high-quality, multicenter validation studies to strengthen the evidence base for ACLF prognostic models. Current models provide moderate to good discrimination but require further confirmation for clinical application.
